Citation
Bioreversible derivatives of 5-fluorouracil (5FU)

Material Information

Title:
Bioreversible derivatives of 5-fluorouracil (5FU) improving dermal and transdermal delivery with prodrugs
Alternate title:
Improving dermal and transdermal delivery with prodrugs
Creator:
Beall, Howard D., 1954-
Publication Date:
Language:
English
Physical Description:
xiii, 160 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
Crystallization ( jstor )
Hydrolysis ( jstor )
Lipids ( jstor )
Melting points ( jstor )
Phosphates ( jstor )
Prodrugs ( jstor )
Receptors ( jstor )
Skin ( jstor )
Solubility ( jstor )
Ultraviolet spectroscopy ( jstor )
Administration, Cutaneous ( mesh )
Antineoplastic Agents -- chemical synthesis ( mesh )
Antineoplastic Agents -- chemistry ( mesh )
Antineoplastic Agents -- pharmacokinetics ( mesh )
Fluorouracil -- analogs & derivatives ( mesh )
Prodrugs -- chemical synthesis ( mesh )
Prodrugs -- chemistry ( mesh )
Prodrugs -- pharmacokinetics ( mesh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1991.
Bibliography:
Includes bibliographical references (leaves 153-159).
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Howard D. Beall.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
49645371 ( OCLC )
ocm49645371
002292959 ( ALEPH )

Downloads

This item has the following downloads:

AA00011216_00001.pdf

AA00011216_00001_0167.txt

bioreversibleder00beal_0098.txt

AA00011216_00001_0042.txt

AA00011216_00001_0146.txt

bioreversibleder00beal_0068.txt

AA00011216_00001_0013.txt

AA00011216_00001_0008.txt

bioreversibleder00beal_0138.txt

bioreversibleder00beal_0096.txt

AA00011216_00001_0066.txt

bioreversibleder00beal_0052.txt

bioreversibleder00beal_0109.txt

bioreversibleder00beal_0129.txt

AA00011216_00001_0157.txt

AA00011216_00001_0072.txt

bioreversibleder00beal_0055.txt

bioreversibleder00beal_0071.txt

bioreversibleder00beal_0073.txt

AA00011216_00001_0065.txt

AA00011216_00001_0047.txt

AA00011216_00001_0018.txt

AA00011216_00001_0055.txt

bioreversibleder00beal_0164.txt

bioreversibleder00beal_0099.txt

bioreversibleder00beal_0028.txt

bioreversibleder00beal_0132.txt

AA00011216_00001_0061.txt

AA00011216_00001_0096.txt

AA00011216_00001_0131.txt

bioreversibleder00beal_0041.txt

AA00011216_00001_0137.txt

AA00011216_00001_0141.txt

AA00011216_00001_0089.txt

AA00011216_00001_0036.txt

AA00011216_00001_0075.txt

AA00011216_00001_0125.txt

AA00011216_00001_0058.txt

bioreversibleder00beal_0002.txt

bioreversibleder00beal_0000.txt

bioreversibleder00beal_0134.txt

bioreversibleder00beal_0031.txt

bioreversibleder00beal_0162.txt

bioreversibleder00beal_0121.txt

AA00011216_00001_0015.txt

AA00011216_00001_0049.txt

bioreversibleder00beal_0027.txt

AA00011216_00001_0081.txt

AA00011216_00001_0003.txt

bioreversibleder00beal_0102.txt

bioreversibleder00beal_0143.txt

AA00011216_00001_0045.txt

bioreversibleder00beal_0040.txt

bioreversibleder00beal_0123.txt

bioreversibleder00beal_0001.txt

bioreversibleder00beal_0142.txt

bioreversibleder00beal_0076.txt

bioreversibleder00beal_0037.txt

AA00011216_00001_0025.txt

bioreversibleder00beal_0117.txt

AA00011216_00001_0019.txt

AA00011216_00001_0135.txt

AA00011216_00001_0144.txt

bioreversibleder00beal_0070.txt

bioreversibleder00beal_0009.txt

bioreversibleder00beal_0083.txt

AA00011216_00001_0164.txt

AA00011216_00001_0085.txt

bioreversibleder00beal_0049.txt

AA00011216_00001_0071.txt

bioreversibleder00beal_0161.txt

AA00011216_00001_0119.txt

AA00011216_00001_0073.txt

bioreversibleder00beal_0140.txt

bioreversibleder00beal_0013.txt

AA00011216_00001_0153.txt

AA00011216_00001_0034.txt

bioreversibleder00beal_0069.txt

bioreversibleder00beal_0048.txt

AA00011216_00001_0056.txt

bioreversibleder00beal_0062.txt

bioreversibleder00beal_0019.txt

AA00011216_00001_0023.txt

bioreversibleder00beal_0033.txt

AA00011216_00001_0122.txt

bioreversibleder00beal_0097.txt

bioreversibleder00beal_0088.txt

AA00011216_00001_0039.txt

AA00011216_00001_0173.txt

bioreversibleder00beal_0094.txt

AA00011216_00001_0130.txt

AA00011216_00001_0118.txt

bioreversibleder00beal_0124.txt

AA00011216_00001_0077.txt

bioreversibleder00beal_0125.txt

AA00011216_00001_0126.txt

bioreversibleder00beal_0090.txt

bioreversibleder00beal_0144.txt

AA00011216_00001_0022.txt

AA00011216_00001_0166.txt

AA00011216_00001_0035.txt

AA00011216_00001_0121.txt

bioreversibleder00beal_0110.txt

AA00011216_00001_0083.txt

AA00011216_00001_0175.txt

bioreversibleder00beal_0077.txt

bioreversibleder00beal_0105.txt

bioreversibleder00beal_0010.txt

bioreversibleder00beal_0148.txt

AA00011216_00001_0154.txt

AA00011216_00001_0138.txt

AA00011216_00001_0143.txt

bioreversibleder00beal_0127.txt

AA00011216_00001_0149.txt

AA00011216_00001_0116.txt

bioreversibleder00beal_0072.txt

AA00011216_00001_0005.txt

bioreversibleder00beal_0029.txt

bioreversibleder00beal_0093.txt

AA00011216_00001_0172.txt

AA00011216_00001_0006.txt

AA00011216_00001_0161.txt

AA00011216_00001_0147.txt

AA00011216_00001_0099.txt

bioreversibleder00beal_0167.txt

bioreversibleder00beal_0092.txt

bioreversibleder00beal_0026.txt

AA00011216_00001_0094.txt

bioreversibleder00beal_0156.txt

AA00011216_00001_0060.txt

AA00011216_00001_0127.txt

bioreversibleder00beal_0057.txt

bioreversibleder00beal_0043.txt

bioreversibleder00beal_0023.txt

bioreversibleder00beal_0006.txt

AA00011216_00001_0087.txt

bioreversibleder00beal_0050.txt

AA00011216_00001_0010.txt

bioreversibleder00beal_0030.txt

AA00011216_00001_0069.txt

bioreversibleder00beal_0119.txt

AA00011216_00001_0120.txt

bioreversibleder00beal_0103.txt

bioreversibleder00beal_0032.txt

AA00011216_00001_0002.txt

AA00011216_00001_0136.txt

bioreversibleder00beal_0175.txt

AA00011216_00001_0038.txt

bioreversibleder00beal_0016.txt

bioreversibleder00beal_0087.txt

AA00011216_00001_0163.txt

AA00011216_00001_0114.txt

AA00011216_00001_0004.txt

AA00011216_00001_0104.txt

AA00011216_00001_0145.txt

bioreversibleder00beal_0149.txt

AA00011216_00001_0033.txt

bioreversibleder00beal_0135.txt

bioreversibleder00beal_0166.txt

AA00011216_00001_0111.txt

bioreversibleder00beal_0067.txt

AA00011216_00001_0123.txt

bioreversibleder00beal_0011.txt

AA00011216_00001_pdf.txt

AA00011216_00001_0097.txt

bioreversibleder00beal_0130.txt

AA00011216_00001_0053.txt

bioreversibleder00beal_0061.txt

bioreversibleder00beal_0174.txt

bioreversibleder00beal_0133.txt

AA00011216_00001_0043.txt

AA00011216_00001_0048.txt

bioreversibleder00beal_0118.txt

AA00011216_00001_0095.txt

bioreversibleder00beal_0018.txt

bioreversibleder00beal_0082.txt

bioreversibleder00beal_0079.txt

AA00011216_00001_0016.txt

bioreversibleder00beal_0086.txt

AA00011216_00001_0079.txt

AA00011216_00001_0024.txt

AA00011216_00001_0029.txt

AA00011216_00001_0080.txt

bioreversibleder00beal_0058.txt

AA00011216_00001_0059.txt

bioreversibleder00beal_0075.txt

bioreversibleder00beal_0159.txt

bioreversibleder00beal_0115.txt

AA00011216_00001_0112.txt

bioreversibleder00beal_0137.txt

AA00011216_00001_0031.txt

bioreversibleder00beal_0035.txt

bioreversibleder00beal_0163.txt

AA00011216_00001_0103.txt

AA00011216_00001_0012.txt

AA00011216_00001_0148.txt

bioreversibleder00beal_0170.txt

AA00011216_00001_0132.txt

bioreversibleder00beal_0154.txt

AA00011216_00001_0140.txt

bioreversibleder00beal_0059.txt

bioreversibleder00beal_0108.txt

AA00011216_00001_0174.txt

AA00011216_00001_0106.txt

bioreversibleder00beal_0136.txt

AA00011216_00001_0158.txt

bioreversibleder00beal_0153.txt

AA00011216_00001_0084.txt

bioreversibleder00beal_0151.txt

AA00011216_00001_0052.txt

bioreversibleder00beal_0095.txt

AA00011216_00001_0162.txt

AA00011216_00001_0062.txt

AA00011216_00001_0133.txt

AA00011216_00001_0041.txt

AA00011216_00001_0101.txt

bioreversibleder00beal_0060.txt

bioreversibleder00beal_0113.txt

AA00011216_00001_0011.txt

AA00011216_00001_0171.txt

AA00011216_00001_0068.txt

bioreversibleder00beal_0104.txt

AA00011216_00001_0046.txt

bioreversibleder00beal_0042.txt

AA00011216_00001_0021.txt

bioreversibleder00beal_0157.txt

bioreversibleder00beal_0122.txt

bioreversibleder00beal_0021.txt

bioreversibleder00beal_0085.txt

bioreversibleder00beal_0004.txt

bioreversibleder00beal_0053.txt

AA00011216_00001_0051.txt

bioreversibleder00beal_0155.txt

AA00011216_00001_0152.txt

AA00011216_00001_0155.txt

AA00011216_00001_0057.txt

bioreversibleder00beal_0107.txt

AA00011216_00001_0082.txt

AA00011216_00001_0142.txt

bioreversibleder00beal_0171.txt

bioreversibleder00beal_0056.txt

AA00011216_00001_0117.txt

AA00011216_00001_0020.txt

AA00011216_00001_0150.txt

bioreversibleder00beal_0014.txt

AA00011216_00001_0151.txt

bioreversibleder00beal_0139.txt

AA00011216_00001_0165.txt

AA00011216_00001_0092.txt

bioreversibleder00beal_0044.txt

bioreversibleder00beal_0158.txt

AA00011216_00001_0110.txt

bioreversibleder00beal_0150.txt

AA00011216_00001_0063.txt

AA00011216_00001_0030.txt

AA00011216_00001_0014.txt

AA00011216_00001_0037.txt

AA00011216_00001_0108.txt

AA00011216_00001_0168.txt

bioreversibleder00beal_0047.txt

AA00011216_00001_0054.txt

bioreversibleder00beal_0165.txt

bioreversibleder00beal_0084.txt

AA00011216_00001_0090.txt

bioreversibleder00beal_0141.txt

bioreversibleder00beal_0078.txt

bioreversibleder00beal_0111.txt

bioreversibleder00beal_0025.txt

bioreversibleder00beal_0039.txt

bioreversibleder00beal_0064.txt

bioreversibleder00beal_0106.txt

bioreversibleder00beal_0160.txt

AA00011216_00001_0007.txt

bioreversibleder00beal_0036.txt

bioreversibleder00beal_0091.txt

bioreversibleder00beal_0017.txt

bioreversibleder00beal_0005.txt

bioreversibleder00beal_0120.txt

bioreversibleder00beal_0126.txt

AA00011216_00001_0156.txt

bioreversibleder00beal_0015.txt

bioreversibleder00beal_0131.txt

bioreversibleder00beal_0024.txt

AA00011216_00001_0139.txt

bioreversibleder00beal_0169.txt

AA00011216_00001_0129.txt

AA00011216_00001_0105.txt

AA00011216_00001_0026.txt

AA00011216_00001_0086.txt

bioreversibleder00beal_0145.txt

bioreversibleder00beal_0063.txt

AA00011216_00001_0001.txt

AA00011216_00001_0102.txt

bioreversibleder00beal_0034.txt

bioreversibleder00beal_0022.txt

AA00011216_00001_0107.txt

bioreversibleder00beal_0081.txt

bioreversibleder00beal_0065.txt

AA00011216_00001_0009.txt

AA00011216_00001_0115.txt

AA00011216_00001_0100.txt

bioreversibleder00beal_0045.txt

AA00011216_00001_0169.txt

AA00011216_00001_0017.txt

bioreversibleder00beal_0147.txt

AA00011216_00001_0159.txt

AA00011216_00001_0076.txt

AA00011216_00001_0093.txt

bioreversibleder00beal_0152.txt

AA00011216_00001_0032.txt

bioreversibleder00beal_0080.txt

bioreversibleder00beal_0116.txt

bioreversibleder00beal_0146.txt

bioreversibleder00beal_0173.txt

AA00011216_00001_0074.txt

bioreversibleder00beal_0172.txt

bioreversibleder00beal_0089.txt

AA00011216_00001_0160.txt

AA00011216_00001_0070.txt

AA00011216_00001_0128.txt

bioreversibleder00beal_0007.txt

bioreversibleder00beal_0003.txt

AA00011216_00001_0088.txt

AA00011216_00001_0078.txt

bioreversibleder00beal_0112.txt

bioreversibleder00beal_0168.txt

AA00011216_00001_0113.txt

bioreversibleder00beal_0054.txt

AA00011216_00001_0028.txt

AA00011216_00001_0067.txt

AA00011216_00001_0134.txt

AA00011216_00001_0091.txt

bioreversibleder00beal_0051.txt

bioreversibleder00beal_0074.txt

bioreversibleder00beal_0128.txt

bioreversibleder00beal_0020.txt

AA00011216_00001_0040.txt

bioreversibleder00beal_0012.txt

bioreversibleder00beal_0066.txt

AA00011216_00001_0044.txt

bioreversibleder00beal_0100.txt

bioreversibleder00beal_0114.txt

AA00011216_00001_0109.txt

AA00011216_00001_0170.txt

AA00011216_00001_0124.txt

bioreversibleder00beal_0101.txt

AA00011216_00001_0027.txt

bioreversibleder00beal_0038.txt

AA00011216_00001_0098.txt

AA00011216_00001_0064.txt

bioreversibleder00beal_0046.txt

bioreversibleder00beal_0008.txt

AA00011216_00001_0176.txt

AA00011216_00001_0050.txt


Full Text












BIOREVERSIBLE DERIVATIVES OF 5-FLUOROURACIL (5FU):
IMPROVING DERMAL AND TRANSDERMAL DELIVERY
WITH PRODRUGS















BY


HOWARD D. BEALL


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY



UNIVERSITY OF FLORIDA

1991




























Copyright 1991

by

Howard D. Beall















ACKNOWLEDGEMENTS


I would like to thank the members of my supervisory

committee, Dr. Kenneth Sloan, Dr Margaret James, Dr. Koppaka

Rao, Dr. Richard Prankerd, and Dr. John Zoltewicz for their

guidance and expert advice over the past four years. My

sincerest thanks go to my research advisor and committee

chairman, Dr. Sloan, for sharing his enthusiasm for teaching

and science. I am especially grateful for his patience and

understanding during my seemingly endless questions and

interruptions.

I would also like to acknowledge the enthusiastic

support of Dr. Noel Meltzer of Hoffmann-La Roche. This

project was partially funded by a grant from Hoffmann-La

Roche.

My special thanks go to my parents for their love and

support throughout my life and to my two-year-old son,

Michael, who could make me laugh when it was the last thing I

felt like doing. But most of all, I want to thank my wife,

Donna, whose love, support, and countless sacrifices made my

return to school and the completion of this project possible.















TABLE OF CONTENTS


ACKNOWLEDGEMENTS ................. ........... ............. iii

LIST OF TABLES .. .......................... ..............vi

LIST OF FIGURES ............................................ viii

KEY TO ABBREVIATIONS ............. ......................... ix

ABSTRACT ............ .......................... ..............x

INTRODUCTION ................ .................... .............. 1

Fluorouracil.. .............................................1
Fluorouracil Derivatives................................... 2
Fluorouracil Metabolism .................................... 3
Structure of the Skin.....................................6
Mechanisms of Transdermal Penetration ..................... 8
Passive Diffusion........................................ 10
Enhancement of Skin Penetration. ..........................11
Prodrugs and Skin Penetration............................13
Dermal versus Transdermal Delivery.......................16
Cutaneous Metabolism......................................17
Hairless Mouse Skin as a Model Membrane ..................18
Research Proposal... ......................................20

1-ALKYLOXYCARBONYL DERIVATIVES ............................ 23

Introduction.. .................................... ........ 23
Materials and Methods. .....................................25
Results and Discussion. ....................................35
Summary ................................................... 51

1-ACYL DERIVATIVES ........................................... 53

Introduction.. .................................... ........53
Materials and Methods. .....................................53
Results and Discussion ...................................64
Summary .......................... .........................80

1,3-BIS-ACYL DERIVATIVES ............. ..................... 82

Introduction.. .................................... ........ 82
Materials and Methods. .....................................84
Results and Discussion ...................................93
Summary ................................................... 107











3-ACYL DERIVATIVES ........................................ 109

Introduction ............................................. 109
Materials and Methods.................................... 111
Results and Discussion................................... 120
Summary .................................... ............. 146

SUMMARY AND CONCLUSIONS ................................... 148

LIST OF REFERENCES .........................................153

BIOGRAPHICAL SKETCH .............. ...................... ..160















LIST OF TABLES


Table 1-1. Structures of 5FU and prodrug derivatives
of 5FU .................... .. ........ ................... 21

Table 2-1. Structures of 1-alkyloxycarbonyl
derivatives................... .......................... 24

Table 2-2. Melting points (MP), lipid solubilities
(SIPM), and aqueous solubilities (SAQ) for
1-alkyloxycarbonyl derivatives............................39

Table 2-3. Solubility ratios (SR), partition
coefficients (PC), and hydrophobicity parameters (9)
for 1-alkyloxycarbonyl derivatives...................... 40

Table 2-4. Pseudo-first-order rate constants (k) and
half-lives (tl/2) for hydrolysis of 1-methyloxy-
carbonyl-5FU in 0.05 M phosphate buffer (pH=7.1,
I=0.12) with and without formaldehyde at 32 OC.......... 44

Table 2-5. Fluxes (J), lag times (tL), and skin
accumulation (SA) values for 1-alkyloxycarbonyl
derivatives ............... ............................... 48

Table 2-6. Second application fluxes (J) and lag times
(tL) for 1-alkyloxycarbonyl derivatives................. 49

Table 3-1. Structures of 1-acyl derivatives. .............. 54

Table 3-2. Melting points (MP), lipid solubilities
(SIPM), and aqueous solubilities (SAQ) for 1-acyl
derivatives ............... ..... .......................... 68

Table 3-3. Partition coefficients (PC) and hydropho-
bicity parameters (c) for 1-acyl derivatives ........... 69

Table 3-4. Pseudo-first-order rate constants (k) and
half-lives (tl/2) for hydrolysis of 1-acyl derivatives
in 0.05 M phosphate buffer (pH=7.1, I=0.12) at 32 C .... 71

Table 3-5. Pseudo-first-order rate constants (k) and
half-lives (ti/2) for hydrolysis of l-acetyl-5FU in
0.05 M phosphate buffer (pH=7.1, 1-0.12) with and
without formaldehyde at 32 C. ........................... 72










Table 3-6. Fluxes (J), lag times (tL), and skin
accumulation (SA) values for 1-acyl derivatives .........77

Table 3-7. Second application fluxes (J) and lag times
(tL) for 1-acyl derivatives.............................. 78

Table 4-1. Structures of 1,3-bis-acyl derivatives. ........ 83

Table 4-2. Melting points (MP), lipid solubilities
(SIPM), and aqueous solubilities (SAQ) for 1,3-bis-
acyl derivatives ........................................ 98

Table 4-3. Pseudo-first-order rate constants (k) and
half-lives (tl/2) for hydrolysis of 1,3-bis-acetyl-5FU
in 0.05 M phosphate buffer (pH=7.1, I=0.12) with and
without formaldehyde at 32 C ......................... .. 100

Table 4-4. Fluxes (J), lag times (tL), and skin
accumulation (SA) values for 1,3-bis-acyl
derivatives............................................. 104

Table 4-5. Second application fluxes (J) and lag times
(tL) for 1,3-bis-acyl derivatives........................ 105

Table 5-1. Structures of 3-acyl derivatives. .............. 110

Table 5-2. Melting points (MP), lipid solubilities
(SIPM), and aqueous solubilities (SAQ) for 3-acyl
derivatives............................................. 127

Table 5-3. Solubility ratios (SR), partition
coefficients (PC), and hydrophobicity parameters (E)
for 3-acyl derivatives.................................. 128

Table 5-4. Pseudo-first-order rate constants (k) and
half-lives (t1/2) for hydrolysis of 3-acyl derivatives
in 0.05 M phosphate buffer (pH=7.1, 1=0.12) with and
without formaldehyde at 32 C ............................. 134

Table 5-5. Reaction products formed during hydrolysis of
3-acetyl-5FU in 0.05 M phosphate buffer (pH=7.1,
I=0.12) with formaldehyde at 32 C.. ..................... 135

Table 5-6. Fluxes (J), lag times (tL), and skin
accumulation (SA) values for 3-acyl derivatives......... 142

Table 5-7. Second application fluxes (J) and lag times
(tL) for 3-acyl derivatives............................. 143















LIST OF FIGURES


Figure 2-1. Plots of ln(C) versus time (min) for
hydrolysis of l-methyloxycarbonyl-5FU in 0.05 M
phosphate buffer (pH=7.1, I=0.12) with and without
formaldehyde at 32 OC....................................43

Figure 2-2. Plots of cumulative amount of total 5FU
species that diffused (Lmol) versus time (h) for
compounds 1, 2, 3, and 5FU. ............................ 46

Figure 2-3. Plots of cumulative amount of total 5FU
species that diffused (Imol) versus time (h) for
compounds 4, 5, 6, and 5FU. ............................ 47

Figure 3-1. X-ray structure of l-acetyl-5FU unprimedd). ... 66

Figure 3-2. X-ray structure of l-acetyl-5FU (primed). ..... 67

Figure 3-3. Plot of pseudo-first-order rate constant (k)
versus formaldehyde concentration (M) for hydrolysis
of l-acetyl-5FU in 0.05 M phosphate buffer (pH=7.1,
1-0.12) at 32 OC........................................ 73

Figure 3-4. Plots of cumulative amount of total 5FU
species that diffused (Jnol) versus time (h) for
compounds 7, 8, 9, and 5FU. ............................ 75

Figure 3-5. Plots of cumulative amount of total 5FU
species that diffused (pmol) versus time (h) for
compounds 10, 11, 12, and 5FU ..........................76

Figure 4-1. X-ray structure of 1,3-bis-acetyl-5FU. ........ 95

Figure 4-2. Plots of In(At-Ao) versus time (min) for
hydrolysis of 1,3-bis-acetyl-5FU in 0.05 M phosphate
buffer (pH=7.1, I=0.12) with and without formaldehyde
at 32 OC.............................................. 99

Figure 4-3. Plots of cumulative amount of total 5FU
species that diffused (pmol) versus time (h) for
compounds 13, 14, and 5FU. ............................. 102

Figure 4-4. Plots of cumulative amount of total 5FU
species that diffused (lnmol) versus time (h) for
compounds 15, 16, and 5FU. ............................. 103










Figure 5-1. Possible scheme for thermal decomposition of
3-acetyl-5FU............................................ 123

Figure 5-2. Possible scheme for thermal intramolecular
rearrangement for 3-acetyl-5FU to l-acetyl-5FU...........124

Figure 5-3. Plot of ln(At-A-) versus time (min) for
hydrolysis of 3-acetyl-SFU in 0.05 M phosphate buffer
(pH=7.1, I=0.12) at 32 oC................................130

Figure 5-4. Plot of ln(At-Ao) versus time (min) for
hydrolysis of 3-propionyl-5FU in 0.05 M phosphate
buffer (pH=7.1, I=0.12) at 32 C ........................ 131

Figure 5-5. Plots of ln(C) versus time (min) for
hydrolysis of 3-acetyl-5FU in 0.05 M phosphate buffer
(pH=7.1, I=0.12) with (n=2) and without (n=3)
formaldehyde at 32 OC................................... 132

Figure 5-6. Plots of ln(C) versus time (min) for
hydrolysis of 3-acetyl-5FU in 0.05 M phosphate buffer
(pH=7.1, 1-0.12) at 32 C using actual concentration
(Ct) and concentration corrected for secondary
degradation (Ccorr) .... ...... ... .... ................. 133

Figure 5-7. Possible scheme for reaction of 3-acetyl-5FU
with formaldehyde to form l-acetyloxymethyl-5FU and
3-acetyloxymethyl-5FU ................................... 136

Figure 5-8. Plots of cumulative amount of total 5FU
species that diffused (IUmol) versus time (h) for
compounds 17, 18, and 5FU.............................. 140

Figure 5-9. Plots of cumulative amount of total 5FU
species that diffused (lmol) versus time (h) for
compounds 19, 20, and 5FU.............................. 141















KEY TO ABBREVIATIONS


bs broad singlet

CDC13 chloroform-d

(CD3)2SO dimethylsulfoxide-d6

CH3CN acetonitrile

d doublet

dec decomposition

dist t distorted triplet

m multiple

q quartet

Rf retention factor

t triplet















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy


BIOREVERSIBLE DERIVATIVES OF 5-FLUOROURACIL (5FU):
IMPROVING DERMAL AND TRANSDERMAL DELIVERY
WITH PRODRUGS

By

Howard D. Beall

December, 1991




Chairman: Kenneth B. Sloan
Major Department: Medicinal Chemistry


Fluorouracil (5FU) is an antineoplastic agent used

topically for treatment of actinic keratoses, superficial

basal cell carcinomas, psoriasis, and other skin conditions,

but treatment is often ineffective due to its poor skin

penetration properties. Four homologous series of

bioreversible derivatives (prodrugs) of 5FU were synthesized,

characterized, and their ability to penetrate through

transdermall) and accumulate in dermall) the skin was

evaluated. Six 1-alkyloxycarbonyl, six 1-acyl, four 1,3-bis-

acyl, and four 3-acyl derivatives were studied.

Lipid solubility was a minimum of 40 times greater for

the derivatives than for 5FU, and aqueous solubility was

actually higher than 5FU for five derivatives. The










partitioning-based method for determining aqueous solubility

gave reliable values for relative solubility in each

homologous series, and overall, the values correlated well

with conventionally determined solubilities. This was the

first known successful attempt to obtain aqueous solubilities

for chemically unstable prodrugs (e.g., 1-acyl and 1,3-bis-

acyl derivatives).

Hydrolysis rates for N1-substituents followed pseudo-

first-order kinetics, but plots for N3-acyl hydrolysis were

biexponential. An 02-acyl intermediate was proposed to

account for the unusual hydrolysis and thermal decomposition

of the N3-acyl group. X-ray crystal analysis showed that the

N3-acyl group was oriented perpendicular to the 5FU ring and

was sterically and electronically hindered from nucleophilic

attack.

Skin penetration was 1.2 to nearly 40 times greater for

the prodrugs than for 5FU. The highest flux was recorded for

l-acetyl-5FU, whereas 3-propionyl-5FU, 1-ethyloxycarbonyl-

5FU, and 1,3-bis-acetyl-5FU exhibited the highest rates of

delivery for their respective series. These derivatives were

also the most aqueous soluble members of each series. This

demonstrates the importance of biphasic solubility for

achieving optimal transdermal delivery of 5FU from a

homologous series of prodrugs.

Skin accumulation was highest for l-acetyl-5FU and

l-propionyl-5FU (>18 times more than 5FU). Rapid hydrolysis

of the 1-acyl series upon partitioning into the skin may


xii










cause retention of highly polar 5FU in the more lipophilic

regions of the epidermis.

Skin damage was only slightly increased with some

prodrugs, all derivatives were stable in formulation, and

overall, l-acetyl-5FU was the best candidate for improving

dermal or transdermal delivery of 5FU.















CHAPTER 1
INTRODUCTION


Fluorouracil


Fluorouracil (5FU) is one of the most widely used and

studied2 anticancer agents. It is used in the palliative

treatment of solid tumors and is most effective in

combination with other antineoplastic agents3 and in

conjunction with radiation therapy.2 It has been used to

treat carcinoma of the colon, rectum, breast, and stomach

and, with less effectiveness, carcinoma of the ovary, cervix,

urinary bladder, liver, and pancreas.4

Fluorouracil (5FU) is used as a single agent in topical

preparations and is effective in treating actinic (solar)

keratoses,5-8 superficial basal cell carcinomas,9-10 and

psoriasis.11-13 Topical 5FU has also been used to treat a

variety of other precancerous conditions, malignant and

benign tumors, and dermatoses.14 Commercial 5FU creams and

solutions may be adequate for treating lesions on the face,

but other areas, especially the forearms and hands, respond

poorly probably due to lack of penetration by 5FU.15 Some

conditions, such as psoriasis, respond to topical 5FU therapy

when the drug is applied under an occlusive dressing.13









However, this is neither the most convenient nor comfortable

treatment option.

The toxicity of 5FU is related to its effect on rapidly

proliferating host cells especially those of the bone marrow

and gastrointestinal lining. Toxicity following systemic

therapy is common while adverse effects from topical therapy

appear to be minimal, other than those associated with the

local inflammatory reactions usually necessary for a

therapeutic response.4


Fluorouracil Derivatives


Since 5FU was first synthesized16 and tested17 as an

antitumor antimetabolite in 1957, numerous attempts have been

made to improve its efficacy and reduce its toxicity.

Derivatives of 5FU and its nucleosides, 5-fluorouridine (FUR)

and 5-fluoro-2'-deoxyuridine (FdUR), have been developed with

this in mind. The majority of these derivatives have been

designed with the expectation that 5FU will be released in

vivo, so they are essentially prodrugs.

A prodrug has been defined as "an agent which must

undergo chemical or enzymatic transformation to the active or

parent drug after administration, so that the metabolic

product or parent drug can subsequently exhibit the desired

pharmacological response" (p. 1275).18 Prodrugs, or

bioreversible derivatives, of 5FU have included hydroxy-

methyl,19 alkyloxyalkyl,20 acyloxyalkyl,21-24 acyl,25-28










alkyloxycarbonyl,25,29-30 alkyloxycarbonyloxyalkyl,31-33 alkyl-

carbamoyl,34-37 aminoalkyl (Mannich base),38-39 phthalid-

yl,40-41 and derivatives containing sulfur rather than oxygen

in some of the above functional groups.42-43 Polymer44-45 and

peptide46 derivatives have also been proposed as sources of

5FU in vivo. Currently, at least two bioreversible

derivatives of 5FU, 1-(tetrahydrofuran-2-yl)-5-fluorouracil47

and l-hexylcarbamoyl-5-fluorouracil,48 have been marketed in

Japan.

The above is just a small sample of the synthetic

literature involving 5FU. In addition, a large number of

studies have been devoted to synthesizing analogs of FUR and

FdUR containing an N1-glycosidic linkage. FdUR, or

floxuridine, is available in the United States, but is only

approved for intraarterial infusion.4 It appears to offer no

advantage over intravenous 5FU, and this therapy is both

hazardous and expensive.3 The nucleosides and their analogs

are converted to 5FU in vivo by a pyrimidine phosphorylase.

Since phosphorylase activity is higher in tumor tissue, it is

believed that development of compounds that are good

substrates for phosphorylase may lead to greater tumor

specificity and decreased toxicity.49


Fluorouracil Metabolism


Fluorouracil (5FU) undergoes extensive catabolic and

anabolic metabolism. Approximately 15% of a single










intravenous dose of 5FU is excreted unchanged in the urine

within six hours, and 90% of that is detected within the

first hour.4 Approximately 80% of the dose is metabolized by

the liver and extrahepatic tissues.1 The principal

catabolite of 5FU is a-fluoro-P-alanine (FBAL), which

accounts for over 95% of the catabolic products found in the

urine. Bile acid conjugates of FBAL represent the major

biliary catabolites of 5FU.1

A number of anabolic pathways have been characterized

for 5FU, and one or more of them may be responsible for the

activation and subsequent cytotoxicity of the drug. The most

direct activation pathway utilizes orotidine monophosphate

phosphoribosyl transferase (OMPT) to form 5-fluorouridine-5'-

monophosphate (FUMP) from 5FU and 5-phosphoribosyl-l-pyro-

phosphate. FUMP can also be generated in two steps with

uridine phosphorylase and uridine kinase. With two more

kinases 5-fluorouridine-5'-triphosphate (FUTP) is formed

which can be incorporated into ribonucleic acid (RNA) leading

to RNA dysfunction.1 Some cell lines appear to favor the

OMPT pathway, while others favor the uridine phosphorylase

pathway, and the FUTP formed from each is added to different

fractions of RNA.50

Some of the 5-fluorouridine-5'-diphosphate (FUDP) that

is generated above as an intermediate can become a substrate

for ribonucleotide reductase. The resulting deoxynucleotide,

5-fluoro-2'-deoxyuridine-5'-diphosphate (FdUDP), can form

5-fluoro-2'-deoxyuridine-5'-triphosphate (FdUTP) using










another kinase. At this point FdUTP can be incorporated into

deoxyribonucleic acid (DNA), but the contribution to 5FU

cytotoxicity by this mechanism is unclear. DNA protective

enzymes, dUTPase and uracil-DNA glycosylase, which are

responsible for keeping uracil residues out of DNA, target

5FU residues as well.50

A diphosphohydrolase converts FdUTP to 5-fluoro-2'-

deoxyuridine-5'-monophosphate (FdUMP). FdUMP can also be

formed from 5FU by thymidine phosphorylase and thymidine

kinase, although this may be quantitatively less important

than other activation steps.1 FdUMP binds covalently to

thymidylate synthase and its cofactor, 5,10-methylene

tetrahydrofolate, thereby preventing synthesis of thymidine-

5'-monophosphate (dTMP) and subsequently, thymidine-5'-

triphosphate (dTTP), an essential component in DNA synthesis.

FdUMP has a greater affinity for thymidylate synthase than

the normal substrate, 2'-deoxyuridine-5'-monophosphate

(dUMP).1

Either inhibition of thymidylate synthase by FdUMP or

incorporation of FUTP into RNA can lead to cell death. The

mechanism that is most important depends on the cell line

being studied.51

Fluorouracil (5FU) may also affect glycoprotein and

glycolipid metabolism, since it is known that FUDP sugars can

be formed.50 The possibility that membrane effects from

altered glycoprotein synthesis could lead to cytotoxicity has

been noted, but has not been well studied.1











Structure of the Skin


Fluorouracil (5FU) is typical of polar, high-melting,

heterocyclic compounds which exhibit poor skin permeability.38

Since the goal of this project is to improve dermal and

transdermal delivery of 5FU, it is necessary to examine the

structure of the barrier to be penetrated, the skin.

The skin is composed of two major tissue layers, the

epidermis and dermis. The epidermis is a continuous, elastic

sheet that is interrupted only by glandular pores and hair

follicles. It consists of four definable sublayers and

averages 75 to 150 gm in thickness. The basal cell layer,

which borders the dermis, is a single layer of keratinocytes.

This is the germinal layer, and all epidermal cells are

initially formed here before moving outward to the next

layer, the stratum spinosum. The basal cells are cuboidal or

columnar in shape, while the stratum spinosum cells are

polyhedral. By the time the cells have migrated to the next

layer, the stratum granulosum, they have become flattened and

contain characteristic keratohyalin granules. The stratum

granulosum marks the transition between nucleated cells and

the anucleated stratum corneum. The stratum corneum cells in

the last layer are highly keratinized, markedly flattened

cells in which cellular components, such as mitochondria and

ribosomes, have degraded along with the nucleus. Stratum

corneum normally consists of about 15 to 20 layers, but each










layer is only about 0.5 pm in thickness. The outer layers

are continuously desquamated and replaced from below. The

entire transit time from basal layer to desquamation is 26 to

42 days. The cells of the epidermis are attached to each

other by desmosomes, which degrade just prior to

desquamation.52

Below the epidermis lies the dermis which constitutes

the bulk of the approximately 2 mm thickness of human skin.

The two layers are anchored to the basal lamina by various

fibrils and microfibrils. The attachment is enhanced by the

interlocking nature of the junction.52

The dermis consists of two regions, the papillary dermis

and reticular dermis. The papillary dermis is the thinner,

outermost region that is molded against the ridges and

grooves of the epidermis. It contains small, loosely

distributed fibrils and encloses the microcirculatory blood

and lymph vessels. In contrast, the collagen bundles and

elastin fibers of the reticular dermis are more densely

packed, and this region is relatively acellular and

avascular. Collagen is the major component of the dermis,

and it gives the skin its tensile strength. The dermis also

contains nerves, excretory and secretary glands, and hair

follicles.52











Mechanisms of Transdermal Penetration


For most substances, the stratum corneum provides the

rate-limiting barrier to skin penetration.53-55 The stratum

corneum is 75-85% protein (dry weight),54 mostly intracellular

keratin, while the intercellular space is a lipid-enriched

region. During the outward migration of the keratinocytes

through the epidermis, lamellar bodies are synthesized, which

contain lipids, polysaccharides, and hydrolytic enzymes. As

the granular cells prepare to enter the stratum corneum, the

lamellar bodies move to the cell periphery and empty their

contents into the spaces between the cells.56 The corneocytes

and intercellular lipids are arranged in a brick and mortar

fashion,57 and the result is a very dense (1.4 g/cm3 in the

dry state),58 highly efficient moisture barrier.

In order for a substance to penetrate the stratum

corneum, it must either (1) cross through the densely packed

corneocytes, (2) traverse through the intercellular region,

thereby avoiding the keratinized cells, or (3) bypass the

stratum corneum completely by diffusing through shunt

pathways such as sweat ducts and hair follicles.53 The shunt

pathways are not considered to be significant, especially

when steady state diffusion has been attained,59 and these

pathways are generally disregarded in discussions of

penetration mechanisms.










Until recently, it was thought that the transcellular

pathway was the primary route through the stratum corneum.

This was based, at least in part, on a gross underestimation

of the volume of the intercellular space.60 The current

belief is that passive diffusion occurs predominantly through

the intercellular channels.61-62 While this seems logical for

lipophilic compounds, this pathway has also been shown to be

the dominant route for mercuric chloride, an ionic compound.

Bodd6 et al.63 showed that mercuric sulfide could be

precipitated in the stratum corneum after topical application

of mercuric chloride. They found that mercuric cation

accumulates initially in the intercellular spaces throughout

the stratum corneum. Intracellular uptake was observed

later, but only in the apical corneocytes by way of

disintegrating desmosomal attachments. Since the

intercellular lipids are arranged in multiple lamellar

bilayers,64 the mercuric cation, and possibly other ions and

hydrophilic molecules, may diffuse through the interlamellar,

hydrophilic channels that are associated with the polar

headgroups of the lipids.63

While most of the current research is directed toward

understanding the intercellular region, Barry59 cautions that

the transcellular route should not be dismissed as

unimportant, especially in penetration enhancer research.











Passive Diffusion


Although the skin is not a homogeneous tissue, it has

the characteristics of a passive diffusion barrier. Fick's

first law of diffusion defines flux as the amount of material

flowing through a unit cross section of a barrier in a unit

time.65 Flux is also proportional to the concentration

gradient, and the first law can be written:

J = -D-dC/dx (1)

where J is the flux, D is the diffusion coefficient of the

penetrant in the barrier, and dC/dx is the concentration

gradient. While the first law is used to describe steady-

state diffusion, Fick's second law describes the change in

concentration with respect to time at a specific location.65

Our interest is in steady-state diffusion, and the second law

will not be discussed further.

The concentration gradient term (dC/dx) from Fick's

first law can be approximated by (C1-C2)/h, where C1 is the

concentration in the barrier on the donor side of the

barrier, while C2 is the concentration in the barrier on the

receptor side, and h is the thickness of the diffusional

barrier. Therefore, the expression becomes

J = D-(CI-C2)/h (2)

For skin penetration studies, "sink" conditions can be

assumed which means that C20O, and

J = D-C1/h (3)










Since C1 is essentially the concentration in the first

layer(s) of skin, C1 will now be referred to as Cs, so

J = D-Cs/h (4)

While Cs is not usually a known quantity, it can be

represented instead by the product of the penetrant

concentration in the donor vehicle (Cv) and the

membrane-vehicle partition coefficient (Km = Cs/Cv), and the

expression now becomes

J = D-Km-Cv/h (5)

This expanded form of Fick's first law is often cited and has

been verified by data from numerous skin penetration

studies.66

Another useful measure of skin penetration is the

permeability coefficient (P). This is simply the

concentration-normalized flux which is expressed:

P = J/Cv = D-Km/h (6)

Permeability coefficients are often used when comparing a

single penetrant in a series of vehicles or when studying an

homologous series of penetrants.


Enhancement of Skin Penetration


The driving force for skin penetration is the membrane-

vehicle partition coefficient (Km). If a penetrant is

delivered in a saturated solution in the presence of excess

solid, then the chemical potential, or escaping tendency, is

maximized, and the actual concentration in the vehicle (Cv)










is irrelevant. Since the membrane thickness (h) and

diffusion coefficient (D) remain relatively constant (D is

inversely proportional to the cube root of the molar volume

according to the Stokes-Einstein equation), Km is the only

variable that can substantially influence the flux.

There are several general approaches for enhancing skin

penetration. Four of these--occlusion, use of penetration

enhancers, iontophoresis, and sonophoresis--produce their

enhancement effects by changing the barrier properties of the

skin. Occlusion involves covering the application site,

impeding transepidermal water loss, and increasing the

hydration state of the skin.67 With penetration enhancers,

accelerants, or promoters in topical formulations, the

reversible reduction of barrier resistance in the stratum

corneum is the goal, and ideally, incorporation of the

enhancer into the skin will not result in cell damage.59

Iontophoresis is a technique in which electroosmotic volume

flow from an applied electric field leads to increases in

mass transfer in excess of passive diffusion.68 Sonophoresis

uses ultrasonic frequencies to increase skin penetration.

The other two methods, formulation (without penetration

enhancers) and the use of prodrugs, do not disrupt the

barrier layer. Essentially, the formulation approach

involves changing the penetrants solubility in the vehicle by

changing the vehicle. The effect on flux of this approach is

indeterminate according to equation (5). If the solubility

in the vehicle (Cv) is increased, then the partition










coefficient (Km) decreases and vice versa. In a series of

papers,69-72 Sloan and coworkers suggested that a parabolic

relationship exists for log Km and log P when they are

plotted against vehicle polarity. Log Km and log P reach a

minimum where vehicle polarity is equal to penetrant polarity

or in other words, where penetrant solubility in the vehicle

is the greatest. While this relationship might be expected,

a more interesting finding was that in most instances fluxes

are also lower from those vehicles in which the penetrants

are most soluble. While a formulation approach may permit

some improvement in skin penetration values, the maximum

achievable levels appear to be limited.

The final method for enhancing skin penetration is the

prodrug approach. This is the only method in which a

substantial increase in Km can be realized, and according to

equation (5), this should translate into substantially

improved skin penetration. It was this knowledge that led to

adoption of the prodrug method for the current project.


Prodrugs and Skin Penetration


The term, prodrug, was defined earlier in this chapter.

Prodrugs, bioreversible derivatives, or latentiated drugs are

simply compounds which undergo biotransformation before

producing their pharmacological effects.73

Generally, prodrugs are designed to overcome some kind

of barrier to a drug's usefulness. These barriers can










include (1) premature metabolism prior to reaching the active

site, (2) too rapid absorption and distribution when

prolonged action is needed, (3) toxicity associated with

(a) local irritation or (b) distribution to tissues other

than the target site, (4) poor site specificity leading to

subtherapeutic levels at the target site, and (5) generally

poor physical chemical properties resulting in (a) solubility

problems in the dosage form or (b) poor absorption across

biological membranes such as the blood brain barrier,

gastrointestinal lining, or skin.74 Obviously, it is the

latter problem that is addressed in the current project.

The idea that improving skin penetration is a solubility

problem has been documented in this chapter, and the

importance of biphasic (lipid and aqueous) solubility is well

recognized.66,75-76 In a recent review of prodrug approaches

for improving dermal delivery, numerous literature examples

are cited in which both solubility and skin penetration data

are presented.77 The evidence strongly suggests that:

although an increase in lipid solubility due to
transient masking of a polar functional group almost
always results in enhanced dermal delivery of the parent
drug, in order to optimize delivery, it is necessary to
use the members of the (homologous) series (of prodrugs)
that are more water soluble than the parent drug or that
are the more water-soluble members) of the series
(p. 68).77

Derivatization to improve solubility characteristics of

polar heterocycles with amide and/or imide functional groups

is relatively easy to accomplish. For example, successive

methylation of uracil at the N1- and N3-positions, while










obviously improving lipid solubility, also increases water

solubility from 3 mg/mL for uracil to 500 mg/mL for 1,3-di-

methyluracil even though the methyl group is hydrophobic

itself. Melting points are also decreased as the two

intermolecular hydrogen-bonding N-H sites are masked.78 Since

alkyl groups such as methyl are stable and therefore not

bioreversible, they are not candidates to function as prodrug

promoieties. However, many other derivatives, such as those

cited earlier, when linked to the N1- or N3-site of 5FU, do

qualify as promoieties and could potentially give 5FU the

improved solubility characteristics necessary for enhanced

skin penetration.

Of the many studies devoted to making bioreversible

derivatives of 5FU, only a few have been directed at

improving its topical delivery. Mollgaard et al.21 looked at

two 1-acyloxymethyl derivatives of 5FU. One of the compounds

delivered 5FU more readily than 5FU itself through excised

human skin. Both compounds showed greatly increased lipid

solubilities with only slightly reduced water solubilities.

Hydrolysis of these derivatives was attributed to cutaneous

metabolism by hydrolytic enzymes.

Three 1,3-bis-aminomethyl (Mannich base) derivatives of

5FU were prepared by Sloan and coworkers,38-39 and their

topical delivery was studied using hairless mouse skin.

Solubilities in lipid and aqueous phases were substantially

increased as were the rates of delivery through the skin of

5FU from these prodrugs. Due to the instability of these










compounds in water, no enzymatic activation was necessary to

release the parent compound. In a later report,79 one of

these prodrugs was compared with a number of 5FU formulations

including four commercially available creams and solutions.

The prodrug outperformed the formulations by a minimum of

four times in terms of 5FU delivered through hairless mouse

skin.

Sasaki et al.37 studied the delivery of 1-alkylcarbamoyl

derivatives of 5FU through rat skin. All three derivatives

(butyl, hexyl, and octyl) were more effective in delivering

5FU than 5FU itself. The lipid solubilities of the three

derivatives were comparable with each other and much higher

than 5FU. The aqueous solubilities, while less than 5FU,

showed an order of magnitude decrease between each derivative

beginning with the butyl derivative. Interestingly, the best

performing compound was the least lipid-soluble and most

water-soluble derivative, the butyl derivative.


Dermal versus Transdermal Delivery


It is important to make a distinction between dermal and

transdermal delivery. Most in vitro skin penetration testing

is done with excised skin to which a drug in a formulation is

applied on the donor side and samples are removed from the

receptor side. These experiments give information on

transdermal rates of delivery.










Topical 5FU works on afflicted cells in the epidermal

region of the skin. This is considered dermal delivery.

Transdermal techniques provide valuable information for

dermally targeted drugs since they indicate how effectively

the drugs penetrate the barrier layer of the skin. Other

experiments can be done to augment the transdermal results

such as measuring accumulation of the drug in the skin.

The correlation between transdermal delivery rates and

epidermal uptake was studied by Sloan et al.79-80 Incorpora-

tion of 3H-deoxyuridine into epidermal DNA of live hairless

mice was quantitated by scintillation counting following

application of various 5FU formulations and a 5FU prodrug.

The prodrug, which had the highest in vitro transdermal flux,

was also the most effective at inhibiting epidermal DNA

synthesis in vivo. The correlation was also good among the

formulations.


Cutaneous Metabolism


When a prodrug is applied topically and targeted for a

dermal site, it is essential that the parent drug is released

before the prodrug leaves the epidermal region. If a prodrug

has a relatively short half-life under physiological

conditions, such as the aforementioned Mannich-base prodrugs

of 5FU, then release of the parent drug will probably occur

either prior to or during its transit through the viable

epidermis. However, if a compound depends on enzymatic











rather than chemical activation, then an appropriate

cutaneous enzyme must be present.

The ability of the skin to metabolize drugs and other

foreign compounds is well known.81-82 Phase I reactions

(oxidation, reduction, and hydrolysis) and phase II

conjugations are known to occur. Drug-metabolizing enzymes

are distributed in all layers of the skin and the appendages.

Of particular interest in prodrug chemistry is the

presence of nonspecific esterases in the skin.82 The

predictable metabolism of esters by these hydrolytic enzymes

makes this functional group a popular choice for prodrug

synthesis.


Hairless Mouse Skin as a Model Membrane


A number of animal skins have been suggested as model

membranes for skin penetration studies. The two most common

models for in vitro diffusion studies are human and hairless

mouse skin. While the advantages of human skin are obvious,

there are also disadvantages. Human skin can be difficult to

obtain and store,83 it can be expensive,83 and it is known to

have high barrier variability.83-84 Factors such as age,

diet, and disease state may not be well controlled with human

skin.85 Hairless mouse skin, on the other hand, is easily

obtained and prepared, and other factors can ordinarily be

controlled.









The importance of hairless mouse skin for in vitro

experimentation has been noted86 despite its generally greater

permeability when compared to human skin. It has been

suggested that this difference may actually be an asset in

that small changes in permeability will be amplified in the

mouse skin model.86

A major criticism of hairless mouse skin involves its

ability to withstand the effects of hydration,87-88 a

necessary condition for controlled in vitro diffusion

experiments. Permeability increases as a function of

hydration time have been attributed to breakdown of the

stratum corneum barrier in these studies. A recent report,89

however, suggested that the absence of an adequate

preservative in the receptor phase may actually be

responsible for breakdown of the skin. Skin penetration data

was collected for delivery of theophylline from a propylene

glycol vehicle following skin hydration periods ranging from

4 to 120 hours. It was found that increased theophylline

flux and loss of barrier function corresponded to the

presence of microbial growth in the receptor phase. When

microbial growth was completely inhibited, fluxes were

essentially constant for all hydration periods.89 Finally,

with regard to hydration, Scheuplein and Ross noted that

"even well-hydrated stratum corneum preferentially dissolves

lipid-soluble molecules, so that the selective permeability

of these molecules is preserved" (p. 353).90










Another difference between human and hairless mouse skin

could be significant in prodrug design. Rat and mouse skins

apparently have higher levels of enzymatic activity for drug

metabolism than human skin. Esterase activity, specifically,

appears to be low in human skin.54


Research Proposal


The overall goal of the present research is to develop

prodrug derivatives of 5FU with solubility characteristics

and skin penetration properties superior to 5FU itself. The

derivatives should be stable in formulation, but should

readily convert to 5FU in the skin since the purpose of

topical 5FU therapy is to deliver 5FU dermally, not

transdermally. In pursuit of this goal, it is hoped that

experimental data will be obtained which supports a

solubility-based method for designing prodrugs with optimized

topical delivery characteristics. The objectives for meeting

this goal are to:

1) synthesize several homologous series of acylated
derivatives of 5FU,

2) verify structures by melting point, elemental
analysis (novel compounds), and standard spectral
techniques,

3) determine physical chemical properties such as lipid
and aqueous solubilities, partition coefficients, and
rates of hydrolysis, and

4) determine skin penetration parameters and quantitate
skin accumulation using hairless mouse skin as the model
membrane.











Table 1-1. Structures of 5FU and prodrug derivatives of 5FU.






O O
HF R F





H Ri






Series R1 R2

l-alkyloxycarbonyl-5FU (I) -(C=O)O(CH2)nCH3 -H
l-acyl-5FU (II) -(C=O) (CH2)nCH3 -H
1,3-bis-acyl-5FU (III) -(C=0) (CH2)nCH3 -(C=0) (CH2)nCH3
3-acyl-5FU (IV) -H -(C=0)(CH2)nCH3










Four series of compounds have been selected as potential

prodrugs of 5FU. They are the l-alkyloxycarbonyl (I), 1-acyl

(II), 1,3-bis-acyl (III), and 3-acyl (IV) derivatives of 5FU

(Table 1-1). Fluorouracil (5FU) has two acidic pKa values,

8.0 and 13.0.91 Spectral studies have shown that the

monoanion is actually a mixture of N1- and N3-anions.30 A

comparison of ionization constants for 5FU derivatives that

are identically substituted at the N1- or N3-positions22,30

suggests that the N3-position is probably the most acidic.

However, the N1-position is the most reactive site for both

synthesis and hydrolysis of acylated derivatives of 5FU.

The three series of acyl derivatives were chosen for two

reasons; they exhibit high lipid and aqueous solubilities,

and they readily hydrolyze without enzymatic activation.26

While rapid hydrolysis is seen as a drawback by some

investigators,26,42 it may actually be an advantage when

developing prodrugs for dermal delivery.38-39,92 The

l-alkyloxycarbonyl series was selected for comparison with

the 1-acyl series since it shows good chemical stability, but

rapid enzymatic hydrolysis.30 Various members of each series

have been synthesized previously,25-27,30,93 but homologous

series have not been examined, and nobody has studied their

applicability for dermal delivery.















CHAPTER 2
1-ALKYLOXYCARBONYL DERIVATIVES


Introduction


Alkyloxycarbonyl derivatives of 5-fluorouracil (5FU)

have previously been studied as potential sources of 5FU in

viva.25,29-30 When substitution is at the N3-position of 5FU,

the derivatives are chemically stable. Their hydrolyses in

human plasma and liver homogenate are also slow enough to

raise questions about their usefulness as prodrugs for the

oral or rectal delivery of 5FU.29 Thus, there is no doubt

that they are too stable to serve as prodrugs for dermal

delivery. On the other hand, substitution at the N1-position

produces compounds that are relatively stable chemically,30

but which are sufficiently labile in the presence of enzymes30

to justify consideration as dermal prodrugs.

Six straight-chain 1-alkyloxycarbonyl derivatives were

selected for study. The derivatives and their structures are

shown in Table 2-1.











Table 2-1. Structures of 1-alkyloxycarbonyl derivatives.







O
H F








RO 0






Compound R

l-methyloxycarbonyl-5FU (1) -CH3
l-ethyloxycarbonyl-5FU (2) -CH2CH3
l-propyloxycarbonyl-5FU (3) -(CH2)2CH3
l-butyloxycarbonyl-5FU (4) -(CH2)3CH3
l-hexyloxycarbonyl-5FU (5) -(CH2)5CH3
l-octyloxycarbonyl-5FU (6) -(CH2)7CH3











Materials and Methods



Synthesis


Melting points (mp) were determined with a Thomas-Hoover

capillary melting point apparatus and are uncorrected.

Elemental microanalyses were obtained for all novel compounds

through Atlantic Microlab, Incorporated in Norcross, Georgia.

Proton nuclear magnetic resonance (1H NMR) spectra were

obtained at 90 MHz on a Varian EM-390 spectrometer. Chemical

shifts (6) are reported in parts per million (ppm) from the

internal standard, tetramethylsilane (TMS). Coupling

constants (J) are expressed in cycles per second (Hz).

Infrared (IR) spectra were recorded with a Perkin-Elmer 1420

spectrophotometer and absorbances are reported in cm-.

Ultraviolet (UV) spectra were obtained with a Cary 210 or

Shimadzu UV-265 spectrophotometer. Maximum absorbances are

reported in nm along with the molar absorptivities (E) in

L/mol.

l-Alkyloxvcarbonyl-5-fluorouracil (general procedure)

To 0.66 g (0.01 mol) of 85% potassium hydroxide

dissolved in methanol (20-50 mL) was added 1.33 g of 5FU

(0.0101 mol). Slightly more than an equivalent of 5FU was

used to prevent excess base from catalyzing the aldol

condensation of acetone. When present, condensation products

complicated product isolation. The methanol suspension was









stirred for 30 minutes, and the methanol was evaporated under

reduced pressure. The potassium salt was suspended in

acetone (25-50 mL) which was evaporated under reduced

pressure to remove residual methanol. The salt was

resuspended in acetone (25-50 mL), and the suspension was

added dropwise over a three minute period to a well stirred

acetone (20 mL) solution containing 1.0 to 1.2 equivalents of

the appropriate alkyl chloroformate. The mixture was stirred

at room temperature for 60 minutes. The mixture was

filtered, and the residue was washed with acetone (20 mL).

The combined acetone solutions were evaporated under reduced

pressure, and the solid residue was crystallized from an

appropriate solvent or solvent combination.1

l-Methyloxycarbonyl-5-fluorouracil (1)

Crystallization from acetone gave 1.36 g of 1 (72%):

mp 158-60 *C (lit.30 mp 159-60 OC); IR (KBr) 1695, 1710, 1740,

and 1760 cm-1 (C=O); 1H NMR [(CD3)2SO] 6 3.86 (s, 3H, C13) and

8.16 (d, J=7 Hz, 1H, C6-f); UVax (CH3CN) 254 nm (E=9.63x103).

l-Ethyloxycarbonyl-5-fluorouracil (2)

Crystallization from acetone/ether gave 1.31 g of 2

(65%): mp 127-8 OC (lit.30 mp 126-8 OC); IR (KBr) 1690, 1730,

and 1750 cm-1 (C=O); 1H NMR [(CD3)2SO] 8 1.31 (t, J=7 Hz, 3H,

Ca3), 4.31 (q, J=7 Hz, 2H, OCH2), and 8.16 (d, J=7 Hz, 1H,
C6-H); UVmax (CH3CN) 254 nm (e=9.86x103).




1Several compounds in this series were provided by Kenneth B.
Sloan, Ph. D. based on the author's procedure.









l-Propyloxycarbonyl-5-fluorouracil (3)

Crystallization from acetone/ether gave 1.37 g of 3

(64%): mp 124-6 oC; IR (KBr) 1690, 1730, and 1755 cm-1 (C=O);

1H NMR [(CD3)2SO] 8 0.95 (t, J=7 Hz, 3H, Cfl3), 1.70 (m, 2H,

OCH2Cf2), 4.23 (t, J=7 Hz, 2H, OCl2), and 8.15 (d, J=7 Hz, 1H,

C6-) ; UV max (CH3CN) 254 nm (E=1.001x104).

Anal. Calc. for C8H9FN204: C, 44.45; H, 4.20; N, 12.96.

Found: C, 44.53; H, 4.23; N, 12.89.

l-Butyloxycarbonyl-5-fluorouracil (4)

Crystallization from dichloromethane/hexane gave 1.33 g

of 4 (58%): mp 97-8 oC (lit.30 mp 102-4 oC); IR (KBr) 1695,

1735, and 1765 cm-1 (C=O); 1H NMR [(CD3)2SO] 8 0.91 (t,

J=7 Hz, 3H, Cl13), 1.3-1.8 (m, 4H, OCH2Cl2Cfl2), 4.27 (t, J=6

Hz, 2H, OCfI2), and 8.13 (d, J=7 Hz, 1H, C6-Ij); UVmax (CH3CN)

254 nm (e=9.93x103).

l-Hexyloxycarbonyl-5-fluorouracil (5)

Crystallization from dichloromethane/hexane gave 1.27 g

of 5 (49%): mp 66-7 C (lit.93 mp 68-9 OC); IR (KBr) 1690,

1730, and 1750 cm-1 (C-O); 1H NMR [(CD3)2SO] 8 0.87 (distd t,

3H, Ca3), 1.1-1.8 (m, 8H, OCH2CIH2C2CHi2CH22), 4.26 (t, J=6 Hz,

2H, OCIH2), and 8.13 (d, J=7 Hz, 1H, C6-Hi); UVmax (CH3CN) 254

nm (e=1.004x104).

l-Octyloxvcarbonyl-5-fluorouracil (6)

Crystallization from dichloromethane/hexane gave 1.80 g

of 6 (61%): mp 97-8 oC; IR (KBr) 1690, 1730, and 1750 cm-1

(C=O); 1H NMR [(CD3)2SO] 8 0.87 (distd t, 3H, CJ3), 1.2-1.9









(m, 12H, OCH2CIj2Ca2Cni2CL2Ci2Ca2), 4.27 (t, J=6 Hz, 2H, OCf2),

and 8.15 (d, J=7 Hz, 1H, C6-i) ; UVmax (CH3CN) 254 nm

(e=1.009x104).

Anal. Calc. for C13H19FN204: C, 54.53; H, 6.69; N, 9.79.

Found: C, 54.46; H, 6.73; N, 9.77.


Lipid Solubility


Lipid solubilities were determined using isopropyl

myristate (IPM), a commercial vehicle used in cosmetics and

topical medicinals,94 as the lipid solvent. The use of IPM as

a model lipophilic vehicle in skin penetration studies is

well established.77,95

Three suspensions of each derivative were stirred at

221 OC for 48 hours. The suspensions were filtered through

0.45 gm nylon filters, and the saturated solutions were

diluted in acetonitrile and analyzed by UV spectroscopy.

Solubilities were calculated using Beer's Law:

A = E-C-d (1)

where A is the absorbance, E is the molar absorptivity, C is

the concentration in mol/L, and d is the path length of the

cuvette in cm. Molar absorptivities were predetermined in

triplicate in acetonitrile at 254 nm.


Aqueous Solubility


For direct measurement of aqueous solubilities, three

suspensions of each derivative were vigorously stirred in










0.05 M acetate buffer (pH=4.0) at 221 oC for 60 minutes.

The suspensions were filtered through 0.45 pLm nylon filters,

and the saturated solutions were diluted in acetonitrile and

analyzed by UV spectroscopy. Solubilities were calculated

using Beer's Law as previously described.


Partition Coefficients


The partitioning-based method for determining aqueous

solubility utilized the saturated IPM solutions from the

lipid solubility study. For most compounds, equal volumes

(1 mL) of saturated IPM solution and 0.05 M acetate buffer

(pH=4.0) were used. The use of equal or near-equal phase

volumes is known to facilitate rapid equilibrium.96 The two

phases were mixed thoroughly for ten seconds and allowed to

separate for 60 seconds. A preliminary study showed that

there was virtually no difference in partition coefficient

(PC) values when partitioning was carried out for 10, 20, or

30 seconds (see Chapter 3). The IPM layers were diluted in

acetonitrile and analyzed by UV spectroscopy. The IPM-buffer

partition coefficients were calculated as follows:

PC = Aafter/(Abefore-Aafter) VAQ/VIPM (2)

where Aafter is the absorbance from the IPM layer after

partitioning, Abefore is the absorbance from the IPM layer

before partitioning, VAQ is the volume of the aqueous phase,

and VIPM is the volume of the IPM phase. Estimated aqueous









solubilities (SAQ) were calculated from the IPM solubility

(SIpM) and the partition coefficient:

SAQ = SIPM/PC (3)

Partitioning was carried out in triplicate for a fixed volume

ratio for each derivative. For those compounds with large

differences in solubility in one phase relative to the other,

volume ratios (IPM:buffer) other than 1:1 were necessary, but

the ratio never exceeded 10:1 or 1:10.


Hvdrolysis Kinetics


Hydrolysis rates have previously been reported for

several members of this homologous series.30,93 In the

present study, hydrolysis rates were determined at 32 C for

l-methyloxycarbonyl-5FU (1) in 0.05 M phosphate buffer

(pH=7.1, 1=0.12) and in the same buffer with 0.11%

formaldehyde (3.6x10-2 M). The rate in the presence of

formaldehyde was determined for comparison with the rate in

plain buffer since formaldehyde was used as a preservative in

the diffusion cell experiments described in the following

section.

The hydrolyses were followed by high performance liquid

chromatography (HPLC). The HPLC system consisted of a

Beckman model 110A pump with a model 153 UV detector, a

Rheodyne 7125 injector with a 20 Im loop, and a Hewlett-

Packard 3392A integrator. The column was a Lichrosorb RP-8

10 pm reversed-phase column, 250 mm x 4.6 mm (inside










diameter). The mobile phase contained 10% methanol and 90%

0.025 M acetate buffer (pH=5.0) and was run at 1.0 mL/min.

The column effluent was monitored at 254 nm, and quantitation

was based on peak areas. Standards chromatographed under the

same conditions were used for calibration.

Hydrolysis was initiated by adding 0.4 mL of a stock

solution of compound 1 in acetonitrile to 25 mL of buffer

prewarmed to 32 OC in a constant temperature water bath to

give final concentrations of ~1.8x10-4 M. Aliquots were

removed at appropriate intervals and chromatographed

immediately. Pseudo-first-order rate constants were

determined from the expression:

In(Ct) = In(Co)-kt (4)

where Ct is the concentration at some time=t, Co is the

concentration at t-0, k is the pseudo-first-order rate

constant, and t is the time. The slopes, -k, of linear plots

of ln(Ct) versus t were determined by linear regression. The

half-lives (tl/2) were calculated from

tl/2 = 0.693/k (5)

Each hydrolysis reaction was run in triplicate and was

followed for a minimum of three half-lives. The correlation

coefficients were >0.999.


Skin Penetration Studies


Diffusion cell experiments were performed to measure the

transdermal delivery of 5FU and the 5FU prodrugs. Franz-type









diffusion cells from Crown Glass in Somerville, NJ with

4.9 cm2 donor surface areas and 20 mL receptor phase volumes

were used for this purpose. The full-thickness skins were

obtained from female hairless mice (SKH-hr-1) from Temple

University Skin and Cancer Hospital.

The mice were killed by cervical dislocation, their

skins were removed immediately by blunt dissection, and

dorsal sections were mounted in the diffusion cells. The

dermal sides of the skins were placed in contact with

receptor phase which contained 0.05 M phosphate buffer

(pH=7.1, I=0.12) with 0.11% formaldehyde as a preservative.

The effectiveness of formaldehyde for this purpose has

recently been documented.89 The receptor phases were stirred

continuously and kept at constant temperature (32 oC) by a

circulating water bath. A preapplication period of 48 hours

was established to uniformly condition the skins and to

remove water-soluble UV-absorbing materials. The receptor

phases were changed three times during this period, and

control experiments from earlier studies have shown that this

procedure effectively removes those materials.97 The

epidermal sides of the skins were exposed to the air and were

left untreated during this period.

After the preapplication period, 0.5 mL aliquots from

suspensions of the prodrugs in IPM were applied to the

epidermal sides of the skins. The IPM suspensions were

stirred at 221 C for 48 hours prior to application to

ensure that saturation was attained. Total concentrations of









the IPM suspensions ranged from 0.3 M to 0.8 M with enough

excess solid present to maintain saturation for the duration

of the application period (see below). Each drug-vehicle

combination was run in triplicate.

Samples were taken from the receptor phases at 4, 8, 12,

21, 24, 27, 30, 33, 36, 45, and 48 hours after donor phase

application. The receptor phases were changed following

removal of each sample so that "sink" conditions were

maintained. Samples were analyzed for total 5FU species that

had diffused by UV spectroscopy (e=7.13x103 at 266 nm) after

allowing at least 72 hours for complete prodrug hydrolysis.

Cumulative amounts of total 5FU species that diffused (pmol)

were plotted against time (h), and the slopes of the linear,

"steady-state" regions were calculated using linear

regression. The slopes, when divided by 4.9 (the area of the

donor surface in cm2), gave the "steady-state" fluxes

(gmol/cm2/h). In a separate experiment, HPLC was used to

determine intact prodrug content in the receptor phases at

each sampling time. Mobile phase containing 18-50% methanol

in 0.025 M acetate buffer (pH=5.0) was used with the system

described earlier. Aliquots were removed and chromatographed

immediately after the samples were taken. Prodrug fluxes

were calculated in the same manner as fluxes for total 5FU.

Donor phases were changed every twelve hours and were

set aside for 1H NMR analysis. Stability of the prodrugs in

IPM was determined from the chemical shift of C6-H. In

dimethylsulfoxide-d6, the C6-H signal for 5FU appears at










8-7.73 ppm. For each of the 1-alkyloxycarbonyl derivatives,

the same signal in dimethylsulfoxide-d6 appears at 8>8.10 ppm.

Since this area of the spectrum is free from interference by

IPM absorbances, the two signals can be identified and

quantified if necessary.

Following removal of the donor phases after the 48-hour

application period, the epidermal sides of the skins were

washed three times with 5 mL portions of methanol to remove

all remnants of prodrug and vehicle from the skin surfaces.

This was accomplished quickly (<3 min) to minimize contact

time between the skins and methanol. The receptor phases

were changed again, and the dermal sides were kept in contact

with the fresh buffer for 23 hours while the epidermal sides

were again left exposed to the air. After this "leaching"

period, another sample was taken from each cell to measure

the skin accumulation of total 5FU species.

Second applications to the epidermal sides of the skins

were made after the "leaching" period with a standard drug-

vehicle suspension Theophylline in propylene glycol

(0.4 M) was applied to assess the damage to the skins from

application of the initial drug-vehicle combinations.

Samples were taken at 1, 2, 3, 5, 7, 9, and 11 or 12 hours

after application. The samples were analyzed for

theophylline by UV spectroscopy (E=1.02x104 at 271 nm) and

second application fluxes were determined as described above.










Results and Discussion



Synthesis and Structure Determination


The known l-alkyloxycarbonyl-5FU derivatives have

melting points25,30,93 and spectral properties (UV30 and 1H

NMR25) in agreement with those reported in the literature.

The structures of the novel compounds were assigned by

comparison of their spectral properties with those of the

known homologs. Elemental microanalyses were obtained for

the novel compounds and were within acceptable limits

(0.4%).

Acylation on the N1- or N3-position can be distinguished

by UV and 1H NMR analysis. Anions of N3-substituted

derivatives undergo a substantial shift of their UVmax to

longer wavelength while anions of N1-substituted compounds do

not.30,98 This is reportedly due to the extended conjugation

possible for the N1-anion.20 Differences in 1H NMR spectra

are also well defined. In chloroform-d, the C6-H signal for

N1-substituted derivatives is a sharp doublet indicating

coupling with C5-F. The same signal in chloroform-d for

N3-substituted derivatives appears as a broad singlet or

triplet-like doublet of doublets from additional coupling of

C6-H with N1-H.20 When the substituents contain a carbonyl

group attached to the N1-position, as they do in the

1-alkyloxycarbonyl series, an anisotropic effect is observed










in which the C6-H signal is shifted downfield relative to 5FU

or the N3-substituted derivatives.25 For example, the C6-H

chemical shift for l-ethyloxycarbonyl-5FU (8=8.00) is 0.77

ppm downfield when compared with 3-ethyloxycarbonyl-5FU

(8=7.23) in chloroform-d.


Solubility


Solubility determinations are generally accomplished by

stirring excess solute in a solvent until saturation is

attained. The excess solid is removed and the saturated

solution is assayed for solute content. This approach is

suitable for stable solutes and for unstable solutes in

aprotic solvents, but another method is needed for measuring

aqueous solubilities of chemically unstable compounds.

An alternative to the direct method for determining

aqueous solubilities is the partitioning-based method. The

advantage of this method is that contact time between the

unstable compound and the aqueous phase can be minimized.

However, several points regarding this procedure require

clarification.

First, partition coefficients are concentration

dependent except when compounds with low associating

tendencies are present in dilute solutions (<10-1 M).96

Solubilities based on partition coefficients can only be

reported as estimates since activity coefficients become more

important at higher solute concentrations.99










Second, the partitioning and phase separation times that

were used in the partitioning-based method for estimating

aqueous solubility were chosen empirically. The times had to

be sufficiently long to allow equilibrium distribution of the

compounds between the phases to occur and to allow subsequent

separation of the phases to occur without substantial

hydrolysis of the prodrugs. Longer times than those chosen

could have been used for these more stable 1-alkyloxycarbonyl

derivatives, but in order to validate the procedure for all

four series, shorter times that were more appropriate for the

less stable derivatives were used.

Finally, due to the experimental design and the liability

of the prodrugs, mutual saturation of the phases prior to

partitioning could not be accomplished. Since the ester

(IPM) that was used as the lipid phase in these experiments

is practically insoluble in water,94 changes in the phase

volumes during partitioning are probably insignificant. This

potential volume change is a common source of error when more

water-soluble lipid solvents are not presaturated with their

corresponding aqueous phases.100

The conventional, direct method was used for determining

lipid solubilities. Since the 1-alkyloxycarbonyl derivatives

are the most chemically stable compounds of the four series,

both the direct and the partitioning-based methods were used

for determining their aqueous solubilities. Thus, the two

methods for determining aqueous solubility were compared

using the 1-alkyloxycarbonyl derivatives as a model series.










Lipid (SIpM) and aqueous (SAQ) solubilities for the

l-alkyloxycarbonyl derivatives are presented in Table 2-2

along with their melting points. Lipid solubilities are

greatly enhanced by making derivatives. Solubility values

range from over 40 times greater than 5FU for 1-methyloxy-

carbonyl-5FU (1) to more than 3000 times greater than 5FU for

l-hexyloxycarbonyl-5FU (5). Increases in lipid solubility

with increasing chain length are accompanied by decreases in

melting point. A change in that trend is observed for

l-octyloxycarbonyl-5FU (6), but this is not unexpected.

Since melting point and solubility depend in part on crystal

lattice energies,101 this result indicates that the crystal

structure is dominated by the 5FU nucleus for lower homologs

and by the aliphatic side chain for higher homologs.

Aqueous solubilities are reported for both the direct

and partitioning methods. The results show that aqueous

solubility peaks for l-ethyloxycarbonyl-5FU (2) and then

decreases. Compounds 1 and 2 have aqueous solubilities

greater than 5FU even though a hydrogen-bonding group (NI-H)

has been masked. Again, this is reflected in the lowered

melting points for the derivatives when compared to 5FU.

Aqueous solubilities determined by the partitioning-

based method underestimated the direct solubilities by 7% for

compounds 5 and 6, 8% for compound 1, 21% for compound 4, 22%

for compound 3, and 34% for compound 2. Relative aqueous

solubility among members of the series is the same with













Table 2-2. Melting points (MP), lipid solubilities (SIPM), and
aqueous solubilities (SAQ) for 1-alkyloxycarbonyl derivatives.


MP
Compound (OC)


SIpMa
(mM)


5FU 280-2 0.049 96 -85

1 158-60 2.1 120 111 124
2 127-8 13 263 174 34
3 124-6 15 55 43
4 97-8 34 29 23 26
5 66-7 153 5.4 5.0 5.8
6 97-8 36 0.14 0.13


aStandard deviations
solubilities.
bStandard deviations


from the mean were within 3% for IPM

from the mean were within 8% for


aqueous solubilities determined by direct method.
cStandard deviations from the mean were within 6% for
experimental values used to calculate partition coefficients
and estimated aqueous solubilities (11% for compound 2).
dLiterature values from references 30 and 93.












Table 2-3. Solubility ratios (SR), partition coefficients
(PC), and hydrophobicity parameters (w) for
l-alkyloxycarbonyl derivatives.


Compound SRa log(SR) Xb


PCc log(PC) Kd


1 0.018 -1.75 0.019 -1.72 0.03
2 0.050 -1.30 0.45 0.075 -1.12 0.60 0.18
3 0.28 -0.56 0.74 0.36 -0.45 0.67 0.11
4 1.2 0.07 0.63 1.4 0.16 0.61 0.09
5 29 1.46 0.70 31 1.48 0.66 0.02
6 257 2.41 0.48 285 2.45 0.49 0.04


log (PC) -
log (SR)


aSolubility ratio calculated from SIPM/SAQ.
bAlog(SR) for compound and preceding compound.
cExperimental partition coefficient (CIPM/CAQ).
dAlog(PC) for compound and preceding compound.










either procedure, and overall, agreement between the two

methods is good.

The direct aqueous solubility values for compounds 1, 4,

and 5 agree with the corresponding literature values included

in Table 2-2. However, the value for compound 2 is much

higher than the literature value. The basis for this

discrepancy is not clear.

In Table 2-3, the solubility ratios (SR) and

experimentally determined partition coefficients (PC) are

compared for the 1-alkyloxycarbonyl series. The values for

log(PC)-log(SR) indicate that the partition coefficients are

somewhat higher than the corresponding solubility ratios.

Generally, the more polar derivatives (log[PC]<0) show the

greatest difference with the exception of compound 1.

Yalkowsky et al.102 studied solubility ratios and

octanol-water partition coefficients for a broad range of

solutes. They concluded that self-association of polar

solutes in octanol increases the ability of octanol to

accommodate the solute which increases the partition

coefficient. Conversely, a nonpolar solute causes a decrease

in the partition coefficient by self-associating in the

aqueous phase.102 The present results can be explained on the

same basis. The lower than expected log(PC)-log(SR) value

for compound 1 could be due to the low concentration in the

IPM phase during partitioning of this derivative. Since the

aqueous solubility of compound 1 is nearly 60 times its lipid

solubility, the concentration in the IPM phase is reduced










well below its already low concentration at saturation.

Therefore, the solute-solute interactions that lead to higher

partition coefficients would also be reduced.

The hydrophobicity parameters100 (t) in Table 2-3 were

calculated from the relationship:103

log(PC)n = log(PC)0+ln (6)

where n is the number of methylene units in a homologous

series using both log(PC) and log(SR) values. Both

calculations yield an average I value equal to 0.60. Values

for I from the literature include 0.54 for silicone oil-water

and 0.66 for hexane-waterl01 indicating that 0.60 is a

reasonable value for the IPM-buffer partitioning system.


Hydrolysis Kinetics


Hydrolysis of l-methyloxycarbonyl-5FU (1) to 5FU in

0.05 M phosphate buffer (pH=7.1, 1=0.12) with and without

0.11% formaldehyde (3.6x10-2 M) was followed by HPLC at 32 OC.

Disappearance of compound 1, indicated by ln(C), is plotted

versus time (min) in Figure 2-1. The linearity of the plots

suggests that hydrolysis of 1 follows first-order kinetics in

the presence and absence of formaldehyde. Pseudo-first-order

rate constants (k) and half-lives (t1/2) from the linear plots

are presented in Table 2-4.

Hydrolysis of compound 1 is clearly faster in the

presence of formaldehyde indicating that general base

catalysis by formaldehyde hydrate may be involved (see
















B Plain Buffer

Formaldehyde Buffer
-9E





B
0 -10.



*
B

-11 0






-12
0 200 400 600 800

Time (min)




Figure 2-1. Plots of In(C) versus time (min) for hydrolysis
of l-methyloxycarbonyl-5FU in 0.05 M phosphate buffer
(pH=7.1, I=0.12) with and without formaldehyde
at 32 oC.













Table 2-4. Pseudo-first-order rate constants (k) and half-
lives (tl/2) for hydrolysis of l-methyloxycarbonyl-5FU
in 0.05 M phosphate buffer (pH=7.1, I=0.12)
with and without formaldehyde at 32 OC.


Compound


Formaldehyde
(M)


k(SD)a
(min-1)


tl/2
(min)


1 0 3.33x10-3(0.04x10-3) 208
1 3.6x10-2 3.78x10-3(0.05x10-3) 183


aMean standard deviation for n=3 values.










Chapter 3). Significant buffer catalysis by phosphate has

also been demonstrated for compounds in this series, and the

reader is referred to the work of Buur and Bundgaard30 for

complete pH-rate profiles and probable hydrolysis mechanisms.

In any case, release of the parent drug from compound 1

by chemical means is too slow for dermal delivery purposes.

Other members of the series would also appear to be poor

candidates, since they hydrolyze chemically over two times

slower than compound 1.30,93 In 80% human plasma, however,

hydrolysis rates are much faster (ti/2=2-3 min)30,93

suggesting enzyme catalysis of the 1-alkyloxycarbonyl series.

Since the skin is metabolically active, further study of

these compounds in diffusion cells was warranted.


Skin Penetration


Skin penetration data from the diffusion cells are

plotted as cumulative amount of total 5FU species that

diffused (JLmol) versus time (h). In Figure 2-2, results for

l-methyloxycarbonyl-5FU (1), l-ethyloxycarbonyl-5FU (2), and

l-propyloxycarbonyl-5FU (3) are compared to 5FU itself. In

Figure 2-3, results for l-butyloxycarbonyl-5FU (4),

l-hexyloxycarbonyl-5FU (5), and l-octyloxycarbonyl-5FU (6)

are compared to 5FU. Error bars correspond to the standard

deviation from the mean for n=3 values.

Fluxes (J), lag times (tL), and skin accumulation (SA)

values for each compound are reported in Table 2-5. Lag time












1000


0 Cpd1

800 Cpd2

Cpd3

E O 5FU
- 600 1

0
E

400 .





200 -




I I

0 10 20 30 40

Time (h)




Figure 2-2. Plots of cumulative amount of total 5FU species
that diffused (Imol) versus time (h) for
compounds 1, 2, 3, and 5FU.





47




300

I Cpd4

Cpds

SI Cpd

t "
E 200 5FU


o









o0 ,I I i II


0 10 20 30 40

Time (h)



Figure 2-3. Plots of cumulative amount of total 5FU species
that diffused (pol) versus time (h) for
compounds 4, 5, 6, and 5FU.












Table 2-5. Fluxes (J), lag times (tL), and skin accumulation
(SA) values for 1-alkyloxycarbonyl derivatives.


J(SD)a
Compound (Lmol/cm2/h)


Prodrug
S-Sb(11 h)c
(%)


tL SA(SD)a
(h) (pmol)


5FU 0.24(0.09) 13 3.7(0.9)

1 2.6(0.6) 42(16) 14 8.3(0.1)
2 5.9(1.3) 90(75) 13 18(4)
3 2.3(0.2) 78(43) 11 5.0(1.4)
4 2.2(0.1) 73(32) 10 4.2(0.5)
5 1.5(0.1) 79(14) 6.2 11(0)
6 0.29(0.02) 8.7 3.2(0.5)


aMean standard deviation for n=3 values.
percent of total 5FU as intact prodrug during "steady-state"
phase in separate experiment (n=l).
CPercent of total 5FU as intact prodrug from 11 h sample in
separate experiment (n=l).













Table 2-6. Second application fluxes (J) and lag times (tL)
for 1-alkyloxycarbonyl derivatives.


Ja(SD) b
(lomol/cm2/h)


Compound


5FU 1.2(0.2) 1.2

1 1.9(0.1) 0.8
2 1.8(0.4) 0.9
3 1.7(0.3) 1.0
4 1.8(0.2) 1.0
5 1.8(0.1) 0.8
6 1.8(0.2) 0.9


aFlux of 0.4 M theopylline from propylene glycol.
bMean standard deviation for n=3 values.










refers to the intersection of the linear, or "steady-state,"

region of each graph with the time (x) axis, and it is the

time required for establishing a uniform concentration

gradient within the skin.65 The percent of total 5FU present

as intact prodrug in the receptor phase is also reported in

Table 2-5. These values were calculated from samples taken

during the "steady-state" phase and from an earlier sample

(11 h) in a separate experiment using HPLC analysis (n=l).

The improvement in skin penetration of 5FU from the

1-alkyloxycarbonyl derivatives is significant except for

l-octyloxycarbonyl-5FU (6). Increases in flux are generally

about one order of magnitude, and l-ethyloxycarbonyl-5FU (2)

with nearly a 25-fold improvement is easily the best

derivative.

The presence of large amounts of intact prodrugs in the

receptor phases is a matter of interest. The high

percentages indicate that the hydrolytic enzymes of the skin

are not effectively converting the prodrugs to 5FU. It is

interesting to note, however, that percentages calculated

from a sample removed prior to "steady-state" are much lower

than the "steady-state" values. It is possible that the

large amounts of diffusant present at "steady-state" are

simply too much for the enzymes to handle. Another

possibility is that the continuous changing of the receptor

phase with each sample eventually depletes the enzymatic

activity.










The trend in skin accumulation is similar to the trend

in flux with the exception of l-hexyloxycarbonyl-5FU (5).

The large skin accumulation value and short lag time for

compound 5 may indicate a high affinity for the lipid regions

of the skin but less affinity for the hydrated regions and

the receptor phase.

Second application fluxes and lag times are presented in

Table 2-6. Skin penetration by theophylline from propylene

glycol, the standard drug-vehicle combination, is

approximately one and one-half times higher for the skins

treated with the 1-alkyloxycarbonyl derivatives than for

those treated with 5FU, and it is consistent throughout the

series. Therefore, skin damage is greater with the prodrugs,

but the difference is small when compared with the general

improvement in delivery of 5FU from the prodrugs.

The stability of the prodrugs in the IPM formulations

was assessed by 1H NMR analysis of the donor phases. After a

minimum of five days from the time the suspensions were

prepared until their 1H NMR spectra were recorded, including

at least twelve hours during which the formulations were in

contact with the skins, the 1-alkyloxycarbonyl derivatives

were found to be intact with no evidence of 5FU formation.


Summary


The 1-alkyloxycarbonyl derivatives of 5FU exhibited

decreased melting points and increased lipid solubilities










when compared to 5FU. Aqueous solubility reached a maximum

for l-ethyloxycarbonyl-5FU (2) and decreased from there with

increasing chain length. Skin penetration and skin

accumulation were also maximized for compound 2 suggesting

that both lipid and aqueous solubilities are important for

predicting transdermal and dermal delivery of these 5FU

prodrugs. The presence of high percentages of prodrugs in

the receptor phases, indicating insufficient release of the

parent drugs in the hairless mouse skin model, may limit the

potential of this series of prodrugs at least for dermal

delivery purposes. Finally, the partitioning-based method

for determining aqueous solubility appears to be a useful

method particularly for determining relative solubilities in

an homologous series.















CHAPTER 3
1-ACYL DERIVATIVES


Introduction


The 1-acyl derivatives of 5-fluorouracil (5FU) are

chemically unstable in aqueous solutions at all pH values.

In fact, their liability has been cited as a limitation to

their usefulness as drugs or prodrugs.26,42 If properly

formulated in an aprotic vehicle, however, this series of 5FU

derivatives may have potential for use as prodrugs for dermal

delivery.

Six straight-chain 1-acyl derivatives were selected for

study. The derivatives and their structures are shown in

Table 3-1.


Materials and Methods



Synthesis


Melting points (mp) were determined with a Thomas-Hoover

capillary melting point apparatus and are uncorrected.

Elemental microanalyses were obtained for all novel compounds

through Atlantic Microlab, Incorporated in Norcross, Georgia.

Proton nuclear magnetic resonance (1H NMR) spectra were

obtained at 90 MHz on a Varian EM-390 spectrometer. Chemical











Table 3-1. Structures of 1-acyl derivatives.






O
H F








R/O 0






Compound R

l-acetyl-5FU (7) -CH3
1-propionyl-5FU (8) -CH2CH3
1-butyryl-5FU (9) -(CH2)2CH3
1-valeryl-5FU (10) -(CH2)3CH3
l-hexanoyl-5FU (11) -(CH2)4CH3
l-octanoyl-5FU (12) -(CH2)6CH3









shifts (6) are reported in parts per million (ppm) from the

internal standard, tetramethylsilane (TMS). Coupling

constants (J) are expressed in cycles per second (Hz).

Infrared (IR) spectra were recorded with a Perkin-Elmer 1420

spectrophotometer and absorbances are reported in cm-1.

Ultraviolet (UV) spectra were obtained with a Cary 210 or

Shimadzu UV-265 spectrophotometer. Maximum absorbances are

reported in nm along with the molar absorptivities (E) in

L/mol. Single-crystal X-ray analysis was obtained for

l-acetyl-5FU through Hoffmann-La Roche in Nutley, NJ.

l-Acyl-5-fluorouracil (general procedure)

To 0.66 g (0.01 mol) of 85% potassium hydroxide

dissolved in methanol (20-50 mL) was added 1.31 g of

5-fluorouracil (0.01 mol). The methanol suspension was

stirred for 30 minutes, and the methanol was evaporated under

reduced pressure. The potassium salt was suspended in

acetonitrile (25-50 mL) which was evaporated under reduced

pressure to remove residual methanol. The salt was

resuspended in acetonitrile (25-50 mL), and the suspension

was added dropwise over 15 to 30 minutes to a well stirred

acetonitrile (25 mL) solution in an ice bath containing 1.0

to 1.2 equivalents of the appropriate acid chloride. The

mixture was stirred at 0 C for 60 minutes. The mixture was

filtered, and the residue was washed with acetonitrile

(25 mL). The combined acetonitrile solutions were evaporated

under reduced pressure, and the solid residue was










crystallized from an appropriate solvent or solvent

combination.

l-Acetvl-5-fluorouracil (7)

Crystallization from dichloromethane gave 0.98 g of 7

(57%): mp 129-30 OC (lit.27 mp 126-7 oC); IR (KBr) 1670, 1695,

1725, and 1770 cm-1 (C=O); 1H NMR (CDC13) 8 2.73 (s, 3H, CH3)

and 8.23 (d, J=7 Hz, 1H, C6-H); UVmax (CH3CN) 261 nm

(E=1.125x104).

l-Propionyl-5-fluorouracil (8)

Crystallization from dichloromethane/hexane gave 1.32 g

of 8 (71%): mp 130-1 OC (lit.27 mp 124-5 oC); IR (KBr) 1695,

1710, and 1740 cm-1 (C=O); 1H NMR (CDC13) 5 1.25 (t, J=7 Hz,

3H, CH13), 3.14 (q, J=7 Hz, 2H, COCH2), and 8.27 (d, J=7 Hz,

1H, C6-H); UVmax (CH3CN) 261 nm (E=1.141x104).

Anal. Calc. for C7H7FN203: C, 45.17; H, 3.79; N, 15.05.

Found: C, 45.26; H, 3.83; N, 14.97.

l-Butyryl-5-fluorouracil (9)

Crystallization from dichlormethane/hexane gave 0.86 g

of 9 (43%): mp 145-6 OC; IR (KBr) 1690, 1710, and 1740 cm-1

(C=O); 1H NMR (CDC13) 8 1.01 (t, J=7 Hz, 3H, CH3), 1.77 (m,

2H, COCH2CH2), 3.09 (t, J=7 Hz, 2H, COCH2), and 8.25 (d,

J=6 Hz, 1H, C6-j); UV max (CH3CN) 261 nm (E=l.168x104).

Anal. Calc. for C8H9FN203: C, 48.00; H, 4.53; N, 14.00.

Found: C, 48.12; H, 4.58; N, 13.91.










l-Valeryl-5-fluorouracil (10)

Crystallization from dichloromethane/hexane gave 1.35 g

of 10 (63%): mp 120-1 C; IR (KBr) 1695, 1715, and 1740 cm-1

(C=O); 1H NMR (CDC13) 8 0.95 (t, J=7 Hz, 3H, CH3), 1.3-1.8 (m,

4H, COCH2CH2CBI2), 3.11 (t, J=7 Hz, 2H, COCH2), and 8.24 (d,

J=7 Hz, 1H, C6-H); UVmax (CH3CN) 261 nm (E=1.175x104).

Anal. Calc. for C9H11FN203: C, 50.47; H, 5.18; N, 13.08.

Found: C, 50.52; H, 5.23; N, 13.03.

l-Hexanoyl-5-fluorouracil (11)

Crystallization from dichloromethane/hexane gave 1.71 g

of 11 (75%): mp 101-2 OC; IR (KBr) 1690, 1715, and 1745 cm-1

(C=O); 1H NMR (CDC13) 8 0.92 (distd t, 3H, Cf3), 1.2-1.9 (m,

6H, COCH2Cl2CH2CH2), 3.09 (t, J=7 Hz, 2H, COCE2), and 8.24 (d,

J=7 Hz, 1H, C6-K) ; UVmax (CH3CN) 261 nm (E=1.158x104) .

Anal. Calc. for Co1H13FN203: C, 52.63; H, 5.74; N, 12.27.

Found: C, 52.69; H, 5.75; N, 12.27.

l-Octanoyl-5-fluorouracil (12)

Crystallization from dichloromethane/hexane gave 1.28 g

of 12 (50%): mp 83-4 oC; IR (KBr) 1685, 1710, and 1745 cm-1

(C=O); 1H NMR (CDC13) 5 0.90 (distd t, 3H, CJ3), 1.2-1.8 (m,

10H, COCH2CHi2C Ii2Ck22C2Ci), 3.10 (t, J=7 Hz, 2H, COCI2), and

8.23 (d, J=7 Hz, 1H, C6-H); UVmax (CH3CN) 261 nm

(E=1.155x104).

Anal. Calc. for C12H17FN203: C, 56.24; H, 6.69; N, 10.93.

Found: C, 56.22; H, 6.73; N, 10.96.











Lipid solubility


Lipid solubilities were determined using isopropyl

myristate (IPM), a commercial vehicle used in cosmetics and

topical medicinals,94 as the lipid solvent. The use of IPM as

a model lipophilic vehicle in skin penetration studies is

well established.77,95

Three suspensions of each derivative were stirred at

221 OC for 48 hours. The suspensions were filtered through

0.45 pm nylon filters, and the saturated solutions were

diluted in acetonitrile and analyzed by UV spectroscopy.

Solubilities were calculated using Beer's Law:

A = E-C-d (1)

where A is the absorbance, E is the molar absorptivity, C is

the concentration in mol/L, and d is the path length of the

cuvette in cm. Molar absorptivities were predetermined in

triplicate in acetonitrile at 261 nm.


Aqueous Solubility


Because of the chemical instability of the 1-acyl

derivatives, direct measurement of aqueous solubilities for

these prodrugs was not attempted. A comparison of the direct

and partitioning-based methods for determining aqueous

solubility was presented in Chapter 2 for the 1-alkyloxy-

carbonyl derivatives.










Partition Coefficients


The partitioning-based method for determining aqueous

solubility utilized the saturated IPM solutions from the

lipid solubility study. For most compounds, equal volumes

(1 mL) of saturated IPM solution and 0.05 M acetate buffer

(pH=4.0) were used. The use of equal or near-equal phase

volumes is known to facilitate rapid equilibrium.96 The two

phases were mixed thoroughly for ten seconds and allowed to

separate for 60 seconds. A preliminary study with 1-acetyl-

5FU (7) showed that there was virtually no difference in

partition coefficient (PC) values when partitioning was

carried out for 10, 20, or 30 seconds (PC=0.1830.006, a

standard deviation of only 3%). The IPM layers were diluted

in acetonitrile and analyzed by UV spectroscopy. The IPM-

buffer partition coefficients were calculated as follows:

PC = Aafter/ (Abefore-Aafter) VAQ/VIPM (2)

where Aafter is the absorbance from the IPM layer after

partitioning, Abefore is the absorbance from the IPM layer

before partitioning, VAQ is the volume of the aqueous phase,

and VIpM is the volume of the IPM phase. Estimated aqueous

solubilities (SAQ) were calculated from the IPM solubility

(SIPM) and the partition coefficient:

SAQ = SIPM/PC (3)

Partitioning was carried out in triplicate for a fixed volume

ratio for each derivative. For those compounds with large










differences in solubility in one phase relative to the other,

volume ratios (IPM:buffer) other than 1:1 were necessary, but

the ratio never exceeded 10:1 or 1:10.


Hydrolysis Kinetics


Hydrolysis rates have previously been reported for one

member of this homologous series.26 In the present study,

hydrolysis rates were determined at 32 OC for all six 1-acyl

derivatives in 0.05 M phosphate buffer (pH=7.1, I=0.12) and

for 1-acetyl-5FU (7) in the same buffer with 0.11%

formaldehyde (3.6x10-2 M). The rate in the presence of

formaldehyde was determined for comparison with the rate in

plain buffer since formaldehyde was used as a preservative in

the diffusion cell experiments described in the following

section. Two other concentrations of formaldehyde (1.8x10-1 M

and 3.6x10-1 M) were studied to assess the catalytic role of

formaldehyde in the hydrolysis of compound 7.

The hydrolyses were followed by UV spectroscopy at

266 nm where the absorbance decrease accompanying hydrolysis

of the 1-acyl derivatives was maximized. Hydrolysis was

initiated by adding 60 to 75 gL of stock solutions of the

derivatives in acetonitrile to 3 mL of buffer prewarmed to

32 OC in a thermostated quartz cuvette to give final

concentrations of 1-2x10-4 M. Absorbances were recorded at

appropriate intervals and pseudo-first-order rate constants










were determined from the expression:

ln(At-A-) = In(Ao-A-)-kt (4)

where At is the absorbance at some time=t, A- is the

absorbance at t=-, Ao is the absorbance at t=0, k is the

pseudo-first-order rate constant, and t is the time. The

hydrolyses were sufficiently fast to allow experimental

determination of A.. The slopes, -k, of linear plots of

In(At-Ao) versus time were determined by linear regression.

The half-lives (t1/2) were calculated from

tl/2 = 0.693/k (5)

Each hydrolysis reaction was run in triplicate and was

followed for a minimum of three half-lives. The correlation

coefficients were 20.999.


Skin Penetration Studies


Diffusion cell experiments were performed to measure the

transdermal delivery of 5FU and the 5FU prodrugs. Franz-type

diffusion cells from Crown Glass in Somerville, NJ with 4.9

cm2 donor surface area and 20 mL receptor phase volume were

used for this purpose. The full-thickness skins were

obtained from female hairless mice (SKH-hr-1) from Temple

University Skin and Cancer Hospital.

The mice were killed by cervical dislocation, their

skins were removed immediately by blunt dissection, and

dorsal sections were mounted in the diffusion cells. The

dermal sides of the skins were placed in contact with










receptor phase which contained 0.05 M phosphate buffer

(pH-7.1, I=0.12) with 0.11% formaldehyde as a preservative.

The effectiveness of formaldehyde for this purpose has

recently been documented.89 The receptor phases were stirred

continuously and kept at constant temperature (32 OC) by a

circulating water bath. A preapplication period of 48 hours

was established to uniformly condition the skins and to

remove water-soluble UV-absorbing materials. The receptor

phases were changed three times during this period, and

control experiments from earlier studies have shown that this

procedure effectively removes those materials.97 The

epidermal sides of the skins were exposed to the air and were

left untreated during this period.

After the preapplication period, 0.5 mL aliquots from

suspensions of the prodrugs in IPM were applied to the

epidermal sides of the skins. The IPM suspensions were

stirred at 221 OC for 48 hours prior to application to

ensure that saturation was attained. Total concentrations of

the IPM suspensions ranged from 0.6 M to 1.0 M with enough

excess solid present to maintain saturation for the duration

of the application period (see below). Each drug-vehicle

combination was run in triplicate.

Samples were taken from the receptor phases at 4, 8, 12,

21, 24, 27, 30, 33, 36, 45, and 48 hours after donor phase

application. The receptor phases were changed following

removal of each sample so that "sink" conditions were

maintained. Samples were analyzed for total 5FU species that









had diffused by UV spectroscopy (E=7.13x103 at 266 nm) after

allowing at least 24 hours for complete prodrug hydrolysis.

Cumulative amounts of total 5FU species that diffused (Imol)

were plotted against time (h), and the slopes of the linear,

"steady-state" regions were calculated using linear

regression. The slopes, when divided by 4.9 (the area of the

donor surface in cm2), gave the "steady-state" fluxes

(pmol/cm2/h). Because of the rapid chemical hydrolysis of

the 1-acyl derivatives, no attempt was made to analyze the

receptor phases for prodrug content.

Donor phases were changed every twelve hours and were

set aside for 1H NMR analysis. Stability of the prodrugs in

IPM was determined from the chemical shift of C6-H. In

dimethylsulfoxide-d6, the C6-H signal for 5FU appears at

5=7.73 ppm. For each of the 1-acyl derivatives, the same

signal in dimethylsulfoxide-d6 appears at 8>8.20 ppm. Since

this area of the spectrum is free from interference by IPM

absorbances, the two signals can be identified and quantified

if necessary.

Following removal of the donor phases after the 48-hour

application period, the epidermal sides of the skins were

washed three times with 5 mL portions of methanol to remove

all remnants of prodrug and vehicle from the skin surfaces.

This was accomplished quickly (<3 min) to minimize contact

time between the skins and methanol. The receptor phases

were changed again, and the dermal sides were kept in contact

with the fresh buffer for 23 hours while the epidermal sides










were again left exposed to the air. After this "leaching"

period, another sample was taken from each cell to measure

the skin accumulation of total 5FU species.

Second applications to the epidermal sides of the skins

were made after the "leaching" period with a standard drug-

vehicle suspension Theophylline in propylene glycol

(0.4 M) was applied to assess the damage to the skins from

application of the initial drug-vehicle combinations.

Samples were taken at 1, 2, 3, 5, 7, 9, and 11 or 12 hours

after application. The samples were analyzed for

theophylline by UV spectroscopy (E=1.02x104 at 271 nm) and

second application fluxes were determined as described above.


Results and Discussion



Synthesis and Structure Determination


The known l-acyl-5FU derivatives have melting points27

and 1H NMR spectra25 in agreement with those reported in the

literature. The structures of novel compounds were assigned

by comparison of their 1H NMR spectra with those of the known

homologs. Elemental microanalyses were obtained for the

novel compounds and were within acceptable limits (0.4%).

The differences in spectral properties between N1- and

N3-acylated derivatives were discussed in detail in

Chapter 2. Ultraviolet (UV) spectra were not obtained for

the 1-acyl derivatives under basic conditions because of










their instability. However, differences in 1H NMR spectra

are very clear. In chloroform-d, the C6-H signal for

l-acetyl-5FU is a sharp doublet at 8=8.23, while the same

signal for 3-acetyl-5FU is a broad singlet at 8=7.23, a

difference of 1.00 ppm.

Results from the single-crystal X-ray analysis show that

the N1-assignment for l-acetyl-5FU is correct. The unit cell

was found to contain two independent molecules (Figure 3-1

and Figure 3-2). Both conformations show that the carbonyl

group from the 1-acyl group is positioned cis to the C6-H.

The circulating K electrons of carbonyl bonds are known to

deshield ortho protons in structurally similar compounds such

as acetophenone.105 Thus, the X-ray results support the

explanation that an anisotropic effect is responsible for the

downfield shift of C6-H in 1H NMR spectra of Nl-acylated

derivatives.


Solubility


Lipid (SIPM) and aqueous (SAQ) solubilities for the

1-acyl derivatives are presented in Table 3-2 along with

their melting points. In general, melting points decrease

and lipid solubilities increase with increasing chain length.

An exception to both of these trends is seen for 1-butyryl-

5FU (9). The basis for this variance from the observed

trends is not clear.























































Figure 3-1. X-ray structure of l-acetyl-5FU unprimedd).
Source: Hoffman-La Roche, Nutley, NJ.























































Figure 3-2. X-ray structure of l-acetyl-5FU (primed).
Source: Hoffmann-La Roche, Nutley, NJ.













Table 3-2. Melting points (MP), lipid solubilities (SipM), and
aqueous solubilities (SAQ) for 1-acyl derivatives.


SIPMa
(mM)


Compound


5FU 280-2 0.049 96

7 129-30 22 119
8 130-1 36 48
9 145-6 17 6.5
10 120-1 39 3.5
11 101-2 112 3.0
12 83-4 111 0.15


aStandard deviations from the mean
solubilities.


were within 5% for IPM


bSolubility determined by direct method.
CStandard deviations from the mean were within 5% for
experimental values used to calculate partition coefficients
and estimated aqueous solubilities.













Table 3-3. Partition coefficients (PC) and hydrophobicity
parameters (t) for 1-acyl derivatives.


Compound


log (PC)


7 0.19 -0.73
8 0.76 -0.12 0.61
9 2.7 0.43 0.55
10 11 1.05 0.62
11 38 1.58 0.53
12 759 2.88 0.65


aExperimental partition coefficient (CIPM/CAQ).
bAlog(PC) for compound and preceding compound.










Aqueous solubilities were determined by the

partitioning-based method for which a theoretical discussion

was presented in Chapter 2. For the 1-acyl series, aqueous

solubility decreases with increasing chain length, and only

l-acetyl-5FU (7) is more water soluble than 5FU.

In Table 3-3, partition coefficients (PC) and

hydrophobicity parameters (t) are listed for the 1-acyl

derivatives. The average E value is 0.60, the same value

that was calculated for the 1-alkyloxycarbonyl series.

Individual values for the 1-acyl derivatives are all within

0.07 of the average. These results further support use of

the partitioning-based method for determining relative

aqueous solubility in a homologous series.


Hydrolysis Kinetics


Hydrolysis of the 1-acyl derivatives was studied in

0.05 M phosphate buffer (pH=7.1, 1=0.12) at 32 OC. Pseudo-

first-order rate constants (k) and half-lives (tl/2) are

presented in Table 3-4. Half-lives are relatively consistent

throughout the series ranging from 3.1 minutes for

l-propionyl-5FU (8) to 4.8 minutes for l-acetyl-5FU (7).

Hydrolysis of compound 7 was also studied in the same

buffer with increasing concentrations of formaldehyde added

to assess the catalytic role of formaldehyde hydrate. These

results are shown in Table 3.5. It is clear that the

hydrolysis rate for compound 7 is faster in the presence of













Table 3-4. Pseudo-first-order rate constants (k) and half-
lives (tl/2) for hydrolysis of 1-acyl derivatives in
0.05 M phosphate buffer (pH=7.1, 1-0.12) at 32 OC.


k(SD)a
(min-1)


Compound


tl/2
(min)


7 0.143(0.002) 4.8
8 0.222(0.004) 3.1
9 0.163(0.003) 4.3
10 0.169(0.006) 4.1
11 0.173(0.003) 4.0
12 0.183(0.003) 3.8


aMean standard deviation for n=3 values.













Table 3-5. Pseudo-first-order rate constants (k) and half-
lives (t1/2) for hydrolysis of l-acetyl-5FU in 0.05 M
phosphate buffer (pH=7.1, I=0.12) with and without
formaldehyde at 32 OC.


Compound


Formaldehyde
(M)


k(SD)
(min-1)


tl/2
(min)


7 0 0.143(0.002)a 4.8
7 3.6x10-2 0.161(0.008)b 4.3
7 1.8x10-1 0.219(0.004)a 3.2
7 3.6x10-1 0.279(0.006)a 2.5


aMean standard deviation for n=3 values.
bMean standard deviation for n=5 values.












0.3

y= 0.14644 + 0.37545x RA2 = 0.996











0.2








a Cpd7




0.1

0.0 0.1 0.2 0.3 0.4

Formaldehyde (M)




Figure 3-3. Plot of pseudo-first-order rate constant (k)
versus formaldehyde concentration (M) for hydrolysis of
l-acetyl-5FU in 0.05 M phosphate buffer
(pH=7.1, 1=0.12) at 32 OC.










formaldehyde and the rate is concentration dependent. In

Figure 3-3, the pseudo-first-order rate constants are plotted

against formaldehyde concentration. The slope of the linear

plot gives the catalytic rate constant (kcat=0.375) for

formaldehyde catalysis of compound 7. Hydrolysis rates for

the 1-acyl derivatives have previously been shown to be

independent of pH at acidic pH values.26 Therefore,

formaldehyde catalysis is likely due to general base rather

than general acid catalysis.


Skin Penetration


Skin penetration data from the diffusion cells are

plotted as cumulative amount of total 5FU species that

diffused (gmol) versus time (h). In Figure 3-4, results for

l-acetyl-5FU (7), l-propionyl-5FU (8), and l-butyryl-5FU (9)

are compared to 5FU itself. In Figure 3-5, results for

l-valeryl-5FU (10), l-hexanoyl-5FU (11), and l-octanoyl-5FU

(12) are compared to 5FU. Error bars correspond to the

standard deviation from the mean for n=3 values.

Fluxes (J), lag times (tL), and skin accumulation (SA)

values for each compound are reported in Table 3-6. Lag time

refers to the intersection of the linear, or "steady-state,"

region of each graph with the time (x) axis, and it is the

time required for establishing a uniform concentration

gradient within the skin.65 The percentages of total 5FU











1200

a Cpd7

1000- Cpd

SCpd 9

O 5FU

0 600-





S 400-
E *


200 -

a iI!!

o gI, I o a -
0 10 20 30 40

Time (h)




Figure 3-4. Plots of cumulative amount of total 5FU species
that diffused (gmol) versus time (h) for
compounds 7, 8, 9, and 5FU.











200




1 Cpd10
Cpd11

E Cpd 12
S 5FU
- 0
E 100



E






0
S' I I



I r i 1 I


0 10 20 30 40
Time (h)



Figure 3-5. Plots of cumulative amount of total 5FU species
that diffused (.mol) versus time (h) for
compounds 10, 11, 12, and 5FU.













Table 3-6. Fluxes (J), lag times (tL), and skin accumulation
(SA) values for 1-acyl derivatives.


Compound


J(SD) a
(imol/cm2/h)


SA(SD)a
(pmol)


5FU 0.24(0.09) 13 3.7(0.9)

7 9.3(0.3) 10 68(10)
8 4.3(0.1) 13 69(10)
9 1.3(0.2) 12 8.2(2.7)
10 1.0(0.1) 9.0 16(4)
11 1.1(0.0) 5.6 11(3)
12 0.60(0.01) 6.5 12(3)


aMean standard deviation for n=3 values.













Table 3-7. Second application fluxes (J) and lag times (tL)
for 1-acyl derivatives.


Ja (SD)b
(Ilmol/cm2/h)


Compound


tL(h)


5FU 1.2(0.2) 1.2

7 1.6(0.0) 0.6
8 1.2(0.2) 0.1
9 1.0(0.0) 0.6
10 0.80(0.03) 0.8
11 0.47(0.02) 0.2
12 0.72(0.11) 0.4


aFlux of 0.4 M theopylline from propylene glycol.
bMean standard deviation for n=3 values.










present as prodrug are presumed to be zero since the half-

lives of these prodrugs under diffusion cell conditions are

only three to five minutes.

The improvement in skin penetration of 5FU from the

1-acyl derivatives is substantial. The best compound,

l-acetyl-5FU (7), shows an increase in flux of nearly 40

times when compared to 5FU. As chain length increases, the

fluxes decrease, but even l-octanoyl-5FU (12) improves

transdermal delivery of 5FU by two and one-half times.

Skin accumulation values are also much higher for the

1-acyl derivatives than for 5FU. They show a decrease with

increasing chain length, but the value for l-butyryl-5FU (9)

is lower than expected. As noted earlier, the lipid

solubility of compound 9 was also less than expected when

compared to the compounds around it in the 1-acyl series.

Therefore, the low skin accumulation value for compound 9 may

be due to its low affinity for the lipid regions of the skin.

Overall, skin accumulation is higher for this series than for

the 1-alkyloxycarbonyl series while skin penetration data for

the two series are similar. Rapid hydrolysis of the more

lipid-soluble 1-acyl derivatives to highly polar 5FU as they

partition into the skin may indicate that only 5FU is

diffusing through the remaining lipid regions of the skin.

This may effectively "lock in" large amounts of 5FU leading

to high skin accumulation values.

Second application fluxes and lag times are reported in

Table 3-7. Skin penetration by theophylline from propylene










glycol, the standard drug-vehicle combination, after

treatment with the 1-acyl derivatives is similar to that

following treatment with 5FU. In fact, the longer-chain

derivatives in this series actually appear to have a

protective effect on the skin since they cause less damage

than 5FU itself.

The stability of the prodrugs in the IPM formulations

was assessed by 1H NMR analysis of the donor phases. After a

minimum of five days from the time the suspensions were

prepared until their 1H NMR spectra were recorded, including

at least twelve hours during which the formulations were in

contact with the skins, the 1-acyl derivatives were found to

be intact with no evidence of 5FU formation.


Summary


The 1-acyl derivatives of 5FU exhibited decreased

melting points and increased lipid solubilities when compared

to 5FU. Aqueous solubility reached a maximum for 1-acetyl-

5FU (7) and decreased from there with increasing chain

length. Skin penetration was also maximized for compound 7

while skin accumulation values were highest and essentially

the same for compound 7 and l-propionyl-5FU (8). As was also

the case for the 1-alkyloxycarbonyl series, this demonstrates

that both lipid and aqueous solubilities are important for

predicting transdermal and dermal delivery of 5FU prodrugs.

The 1-acyl derivatives may be better candidates for dermal






81



delivery purposes than the 1-alkyloxycarbonyl derivatives

because of their rapid hydrolysis and higher skin

accumulation values. The partitioning-based method for

determining aqueous solubilities of chemically unstable

prodrugs was used to estimate the aqueous solubilities of the

1-acyl derivatives since this could not be accomplished by

other methods.















CHAPTER 4
1,3-BIS-ACYL DERIVATIVES


Introduction


The 1,3-bis-acyl derivatives of 5FU are double prodrugs.

The acyl groups at the N1-position are rapidly hydrolyzed

leaving the 3-acyl derivatives. The 3-acyl derivatives will

be discussed in Chapter 5. The rate of N1-deacylation for

the 1,3-bis-acyl derivatives is even faster than hydrolysis

of the 1-acyl derivatives. This can best be explained by

comparing pKa values for the 3-acyl derivatives and 5FU. The

3-acyl derivatives have reported pKa values of 7.1 to 7.226

while the pKa for the first ionization of 5FU is 8.0. Thus,

the leaving group potential of the N3-acyl anions is greater

than the 5FU anion, and the rates of N1-deacylation are

faster for the 1,3-bis-acyl derivatives.

The 1,3-bis-acyl derivatives were chosen as prodrug

candidates in order to study a series in which both hydrogen-

bonding groups were masked. It was anticipated that these

compounds would be more lipid soluble and less water soluble

than the other series and would be useful for comparison with

the other series.











Table 4-1. Structures of 1,3-bis-acyl derivatives.


R2


A


RO/


Compound RI,R2


1,3-bis-acetyl-5FU (13)
1,3-bis-propionyl-5FU (14)
1,3-bis-butyryl-5FU (15)
1,3-bis-valeryl-5FU (16)


-CH3
-CH2CH3
-(CH2)2CH3
-(CH2)3CH3










Four straight-chain 1,3-bis-acyl derivatives were

selected for study. The derivatives and their structures are

shown in Table 4-1.


Materials and Methods



Synthesis


Melting points (mp) were determined with a Thomas-Hoover

capillary melting point apparatus and are uncorrected.

Elemental microanalyses were obtained for all novel compounds

through Atlantic Microlab, Incorporated in Norcross, Georgia.

Proton nuclear magnetic resonance (1H NMR) spectra were

obtained at 90 MHz on a Varian EM-390 spectrometer. Chemical

shifts (6) are reported in parts per million (ppm) from the

internal standard, tetramethylsilane (TMS). Coupling

constants (J) are expressed in cycles per second (Hz).

Infrared (IR) spectra were recorded with a Perkin-Elmer 1420

spectrophotometer and absorbances are reported in cm-1.

Ultraviolet (UV) spectra were obtained with a Cary 210 or

Shimadzu UV-265 spectrophotometer. Maximum absorbances are

reported in nm along with the molar absorptivities (e) in

L/mol. Single-crystal X-ray analysis was obtained for

1,3-bis-acetyl-5FU through Hoffmann-La Roche in Nutley, NJ.

1.3-bis-Acyl-5-fluorouracil (general procedure)

To 1.31 g (0.01 mol) of 5FU suspended in acetonitrile

(20 mL) was added 1.01 g of triethylamine (0.01 mol) in









acetonitrile (5 mL). The mixture was stirred continuously at

0 *C while 0.011 mol of the appropriate acid chloride in

acetonitrile (5 mL) was added dropwise over 5-10 minutes.

The above sequence was repeated until 0.03 mol of

triethylamine and 0.033 mol of acid chloride were added, then

the mixture was stirred for an additional 30 minutes at 0 C.

Alternate addition of base and acylating agent was found to

increase yield and decrease formation of colored side

products that were difficult to remove. The mixture was

filtered, and the residue was washed with acetonitrile

(25 mL). The combined acetonitrile solutions were evaporated

under reduced pressure, and the solid residue was

crystallized from an appropriate solvent or solvent

combination.

1.3-bis-Acetyl-5-fluorouracil (131

Crystallization from ether gave 1.69 g of 13 (79%): mp

112-3 OC (lit.25 mp 111-3 oC); IR (KBr) 1680, 1695, 1740,

1750, and 1795 cm-1 (C=0); 1H NMR (CDC13) 8 2.58 (s, 3H,

3-CH3), 2.72 (s, 3H, 1-Cl13), and 8.23 (d, J=7 Hz, 1H, C6-1j);

UVmax (CH3CN) 262 nm (E=9.75x103).

1.3-bis-Propionyl-5-fluorouracil (14)

Crystallization from ether gave 1.82 g of 14 (75%): mp

100-1 OC; IR (KBr) 1690, 1725, and 1795 cm-1 (C=0); 1H NMR

(CDC13) 8 1.25 (t, J=7 Hz, 3H, 3-CH3), 1.28 (t, J=7 Hz, 3H,

1-CH3), 2.85 (q, J=7 Hz, 2H, 3-COCf2), 3.11 (q, J=7 Hz, 2H,










1-COCfl2), and 8.25 (d, J=7 Hz, 1H, C6-j); UVmax (CH3CN) 262 nm

(e=9.65x103).

Anal. Calc. for C10H11FN204: C, 49.59; H, 4.58; N, 11.57.

Found: C, 49.44; H, 4.59; N, 11.53.

1.3-bis-Butyryl-5-fluorouracil (151

Extraction of the residue with hot low-boiling petroleum

ether gave a cloudy solution which cleared upon cooling to

0 oC and produced a yellow resinous sediment. The mixture

was allowed to warm to room temperature, and the clear

solution was decanted from the sediment. The solution was

cooled again to 0 oC and crystallization gave 1.70 g of 15

(63%): mp 48-9 OC (lit.26 mp 47.5-48.5 OC); IR (KBr) 1685,

1710, 1735, and 1795 cm-1 (C=O); 1H NMR (CDCl3) 8 1.00 (t,

J=7 Hz, 3H, 3-Cl3), 1.03 (t, J=7 Hz, 3H, l-Cl3), 1.6-1.9 (m,

4H, 3-COCH2CH2 and 1-COCH2CIi2), 2.80 (t, J=7 Hz, 2H, 3-COCfl2),

3.07 (t, J=7 Hz, 2H, 1-COCH2), and 8.23 (d, J=6 Hz, 1H, C6-a);

UV max (CH3CN) 262 nm (E=1.051x104).

1.3-bis-Valeryl-5-fluorouracil (16)

Extraction of the residue with hot low-boiling petroleum

ether gave a cloudy solution which cleared upon cooling to

0 C and produced a brown resinous sediment. The mixture was

allowed to warm to room temperature, and the clear solution

was decanted from the sediment. The solution was cooled

again to 0 oC and crystallization gave 2.24 g of 16 (75%):

mp 47-8 *C; IR (KBr) 1685, 1710, 1735, and 1795 cm-1 (C=O);

1H NMR (CDC13) 8 0.95 (t, J=7 Hz, 6H, 3-CH3 and 1-Ca3),










1.3-1.8 (m, 8H, 3-COCH2CHi2CI2 and 1-COCH2CZCH2), 2.83 (t,

J=7 Hz, 2H, 3-COCj2), 3.08 (t, J=7 Hz, 2H, 1-COCH2), and 8.22

(d, J=7 Hz, 1H, C6-H); UVmax (CH3CN) 262 nm (e=1.071xl04).

Anal. Calc. for C14H19FN204: C, 56.37; H, 6.42; N, 9.39.

Found: C, 56.29; H, 6.48; N, 9.33.


Lipid solubility


Lipid solubilities were determined using isopropyl

myristate (IPM), a commercial vehicle used in cosmetics and

topical medicinals,94 as the lipid solvent. The use of IPM as

a model lipophilic vehicle in skin penetration studies is

well established.77,95

Three suspensions of each derivative were stirred at

221 OC for 48 hours. The suspensions were filtered through

0.45 pm nylon filters, and the saturated solutions were

diluted in acetonitrile and analyzed by UV spectroscopy.

Solubilities were calculated using Beer's Law:

A = E-C-d (1)

where A is the absorbance, E is the molar absorptivity, C is

the concentration in mol/L,and d is the path length of the

cuvette in cm. Molar absorptivities were predetermined in

triplicate in acetonitrile at 262 nm.


Aqueous Solubility


Because of the chemical instability of the 1,3-bis-acyl

derivatives, direct measurement of aqueous solubilities for




Full Text
154
16. Duschinsky, R.; Pleven, E.; Heidelberger, C. J. Am.
Chem. S no. 1957, 22, 4559.
17. Heidelberger, C.: Chaudhuri, N. K.; Danneberg, P.;
Mooren, D.; Griesbach, L.; Duschinsky, R.; Schnitzer,
R. J.; Pleven, E.; Scheiner, J. Nature 1957, 179. 663.
18 .
Stella, V.
, J.
; Himmelstein, K. J.
J,
Med,
. Chem. 1980r
22, 1275.
19.
Bansal, P.
, C.
; Pitman, I. H.; Tam,
. J
. N.
S.; Mertes, M
Kaminski,
J.
J. J. Pharm. 1981,
22,
850 .
20.
Ozaki, S.;
Watanabe, Y.; Hoshiko,
T.;
: Nagase, T.;
Ogasawara, T.; Furukawa, H.; Uemura, A.; Ishikawa, K.;
Mori, H.; Hoshi, A.; ligo, M.; Tokuzen, R. Chem. Pharm.
Bull. 1986, 21, 150.
21. Mollgaard, B.; Hoelgaard, A.; Bundgaard, H. Tnt. J.
Pharm. 1982, 12., 153.
22. Buur, A.; Bundgaard, H.; Falch, E. Int. J. Pharm. 1985,
21, 43.
23. Ozaki, S.; Watanabe, Y.; Hoshiko, T.; Mizuno, H.;
Ishikawa, K.; Mori, H. Chem. Pharm. Bull. 1984, 22, 733
24. Ahmad, S.; Ozaki, S.; Nagase, T.; ligo, M.; Tokuzen, R.
Hoshi, A. Chem. Pharm. Bull. 1987, 22, 4137.
25. Kametani, T.; Kigasawa, K.; Hiiragi, M.; Wakisaka, K.;
Haga, S.; Nagamatsu, Y.; Sugi, H.; Fukawa, K.; Irino,
0.; Yamamoto, T.; Nishimura, N.; Taguchi, A. J. Med.
Chem. 1980, 22, 1324.
26
Buur,
A.;
Bundgaard,
H. Int. J. Pharm. 1984, 21,
349.
27 .
Masao,
T.
Chem. Lett.
1975, 129.
28.
Buur,
A.;
Bundgaard,
H. Arch. Pharm.
Chem.. Sci.
Ed.
1986,
ii,
99.
29.
Buur,
A.;
Bundgaard,
H. Arch. Pharm.
Chem.. Sel.
Ed.
1984,
12,
37.
30.
Buur,
A. ;
Bundgaard,
H. J. Pharm. Scj
^ 1986, 22,
522.
31.
Nagase
i, T
.; Seike, K.
; Shiraishi, K.;
Yamada, Y.
; Ozak
S. Chem.
Lett. 1988,
1381.
32.
Nagase
!, T
.; Shiraishi
, K.; Yamada, Y.
; Ozaki, S.
Heterocvcles 1988, 22, 1155.


29
0.05 M acetate buffer (pH=4.0) at 221 C for 60 minutes.
The suspensions were filtered through 0.45 |!m nylon filters,
and the saturated solutions were diluted in acetonitrile and
analyzed by UV spectroscopy. Solubilities were calculated
using Beer's Law as previously described.
Partition Coefficients
The partitioning-based method for determining aqueous
solubility utilized the saturated IPM solutions from the
lipid solubility study. For most compounds, equal volumes
(1 mL) of saturated IPM solution and 0.05 M acetate buffer
(pH=4.0) were used. The use of equal or near-equal phase
volumes is known to facilitate rapid equilibrium.96 The two
phases were mixed thoroughly for ten seconds and allowed to
separate for 60 seconds. A preliminary study showed that
there was virtually no difference in partition coefficient
(PC) values when partitioning was carried out for 10, 20, or
30 seconds (see Chapter 3). The IPM layers were diluted in
acetonitrile and analyzed by UV spectroscopy. The IPM-buffer
partition coefficients were calculated as follows:
PC = Aafter/(Abefore-Aafter) -Vaq/Vxpm (2)
where Aafter is the absorbance from the IPM layer after
partitioning, Abefore is the absorbance from the IPM layer
before partitioning, Vrq is the volume of the aqueous phase,
and Vipm is the volume of the IPM phase. Estimated aqueous


133
time (min)
Figure 5-6. Plots of ln(C) versus time (min) for hydrolysis
of 3-acetyl-5FU in 0.05 M phosphate buffer (pH=7.1, 1=0.12)
at 32 C using actual concentration (Ct) and concentration
corrected for secondary degradation (Ccorr)


cause retention of highly polar 5FU in the more lipophilic
regions of the epidermis.
Skin damage was only slightly increased with some
prodrugs, all derivatives were stable in formulation, and
overall, l-acetyl-5FD was the best candidate for improving
dermal or transdermal delivery of 5FU.


LIST OF FIGURES
Figure 2-1. Plots of ln(C) versus time (min) for
hydrolysis of l-methyloxycarbonyl-5FU in 0.05 M
phosphate buffer (pH=7.1, 1=0.12) with and without
formaldehyde at 32 C 43
Figure 2-2. Plots of cumulative amount of total 5FU
species that diffused 4lmol) versus time (h) for
compounds 1, 2, 3, and 5FU 46
Figure 2-3. Plots of cumulative amount of total 5FU
species that diffused Oimol) versus time (h) for
compounds 4, 5, 6, and 5FU 47
Figure 3-1. X-ray structure of l-acetyl-5FU (unprimed). ... 66
Figure 3-2. X-ray structure of l-acetyl-5FU (primed) 67
Figure 3-3. Plot of pseudo-first-order rate constant (k)
versus formaldehyde concentration (M) for hydrolysis
of l-acetyl-5FU in 0.05 M phosphate buffer (pH=7.1,
1=0.12) at 32 C 73
Figure 3-4. Plots of cumulative amount of total 5FU
species that diffused (|imol) versus time (h) for
compounds 7, 8, 9, and 5FU 75
Figure 3-5. Plots of cumulative amount of total 5FU
species that diffused (|imol) versus time (h) for
compounds 10, 11, 12, and 5FU 76
Figure 4-1. X-ray structure of 1,3-bis-acetyl-5FU 95
Figure 4-2. Plots of ln(At-A=) versus time (min) for
hydrolysis of 1,3-bis-acetyl-5FU in 0.05 M phosphate
buffer (pH=7.1, 1=0.12) with and without formaldehyde
at 32 C 99
Figure 4-3. Plots of cumulative amount of total 5FU
species that diffused (|imol) versus time (h) for
compounds 13, 14, and 5FU 102
Figure 4-4. Plots of cumulative amount of total 5FU
species that diffused (|imol) versus time (h) for
compounds 15, 16, and 5FU 103
viii


CHAPTER 3
1-ACYL DERIVATIVES
Introduction
The 1-acyl derivatives of 5-fluorouracil (5FU) are
chemically unstable in aqueous solutions at all pH values.
In fact, their lability has been cited as a limitation to
their usefulness as drugs or prodrugs.26'42 If properly
formulated in an aprotic vehicle, however, this series of 5FU
derivatives may have potential for use as prodrugs for dermal
delivery.
Six straight-chain 1-acyl derivatives were selected for
study. The derivatives and their structures are shown in
Table 3-1.
Materials and Methods
Synthesis
Melting points (mp) were determined with a Thomas-Hoover
capillary melting point apparatus and are uncorrected.
Elemental microanalyses were obtained for all novel compounds
through Atlantic Microlab, Incorporated in Norcross, Georgia.
Proton nuclear magnetic resonance (iH NMR) spectra were
obtained at 90 MHz on a Varian EM-390 spectrometer. Chemical
53


144
alone and hydrolysis studies showed that products other than
5F0 were formed during hydrolysis of the 3-acyl derivatives
in formaldehyde buffer, it was necessary to compare the molar
absorptivity (£) at 266 nm for 5FU with e values at 266 nm for
hydrolyzed 3-acyl and 1,3-bis-acyl derivatives in solutions
containing receptor phase. Solutions containing known
concentrations of each of the 3-acyl and 1,3-bis-acyl
derivatives were incubated for 24 hours at 32 C at which
time HPLC analysis showed that only 5FU and the two
formaldehyde reaction products were present. Subsequent
analysis by UV spectroscopy showed that the multiple
component solutions had e values that were within 5% of the
£ value for 5FU alone at 266 nm. This verified that the UV
assay for determining cumulative amount of total 5FU species
that diffused and skin accumulation of total 5FU species is
valid.
Fluxes (J), lag times (ti.) and skin accumulation (SA)
values for each compound are reported in Table 5-6. Lag time
refers to the intersection of the linear, or "steady-state,"
region of each graph with the time (x) axis, and it is the
time required for establishing a uniform concentration
gradient within the skin.65 The percentages of total 5FU
present as intact prodrugs in the receptor phases are also
reported in Table 5-6. These values were calculated from
samples taken during the "steady-state" phase and from an
earlier sample (11 h) in a separate experiment using HPLC
analysis (n=l).


59
Partition Coefficipnts
The partitioning-based method for determining aqueous
solubility utilized the saturated IPM solutions from the
lipid solubility study. For most compounds, equal volumes
(1 mL) of saturated IPM solution and 0.05 M acetate buffer
(pH=4.0) were used. The use of equal or near-equal phase
volumes is known to facilitate rapid equilibrium.96 The two
phases were mixed thoroughly for ten seconds and allowed to
separate for 60 seconds. A preliminary study with 1-acetyl-
5FD (7) showed that there was virtually no difference in
partition coefficient (PC) values when partitioning was
carried out for 10, 20, or 30 seconds (PC=0.183+0.006, a
standard deviation of only 3%). The IPM layers were diluted
in acetonitrile and analyzed by UV spectroscopy. The IPM-
buffer partition coefficients were calculated as follows:
PC = Aafter/ (Abefore^after) 'Vaq/^IPM (2)
where Aafter is the absorbance from the IPM layer after
partitioning, Abefore is the absorbance from the IPM layer
before partitioning, Vaq is the volume of the aqueous phase,
and Vipm is the volume of the IPM phase. Estimated aqueous
solubilities (Saq) were calculated from the IPM solubility
(Sipm) and the partition coefficient:
Saq = Sjpm/PC (3)
Partitioning was carried out in triplicate for a fixed volume
ratio for each derivative. For those compounds with large


52
when compared to 5FU. Aqueous solubility reached a maximum
for l-ethyloxycarbonyl-5FU (2) and decreased from there with
increasing chain length. Skin penetration and skin
accumulation were also maximized for compound 2 suggesting
that both lipid and aqueous solubilities are important for
predicting transdermal and dermal delivery of these 5FU
prodrugs. The presence of high percentages of prodrugs in
the receptor phases, indicating insufficient release of the
parent drugs in the hairless mouse skin model, may limit the
potential of this series of prodrugs at least for dermal
delivery purposes. Finally, the partitioning-based method
for determining aqueous solubility appears to be a useful
method particularly for determining relative solubilities in
an homologous series.


34
8=7.73 ppm. For each of the 1-alkyloxycarbonyl derivatives,
the same signal in dimethylsulfoxide-d6 appears at 8>8.10 ppm.
Since this area of the spectrum is free from interference by
IPM absorbances, the two signals can be identified and
quantified if necessary.
Following removal of the donor phases after the 48-hour
application period, the epidermal sides of the skins were
washed three times with 5 mL portions of methanol to remove
all remnants of prodrug and vehicle from the skin surfaces.
This was accomplished quickly (<3 min) to minimize contact
time between the skins and methanol. The receptor phases
were changed again, and the dermal sides were kept in contact
with the fresh buffer for 23 hours while the epidermal sides
were again left exposed to the air. After this "leaching"
period, another sample was taken from each cell to measure
the skin accumulation of total 5FU species.
Second applications to the epidermal sides of the skins
were made after the "leaching" period with a standard drug-
vehicle suspension Theophylline in propylene glycol
(0.4 M) was applied to assess the damage to the skins from
application of the initial drug-vehicle combinations.
Samples were taken at 1, 2, 3, 5, 7, 9, and 11 or 12 hours
after application. The samples were analyzed for
theophylline by UV spectroscopy (E=1.02xl04 at 271 nm) and
second application fluxes were determined as described above.


5
another kinase. At this point FdUTP can be incorporated into
deoxyribonucleic acid (DNA), but the contribution to 5FU
cytotoxicity by this mechanism is unclear. DNA protective
enzymes, dUTPase and uracil-DNA glycosylase, which are
responsible for keeping uracil residues out of DNA, target
5FU residues as well.50
A diphosphohydrolase converts FdUTP to 5-fluoro-2'-
deoxyuridine-5'-monophosphate (FdUMP). FdUMP can also be
formed from 5FU by thymidine phosphorylase and thymidine
kinase, although this may be quantitatively less important
than other activation steps.1 FdUMP binds covalently to
thymidylate synthase and its cofactor, 5,10-methylene
tetrahydrofolate, thereby preventing synthesis of thymidine
s' -monophosphate (dTMP) and subsequently, thymidine-5'-
triphosphate (dTTP), an essential component in DNA synthesis.
FdUMP has a greater affinity for thymidylate synthase than
the normal substrate, 21-deoxyuridine-5'-monophosphate
(dUMP).1
Either inhibition of thymidylate synthase by FdUMP or
incorporation of FUTP into RNA can lead to cell death. The
mechanism that is most important depends on the cell line
being studied.51
Fluorouracil (5FU) may also affect glycoprotein and
glycolipid metabolism, since it is known that FUDP sugars can
be formed.50 The possibility that membrane effects from
altered glycoprotein synthesis could lead to cytotoxicity has
been noted, but has not been well studied.1


42
well below its already low concentration at saturation.
Therefore, the solute-solute interactions that lead to higher
partition coefficients would also be reduced.
The hydrophobicity parameters100 (it) in Table 2-3 were
calculated from the relationship:103
log (PC) n = log (PC) o+Itn (6)
where n is the number of methylene units in a homologous
series using both log(PC) and log(SR) values. Both
calculations yield an average 7C value equal to 0.60. Values
for 71 from the literature include 0.54 for silicone oil-water
and 0.66 for hexane-water101 indicating that 0.60 is a
reasonable value for the IPM-buffer partitioning system.
Hydrolysis Kinetics
Hydrolysis of l-methyloxycarbonyl-5FU (1) to 5FU in
0.05 M phosphate buffer (pH=7.1, 1=0.12) with and without
0.11% formaldehyde (3.6xl0-2 M) was followed by HPLC at 32 C.
Disappearance of compound 1, indicated by ln(C), is plotted
versus time (min) in Figure 2-1. The linearity of the plots
suggests that hydrolysis of 1 follows first-order kinetics in
the presence and absence of formaldehyde. Pseudo-first-order
rate constants (k) and half-lives (ti/2) from the linear plots
are presented in Table 2-4.
Hydrolysis of compound 1 is clearly faster in the
presence of formaldehyde indicating that general base
catalysis by formaldehyde hydrate may be involved (see


48
Table 2-5. Fluxes
(SA) values
(J), lag times (tl), and skin accumulation
for 1-alkyloxycarbonyl derivatives.
Compound
J(+SD) a
(p.mol/cm2/h)
Prodrug
S-Sb(ll h)c
(%)
tL
(h)
SA(SD)a
Oimol)
5FU
0.24 (0.09)
-
13
3.7 (0.9)
1
2.6(0.6)
42(16)
14
8.3(0.1)
2
5.9(1.3)
90(75)
13
18(4)
3
2.3(0.2)
78(43)
11
5.0(1.4)
4
2.2(0.1)
73(32)
10
4.2(0.5)
5
1.5(0.1)
79(14)
6.2
11(0)
6
0.29(0.02)
-
8.7
3.2(0.5)
aMean standard deviation for n=3 values.
bPercent of total 5FU as intact prodrug during "steady-state"
phase in separate experiment (n=l).
cPercent of total 5F0 as intact prodrug from 11 h sample in
separate experiment (n1).


83
Table 4-1. Structures of 1,3-bis-acyl derivatives.
Compound Ri,R2
1.3-bis-acetyl-5FU (13) -CH3
1.3-bis-propionyl-5FO (14) -CH2CH3
1.3-bis-butyryl-5FU (15) -(CH2)2CH3
1.3-bis-valeryl-5FO (16) -(CH2)3CH3


118
dermal sides of the skins were placed in contact with
receptor phase which contained 0.05 M phosphate buffer
(pH=7.1, 1=0.12) with 0.11% formaldehyde as a preservative.
The effectiveness of formaldehyde for this purpose has
recently been documented.89 The receptor phases were stirred
continuously and kept at constant temperature (32 C) by a
circulating water bath. A preapplication period of 48 hours
was established to uniformly condition the skins and to
remove water-soluble UV-absorbing materials. The receptor
phases were changed three times during this period, and
control experiments from earlier studies have shown that this
procedure effectively removes those materials.97 The
epidermal sides of the skins were exposed to the air and were
left untreated during this period.
After the preapplication period, 0.5 mL aliquots from
suspensions of the prodrugs in IPM were applied to the
epidermal sides of the skins. The IPM suspensions were
stirred at 221 C for 48 hours prior to application to
ensure that saturation was attained. Total concentrations of
the IPM suspensions ranged from 0.4 M to 0.6 M with enough
excess solid present to maintain saturation for the duration
of the application period (see below). Each drug-vehicle
combination was run in triplicate.
Samples were taken from the receptor phases at 4, 8, 12,
21, 24, 27, 30, 33, 36, 45, and 48 hours after donor phase
application. The receptor phases were changed following
removal of each sample so that "sink" conditions were


124
Figure 5-2. Possible scheme for thermal intramolecular
rearrangement for 3-acetyl-5FU to l-acetyl-5FU.


128
Table 5-3. Solubility ratios (SR), partition coefficients
(PC) and hydrophobicity parameters (Jl) for
3-acyl derivatives.
log (PC) -
Compound
SRa
log(SR)
7lb
PCC
log (PC)
7ld
log(SR)
17
0.02 6
-1.59
0.041
-1.39
0.20
18
0.072
-1.14
0.45
0.11
-0.97
0.42
0.17
19
0.42
-0.38
0.76
0.98
-0.01
0.96
0.37
20
1.9
0.27
0.65
1.7
0.22
0.23
-0.05
aSolubility ratio calculated from Sipm/Saq.
bAlog(SR) for compound and previous compound.
Experimental partition coefficient (Cipm/Caq)
dAlog(PC) for compound and previous compound.


Cumulative Amount (nmol)
76
Time (h)
Figure 3-5. Plots of cumulative amount of total 5FU species
that diffused (nmol) versus time (h) for
compounds 10, 11, 12, and 5FU.


CHAPTER 2
1-ALKYLOXYCARBONYL DERIVATIVES
Introduction
Alkyloxycarbonyl derivatives of 5-fluorouracil (5FU)
have previously been studied as potential sources of 5FD in
vivo.25,29-30 when substitution is at the N3-position of 5FU,
the derivatives are chemically stable. Their hydrolyses in
human plasma and liver homogenate are also slow enough to
raise questions about their usefulness as prodrugs for the
oral or rectal delivery of 5FU,29 Thus, there is no doubt
that they are too stable to serve as prodrugs for dermal
delivery. On the other hand, substitution at the Ni-position
produces compounds that are relatively stable chemically,30
but which are sufficiently labile in the presence of enzymes30
to justify consideration as dermal prodrugs.
Six straight-chain 1-alkyloxycarbonyl derivatives were
selected for study. The derivatives and their structures are
shown in Table 2-1.
23


62
receptor phase which contained 0.05 M phosphate buffer
(pH=7.1, 1=0.12) with 0.11% formaldehyde as a preservative.
The effectiveness of formaldehyde for this purpose has
recently been documented.89 The receptor phases were stirred
continuously and kept at constant temperature (32 C) by a
circulating water bath. A preapplication period of 48 hours
was established to uniformly condition the skins and to
remove water-soluble UV-absorbing materials. The receptor
phases were changed three times during this period, and
control experiments from earlier studies have shown that this
procedure effectively removes those materials.97 The
epidermal sides of the skins were exposed to the air and were
left untreated during this period.
After the preapplication period, 0.5 mL aliquots from
suspensions of the prodrugs in IPM were applied to the
epidermal sides of the skins. The IPM suspensions were
stirred at 221 C for 48 hours prior to application to
ensure that saturation was attained. Total concentrations of
the IPM suspensions ranged from 0.6 M to 1.0 M with enough
excess solid present to maintain saturation for the duration
of the application period (see below). Each drug-vehicle
combination was run in triplicate.
Samples were taken from the receptor phases at 4, 8, 12,
21, 24, 27, 30, 33, 36, 45, and 48 hours after donor phase
application. The receptor phases were changed following
removal of each sample so that "sink" conditions were
maintained. Samples were analyzed for total 5FU species that


112
under reduced pressure at room temperature until the volume
was reduced to 25-50 mL. The solution was cooled again to
0 C until crystallization occurred.
3-ftcetyl-5-fluorouracil LH1
Crystallization gave 1.22 g of 17 (71%) : mp 115-7 C dec
(lit.25 mp 114-7 C) ; IR (KBr) 1650, 1685, 1720, and 1805 cm"1
(C=0) ; 1H NMR (CDCI3) 8 2.58 (s, 3H, Cfl3) and 7.23 (bs, 1H,
C6-a) ; UVmax (CH3CN) 267 nm (e=6.57xl03) .
3-Propionyl-5-fluorouracil (18)
Crystallization gave 0.89 g of 18 (48%) : mp 102-3 C dec
(lit.25 mp 99-102 C); IR (KBr) 1655, 1680, 1730, and
1815 cm-1 (C=0); XH NMR (CDCI3) 8 1.27 (t, J=7 Hz, 3H, CH3),
2.86 (q, J=7 Hz, 2H, C0Cfi2) > and 7.26 (bs, 1H, C6-E) ; V^x
(CH3CN) 267 nm (e=6.66xl03) .
3-Butvrvl-5-fluorouracil (19)
Crystallization gave 1.22 g of 19 (61%); mp 111-2 C
(lit.26 mp 132-4 C) ; IR (KBr) 1660, 1730, and 1810 cm'1
(C=0); XH NMR (CDCI3) 8 1.02 (t, J=7 Hz, 3H, CH3), 1.6-2.0
(m, 2H, COCH2CE2) 2.82 (t, J=7 Hz, 2H, COCH2), and 7.26
(bs, 1H, C6-H); UV max (CH3CN) 267 nm (e=6.55xl03) .
Anal. Calc, for C8H9FN2C>3: C, 48.00; H, 4.53; N, 14.00.
Found: C, 48.08; H, 4.55; N, 13.93.
3-Valervl-5-fluorouracil (20)
Crystallization gave 1.54 g of 20 (72%) : mp 110-1 C
dec; IR (KBr) 1665, 1730, and 1815 cm-1 (C=0); !h NMR (CDCI3)
8 0.97 (t, J=7 Hz, 3H, C3) 1.3-1.9 (m, 4H, COCH2C2CH2> <


45
Chapter 3). Significant buffer catalysis by phosphate has
also been demonstrated for compounds in this series, and the
reader is referred to the work of Buur and Bundgaard30 for
complete pH-rate profiles and probable hydrolysis mechanisms.
In any case, release of the parent drug from compound 1
by chemical means is too slow for dermal delivery purposes.
Other members of the series would also appear to be poor
candidates, since they hydrolyze chemically over two times
slower than compound 1.30>93 In 80% human plasma, however,
hydrolysis rates are much faster (ti/2=2-3 min)3093
suggesting enzyme catalysis of the 1-alkyloxycarbonyl series.
Since the skin is metabolically active, further study of
these compounds in diffusion cells was warranted.
Skin Penetration
Skin penetration data from the diffusion cells are
plotted as cumulative amount of total 5FU species that
diffused ((Imol) versus time (h) In Figure 2-2, results for
l-methyloxycarbonyl-5FU (1), l-ethyloxycarbonyl-5FU (2), and
l-propyloxycarbonyl-5FU (3) are compared to 5FU itself. In
Figure 2-3, results for l-butyloxycarbonyl-5FU (4),
l-hexyloxycarbonyl-5FU (5), and l-octyloxycarbonyl-5FO (6)
are compared to 5FU. Error bars correspond to the standard
deviation from the mean for n=3 values.
Fluxes (J) lag times (ti,) and skin accumulation (SA)
values for each compound are reported in Table 2-5. Lag time


2
However, this is neither the most convenient nor comfortable
treatment option.
The toxicity of 5FD is related to its effect on rapidly
proliferating host cells especially those of the bone marrow
and gastrointestinal lining. Toxicity following systemic
therapy is common while adverse effects from topical therapy
appear to be minimal, other than those associated with the
local inflammatory reactions usually necessary for a
therapeutic response.4
Fluorouracil Derivatives
Since 5F was first synthesized16 and tested17 as an
antitumor antimetabolite in 1957, numerous attempts have been
made to improve its efficacy and reduce its toxicity.
Derivatives of 5FD and its nucleosides, 5-fluorouridine (FUR)
and 5-fluoro-2'-deoxyuridine (FdUR), have been developed with
this in mind. The majority of these derivatives have been
designed with the expectation that 5FU will be released in
vivo. so they are essentially prodrugs.
A prodrug has been defined as "an agent which must
undergo chemical or enzymatic transformation to the active or
parent drug after administration, so that the metabolic
product or parent drug can subsequently exhibit the desired
pharmacological response" (p. 1275).18 Prodrugs, or
bioreversible derivatives, of 5FU have included hydroxy
methyl,19 alkyloxyalkyl,20 acyloxyalkyl,21-24 acyl,25-28


36
in which the C6-H signal is shifted downfield relative to 5FU
or the N3-substituted derivatives.25 For example, the C-H
chemical shift for l-ethyloxycarbonyl-5FU (5=8.00) is 0.77
ppm downfield when compared with 3-ethyloxycarbonyl-5FU
(5=7.23) in chloroform-d.
Solubility
Solubility determinations are generally accomplished by
stirring excess solute in a solvent until saturation is
attained. The excess solid is removed and the saturated
solution is assayed for solute content. This approach is
suitable for stable solutes and for unstable solutes in
aprotic solvents, but another method is needed for measuring
aqueous solubilities of chemically unstable compounds.
An alternative to the direct method for determining
aqueous solubilities is the partitioning-based method. The
advantage of this method is that contact time between the
unstable compound and the aqueous phase can be minimized.
However, several points regarding this procedure require
clarification.
First, partition coefficients are concentration
dependent except when compounds with low associating
tendencies are present in dilute solutions (<10_1 M) .55
Solubilities based on partition coefficients can only be
reported as estimates since activity coefficients become more
important at higher solute concentrations."


68
Table 3-2. Melting points (MP), lipid solubilities (Sipm) and
aqueous solubilities (Saq) for 1-acyl derivatives.
Compound
MP
(C)
Sipm3
(mM)
SAQb
(mM)
saqc
(mM)
5FU
280-2
0.049
96
-
7
129-30
22
_
119
8
130-1
36
-
48
9
145-6
17
-
6.5
10
120-1
39
-
3.5
11
101-2
112
-
3.0
12
83-4
111
0.15
aStandard
deviations from
the mean
were within
5% for IPM
solubilities.
bSolubility determined by direct method.
cStandard deviations from the mean were within +5% for
experimental values used to calculate partition coefficients
and estimated aqueous solubilities.


ACKNOWLEDGEMENTS
I would like to thank the members of my supervisory
committee, Dr. Kenneth Sloan, Dr Margaret James, Dr. Koppaka
Rao, Dr. Richard Prankerd, and Dr. John Zoltewicz for their
guidance and expert advice over the past four years. My
sincerest thanks go to my research advisor and committee
chairman, Dr. Sloan, for sharing his enthusiasm for teaching
and science. I am especially grateful for his patience and
understanding during my seemingly endless questions and
interruptions.
I would also like to acknowledge the enthusiastic
support of Dr. Noel Meltzer of Hoffmann-La Roche. This
project was partially funded by a grant from Hoffmann-La
Roche.
My special thanks go to my parents for their love and
support throughout my life and to my two-year-old son,
Michael, who could make me laugh when it was the last thing I
felt like doing. But most of all, I want to thank my wife,
Donna, whose love, support, and countless sacrifices made my
return to school and the completion of this project possible.


32
diffusion cells from Crown Glass in Somerville, NJ with
4.9 cm2 donor surface areas and 20 mL receptor phase volumes
were used for this purpose. The full-thickness skins were
obtained from female hairless mice (SKH-hr-1) from Temple
University Skin and Cancer Hospital.
The mice were killed by cervical dislocation, their
skins were removed immediately by blunt dissection, and
dorsal sections were mounted in the diffusion cells. The
dermal sides of the skins were placed in contact with
receptor phase which contained 0.05 M phosphate buffer
(pH=7.1, 1=0.12) with 0.11% formaldehyde as a preservative.
The effectiveness of formaldehyde for this purpose has
recently been documented.89 The receptor phases were stirred
continuously and kept at constant temperature (32 C) by a
circulating water bath. A preapplication period of 48 hours
was established to uniformly condition the skins and to
remove water-soluble UV-absorbing materials. The receptor
phases were changed three times during this period, and
control experiments from earlier studies have shown that this
procedure effectively removes those materials.97 The
epidermal sides of the skins were exposed to the air and were
left untreated during this period.
After the preapplication period, 0.5 mL aliquots from
suspensions of the prodrugs in IPM were applied to the
epidermal sides of the skins. The IPM suspensions were
stirred at 22+1 C for 48 hours prior to application to
ensure that saturation was attained. Total concentrations of


12
is irrelevant. Since the membrane thickness (h) and
diffusion coefficient (D) remain relatively constant (D is
inversely proportional to the cube root of the molar volume
according to the Stokes-Einstein equation), Km is the only
variable that can substantially influence the flux.
There are several general approaches for enhancing skin
penetration. Four of theseocclusion, use of penetration
enhancers, iontophoresis, and sonophoresisproduce their
enhancement effects by changing the barrier properties of the
skin. Occlusion involves covering the application site,
impeding transepiderraal water loss, and increasing the
hydration state of the skin.67 With penetration enhancers,
accelerants, or promoters in topical formulations, the
reversible reduction of barrier resistance in the stratum
corneum is the goal, and ideally, incorporation of the
enhancer into the skin will not result in cell damage.59
Iontophoresis is a technique in which electroosmotic volume
flow from an applied electric field leads to increases in
mass transfer in excess of passive diffusion.68 Sonophoresis
uses ultrasonic frequencies to increase skin penetration.
The other two methods, formulation (without penetration
enhancers) and the use of prodrugs, do not disrupt the
barrier layer. Essentially, the formulation approach
involves changing the penetrants solubility in the vehicle by
changing the vehicle. The effect on flux of this approach is
indeterminate according to equation (5). If the solubility
in the vehicle (Cv) is increased, then the partition


6
Structure of the Skin
Fluorouracil (5F) is typical of polar, high-melting,
heterocyclic compounds which exhibit poor skin permeability.3
Since the goal of this project is to improve dermal and
transdermal delivery of 5FU, it is necessary to examine the
structure of the barrier to be penetrated, the skin.
The skin is composed of two major tissue layers, the
epidermis and dermis. The epidermis is a continuous, elastic
sheet that is interrupted only by glandular pores and hair
follicles. It consists of four definable sublayers and
averages 75 to 150 |lm in thickness. The basal cell layer,
which borders the dermis, is a single layer of keratinocytes.
This is the germinal layer, and all epidermal cells are
initially formed here before moving outward to the next
layer, the stratum spinosum. The basal cells are cuboidal or
columnar in shape, while the stratum spinosum cells are
polyhedral. By the time the cells have migrated to the next
layer, the stratum granulosum, they have become flattened and
contain characteristic keratohyalin granules. The stratum
granulosum marks the transition between nucleated cells and
the anucleated stratum corneum. The stratum corneum cells in
the last layer are highly keratinized, markedly flattened
cells in which cellular components, such as mitochondria and
ribosomes, have degraded along with the nucleus. Stratum
corneum normally consists of about 15 to 20 layers, but each


122
evaporated at which time gradual disappearance of compound 17
(Rf=0.43) began with formation of both l-acetyl-5FU (Rf=0.62)
and 1,3-bis-acetyl-5FU (Rf=0.66) as well as 5FU (Rf=0.30).
Opon resolidification of the melt, its 1H NMR spectrum showed
that compound 17 had been converted to l-acetyl-5FU (~70%) ,
5FU (-30%) and a trace of 1,3-bis-acetyl-5FU. Percentages
were calculated from integration of the corresponding C6-H
signals.
Similar studies showed that melting 1,3-bis-acetyl-5FO
alone produced l-acetyl-5FU (-40%) and 1,3-bis-acetyl-5FU
(-60%) while melting equivalent amounts of 1,3-bis-acetyl-5FO
and 5FU produced l-acetyl-5FU (-75%), 5FU (-20%), and 1,3-
bis-acetyl-5FU (-5%).
Since rate and duration of heating were not controlled
in these experiments, conclusions regarding the relative
amounts of decomposition products formed can not be drawn.
However, the data do suggest that thermal N3-deacylation
occurs from both 3-acetyl-5FU and 1,3-bis-acetyl-5FU, and
overall, there is a net loss of acetyl groups. Some of the
N3-acetyl groups (or even N1-acetyl groups) could be lost to
ketene formation as shown in Figure 5-1. There is also a net
gain of N1-acetyl groups when 3-acetyl-5FU is heated alone or
a mixture of 1,3-bis-acetyl-5FU and 5FU is heated. This
indicates that a rearrangement has occurred. The TLC and
3H NMR data suggest that the rearrangement is to some degree
intermolecular since some 1,3-bis-acetyl-5FU is produced when
3-acetyl-5FU is heated alone, but an intramolecular


131
Time (min)
Figure 5-4. Plot of In (At-Aoo) versus time (min) for hydrolysis
of 3-propionyl-5FU in 0.05 M phosphate buffer
(pH-7.1, 1-0.12) at 32 C (n=2).


151
the exception of 3-acetyl-5FU which was much lower. Since
the 1,3-bis-acyl derivatives hydrolyzed to give the
corresponding 3-acyl derivatives almost immediately, they
should behave in a similar manner to the 3-acyl derivatives.
Testing of the 3-acyl, 1,3-bis-acyl, and 1-alkyloxycarbonyl
series in other skin models should be carried out before they
are dismissed as too stable for use as prodrugs for dermal as
opposed to transdermal delivery.
Skin accumulation of total 5FU species was highest for
the 1-acyl derivatives particularly for l-acetyl-5F0 and
l-propionyl-5FU. These two prodrugs had skin accumulation
values more than 18 times the value for 5FU and nearly four
times more than the next best derivative. It was suggested
that rapid hydrolysis of the 1-acyl derivatives upon
partitioning into the skin effectively "locks in" large
amounts of 5FU since it is the highly polar 5FU molecule that
must diffuse through the remaining lipid regions of the skin.
Second application studies using the standard drug-
vehicle combination showed that skin damage was at most one
and one-half times greater for the prodrugs than for 5FU, a
small increase when compared to the much greater improvement
in skin penetration and skin accumulation that was achieved.
The stability of the prodrugs in the IPM formulations
was documented and no decomposition was observed even after a
twelve-hour period during which the formulations were in
contact with the skins. This shows that even chemically


72
Table 3-5. Pseudo-first-order rate constants (k) and half-
lives (ti/2) for hydrolysis of l-acetyl-5FU in 0.05 M
phosphate buffer (pH=7.1, 1=0.12) with and without
formaldehyde at 32 C.
Compound
Formaldehyde
(M)
k(SD)
(min-1)
tl/2
(min)
7
0
0.143(0.002)a
4.8
7
3.6xl0'2
0.161(0.008)b
4.3
7
1.8X10'1
0.219(0.004)a
3.2
7
3.6xl0_1
0.279(0.006)a
2.5
aMean standard deviation for n=3 values.
bMean standard deviation for n=5 values.


58
Lipid solubility
Lipid solubilities were determined using isopropyl
myristate (IPM), a commercial vehicle used in cosmetics and
topical medicinis,94 as the lipid solvent. The use of IPM as
a model lipophilic vehicle in skin penetration studies is
well established.77'95
Three suspensions of each derivative were stirred at
221 C for 48 hours. The suspensions were filtered through
0.45 [lm nylon filters, and the saturated solutions were
diluted in acetonitrile and analyzed by UV spectroscopy.
Solubilities were calculated using Beer's Law:
A = e-C-d (1)
where A is the absorbance, £ is the molar absorptivity, C is
the concentration in mol/L, and d is the path length of the
cuvette in cm. Molar absorptivities were predetermined in
triplicate in acetonitrile at 261 nm.
Aqueous Solubility
Because of the chemical instability of the 1-acyl
derivatives, direct measurement of aqueous solubilities for
these prodrugs was not attempted. A comparison of the direct
and partitioning-based methods for determining aqueous
solubility was presented in Chapter 2 for the 1-alkyloxy-
carbonyl derivatives.


106
are much less effective at delivering 5FU transdermally than
the 1-alkyloxycarbonyl and 1-acyl series. The results for
this highly lipophilic, but poorly aqueous-soluble series
demonstrate the importance of biphasic solubility for
maximizing skin penetration.
The negative lag times observed for 1,3-bis-butyryl-5FU
(15) and 1,3-bis-valeryl-5FU (16) may have been caused by a
change in the physical state of the donor phase. A large
amount of solid was required in formulating the donor phases
for compounds 15 and 16 due to their high solubility in IPM.
After a few hours of skin contact at 32 C, it became
difficult to determine if excess solid from these low-melting
derivatives was still present in the donor phases since their
appearance was more like a gel than a suspension.
Skin accumulation values are greater for the 1,3-bis-
acyl series than for 5FU, but the improvement is small. The
best derivatives, compounds 13, 15, and 16, have skin
accumulation values less than three times higher than 5FU.
Second application fluxes and lag times are reported in
Table 4-5. Skin penetration by theophylline from propylene
glycol, the standard drug-vehicle combination, after
treatment with the 1,3-bis-acyl derivatives is similar to
that following treatment with 5FU.
The stability of the prodrugs in the IPM formulations
was assessed by 1H NMR analysis of the donor phases. After a
minimum of five days from the time the suspensions were
prepared until their *H NMR spectra were recorded, including


60
differences in solubility in one phase relative to the other,
volume ratios (IPM:buffer) other than 1:1 were necessary, but
the ratio never exceeded 10:1 or 1:10.
Hydrolysis Kinetics
Hydrolysis rates have previously been reported for one
member of this homologous series.26 In the present study,
hydrolysis rates were determined at 32 C for all six 1-acyl
derivatives in 0.05 M phosphate buffer (pH=7.1, 1=0.12) and
for l-acetyl-5FU (7) in the same buffer with 0.11%
formaldehyde (3.6xl0~2 M). The rate in the presence of
formaldehyde was determined for comparison with the rate in
plain buffer since formaldehyde was used as a preservative in
the diffusion cell experiments described in the following
section. Two other concentrations of formaldehyde (1.8xl0_1 M
and 3.6xl0_1 M) were studied to assess the catalytic role of
formaldehyde in the hydrolysis of compound 7.
The hydrolyses were followed by V spectroscopy at
266 nm where the absorbance decrease accompanying hydrolysis
of the 1-acyl derivatives was maximized. Hydrolysis was
initiated by adding 60 to 75 |XL of stock solutions of the
derivatives in acetonitrile to 3 mL of buffer prewarmed to
32 C in a thermostated quartz cuvette to give final
concentrations of l-2xl0~4 M. Absorbances were recorded at
appropriate intervals and pseudo-first-order rate constants


Cumulative Amount (pmol)
140
Time (h)
Figure 5-8. Plots of cumulative amount of total 5FO species
that diffused (pjnol) versus time (h) for
compounds 17, 18, and 5FU.


21
Table 1-1. Structures of 5FU and prodrug derivatives of 5FU.
Series
Rl
R2
l-alkyloxycarbonyl-5FU (I)
l-acyl-5FU (II)
1,3-bis-acyl-5FU (III)
3-acyl-5FU (IV)
-(C=0)0(CH2)nCH3
-(C=0)(CH2)nCH3
-(C=0)(CH2)nCH3
-H
-H
-H
-(C=0) (CH2)nCH3
(C=0)(CH2)nCH3


43
O -10
c
-11 .
-12
200
q Plain Buller
t Formaldehyde Buffer
T
400
Time (min)
600
800
Figure 21. Plots of ln(C) versus time (min) for hydrolysis
of l-methyloxycarbonyl-5FU in 0.05 M phosphate buffer
(pH=7.1, 1=0.12) with and without formaldehyde
at 32 C.


10
Passive Diffusion
Although the skin is not a homogeneous tissue, it has
the characteristics of a passive diffusion barrier. Fick's
first law of diffusion defines flux as the amount of material
flowing through a unit cross section of a barrier in a unit
time.65 Flux is also proportional to the concentration
gradient, and the first law can be written:
J = -D-dC/dx (1)
where J is the flux, D is the diffusion coefficient of the
penetrant in the barrier, and dC/dx is the concentration
gradient. While the first law is used to describe steady-
state diffusion, Fick's second law describes the change in
concentration with respect to time at a specific location.65
Our interest is in steady-state diffusion, and the second law
will not be discussed further.
The concentration gradient term (dC/dx) from Fick's
first law can be approximated by (C1-C2) /h, where Ci is the
concentration in the barrier on the donor side of the
barrier, while C2 is the concentration in the barrier on the
receptor side, and h is the thickness of the diffusional
barrier. Therefore, the expression becomes
J = D- (C1-C2) /h (2)
For skin penetration studies, "sink" conditions can be
assumed which means that C2=0, and
J = D-Ci/h
(3)


CHAPTER 5
3-ACYL DERIVATIVES
Introduction
The 3-acyl derivatives of 5FU are more stable chemically
than either the 1-acyl or 1,3-bis-acyl derivatives.26
Enzymatic hydrolysis proceeds at a faster rate than chemical
hydrolysis, and this difference is more pronounced for the
short-chain derivatives of the 3-acyl series26 some of which
may be useful as prodrugs for dermal delivery.
The 3-acyl derivatives were chosen for evaluation
because of their chemical stability and for comparison with
the 1-acyl series. It was known that N1- and N3-substituted
derivatives have different physical-chemical properties,22'29"
30 and it was likely that they would exhibit differences in
skin penetration as well.
Four straight chain 3-acyl derivatives were selected for
study. The derivatives and their structures are presented in
Table 5-1.
109


26
stirred for 30 minutes, and the methanol was evaporated under
reduced pressure. The potassium salt was suspended in
acetone (25-50 mL) which was evaporated under reduced
pressure to remove residual methanol. The salt was
resuspended in acetone (25-50 mL), and the suspension was
added dropwise over a three minute period to a well stirred
acetone (20 mL) solution containing 1.0 to 1.2 equivalents of
the appropriate alkyl chloroformate. The mixture was stirred
at room temperature for 60 minutes. The mixture was
filtered, and the residue was washed with acetone (20 mL).
The combined acetone solutions were evaporated under reduced
pressure, and the solid residue was crystallized from an
appropriate solvent or solvent combination.1
l-Methyloxycarbonyl-5-fluorouracil (1)
Crystallization from acetone gave 1.36 g of 1 (72%) :
mp 158-60 C (lit.30 mp 159-60 C) ; IR (KBr) 1695, 1710, 1740,
and 1760 cm'1 (C=0); 3H NMR t(CD3)2SO] 5 3.86 (s, 3H, Cfi3) and
8.16 (d, J=7 Hz, 1H, C6-H) , (IVmI (CH3CN) 254 nm (E-9.63xl03) .
l-Ethvloxvcarbonvl-5-fluorouracil (2)
Crystallization from acetone/ether gave 1.31 g of 2
(65%) : mp 127-8 C (lit.30 mp 126-8 C); IR (KBr) 1690, 1730,
and 1750 cm'1 (C=0); 3H NMR [ CH3), 4.31 (q, J=7 Hz, 2H, OCH2), and 8.16 (d, J=7 Hz, 1H,
C6-H); UVmav (CH3CN) 254 nm (e=9.86xl03).
^Several compounds in this series were provided by Kenneth B.
Sloan, Ph. D. based on the author's procedure.


BIOGRAPHICAL SKETCH
Howard D. Beall was born July 6, 1954, in Salt Lake
City, Utah. The Beall family returned to Montana before he
was two years old, and he spent his school years in several
Montana cities.
He graduated with honors from the University of Montana
in 1977 with a Bachelor of Science degree in pharmacy. He
practiced pharmacy in various settings in Montana and Oregon
before moving to Florida and earning a Master of Science
degree in pharmacy from the University of Florida in 1983.
He entered the doctoral program in the Department of
Medicinal Chemistry at the University of Florida in 1987.
He is a member of Rho Chi National Honor Society and the
American Association of Pharmaceutical Scientists. He
received the Merck Award for academic achievement in 1977 and
was a fellow of the American Foundation for Pharmaceutical
Education from 1990-91.
He married Donna Goyette in 1984, and they have a son,
Michael, born in 1989, and a daughter, Allison, born in 1991.
160


117
recorded at appropriate intervals and pseudo-first-order rate
constants were determined from the expression:
In (Atfioo) = lnfAo-A^J-kt (6)
where At is the absorbance at some time=t, A, is the
absorbance at t=, A0 is the absorbance at t=0, k is the
pseudo-first-order rate constant, and t is the time. The
hydrolyses were sufficiently fast to allow experimental
determination of A,. The slopes, -k, of linear plots of
ln(At~A,) versus time were determined by linear regression.
The half-lives (ti/2) were calculated from equation (5).
The hydrolysis reactions were run in duplicate or
triplicate and were followed for a minimum of three half-
lives. The correlation coefficients for all reported rate
constants were >0.999 except where noted.
Skin Penetration Studies
Diffusion cell experiments were performed to measure the
transdermal delivery of 5FU and the 5FU prodrugs. Franz-type
diffusion cells from Crown Glass in Somerville, NJ with 4.9
cm2 donor surface area and 20 mL receptor phase volume were
used for this purpose. The full-thickness skins were
obtained from female hairless mice (SKH-hr-1) from Temple
University Skin and Cancer Hospital.
The mice were killed by cervical dislocation, their
skins were removed immediately by blunt dissection, and
dorsal sections were mounted in the diffusion cells. The


105
Table 4-5. Second application fluxes (J) and lag times (tl)
for 1,3-bis-acyl derivatives.
Ja(+SD)b
Compound (|lmol/cm2/h) tt,(h)
5FU
1.2(0.2)
1.2
13
1.6(0.1)
0.7
14
1.6(0.3)
0.7
15
1.1(0.1)
0.9
16
0.87 (0.14)
0.8
aFlux of 0.4 M theopylline from propylene glycol.
bMean standard deviation for n=3 values.


64
were again left exposed to the air. After this "leaching"
period, another sample was taken from each cell to measure
the skin accumulation of total 5FU species.
Second applications to the epidermal sides of the skins
were made after the "leaching" period with a standard drug-
vehicle suspension Theophylline in propylene glycol
(0.4 M) was applied to assess the damage to the skins from
application of the initial drug-vehicle combinations.
Samples were taken at 1, 2, 3, 5, 7, 9, and 11 or 12 hours
after application. The samples were analyzed for
theophylline by UV spectroscopy (e=1.02xl04 at 271 nm) and
second application fluxes were determined as described above.
Results and Discussion
Synthesis and Structure Determination
The known l-acyl-5FU derivatives have melting points27
and 7H NMR spectra25 in agreement with those reported in the
literature. The structures of novel compounds were assigned
by comparison of their NMR spectra with those of the known
homologs. Elemental microanalyses were obtained for the
novel compounds and were within acceptable limits (0.4%).
The differences in spectral properties between N1- and
N3-acylated derivatives were discussed in detail in
Chapter 2. Ultraviolet (UV) spectra were not obtained for
the 1-acyl derivatives under basic conditions because of


113
2.84 (t, J=7 Hz, 2H, COCH2) and 7.23 (bs, 1H, C6-H); UVmax
(CH3CN) 267 nm (e=6.68xl03) .
Anal. Calc, for C9H11FN2O3: C, 50.47; H, 5.18; N, 13.08.
Found: C, 50.73; H, 5.24; N, 12.95.
T.i ni d sol nbi 1 i t~ y
Lipid solubilities were determined using isopropyl
myristate (IPM), a commercial vehicle used in cosmetics and
topical medicinis,94 as the lipid solvent. The use of IPM as
a model lipophilic vehicle in skin penetration studies is
well established.77, 95
Three suspensions of each derivative were stirred at
221 C for 48 hours. The suspensions were filtered through
0.45 |lm nylon filters, and the saturated solutions were
diluted in acetonitrile and analyzed by UV spectroscopy.
Solubilities were calculated using Beer's Law:
A = e-C-d (1)
where A is the absorbance, E is the molar absorptivity, C is
the concentration in mol/L,and d is the path length of the
cuvette in cm. Molar absorptivities were predetermined in
triplicate in acetonitrile at 267 nm.
Aqnpons Solubility
For direct measurement of aqueous solubilities, two or
more suspensions of each derivative were vigorously stirred
in 0.05 M acetate buffer (pH=4.0) at 221 C for 60 minutes.


9
Until recently, it was thought that the transcellular
pathway was the primary route through the stratum corneum.
This was based, at least in part, on a gross underestimation
of the volume of the intercellular space.60 The current
belief is that passive diffusion occurs predominantly through
the intercellular channels.61-62 While this seems logical for
lipophilic compounds, this pathway has also been shown to be
the dominant route for mercuric chloride, an ionic compound.
Bodd et al.63 showed that mercuric sulfide could be
precipitated in the stratum corneum after topical application
of mercuric chloride. They found that mercuric cation
accumulates initially in the intercellular spaces throughout
the stratum corneum. Intracellular uptake was observed
later, but only in the apical corneocytes by way of
disintegrating desmosomal attachments. Since the
intercellular lipids are arranged in multiple lamellar
bilayers,64 the mercuric cation, and possibly other ions and
hydrophilic molecules, may diffuse through the interlamellar,
hydrophilic channels that are associated with the polar
headgroups of the lipids.63
While most of the current research is directed toward
understanding the intercellular region, Barry59 cautions that
the transcellular route should not be dismissed as
unimportant, especially in penetration enhancer research.


LIST OF REFERENCES
1. Diasio, R. B.; Harris, B. E. Clin. Pharmacokin. 1989,
12, 215.
2. Kovach, J. S.; Beart, R. W. Invest. New Drugs 1989, 7,
13.
3. Bennett, D. R., Ed. Drug Evaluations Annual 1991;
American Medical Association: Milwaukee, WI, 1990;
p. 1772.
4. McEvoy, G. K., Ed. AHFS Drug Information 90: American
Society of Hospital Pharmacists: Bethesda, MD, 1990;
p. 513.
5.
Dillaha,
C. J.;
Jansen
, G. T.;
Honeycutt,
w.
M. ;
Bradford,
A. C.
Arch.
Dermatol
^ 1963, 22,
247 .
6.
Dillaha,
C. J.;
Jansen
, G T .;
Honeycutt,
w.
M. ;
Holt
G. A. Arch. Dermatol. 1965, 52, 410.
7. Simmonds, W. L. Cutis 1973, 12, 615.
8. Simmonds, W. L. Cutis 1976, 12, 298.
9. Klein, E.; Stoll, H. L.; Milgrom, H.; Traenkle, H. L.;
Case, R. W.; Laor, Y.; Helm, F.; Nadel, R. S. J. Invest
Dermatol. 1965, 12, 489.
10.
Stoll,
H.
L.; Klein, E. L.
Invest. Dermatol. 1969, 52
304.
11.
Tsuji,
T. ;
; Sugai, T. Arch.
Dermatol. 1972. 105, 208.
12.
L junggren,
. B.; Moller, H.
Arch.
Dermatol. 1972, 122,
263.
13.
Pearlman,
D. L.; Youngberg
', B.;
Engelhard. C. J. Am.
Acad.
Dermatol. 1986, 12,
1247 .
14 .
Goette
/ D
. K. J. Am. Acad.
Dermatol. 1981, , 633.
15.
Robinson,
T. A.; Kligman,
A. M.
Br. J. Dermatol. 1975
52. 703.
153


22
Four series of compounds have been selected as potential
prodrugs of 5FU. They are the 1-alkyloxycarbonyl (I), 1-acyl
(II), 1,3-bis-acyl (III), and 3-acyl (IV) derivatives of 5F
(Table 1-1). Fluorouracil (5FD) has two acidic pKa values,
8.0 and 13.0.91 Spectral studies have shown that the
monoanion is actually a mixture of N1- and N3-anions.30 A
comparison of ionization constants for 5FU derivatives that
are identically substituted at the N1- or N3-positions22'30
suggests that the N3-position is probably the most acidic.
However, the N1-position is the most reactive site for both
synthesis and hydrolysis of acylated derivatives of 5FU.
The three series of acyl derivatives were chosen for two
reasons; they exhibit high lipid and aqueous solubilities,
and they readily hydrolyze without enzymatic activation.26
While rapid hydrolysis is seen as a drawback by some
investigators,26'42 it may actually be an advantage when
developing prodrugs for dermal delivery.38-39'92 The
1-alkyloxycarbonyl series was selected for comparison with
the 1-acyl series since it shows good chemical stability, but
rapid enzymatic hydrolysis.30 Various members of each series
have been synthesized previously,25-27'30'93 but homologous
series have not been examined, and nobody has studied their
applicability for dermal delivery.


108
group is oriented perpendicular to the plane of the 5FU ring
and is sterically and electronically hindered from
nucleophilic attack.


70
Aqueous solubilities were determined by the
partitioning-based method for which a theoretical discussion
was presented in Chapter 2. For the 1-acyl series, aqueous
solubility decreases with increasing chain length, and only
l-acetyl-5FU (7) is more water soluble than 5FU.
In Table 3-3, partition coefficients (PC) and
hydrophobicity parameters (Jl) are listed for the 1-acyl
derivatives. The average it value is 0.60, the same value
that was calculated for the 1-alkyloxycarbonyl series.
Individual values for the 1-acyl derivatives are all within
0.07 of the average. These results further support use of
the partitioning-based method for determining relative
aqueous solubility in a homologous series.
Hydrolysis Kinetics
Hydrolysis of the 1-acyl derivatives was studied in
0.05 M phosphate buffer (pH=7.1, 1=0.12) at 32 C. Pseudo-
first-order rate constants (k) and half-lives (11/2> are
presented in Table 3-4. Half-lives are relatively consistent
throughout the series ranging from 3.1 minutes for
l-propionyl-5FU (8) to 4.8 minutes for l-acetyl-5FU (7).
Hydrolysis of compound 7 was also studied in the same
buffer with increasing concentrations of formaldehyde added
to assess the catalytic role of formaldehyde hydrate. These
results are shown in Table 3.5. It is clear that the
hydrolysis rate for compound 7 is faster in the presence of


I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
jjhn^A. Zoltewicz ^0
Professor of Chemistry
This dissertation was submitted to the Graduate Faculty
of the College of Pharmacy and to the Graduate School and was
accepted as partial fulfillment of the requirements for the
degree of Doctor of Philosophy.
December, 1991
%/,
Dean, College of Pharmacy'-
Dean, Graduate School


(loiurl) lunouiv aAjjeiniuno
141
Time (h)
Figure 5-9. Plots of cumulative amount of total 5FU species
that diffused (|imol) versus time (h) for
compounds 19, 20, and 5FU.


125
rearrangement from N3- to N1-acetyl is possible in this case
as well. In Figure 5-2, an intramolecular rearrangement
scheme is presented in which a transient 02-acyl intermediate
is proposed.
The existence of intramolecular 1,3 acyl migrations from
both N to 0 and 0 to N has been documented.109 Of the several
possible mechanisms for the 0 to N thermal rearrangement, one
that is preferred involves rate-determining nucleophilic
attack by the nitrogen lone pair on the carbonyl carbon with
partial C-0 bond cleavage to relieve the strain associated
with formation of a four-membered ring.113 This synchronous
mechanism is depicted in Figure 5-2 for both the N to 0 and
0 to N rearrangements. The perpendicular orientation of the
N3-acetyl group to the 5F0 ring would facilitate a
rearrangement of this type. All steps are shown as
reversible since acyl or aroyl rearrangements are potentially
reversible whether they are intramolecular or
intermolecular.111'3 It is worth noting that similar melting
behavior to that seen for the 3-acyl derivatives in the
present study was also observed for the 0 to N benzoyl
migration in 6-phenanthridinone.113 Further discussion
regarding the proposed 02-acyl intermediate will be presented
in the hydrolysis kinetics section.


In (At-A.
130
Time (min)
Figure 5-3. Plot of ln(At-A.) versus time (min) for hydrolysis
of 3-acetyl-5FU in 0.05 M phosphate buffer
(pH=7.1, 1=0.12) at 32 C (n=3).


136
ch2o
o
ch2o
HO H
0=
I
H3C
Figure 5-7. Possible scheme for reaction of 3-acetyl-5FU with
formaldehyde to form l-acetyloxymethyl-5FU and
3-acetyloxymethyl-5FD.


Cumulative Amount (jimol)
103
Time (h)
Figure 4-4. Plots of cumulative amount of total 5FU species
that diffused (|lmol) versus time (h) for
compounds 15, 16, and 5FU.


3-ACYL DERIVATIVES
109
Introduction 109
Materials and Methods Ill
Results and Discussion 120
Summary 146
SUMMARY AND CONCLUSIONS 148
LIST OF REFERENCES 153
BIOGRAPHICAL SKETCH 160
v


159
100. Hansch, C.; Leo, A. In Substituent Constants for
Corre! at i nn Analysis in Chemistry and Riolngy: John
Wiley and Sons: New York, 1979; p. 13.
101. Yalkowsky, S. H.; Flynn, G. L.; Slunick, T. G. J. Pharm
Sci. 1972, 11, 852.
102.
Yalkowsky,
J. Pharm.
. s.
Sci
H.; Valvani, S. C.;
1983, 72, 866.
Roseman, T.
J.
103.
Flynn, G.
L.;
Yalkowsky, S. H. J.
Pharm. Sci.
1972
838.
104. Chem. Abstr. 1963, 11, 1477a.
105. Silverstein, R. M.; Bassler, G. C.; Morrill, T. C. In
Spentrometrio Identification of Organic Compounds: John
Wiley and Sons: New York, 1981; p. 95, 181.
106. Wierzchowski, K. L.; Litonska, E.; Shugar, D. J. Am.
Chem. Soc. 1965, 11, 4621.
107. Wempen, I.; Fox, J. J. J. Am. Chem. Soc. 1964, !£, 2474.
108. Stolarski, R.; Remin, M.; Shugar, D. 7,. Natnrforsnh.
1977, 12£, 894.
109. Curtin, D. Y.; Miller, L. L. J. Am. Chem. Soc. 1967, H,
637 .
110. McCarthy, D. G.; Hegarty, A. F. J. Chem. Soc. Perkin II
1977, 1085.
111. Minkin, V. I.; Olekhnovich, L. P.; Zhdanov, Y. A. A or..
Chem. Res. 1981, 14, 210.
112. Fleming, I.; Philippides, D. J. Chem. Soc. 1C) 1970,
2426.
113. Curtin, D. Y.; Engelmann, J. H. Tetrahedron Letters
1968, 3911.
114. Prankerd, R. J.; Stella, V. J. Int. J. Pharm. 1989, 2,
71.
115.Prankerd, R. J.; Walters, J. M.; Parns, J. H. Int. J
Pharm. in press.


LIST OF TABLES
Table 1-1. Structures of 5FU and prodrug derivatives
of 5FU 21
Table 2-1. Structures of 1-alkyloxycarbonyl
derivatives 24
Table 2-2. Melting points (MP), lipid solubilities
(Sjpm) and aqueous solubilities (Saq) for
1-alkyloxycarbonyl derivatives 39
Table 2-3. Solubility ratios (SR), partition
coefficients (PC) and hydrophobicity parameters (It)
for 1-alkyloxycarbonyl derivatives 40
Table 2-4. Pseudo-first-order rate constants (k) and
half-lives (ti/2) for hydrolysis of 1-methyloxy-
carbonyl-5FU in 0.05 M phosphate buffer (pH=7.1,
1=0.12) with and without formaldehyde at 32 C 44
Table 2-5. Fluxes (J) lag times (ti.) and skin
accumulation (SA) values for 1-alkyloxycarbonyl
derivatives 48
Table 2-6. Second application fluxes (J) and lag times
(tj,) for 1-alkyloxycarbonyl derivatives 49
Table 3-1. Structures of 1-acyl derivatives 54
Table 3-2. Melting points (MP), lipid solubilities
(Sjpm) and aqueous solubilities (Saq) for 1-acyl
derivatives 68
Table 3-3. Partition coefficients (PC) and hydropho
bicity parameters (It) for 1-acyl derivatives 69
Table 3-4. Pseudo-first-order rate constants (k) and
half-lives (ti/2) for hydrolysis of 1-acyl derivatives
in 0.05 M phosphate buffer (pH=7.1, 1=0.12) at 32 C.... 71
Table 3-5. Pseudo-first-order rate constants (k) and
half-lives (11/2) for hydrolysis of l-acetyl-5FU in
0.05 M phosphate buffer (pH=7.1, 1=0.12) with and
without formaldehyde at 32 C 72
vi


CHAPTER 6
SUMMARY AND CONCLUSIONS
Fluorouracil (5FU) is a commercially available
antineoplastic agent that has been used topically for
treatment of actinic keratoses, superficial basal cell
carcinomas, psoriasis, and other precancerous conditions,
malignant and benign tumors, and dermatoses. As a typical
polar, heterocyclic compound, however, 5FU has poor
solubility and skin penetration properties, and topical
treatment with 5FU is often ineffective.
In an attempt to improve the topical delivery of 5FU,
four homologous series of bioreversible derivatives
(prodrugs) of 5FU were synthesized and characterized, and
their ability to penetrate through and accumulate in the skin
was evaluated. The four series included six 1-alkyloxy-
carbonyl, six 1-acyl, four 1,3-bis-acyl, and four 3-acyl
derivatives.
Solubilities and melting points for the four series of
prodrugs showed the expected trends. Melting points
generally decreased and lipid solubilities generally
increased with increasing chain length, and in fact, lipid
solubilities were a minimum of 40 times greater for the
derivatives than for 5FU. Aqueous solubilities were
maximized for the first or second member of each series and
then decreased with increasing chain length. Five
148


134
Table 5-4. Pseudo-first-order rate constants (k) and half-
lives (ti/2) for hydrolysis of 3-acyl derivatives in
0.05 M phosphate buffer (pH=7.1, 1=0.12) with and
without formaldehyde at 32 C.
Compound
(Assay)
Phase
Formaldehyde
(M)
k(SD)a
(min-1)
tl/2
(min)
17 (UV)
initial
0
7.01x10-3 (0.10x10-3)
99
17 (UV)
terminal
0
3.91x10-3(0.23x10-3)
177
18 (UV)
initial
0
6.20x10-3(0.14x10-3)
112
18(UV)
terminal
0
2.31x10-3(0.01x10-3)
300
17(HPLC)
initial
0
6.83x10-3(0.31xl0-3)b
101
17(HPLC)
terminal
0
3.92x10-3(0.32x10-3)
177
17 (HPLC)
initial
3.6xl0~4
7.20x10-3(0.17x10-3)
96
17(HPLC)
terminal
3.6xl0-4
5.07x10-3(0.08x10-3)
137
17(HPLC)
all
3.6xl0-3
7.19x10-3(0.14x10-3)
96
17(HPLC)
all
3.6xl0-2
7.00x10-3(0.28x10-3)
99
aMean standard deviation for n=2-3 values (see Figures).
Correlation coefficient <0.999 (=0.998).


86
1-C0C2), and 8.25 (d, J=7 Hz, 1H, C6-) ; UVmax (CH3CN) 262 nm
(6=9.65xl03) .
Anal. Calc, for C10H11FN2O4: C, 49.59; H, 4.58; N, 11.57.
Found: C, 49.44; H, 4.59; N, 11.53.
1.3-bis-Butyryl-5-fluorouracil (JJLL
Extraction of the residue with hot low-boiling petroleum
ether gave a cloudy solution which cleared upon cooling to
0 C and produced a yellow resinous sediment. The mixture
was allowed to warm to room temperature, and the clear
solution was decanted from the sediment. The solution was
cooled again to 0 C and crystallization gave 1.70 g of 15
(63%) : mp 48-9 C (lit.26 mp 47.5-48.5 C) ; IR (KBr) 1685,
1710, 1735, and 1795 cm'1 (C=0); iH NMR (CDCI3) 8 1.00 (t,
J=7 Hz, 3H, 3-CH3), 1.03 (t, J=7 Hz, 3H, I-CH3) 1.6-1.9 (m,
4H, 3-COCH2C2 and I-COCH2CI2) 2.80 (t, J=7 Hz, 2H, 3-COC2) ,
3.07 (t, J=7 Hz, 2H, I-COCH2) and 8.23 (d, J=6 Hz, 1H, C6-H);
UV max (CH3CN) 262 nm (e=l. 051xl04) .
1.3-bis-Valery1-5-fluorouracil (16)
Extraction of the residue with hot low-boiling petroleum
ether gave a cloudy solution which cleared upon cooling to
0 C and produced a brown resinous sediment. The mixture was
allowed to warm to room temperature, and the clear solution
was decanted from the sediment. The solution was cooled
again to 0 C and crystallization gave 2.24 g of 16 (75%) :
mp 47-8 C; IR (KBr) 1685, 1710, 1735, and 1795 cm'1 (C=0);
1H NMR (CDCI3) 5 0.95 (t, J=7 Hz, 6H, 3-CH3 and I-CH3) ,


81
delivery purposes than the 1-alkyloxycarbonyl derivatives
because of their rapid hydrolysis and higher skin
accumulation values. The partitioning-based method for
determining aqueous solubilities of chemically unstable
prodrugs was used to estimate the aqueous solubilities of the
1-acyl derivatives since this could not be accomplished by
other methods.


Cumulative Amount (pmol)
47
Figure 2-3. Plots of cumulative amount of total 5FU species
that diffused (Umol) versus time (h) for
compounds 4, 5, 6, and 5FU.


114
The suspensions were filtered through 0.45 (im nylon filters,
and the saturated solutions were diluted in acetonitrile and
analyzed by UV spectroscopy. Solubilities were calculated
using Beer's Law as previously described.
Partition Coefficients
The partitioning-based method for determining aqueous
solubility utilized the saturated IPM solutions from the
lipid solubility study. For most compounds, equal volumes
(1 mL) of saturated IPM solution and 0.05 M acetate buffer
(pH=4.0) were used. The use of equal or near-equal phase
volumes is known to facilitate rapid equilibrium.96 The two
phases were mixed thoroughly for ten seconds and allowed to
separate for 60 seconds. A preliminary study showed that
there was virtually no difference in partition coefficient
(PC) values when partitioning was carried out for 10, 20, or
30 seconds (see Chapter 3). The IPM layers were diluted in
acetonitrile and analyzed by UV spectroscopy. The IPM-buffer
partition coefficients were calculated as follows:
PC = Aafter/ ^beforeAafter) '^aq/Vipm (2)
where Aafter is the absorbance from the IPM layer after
partitioning, Abefore is the absorbance from the IPM layer
before partitioning, V^q is the volume of the aqueous phase,
and Vjpm is the volume of the IPM phase. Estimated aqueous
solubilities (S*q) were calculated from the IPM solubility


74
formaldehyde and the rate is concentration dependent. In
Figure 3-3, the pseudo-first-order rate constants are plotted
against formaldehyde concentration. The slope of the linear
plot gives the catalytic rate constant (kcat=0.375) for
formaldehyde catalysis of compound 7. Hydrolysis rates for
the 1-acyl derivatives have previously been shown to be
independent of pH at acidic pH values.26 Therefore,
formaldehyde catalysis is likely due to general base rather
than general acid catalysis.
Skin penetration data from the diffusion cells are
plotted as cumulative amount of total 5F0 species that
diffused (Umol) versus time (h). In Figure 3-4, results for
l-acetyl-5FU (7), l-propionyl-5FU (8), and l-butyryl-5FU (9)
are compared to 5FU itself. In Figure 3-5, results for
l-valeryl-5FD (10), l-hexanoyl-5FU (11), and l-octanoyl-5FU
(12) are compared to 5FU. Error bars correspond to the
standard deviation from the mean for n=3 values.
Fluxes (J), lag times (ti,) and skin accumulation (SA)
values for each compound are reported in Table 3-6. Lag time
refers to the intersection of the linear, or "steady-state,"
region of each graph with the time (x) axis, and it is the
time required for establishing a uniform concentration
gradient within the skin.65 The percentages of total 5FD


KEY TO ABBREVIATIONS
bs
broad singlet
CDCI3
chloroform-d
(CD3) 2SO
dimethylsulfoxide-d6
CH3CN
acetonitrile
d
doublet
dec
decomposition
dist t
distorted triplet
m
multiplet
q
quartet
Rf
retention factor
t
triplet
x


56
crystallized from an appropriate solvent or solvent
combination.
l-Acetyl-5-fluorouracil ill
Crystallization from dichloromethane gave 0.98 g of 7
(57%) : mp 129-30 C (lit.27 mp 126-7 C) ; IR (KBr) 1670, 1695,
1725, and 1770 cm"1 and 8.23 (d, J=7 Hz, 1H, C6-fl> ; UVmax (CH3CN) 261 nm
(E=1.125x104) .
l-Propionyl-5-fluorouracil (8)
Crystallization from dichloromethane/hexane gave 1.32 g
of 8 (71%); mp 130-1 C (lit.27 mp 124-5 C) ; IR (KBr) 1695,
1710, and 1740 cm'1 (C=0); 7H NMR (CDCI3) 8 1.25 (t, J=7 Hz,
3H, CH3), 3.14 (q, J=7 Hz, 2H, COCH2) / and 8.27 (d, J=7 Hz,
1H, C6-H) ; UVmax (CH3CN) 261 nm (6=1.141xl04) .
Anal. Calc, for C7H7FN2O3: C, 45.17; H, 3.79; N, 15.05.
Found: C, 45.26; H, 3.83; N, 14.97.
l-Butvrvl-5-fluorouracil (9)
Crystallization from dichlormethane/hexane gave 0.86 g
of 9 (43%): mp 145-6 C; IR (KBr) 1690, 1710, and 1740 cm'1
(C=0); XH NMR (CDCI3) 8 1.01 (t, J=7 Hz, 3H, CH3) 1.77 (m,
2H, COCH2CH2), 3.09 (t, J=7 Hz, 2H, COCH2) and 8.25 (d,
J=6 Hz, 1H, C6-fi) ; UV max (CH3CN) 261 nm (6=1.168xl04) .
Anal. Calc, for C8H9FN2O3: C, 48.00; H, 4.53; N, 14.00.
Found: C, 48.12; H, 4.58; N, 13.91.


107
at least twelve hours during which the formulations were in
contact with the skins, the 1,3-bis-acyl derivatives were
found to be intact with no evidence of formation of the
corresponding 3-acyl derivatives or 5FU.
Snmma ry
The 1,3-bis-acyl derivatives of 5FU exhibited decreased
melting points and greatly increased lipid solubilities when
compared to 5F. Aqueous solubility was at best an order of
magnitude less than 5FU and was highest for 1,3-bis-acetyl-
5FU (13). Flux was also maximized for compound 13 which
again shows the importance of aqueous solubility for
optimizing skin penetration within a homologous series of
more lipid-soluble prodrugs. Biphasic solubility data
appears to be useful for series to series comparisons as
well. The highly lipophilic 1,3-bis-acyl derivatives were
less effective at delivering 5FU transdermally than the less
lipid-soluble, but more aqueous-soluble series such as the
1-alkyloxycarbonyl and 1-acyl series. The partitioning-based
method for determining aqueous solubilities of chemically
unstable prodrugs was used to estimate the aqueous
solubilities of two of the 1,3-bis-acyl derivatives since
this could not be accomplished by other methods. The
difference in the stabilities of the N1- and N3-acyl groups of
the 1,3-bis-acyl derivatives has been rationalized based on
an X-ray crystallography study which showed that the N3-acyl


Copyright
by
Howard D.
1991
Beall


123
H
Figure 5-1. Possible scheme for thermal decomposition of
3-acetyl-5FU to 5FU.


25
Materials and Methods
Synthes js
Melting points (mp) were determined with a Thomas-Hoover
capillary melting point apparatus and are uncorrected.
Elemental microanalyses were obtained for all novel compounds
through Atlantic Microlab, Incorporated in Norcross, Georgia.
Proton nuclear magnetic resonance (H NMR) spectra were
obtained at 90 MHz on a Varian EM-390 spectrometer. Chemical
shifts (8) are reported in parts per million (ppm) from the
internal standard, tetramethylsilane (TMS). Coupling
constants (J) are expressed in cycles per second (Hz).
Infrared (IR) spectra were recorded with a Perkin-Elmer 1420
spectrophotometer and absorbances are reported in cm-1.
Ultraviolet (UV) spectra were obtained with a Cary 210 or
Shimadzu UV-265 spectrophotometer. Maximum absorbances are
reported in nm along with the molar absorptivities (E) in
L/mol.
l-Alkvloxvcarbonvl-5-fluorouracil (general procedure)
To 0.66 g (0.01 mol) of 85% potassium hydroxide
dissolved in methanol (20-50 mL) was added 1.33 g of 5FU
(0.0101 mol). Slightly more than an equivalent of 5FU was
used to prevent excess base from catalyzing the aldol
condensation of acetone. When present, condensation products
complicated product isolation. The methanol suspension was


150
Hydrolysis of the 3-acyl derivatives was biexponential.
A reversible reaction involving an O-acyl intermediate was
proposed for the initial phase followed by terminal phase
hydrolysis of the 3-acyl derivative, an O-acyl intermediate,
or both to 5FU. Formation of nearly equal amounts of 1- and
3-acetyloxymethyl-5FU during hydrolysis of 3-acetyl-5FU in
buffer with formaldehyde suggested that the intermediate may
be the 02-acetyl derivative.
An 02-acyl intermediate was also proposed to explain the
thermal decomposition and rearrangement of the 3-acyl
derivatives to form N2-acylated compounds. Ketene formation
was proposed to account for the formation of 5FU from the
3-acyl derivatives (and possibly the 1-acyl derivatives)
during their thermal decomposition.
Skin penetration was 1.2 to nearly 40 times greater for
the prodrugs than for 5FU itself. The highest flux was
recorded for l-acetyl-5F0,whereas 3-propionyl-5FU,
l-ethyloxycarbonyl-5FU, and 1,3-bis-acetyl-5FU exhibited the
highest rates of delivery of 5FU for their respective series.
Without exception, the highest fluxes were obtained for the
most aqueous soluble derivative in each series. The skin
penetration results demonstrate the importance of biphasic
solubility for achieving optimal transdermal delivery of 5FU
from a homologous series of prodrugs.
Intact prodrug content in the receptor phases
constituted over 40% of total 5F species that had diffused
for the 1-alkyloxycarbonyl and the 3-acyl derivatives with


101
3-acyl derivatives compared to the 1,3-bis-acyl derivatives
allows the rates of hydrolysis for the N1-acyl groups to be
followed for three to four half-lives before detectable
hydrolysis of the N3-acyl groups is observed. The hydrolysis
of the 3-acyl derivatives will be discussed in Chapter 5.
Hydrolysis of compound 13 is clearly faster in the
presence of formaldehyde indicating that general base
catalysis by formaldehyde hydrate may be involved in this
series as well (see Chapter 3). Complete pH-rate profiles
for selected acyl derivatives of 5FU can be found in the work
of Buur and Bundgaard.26
Skin Penetration
Skin penetration data from the diffusion cells are
plotted as cumulative amount of total 5FU species that
diffused (|lmol) versus time (h) In Figure 4-3, results for
1,3-bis-acetyl-5FU (13) and 1,3-bis-propionyl-5FD (14) are
compared to 5FU itself. In Figure 4-4, results for 1,3-bis-
butyryl-5FU (15) and 1,3-bis-valeryl-5FU (16) are compared to
5FO. Error bars correspond to the standard deviation from
the mean for n=3 values.
Fluxes (J), lag times (tx,) and skin accumulation (SA)
values for each compound are reported in Table 4-4. The
highest flux obtained for the 1,3-bis-acyl series is for
1,3-bis-acetyl-5FU (13) with nearly an order of magnitude
improvement over 5FU. Overall, the 1,3-bis-acyl derivatives


TABLE OF CONTENTS
ACKNOWLEDGEMENTS iii
LIST OF TABLES vi
LIST OF FIGURES viii
KEY TO ABBREVIATIONS ix
ABSTRACT x
INTRODUCTION 1
Fluorouracil 1
Fluorouracil Derivatives 2
Fluorouracil Metabolism 3
Structure of the Skin 6
Mechanisms of Transdermal Penetration 8
Passive Diffusion 10
Enhancement of Skin Penetration 11
Prodrugs and Skin Penetration 13
Dermal versus Transdermal Delivery 16
Cutaneous Metabolism 17
Hairless Mouse Skin as a Model Membrane 18
Research Proposal 20
1-ALKYLOXYCARBONYL DERIVATIVES 23
Introduction 23
Materials and Methods 25
Results and Discussion 35
Summary 51
1-ACYL DERIVATIVES 53
Introduction 53
Materials and Methods 53
Results and Discussion 64
Summary 80
1,3-BIS-ACYL DERIVATIVES 82
Introduction 82
Materials and Methods 84
Results and Discussion 93
Summary 107
iv


91
dorsal sections were mounted in the diffusion cells. The
dermal sides of the skins were placed in contact with
receptor phase which contained 0.05 M phosphate buffer
(pH=7.1, 1=0.12) with 0.11% formaldehyde as a preservative.
The effectiveness of formaldehyde for this purpose has
recently been documented.89 The receptor phases were stirred
continuously and kept at constant temperature (32 C) by a
circulating water bath. A preapplication period of 48 hours
was established to uniformly condition the skins and to
remove water-soluble UV-absorbing materials. The receptor
phases were changed three times during this period, and
control experiments from earlier studies have shown that this
procedure effectively removes those materials.97 The
epidermal sides of the skins were exposed to the air and were
left untreated during this period.
After the preapplication period, 0.5 mL aliquots from
suspensions of the prodrugs in IPM were applied to the
epidermal sides of the skins. The IPM suspensions were
stirred at 221 C for 48 hours prior to application to
ensure that saturation was attained. Total concentrations of
the IPM suspensions ranged from 0.5 M to 2.0 M with enough
excess solid present to maintain saturation for the duration
of the application period (see below). Each drug-vehicle
combination was run in triplicate.
Samples were taken from the receptor phases at 4, 8, 12,
21, 24, 27, 30, 33, 36, 45, and 48 hours after donor phase
application. The receptor phases were changed following


132
-9.0
O
c
-10.0 .
-11.0
1 00
Plain Buffer
t 3.6e-4 M Formaldehyde
I 3.6e-3 M Formaldehyde
3.6e-2 M Formaldehyde
0 O


I
200
time (min)
300
400
Figure 5-5. Plots of ln(C) versus time (min) for hydrolysis
of 3-acetyl-5FU in 0.05 M phosphate buffer (pH=7.1, 1=0.12)
with (n=2) and without (n=3) formaldehyde at 32 C.


20
Another difference between human and hairless mouse skin
could be significant in prodrug design. Rat and mouse skins
apparently have higher levels of enzymatic activity for drug
metabolism than human skin. Esterase activity, specifically,
appears to be low in human skin.54
Research. Proposal
The overall goal of the present research is to develop
prodrug derivatives of 5FU with solubility characteristics
and skin penetration properties superior to 5FU itself. The
derivatives should be stable in formulation, but should
readily convert to 5FU in the skin since the purpose of
topical 5F therapy is to deliver 5FU dermally, not
transdermally. In pursuit of this goal, it is hoped that
experimental data will be obtained which supports a
solubility-based method for designing prodrugs with optimized
topical delivery characteristics. The objectives for meeting
this goal are to:
1) synthesize several homologous series of acylated
derivatives of 5FD,
2) verify structures by melting point, elemental
analysis (novel compounds) and standard spectral
techniques,
3) determine physical chemical properties such as lipid
and aqueous solubilities, partition coefficients, and
rates of hydrolysis, and
4) determine skin penetration parameters and quantitate
skin accumulation using hairless mouse skin as the model
membrane.


98
Table 4-2. Melting points (MP), lipid solubilities (Sjpm) / and
aqueous solubilities (Saq) for 1,3-bis-acyl derivatives.
Compound
MP
(C)
sIPMa
(mM)
Saq5
(mM)
Saqc
(mM)
5FU
280-2
0.049
96
-
13
112-3
26
-
9.0
14
100-1
72
-
3.3
15
48-9
625
-
-
16
47-8
1180

~
aStandard
deviation from
the means
were within 4%
for IPM
solubilities.
bSolubility determined by direct method.
cStandard deviations from the mean were within 4% for
experimental values used to calculate partition coefficients
and estimated aqueous solubilities.


110
Table 5-1. Structures of 3-acyl derivatives.
R
H
Compound R
3-acetyl-5FU (17) -CH3
3-propionyl-5FU (18) -CH2CH3
3-butyryl-5FO (19) -(CH2)2CH3
3-valeryl-5FU (20) -(CH2)3CH3


BIOREVERSIBLE DERIVATIVES OF 5-FLUOROURACIL (5FU):
IMPROVING DERMAL AND TRANSDERMAL DELIVERY
WITH PRODRUGS
BY
HOWARD D. BEALL
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1991

Copyright
by
Howard D.
1991
Beall

ACKNOWLEDGEMENTS
I would like to thank the members of my supervisory
committee, Dr. Kenneth Sloan, Dr Margaret James, Dr. Koppaka
Rao, Dr. Richard Prankerd, and Dr. John Zoltewicz for their
guidance and expert advice over the past four years. My
sincerest thanks go to my research advisor and committee
chairman, Dr. Sloan, for sharing his enthusiasm for teaching
and science. I am especially grateful for his patience and
understanding during my seemingly endless questions and
interruptions.
I would also like to acknowledge the enthusiastic
support of Dr. Noel Meltzer of Hoffmann-La Roche. This
project was partially funded by a grant from Hoffmann-La
Roche.
My special thanks go to my parents for their love and
support throughout my life and to my two-year-old son,
Michael, who could make me laugh when it was the last thing I
felt like doing. But most of all, I want to thank my wife,
Donna, whose love, support, and countless sacrifices made my
return to school and the completion of this project possible.

TABLE OF CONTENTS
ACKNOWLEDGEMENTS iii
LIST OF TABLES vi
LIST OF FIGURES viii
KEY TO ABBREVIATIONS ix
ABSTRACT x
INTRODUCTION 1
Fluorouracil 1
Fluorouracil Derivatives 2
Fluorouracil Metabolism 3
Structure of the Skin 6
Mechanisms of Transdermal Penetration 8
Passive Diffusion 10
Enhancement of Skin Penetration 11
Prodrugs and Skin Penetration 13
Dermal versus Transdermal Delivery 16
Cutaneous Metabolism 17
Hairless Mouse Skin as a Model Membrane 18
Research Proposal 20
1-ALKYLOXYCARBONYL DERIVATIVES 23
Introduction 23
Materials and Methods 25
Results and Discussion 35
Summary 51
1-ACYL DERIVATIVES 53
Introduction 53
Materials and Methods 53
Results and Discussion 64
Summary 80
1,3-BIS-ACYL DERIVATIVES 82
Introduction 82
Materials and Methods 84
Results and Discussion 93
Summary 107
iv

3-ACYL DERIVATIVES
109
Introduction 109
Materials and Methods Ill
Results and Discussion 120
Summary 146
SUMMARY AND CONCLUSIONS 148
LIST OF REFERENCES 153
BIOGRAPHICAL SKETCH 160
v

LIST OF TABLES
Table 1-1. Structures of 5FU and prodrug derivatives
of 5FU 21
Table 2-1. Structures of 1-alkyloxycarbonyl
derivatives 24
Table 2-2. Melting points (MP), lipid solubilities
(Sjpm) , and aqueous solubilities (Saq) for
1-alkyloxycarbonyl derivatives 39
Table 2-3. Solubility ratios (SR), partition
coefficients (PC) , and hydrophobicity parameters (It)
for 1-alkyloxycarbonyl derivatives 40
Table 2-4. Pseudo-first-order rate constants (k) and
half-lives (ti/2) for hydrolysis of 1-methyloxy-
carbonyl-5FU in 0.05 M phosphate buffer (pH=7.1,
1=0.12) with and without formaldehyde at 32 °C 44
Table 2-5. Fluxes (J) , lag times (ti.) , and skin
accumulation (SA) values for 1-alkyloxycarbonyl
derivatives 48
Table 2-6. Second application fluxes (J) and lag times
(tj,) for 1-alkyloxycarbonyl derivatives 49
Table 3-1. Structures of 1-acyl derivatives 54
Table 3-2. Melting points (MP), lipid solubilities
(Sjpm) , and aqueous solubilities (Saq) for 1-acyl
derivatives 68
Table 3-3. Partition coefficients (PC) and hydropho¬
bicity parameters (It) for 1-acyl derivatives 69
Table 3-4. Pseudo-first-order rate constants (k) and
half-lives (ti/2) for hydrolysis of 1-acyl derivatives
in 0.05 M phosphate buffer (pH=7.1, 1=0.12) at 32 °C.... 71
Table 3-5. Pseudo-first-order rate constants (k) and
half-lives (11/2) for hydrolysis of l-acetyl-5FU in
0.05 M phosphate buffer (pH=7.1, 1=0.12) with and
without formaldehyde at 32 °C 72
vi

Table 3-6. Fluxes (J), lag times (ti.), and skin
accumulation (SA) values for 1-acyl derivatives 77
Table 3-7. Second application fluxes (J) and lag times
(ti) for 1-acyl derivatives 78
Table 4-1. Structures of 1,3-bis-acyl derivatives 83
Table 4-2. Melting points (MP), lipid solubilities
(Sipm) » and aqueous solubilities (Saq) for 1,3-bis-
acyl derivatives 98
Table 4-3. Pseudo-first-order rate constants (k) and
half-lives (tj./2) for hydrolysis of 1,3-bis-acetyl-5FU
in 0.05 M phosphate buffer (pH=7.1, 1=0.12) with and
without formaldehyde at 32 °C 100
Table 4-4. Fluxes (J), lag times (tl), and skin
accumulation (SA) values for 1,3-bis-acyl
derivatives 104
Table 4-5. Second application fluxes (J) and lag times
(tl) for 1,3-bis-acyl derivatives 105
Table 5-1. Structures of 3-acyl derivatives 110
Table 5-2. Melting points (MP), lipid solubilities
(Sipm) / and aqueous solubilities (Saq) for 3-acyl
derivatives 127
Table 5-3. Solubility ratios (SR), partition
coefficients (PC) , and hydrophobicity parameters (7C)
for 3-acyl derivatives 128
Table 5-4. Pseudo-first-order rate constants (k) and
half-lives (ti/2) for hydrolysis of 3-acyl derivatives
in 0.05 M phosphate buffer (pH=7.1, 1=0.12) with and
without formaldehyde at 32 °C 134
Table 5-5. Reaction products formed during hydrolysis of
3-acetyl-5FU in 0.05 M phosphate buffer (pH=7.1,
1=0.12) with formaldehyde at 32 °C 135
Table 5-6. Fluxes (J) , lag times (ti,) , and skin
accumulation (SA) values for 3-acyl derivatives 142
Table 5-7. Second application fluxes (J) and lag times
(tl) for 3-acyl derivatives 143
vii

LIST OF FIGURES
Figure 2-1. Plots of ln(C) versus time (min) for
hydrolysis of l-methyloxycarbonyl-5FU in 0.05 M
phosphate buffer (pH=7.1, 1=0.12) with and without
formaldehyde at 32 °C 43
Figure 2-2. Plots of cumulative amount of total 5FU
species that diffused 4lmol) versus time (h) for
compounds 1, 2, 3, and 5FU 46
Figure 2-3. Plots of cumulative amount of total 5FU
species that diffused Oimol) versus time (h) for
compounds 4, 5, 6, and 5FU 47
Figure 3-1. X-ray structure of l-acetyl-5FU (unprimed). ... 66
Figure 3-2. X-ray structure of l-acetyl-5FU (primed) 67
Figure 3-3. Plot of pseudo-first-order rate constant (k)
versus formaldehyde concentration (M) for hydrolysis
of l-acetyl-5FU in 0.05 M phosphate buffer (pH=7.1,
1=0.12) at 32 °C 73
Figure 3-4. Plots of cumulative amount of total 5FU
species that diffused (|imol) versus time (h) for
compounds 7, 8, 9, and 5FU 75
Figure 3-5. Plots of cumulative amount of total 5FU
species that diffused (|imol) versus time (h) for
compounds 10, 11, 12, and 5FU 76
Figure 4-1. X-ray structure of 1,3-bis-acetyl-5FU 95
Figure 4-2. Plots of ln(At-A=) versus time (min) for
hydrolysis of 1,3-bis-acetyl-5FU in 0.05 M phosphate
buffer (pH=7.1, 1=0.12) with and without formaldehyde
at 32 °C 99
Figure 4-3. Plots of cumulative amount of total 5FU
species that diffused (|imol) versus time (h) for
compounds 13, 14, and 5FU 102
Figure 4-4. Plots of cumulative amount of total 5FU
species that diffused (|imol) versus time (h) for
compounds 15, 16, and 5FU 103
viii

Figure 5-1. Possible scheme for thermal decomposition of
3-acetyl-5FU 123
Figure 5-2. Possible scheme for thermal intramolecular
rearrangement for 3-acetyl-5FU to l-acetyl-5FU 124
Figure 5-3. Plot of In (At-Aod versus time (min) for
hydrolysis of 3-acetyl-5FU in 0.05 M phosphate buffer
(pH-7.1, 1=0.12) at 32 °C 130
Figure 5-4. Plot of ln(At-A°°) versus time (min) for
hydrolysis of 3-propionyl-5FU in 0.05 M phosphate
buffer (pH=7.1, 1=0.12) at 32 °C 131
Figure 5-5. Plots of ln(C) versus time (min) for
hydrolysis of 3-acetyl-5FU in 0.05 M phosphate buffer
(pH=7.1, 1=0.12) with (n=2) and without (n=3)
formaldehyde at 32 °C 132
Figure 5-6. Plots of ln(C) versus time (min) for
hydrolysis of 3-acetyl-5FO in 0.05 M phosphate buffer
(pH=7.1, 1=0.12) at 32 “C using actual concentration
(Ct) and concentration corrected for secondary
degradation (CCorr) 133
Figure 5-7. Possible scheme for reaction of 3-acetyl-5FU
with formaldehyde to form l-acetyloxymethyl-5FU and
3-acetyloxymethyl-5FU 136
Figure 5-8. Plots of cumulative amount of total 5FU
species that diffused (nmol) versus time (h) for
compounds 17, 18, and 5FU 140
Figure 5-9. Plots of cumulative amount of total 5FU
species that diffused (|lmol) versus time (h) for
compounds 19, 20, and 5FU 141
ix

KEY TO ABBREVIATIONS
bs
broad singlet
CDCI3
chloroform-d
(CD3) 2SO
dimethylsulfoxide-d6
CH3CN
acetonitrile
d
doublet
dec
decomposition
dist t
distorted triplet
m
multiplet
q
quartet
Rf
retention factor
t
triplet
x

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
BIOREVERSIBLE DERIVATIVES OF 5-FLU0R0URACIL (5FU) :
IMPROVING DERMAL AND TRANSDERMAL DELIVERY
WITH PRODRUGS
By
Howard D. Beall
December, 1991
Chairman: Kenneth B. Sloan
Major Department: Medicinal Chemistry
Fluorouracil (5FU) is an antineoplastic agent used
topically for treatment of actinic keratoses, superficial
basal cell carcinomas, psoriasis, and other skin conditions,
but treatment is often ineffective due to its poor skin
penetration properties. Four homologous series of
bioreversible derivatives (prodrugs) of 5FU were synthesized
characterized, and their ability to penetrate through
(transdermal) and accumulate in (dermal) the skin was
evaluated. Six 1-alkyloxycarbonyl, six 1-acyl, four 1,3-bis
acyl, and four 3-acyl derivatives were studied.
Lipid solubility was a minimum of 40 times greater for
the derivatives than for 5FU, and aqueous solubility was
actually higher than 5FU for five derivatives. The
xi

partitioning-based method for determining aqueous solubility
gave reliable values for relative solubility in each
homologous series, and overall, the values correlated well
with conventionally determined solubilities. This was the
first known successful attempt to obtain aqueous solubilities
for chemically unstable prodrugs (e.g., 1-acyl and 1,3-bis-
acyl derivatives).
Hydrolysis rates for N^-substituents followed pseudo-
first-order kinetics, but plots for N3-acyl hydrolysis were
biexponential. An 02-acyl intermediate was proposed to
account for the unusual hydrolysis and thermal decomposition
of the N3-acyl group. X-ray crystal analysis showed that the
N3-acyl group was oriented perpendicular to the 5FU ring and
was sterically and electronically hindered from nucleophilic
attack.
Skin penetration was 1.2 to nearly 40 times greater for
the prodrugs than for 5FU. The highest flux was recorded for
l-acetyl-5FD, whereas 3-propionyl-5FU, 1-ethyloxycarbonyl-
5FÜ, and 1,3-bis-acetyl-5FU exhibited the highest rates of
delivery for their respective series. These derivatives were
also the most aqueous soluble members of each series. This
demonstrates the importance of biphasic solubility for
achieving optimal transdermal delivery of 5FU from a
homologous series of prodrugs.
Skin accumulation was highest for l-acetyl-5FU and
l-propionyl-5FD <>18 times more than 5FU) . Rapid hydrolysis
of the 1-acyl series upon partitioning into the skin may
xii

cause retention of highly polar 5FU in the more lipophilic
regions of the epidermis.
Skin damage was only slightly increased with some
prodrugs, all derivatives were stable in formulation, and
overall, l-acetyl-5FD was the best candidate for improving
dermal or transdermal delivery of 5FU.

CHAPTER 1
INTRODUCTION
Fluorouracil
Fluorouracil (5FU) is one of the most widely used1 and
studied2 anticancer agents. It is used in the palliative
treatment of solid tumors and is most effective in
combination with other antineoplastic agents2 and in
conjunction with radiation therapy.2 It has been used to
treat carcinoma of the colon, rectum, breast, and stomach
and, with less effectiveness, carcinoma of the ovary, cervix,
urinary bladder, liver, and pancreas.4
Fluorouracil (5FU) is used as a single agent in topical
preparations and is effective in treating actinic (solar)
keratoses,5-8 superficial basal cell carcinomas,9-10 and
psoriasis.11'12 Topical 5FU has also been used to treat a
variety of other precancerous conditions, malignant and
benign tumors, and dermatoses.14 Commercial 5FU creams and
solutions may be adequate for treating lesions on the face,
but other areas, especially the forearms and hands, respond
poorly probably due to lack of penetration by 5FU.15 Some
conditions, such as psoriasis, respond to topical 5FU therapy
when the drug is applied under an occlusive dressing.12
1

2
However, this is neither the most convenient nor comfortable
treatment option.
The toxicity of 5FD is related to its effect on rapidly
proliferating host cells especially those of the bone marrow
and gastrointestinal lining. Toxicity following systemic
therapy is common while adverse effects from topical therapy
appear to be minimal, other than those associated with the
local inflammatory reactions usually necessary for a
therapeutic response.4
Fluorouracil Derivatives
Since 5FU was first synthesized16 and tested17 as an
antitumor antimetabolite in 1957, numerous attempts have been
made to improve its efficacy and reduce its toxicity.
Derivatives of 5FD and its nucleosides, 5-fluorouridine (FUR)
and 5-fluoro-2'-deoxyuridine (FdUR), have been developed with
this in mind. The majority of these derivatives have been
designed with the expectation that 5FU will be released in
vivo. so they are essentially prodrugs.
A prodrug has been defined as "an agent which must
undergo chemical or enzymatic transformation to the active or
parent drug after administration, so that the metabolic
product or parent drug can subsequently exhibit the desired
pharmacological response" (p. 1275).18 Prodrugs, or
bioreversible derivatives, of 5FU have included hydroxy¬
methyl,19 alkyloxyalkyl,20 acyloxyalkyl,21-24 acyl,25-28

3
alkyloxycarbonyl,25'29-30 alkyloxycarbonyloxyalkyl,31-33 alkyl-
carbamoyl, 34-37 aminoalkyl (Mannich base),38-39 phthalid-
yl, 40-41 anc¿ derivatives containing sulfur rather than oxygen
in some of the above functional groups.42-43 Polymer44-45 and
peptide46 derivatives have also been proposed as sources of
5FU in vivo. Currently, at least two bioreversible
derivatives of 5FU, 1-(tetrahydrofuran-2-yl)-5-fluorouracil47
and l-hexylcarbamoyl-5-fluorouracil,48 have been marketed in
Japan.
The above is just a small sample of the synthetic
literature involving 5FU. In addition, a large number of
studies have been devoted to synthesizing analogs of FUR and
FdUR containing an N1-glycosidic linkage. FdUR, or
floxuridine, is available in the United States, but is only
approved for intraarterial infusion.4 It appears to offer no
advantage over intravenous 5FU, and this therapy is both
hazardous and expensive.3 The nucleosides and their analogs
are converted to 5FU in vivo by a pyrimidine phosphorylase.
Since phosphorylase activity is higher in tumor tissue, it is
believed that development of compounds that are good
substrates for phosphorylase may lead to greater tumor
specificity and decreased toxicity.49
Fluorouracil Metabolism
Fluorouracil (5FU) undergoes extensive catabolic and
anabolic metabolism. Approximately 15% of a single

4
intravenous dose of 5FU is excreted unchanged in the urine
within six hours, and 90% of that is detected within the
first hour.4 Approximately 80% of the dose is metabolized by
the liver and extrahepatic tissues.1 The principal
catabolite of 5FU is a-fluoro-p-alanine (FBAL), which
accounts for over 95% of the catabolic products found in the
urine. Bile acid conjugates of FBAL represent the major
biliary catabolites of 5FU.1
A number of anabolic pathways have been characterized
for 5FU, and one or more of them may be responsible for the
activation and subsequent cytotoxicity of the drug. The most
direct activation pathway utilizes orotidine monophosphate
phosphoribosyl transferase (OMPT) to form 5-fluorouridine-5'-
monophosphate (FUMP) from 5FU and 5-phosphoribosyl-l-pyro-
phosphate. FUMP can also be generated in two steps with
uridine phosphorylase and uridine kinase. With two more
kinases 5-fluorouridine-5'-triphosphate (FUTP) is formed
which can be incorporated into ribonucleic acid (RNA) leading
to RNA dysfunction.1 Some cell lines appear to favor the
OMPT pathway, while others favor the uridine phosphorylase
pathway, and the FUTP formed from each is added to different
fractions of RNA.50
Some of the 5-fluorouridine-5'-diphosphate (FUDP) that
is generated above as an intermediate can become a substrate
for ribonucleotide reductase. The resulting deoxynucleotide,
5-fluoro-2'-deoxyuridine-5'-diphosphate (FdUDP), can form
5-fluoro-2'-deoxyuridine-5'-triphosphate (FdUTP) using

5
another kinase. At this point FdUTP can be incorporated into
deoxyribonucleic acid (DNA), but the contribution to 5FU
cytotoxicity by this mechanism is unclear. DNA protective
enzymes, dUTPase and uracil-DNA glycosylase, which are
responsible for keeping uracil residues out of DNA, target
5FU residues as well.50
A diphosphohydrolase converts FdUTP to 5-fluoro-2'-
deoxyuridine-5'-monophosphate (FdUMP). FdUMP can also be
formed from 5FU by thymidine phosphorylase and thymidine
kinase, although this may be quantitatively less important
than other activation steps.1 FdUMP binds covalently to
thymidylate synthase and its cofactor, 5,10-methylene
tetrahydrofoíate, thereby preventing synthesis of thymidine¬
s' -monophosphate (dTMP) and subsequently, thymidine-5'-
triphosphate (dTTP), an essential component in DNA synthesis.
FdUMP has a greater affinity for thymidylate synthase than
the normal substrate, 21-deoxyuridine-5'-monophosphate
(dUMP).1
Either inhibition of thymidylate synthase by FdUMP or
incorporation of FUTP into RNA can lead to cell death. The
mechanism that is most important depends on the cell line
being studied.51
Fluorouracil (5FU) may also affect glycoprotein and
glycolipid metabolism, since it is known that FUDP sugars can
be formed.50 The possibility that membrane effects from
altered glycoprotein synthesis could lead to cytotoxicity has
been noted, but has not been well studied.1

6
Structure of the Skin
Fluorouracil (5FU) is typical of polar, high-melting,
heterocyclic compounds which exhibit poor skin permeability.3®
Since the goal of this project is to improve dermal and
transdermal delivery of 5FU, it is necessary to examine the
structure of the barrier to be penetrated, the skin.
The skin is composed of two major tissue layers, the
epidermis and dermis. The epidermis is a continuous, elastic
sheet that is interrupted only by glandular pores and hair
follicles. It consists of four definable sublayers and
averages 75 to 150 |lm in thickness. The basal cell layer,
which borders the dermis, is a single layer of keratinocytes.
This is the germinal layer, and all epidermal cells are
initially formed here before moving outward to the next
layer, the stratum spinosum. The basal cells are cuboidal or
columnar in shape, while the stratum spinosum cells are
polyhedral. By the time the cells have migrated to the next
layer, the stratum granulosum, they have become flattened and
contain characteristic keratohyalin granules. The stratum
granulosum marks the transition between nucleated cells and
the anucleated stratum corneum. The stratum corneum cells in
the last layer are highly keratinized, markedly flattened
cells in which cellular components, such as mitochondria and
ribosomes, have degraded along with the nucleus. Stratum
corneum normally consists of about 15 to 20 layers, but each

7
layer is only about 0.5 Hm in thickness. The outer layers
are continuously desquamated and replaced from below. The
entire transit time from basal layer to desquamation is 26 to
42 days. The cells of the epidermis are attached to each
other by desmosomes, which degrade just prior to
desquamation.52
Below the epidermis lies the dermis which constitutes
the bulk of the approximately 2 mm thickness of human skin.
The two layers are anchored to the basal lamina by various
fibrils and microfibrils. The attachment is enhanced by the
interlocking nature of the junction.52
The dermis consists of two regions, the papillary dermis
and reticular dermis. The papillary dermis is the thinner,
outermost region that is molded against the ridges and
grooves of the epidermis. It contains small, loosely
distributed fibrils and encloses the microcirculatory blood
and lymph vessels. In contrast, the collagen bundles and
elastin fibers of the reticular dermis are more densely
packed, and this region is relatively acellular and
avascular. Collagen is the major component of the dermis,
and it gives the skin its tensile strength. The dermis also
contains nerves, excretory and secretory glands, and hair
follicles.52

Mechanisms of Transdermal Penetration
For most substances, the stratum corneum provides the
rate-limiting barrier to skin penetration.53-55 The stratum
corneum is 75-85% protein (dry weight),54 mostly intracellular
keratin, while the intercellular space is a lipid-enriched
region. During the outward migration of the keratinocytes
through the epidermis, lamellar bodies are synthesized, which
contain lipids, polysaccharides, and hydrolytic enzymes. As
the granular cells prepare to enter the stratum corneum, the
lamellar bodies move to the cell periphery and empty their
contents into the spaces between the cells.56 The corneocytes
and intercellular lipids are arranged in a brick and mortar
fashion,57 and the result is a very dense (1.4 g/cm3 in the
dry state),58 highly efficient moisture barrier.
In order for a substance to penetrate the stratum
corneum, it must either (1) cross through the densely packed
corneocytes, (2) traverse through the intercellular region,
thereby avoiding the keratinized cells, or (3) bypass the
stratum corneum completely by diffusing through shunt
pathways such as sweat ducts and hair follicles.53 The shunt
pathways are not considered to be significant, especially
when steady state diffusion has been attained,59 and these
pathways are generally disregarded in discussions of
penetration mechanisms.

9
Until recently, it was thought that the transcellular
pathway was the primary route through the stratum corneum.
This was based, at least in part, on a gross underestimation
of the volume of the intercellular space.60 The current
belief is that passive diffusion occurs predominantly through
the intercellular channels.61-62 While this seems logical for
lipophilic compounds, this pathway has also been shown to be
the dominant route for mercuric chloride, an ionic compound.
Boddé et al.63 showed that mercuric sulfide could be
precipitated in the stratum corneum after topical application
of mercuric chloride. They found that mercuric cation
accumulates initially in the intercellular spaces throughout
the stratum corneum. Intracellular uptake was observed
later, but only in the apical corneocytes by way of
disintegrating desmosomal attachments. Since the
intercellular lipids are arranged in multiple lamellar
bilayers,64 the mercuric cation, and possibly other ions and
hydrophilic molecules, may diffuse through the interlamellar,
hydrophilic channels that are associated with the polar
headgroups of the lipids.63
While most of the current research is directed toward
understanding the intercellular region, Barry59 cautions that
the transcellular route should not be dismissed as
unimportant, especially in penetration enhancer research.

10
Passive Diffusion
Although the skin is not a homogeneous tissue, it has
the characteristics of a passive diffusion barrier. Fick's
first law of diffusion defines flux as the amount of material
flowing through a unit cross section of a barrier in a unit
time.65 Flux is also proportional to the concentration
gradient, and the first law can be written:
J = -D-dC/dx (1)
where J is the flux, D is the diffusion coefficient of the
penetrant in the barrier, and dC/dx is the concentration
gradient. While the first law is used to describe steady-
state diffusion, Fick's second law describes the change in
concentration with respect to time at a specific location.65
Our interest is in steady-state diffusion, and the second law
will not be discussed further.
The concentration gradient term (dC/dx) from Fick's
first law can be approximated by (C1-C2) /h, where Ci is the
concentration in the barrier on the donor side of the
barrier, while C2 is the concentration in the barrier on the
receptor side, and h is the thickness of the diffusional
barrier. Therefore, the expression becomes
J = D- (C1-C2) /h (2)
For skin penetration studies, "sink" conditions can be
assumed which means that C2=0, and
J = D-Ci/h
(3)

11
Since Ci is essentially the concentration in the first
layer(s) of skin, Ci will now be referred to as Cs, so
J = D-Cs/h (4)
While Cs is not usually a known quantity, it can be
represented instead by the product of the penetrant
concentration in the donor vehicle (Cv) and the
membrane-vehicle partition coefficient (Km = Cs/Cv), and the
expression now becomes
J = D-Km-Cv/h (5)
This expanded form of Fick's first law is often cited and has
been verified by data from numerous skin penetration
studies.66
Another useful measure of skin penetration is the
permeability coefficient (P). This is simply the
concentration-normalized flux which is expressed:
P = J/Cv = D•Km/h (6)
Permeability coefficients are often used when comparing a
single penetrant in a series of vehicles or when studying an
homologous series of penetrants.
Enhancement of Skin Penetration
The driving force for skin penetration is the membrane-
vehicle partition coefficient (Km). If a penetrant is
delivered in a saturated solution in the presence of excess
solid, then the chemical potential, or escaping tendency, is
maximized, and the actual concentration in the vehicle (Cv)

12
is irrelevant. Since the membrane thickness (h) and
diffusion coefficient (D) remain relatively constant (D is
inversely proportional to the cube root of the molar volume
according to the Stokes-Einstein equation), Km is the only
variable that can substantially influence the flux.
There are several general approaches for enhancing skin
penetration. Four of these—occlusion, use of penetration
enhancers, iontophoresis, and sonophoresis—produce their
enhancement effects by changing the barrier properties of the
skin. Occlusion involves covering the application site,
impeding transepiderraal water loss, and increasing the
hydration state of the skin.67 With penetration enhancers,
accelerants, or promoters in topical formulations, the
reversible reduction of barrier resistance in the stratum
corneum is the goal, and ideally, incorporation of the
enhancer into the skin will not result in cell damage.59
Iontophoresis is a technique in which electroosmotic volume
flow from an applied electric field leads to increases in
mass transfer in excess of passive diffusion.68 Sonophoresis
uses ultrasonic frequencies to increase skin penetration.
The other two methods, formulation (without penetration
enhancers) and the use of prodrugs, do not disrupt the
barrier layer. Essentially, the formulation approach
involves changing the penetrants solubility in the vehicle by
changing the vehicle. The effect on flux of this approach is
indeterminate according to equation (5). If the solubility
in the vehicle (Cv) is increased, then the partition

13
coefficient (Km) decreases and vice versa. In a series of
papers,69-72 Sloan and coworkers suggested that a parabolic
relationship exists for log Km and log P when they are
plotted against vehicle polarity. Log Km and log P reach a
minimum where vehicle polarity is equal to penetrant polarity
or in other words, where penetrant solubility in the vehicle
is the greatest. While this relationship might be expected,
a more interesting finding was that in most instances fluxes
are also lower from those vehicles in which the penetrants
are most soluble. While a formulation approach may permit
some improvement in skin penetration values, the maximum
achievable levels appear to be limited.
The final method for enhancing skin penetration is the
prodrug approach. This is the only method in which a
substantial increase in Km can be realized, and according to
equation (5), this should translate into substantially
improved skin penetration. It was this knowledge that led to
adoption of the prodrug method for the current project.
Prodruas and Skin Penetration
The term, prodrug, was defined earlier in this chapter.
Prodrugs, bioreversible derivatives, or latentiated drugs are
simply compounds which undergo biotransformation before
producing their pharmacological effects.72
Generally, prodrugs are designed to overcome some kind
of barrier to a drug's usefulness. These barriers can

14
include (1) premature metabolism prior to reaching the active
site, (2) too rapid absorption and distribution when
prolonged action is needed, (3) toxicity associated with
(a) local irritation or (b) distribution to tissues other
than the target site, (4) poor site specificity leading to
subtherapeutic levels at the target site, and (5) generally
poor physical chemical properties resulting in (a) solubility
problems in the dosage form or (b) poor absorption across
biological membranes such as the blood brain barrier,
gastrointestinal lining, or skin.74 Obviously, it is the
latter problem that is addressed in the current project.
The idea that improving skin penetration is a solubility
problem has been documented in this chapter, and the
importance of biphasic (lipid and aqueous) solubility is well
recognized.7S_7® In a recent review of prodrug approaches
for improving dermal delivery, numerous literature examples
are cited in which both solubility and skin penetration data
are presented.77 The evidence strongly suggests that:
although an increase in lipid solubility due to
transient masking of a polar functional group almost
always results in enhanced dermal delivery of the parent
drug, in order to optimize delivery, it is necessary to
use the members of the (homologous) series (of prodrugs)
that are more water soluble than the parent drug or that
are the more water-soluble member(s) of the series
(p. 68) ,77
Derivatization to improve solubility characteristics of
polar heterocycles with amide and/or imide functional groups
is relatively easy to accomplish. For example, successive
methylation of uracil at the N1- and N3-positions, while

15
obviously improving lipid solubility, also increases water
solubility from 3 mg/mL for uracil to 500 mg/mL for 1,3-di-
methyluracil even though the methyl group is hydrophobic
itself. Melting points are also decreased as the two
intermolecular hydrogen-bonding N-H sites are masked.78 Since
alkyl groups such as methyl are stable and therefore not
bioreversible, they are not candidates to function as prodrug
promoieties. However, many other derivatives, such as those
cited earlier, when linked to the N1- or N8-site of 5FU, do
qualify as promoieties and could potentially give 5FU the
improved solubility characteristics necessary for enhanced
skin penetration.
Of the many studies devoted to making bioreversible
derivatives of 5FU, only a few have been directed at
improving its topical delivery. Mollgaard et al.21 looked at
two 1-acyloxymethyl derivatives of 5FO. One of the compounds
delivered 5FU more readily than 5FU itself through excised
human skin. Both compounds showed greatly increased lipid
solubilities with only slightly reduced water solubilities.
Hydrolysis of these derivatives was attributed to cutaneous
metabolism by hydrolytic enzymes.
Three 1,3-bis-aminomethyl (Mannich base) derivatives of
5FU were prepared by Sloan and coworkers,88-88 and their
topical delivery was studied using hairless mouse skin.
Solubilities in lipid and aqueous phases were substantially
increased as were the rates of delivery through the skin of
5FU from these prodrugs. Due to the instability of these

16
compounds in water, no enzymatic activation was necessary to
release the parent compound. In a later report,79 one of
these prodrugs was compared with a number of 5FU formulations
including four commercially available creams and solutions.
The prodrug outperformed the formulations by a minimum of
four times in terms of 5FU delivered through hairless mouse
skin.
Sasaki et al.37 studied the delivery of 1-alkylcarbamoyl
derivatives of 5FÜ through rat skin. All three derivatives
(butyl, hexyl, and octyl) were more effective in delivering
5FÜ than 5FU itself. The lipid solubilities of the three
derivatives were comparable with each other and much higher
than 5FU. The aqueous solubilities, while less than 5FU,
showed an order of magnitude decrease between each derivative
beginning with the butyl derivative. Interestingly, the best
performing compound was the least lipid-soluble and most
water-soluble derivative, the butyl derivative.
Dermal versus Transdermal. .Delivery
It is important to make a distinction between dermal and
transdermal delivery. Most in vitro skin penetration testing
is done with excised skin to which a drug in a formulation is
applied on the donor side and samples are removed from the
receptor side. These experiments give information on
transdermal rates of delivery.

17
Topical 5FU works on afflicted cells in the epidermal
region of the skin. This is considered dermal delivery.
Transdermal techniques provide valuable information for
dermally targeted drugs since they indicate how effectively
the drugs penetrate the barrier layer of the skin. Other
experiments can be done to augment the transdermal results
such as measuring accumulation of the drug in the skin.
The correlation between transdermal delivery rates and
epidermal uptake was studied by Sloan et al.79'60 Incorpora¬
tion of 3H-deoxyuridine into epidermal DNA of live hairless
mice was quantitated by scintillation counting following
application of various 5FD formulations and a 5FU prodrug.
The prodrug, which had the highest in vitro transdermal flux,
was also the most effective at inhibiting epidermal DNA
synthesis in vivo. The correlation was also good among the
formulations.
Cutaneous Metabolism
When a prodrug is applied topically and targeted for a
dermal site, it is essential that the parent drug is released
before the prodrug leaves the epidermal region. If a prodrug
has a relatively short half-life under physiological
conditions, such as the aforementioned Mannich-base prodrugs
of 5FU, then release of the parent drug will probably occur
either prior to or during its transit through the viable
epidermis. However, if a compound depends on enzymatic

18
rather than chemical activation, then an appropriate
cutaneous enzyme must be present.
The ability of the skin to metabolize drugs and other
foreign compounds is well known.81-82 phase I reactions
(oxidation, reduction, and hydrolysis) and phase II
conjugations are known to occur. Drug-metabolizing enzymes
are distributed in all layers of the skin and the appendages.
Of particular interest in prodrug chemistry is the
presence of nonspecific esterases in the skin.82 The
predictable metabolism of esters by these hydrolytic enzymes
makes this functional group a popular choice for prodrug
synthesis.
Hairless Mouse Skin as a Model Membrane
A number of animal skins have been suggested as model
membranes for skin penetration studies. The two most common
models for in vitro diffusion studies are human and hairless
mouse skin. While the advantages of human skin are obvious,
there are also disadvantages. Human skin can be difficult to
obtain and store,83 it can be expensive,83 and it is known to
have high barrier variability.83-84 Factors such as age,
diet, and disease state may not be well controlled with human
skin.85 Hairless mouse skin, on the other hand, is easily
obtained and prepared, and other factors can ordinarily be
controlled.

19
The importance of hairless mouse skin for in vitro
experimentation has been noted86 despite its generally greater
permeability when compared to human skin. It has been
suggested that this difference may actually be an asset in
that small changes in permeability will be amplified in the
mouse skin model.86
A major criticism of hairless mouse skin involves its
ability to withstand the effects of hydration,87-88 a
necessary condition for controlled in vitro diffusion
experiments. Permeability increases as a function of
hydration time have been attributed to breakdown of the
stratum corneum barrier in these studies. A recent report,89
however, suggested that the absence of an adequate
preservative in the receptor phase may actually be
responsible for breakdown of the skin. Skin penetration data
was collected for delivery of theophylline from a propylene
glycol vehicle following skin hydration periods ranging from
4 to 120 hours. It was found that increased theophylline
flux and loss of barrier function corresponded to the
presence of microbial growth in the receptor phase. When
microbial growth was completely inhibited, fluxes were
essentially constant for all hydration periods.89 Finally,
with regard to hydration, Scheuplein and Ross noted that
"even well-hydrated stratum corneum preferentially dissolves
lipid-soluble molecules, so that the selective permeability
of these molecules is preserved" (p. 353).90

20
Another difference between human and hairless mouse skin
could be significant in prodrug design. Rat and mouse skins
apparently have higher levels of enzymatic activity for drug
metabolism than human skin. Esterase activity, specifically,
appears to be low in human skin.54
Research. Proposal
The overall goal of the present research is to develop
prodrug derivatives of 5FU with solubility characteristics
and skin penetration properties superior to 5FU itself. The
derivatives should be stable in formulation, but should
readily convert to 5FU in the skin since the purpose of
topical 5FÜ therapy is to deliver 5FU dermally, not
transdermally. In pursuit of this goal, it is hoped that
experimental data will be obtained which supports a
solubility-based method for designing prodrugs with optimized
topical delivery characteristics. The objectives for meeting
this goal are to:
1) synthesize several homologous series of acylated
derivatives of 5FD,
2) verify structures by melting point, elemental
analysis (novel compounds) , and standard spectral
techniques,
3) determine physical chemical properties such as lipid
and aqueous solubilities, partition coefficients, and
rates of hydrolysis, and
4) determine skin penetration parameters and quantitate
skin accumulation using hairless mouse skin as the model
membrane.

21
Table 1-1. Structures of 5F0 and prodrug derivatives of 5FU.
Series
Rl
R2
l-alkyloxycarbonyl-5FU (I)
l-acyl-5FU (II)
1,3-bis-acyl-5FU (III)
3-acyl-5FU (IV)
-(C=0)0(CH2)nCH3
-(C=0)(CH2)nCH3
-(C=0) (CH2)nCH3
-H
-H
-H
- —(C=0)(CH2)nCH3

22
Four series of compounds have been selected as potential
prodrugs of 5FU. They are the 1-alkyloxycarbonyl (I), 1-acyl
(II), 1,3-bis-acyl (III), and 3-acyl (IV) derivatives of 5FÜ
(Table 1-1). Fluorouracil (5FD) has two acidic pKa values,
8.0 and 13.0.91 Spectral studies have shown that the
monoanion is actually a mixture of N1- and N3-anions.30 A
comparison of ionization constants for 5FU derivatives that
are identically substituted at the N1- or N3-positions22'30
suggests that the N3-position is probably the most acidic.
However, the N1-position is the most reactive site for both
synthesis and hydrolysis of acylated derivatives of 5FU.
The three series of acyl derivatives were chosen for two
reasons; they exhibit high lipid and aqueous solubilities,
and they readily hydrolyze without enzymatic activation.26
While rapid hydrolysis is seen as a drawback by some
investigators,26'42 it may actually be an advantage when
developing prodrugs for dermal delivery.38-39'92 The
1-alkyloxycarbonyl series was selected for comparison with
the 1-acyl series since it shows good chemical stability, but
rapid enzymatic hydrolysis.30 Various members of each series
have been synthesized previously,25-27'30'93 but homologous
series have not been examined, and nobody has studied their
applicability for dermal delivery.

CHAPTER 2
1-ALKYLOXYCARBONYL DERIVATIVES
Introduction
Alkyloxycarbonyl derivatives of 5-fluorouracil (5FU)
have previously been studied as potential sources of 5FU in
vivo.25,29-30 when substitution is at the N3-position of 5FU,
the derivatives are chemically stable. Their hydrolyses in
human plasma and liver homogenate are also slow enough to
raise questions about their usefulness as prodrugs for the
oral or rectal delivery of 5FU,29 Thus, there is no doubt
that they are too stable to serve as prodrugs for dermal
delivery. On the other hand, substitution at the Ni-position
produces compounds that are relatively stable chemically,30
but which are sufficiently labile in the presence of enzymes30
to justify consideration as dermal prodrugs.
Six straight-chain 1-alkyloxycarbonyl derivatives were
selected for study. The derivatives and their structures are
shown in Table 2-1.
23

24
Table 2-1. Structures of 1-alkyloxycarbonyl derivatives.
Compound
R
l-methyloxycarbonyl-5FU
l-ethyloxycarbonyl-5FU
l-propyloxycarbonyl-5FU
l-butyloxycarbonyl-5FD
l-hexyloxycarbonyl-5F0
l-octyloxycarbonyl-5FU
(1)
-ch3
(2)
-CH2CH3
(3)
-(CH2)2CH3
(4)
-(CH2)3CH3
(5)
-(ch2)5ch3
(6)
-
25
Materials and Methods
Synthesis
Melting points (mp) were determined with a Thomas-Hoover
capillary melting point apparatus and are uncorrected.
Elemental microanalyses were obtained for all novel compounds
through Atlantic Microlab, Incorporated in Norcross, Georgia.
Proton nuclear magnetic resonance (ÍH NMR) spectra were
obtained at 90 MHz on a Varian EM-390 spectrometer. Chemical
shifts (8) are reported in parts per million (ppm) from the
internal standard, tetramethylsilane (TMS). Coupling
constants (J) are expressed in cycles per second (Hz).
Infrared (IR) spectra were recorded with a Perkin-Elmer 1420
spectrophotometer and absorbances are reported in cm-1.
Ultraviolet (UV) spectra were obtained with a Cary 210 or
Shimadzu UV-265 spectrophotometer. Maximum absorbances are
reported in nm along with the molar absorptivities (E) in
L/mol.
l-Alkvloxvcarbonvl-5-fluorouracil (general procedure)
To 0.66 g (0.01 mol) of 85% potassium hydroxide
dissolved in methanol (20-50 mL) was added 1.33 g of 5FU
(0.0101 mol). Slightly more than an equivalent of 5FU was
used to prevent excess base from catalyzing the aldol
condensation of acetone. When present, condensation products
complicated product isolation. The methanol suspension was

26
stirred for 30 minutes, and the methanol was evaporated under
reduced pressure. The potassium salt was suspended in
acetone (25-50 mL) which was evaporated under reduced
pressure to remove residual methanol. The salt was
resuspended in acetone (25-50 mL), and the suspension was
added dropwise over a three minute period to a well stirred
acetone (20 mL) solution containing 1.0 to 1.2 equivalents of
the appropriate alkyl chloroformate. The mixture was stirred
at room temperature for 60 minutes. The mixture was
filtered, and the residue was washed with acetone (20 mL).
The combined acetone solutions were evaporated under reduced
pressure, and the solid residue was crystallized from an
appropriate solvent or solvent combination.1
l-Methyloxycarbonyl-5-fluorouracil (1)
Crystallization from acetone gave 1.36 g of 1 (72%) :
mp 158-60 °C (lit.30 mp 159-60 °C) ; IR (KBr) 1695, 1710, 1740,
and 1760 cm'1 (C=0); 3H NMR t(CD3)2SO] 5 3.86 (s, 3H, Cfi3) and
8.16 (d, J=7 Hz, 1H, C6-H) ,• (IVmI (CH3CN) 254 nm (E-9.63xl03) .
l-Ethvloxvcarbonvl-5-fluorouracil (2)
Crystallization from acetone/ether gave 1.31 g of 2
(65%) : mp 127-8 °C (lit.30 mp 126-8 °C); IR (KBr) 1690, 1730,
and 1750 cm'1 (C=0); 3H NMR [ CH3), 4.31 (q, J=7 Hz, 2H, OCÜ2), and 8.16 (d, J=7 Hz, 1H,
C6-H) ; UVmav (CH3CN) 254 nm (e=9.86xl03) .
^â– Several compounds in this series were provided by Kenneth B.
Sloan, Ph. D. based on the author's procedure.

27
1-Propvloxvcarbonv1-5-fluorouraci] L31
Crystallization from acetone/ether gave 1.37 g of 3
(64%): mp 124-6 °C; IR (KBr) 1690, 1730, and 1755 cm'1 (C=0);
XH NMR [ (CD3) 2SO] 8 0.95 (t, J=7 Hz, 3H, CH3) , 1.70 (m, 2H,
OCH2CH2), 4.23 (t, J=7 Hz, 2H, OCÜ2) , and 8.15 (d, J=7 Hz, 1H,
C6-H) ; UV max (CH3CN) 254 nm (E=l. OOlxlO4) .
Anal. Calc, for C8H9FN2O4: C, 44.45; H, 4.20; N, 12.96.
Found: C, 44.53; H, 4.23; N, 12.89.
1-Butyloxycarbony1-5-fluorouraci1 14)
Crystallization from dichloromethane/hexane gave 1.33 g
of 4 (58%): mp 97-8 °C (lit.30 mp 102-4 °C); IR (KBr) 1695,
1735, and 1765 cm-1 (C=0); XH NMR [(CD3)2SO] 8 0.91 (t,
J=7 Hz, 3H, CÜ3), 1.3-1.8 (m, 4H, OCH2CH2CH2) , 4.27 (t, J=6
Hz, 2H, OCH2), and 8.13 (d, J=7 Hz, 1H, C6-fl) ; (CH3CN)
254 nm <£=9.93xl03) .
1-Hexyloxvcarbonyl-5-fluorouraci1 (5)
Crystallization from dichloromethane/hexane gave 1.27 g
of 5 (49%) : mp 66-7 °C (lit.93 mp 68-9 °C) ; IR (KBr) 1690,
1730, and 1750 cm-1 (C=0); 3H NMR [ (CD3)2SO] 8 0.87 (distd t,
3H, CH3), 1.1-1.8 (m, 8H, OCH2CH2CH2CH2CH2) , 4.26 (t, J=6 Hz,
2H, OCH2), and 8.13 (d, J=7 Hz, 1H, C6-H); UVmax (CH3CN) 254
nm (6=1.004xl04).
l-0ctvloxvcarbonvl-5-fluorouracil (6)
Crystallization from dichloromethane/hexane gave 1.80 g
of 6 (61%): mp 97-8 °C; IR (KBr) 1690, 1730, and 1750 cm"l
(C=0); 3H NMR [(CD3)2S0] 8 0.87 (distd t, 3H, CH3) , 1.2-1.9

28
(m, 12H, OCH2CH2CH2CÍI2CÍI2CH2CII2) ' 4-27 (t/ J=6 Hz, 2H, 0CÜ2) ,
and 8.15 (d, J=7 Hz, 1H, C6-H) ; UVmax (CH3CN) 254 nm
(e=1.009xl04) .
Anal. Calc, for C13H19FN2O4: C, 54.53; H, 6.69; N, 9.79.
Found: C, 54.46; H, 6.73; N, 9.77.
Lipid Solubility
Lipid solubilities were determined using isopropyl
myristate (IPM), a commercial vehicle used in cosmetics and
topical medicináis,94 as the lipid solvent. The use of IPM as
a model lipophilic vehicle in skin penetration studies is
well established.77'95
Three suspensions of each derivative were stirred at
22±1 °C for 48 hours. The suspensions were filtered through
0.45 |lm nylon filters, and the saturated solutions were
diluted in acetonitrile and analyzed by UV spectroscopy.
Solubilities were calculated using Beer's Law:
A = e-C-d (1)
where A is the absorbance, £ is the molar absorptivity, C is
the concentration in mol/L, and d is the path length of the
cuvette in cm. Molar absorptivities were predetermined in
triplicate in acetonitrile at 254 nm.
Aqueous Solubility
For direct measurement of aqueous solubilities, three
suspensions of each derivative were vigorously stirred in

29
0.05 M acetate buffer (pH=4.0) at 22±1 °C for 60 minutes.
The suspensions were filtered through 0.45 |!m nylon filters,
and the saturated solutions were diluted in acetonitrile and
analyzed by UV spectroscopy. Solubilities were calculated
using Beer's Law as previously described.
Partition Coefficients
The partitioning-based method for determining aqueous
solubility utilized the saturated IPM solutions from the
lipid solubility study. For most compounds, equal volumes
(1 mL) of saturated IPM solution and 0.05 M acetate buffer
(pH=4.0) were used. The use of equal or near-equal phase
volumes is known to facilitate rapid equilibrium.96 The two
phases were mixed thoroughly for ten seconds and allowed to
separate for 60 seconds. A preliminary study showed that
there was virtually no difference in partition coefficient
(PC) values when partitioning was carried out for 10, 20, or
30 seconds (see Chapter 3). The IPM layers were diluted in
acetonitrile and analyzed by UV spectroscopy. The IPM-buffer
partition coefficients were calculated as follows:
PC = Aafter/(Abefore“^after) ‘Vaq/Vjpm (2)
where Aafter is the absorbance from the IPM layer after
partitioning, Abefore is the absorbance from the IPM layer
before partitioning, Vaq is the volume of the aqueous phase,
and Vipm is the volume of the IPM phase. Estimated aqueous

30
solubilities (Saq) were calculated from the IPM solubility
(Sipm) and the partition coefficient:
Saq = Sipm/PC (3)
Partitioning was carried out in triplicate for a fixed volume
ratio for each derivative. For those compounds with large
differences in solubility in one phase relative to the other,
volume ratios (IPM:buffer) other than 1:1 were necessary, but
the ratio never exceeded 10:1 or 1:10.
Hydrolysis Kinetics
Hydrolysis rates have previously been reported for
several members of this homologous series.30,93 In the
present study, hydrolysis rates were determined at 32 °C for
l-methyloxycarbonyl-5FU (1) in 0.05 M phosphate buffer
(pH=7.1, 1=0.12) and in the same buffer with 0.11%
formaldehyde (3.6xl0-2 M). The rate in the presence of
formaldehyde was determined for comparison with the rate in
plain buffer since formaldehyde was used as a preservative in
the diffusion cell experiments described in the following
section.
The hydrolyses were followed by high performance liquid
chromatography (HPLC). The HPLC system consisted of a
Beckman model 110A pump with a model 153 ÃœV detector, a
Rheodyne 7125 injector with a 20 (Im loop, and a Hewlett-
Packard 3392A integrator. The column was a Lichrosorb RP-8
10 )lm reversed-phase column, 250 mm x 4.6 mm (inside

31
diameter). The mobile phase contained 10% methanol and 90%
0.025 M acetate buffer (pH=5.0) and was run at 1.0 mL/min.
The column effluent was monitored at 254 nm, and quantitation
was based on peak areas. Standards chromatographed under the
same conditions were used for calibration.
Hydrolysis was initiated by adding 0.4 mL of a stock
solution of compound 1 in acetonitrile to 25 mL of buffer
prewarmed to 32 °C in a constant temperature water bath to
give final concentrations of ~1.8xl0~4 M. Aliquots were
removed at appropriate intervals and chromatographed
immediately. Pseudo-first-order rate constants were
determined from the expression:
ln(Ct) = In (C0) -kt (4)
where Ct is the concentration at some time=t, C0 is the
concentration at t=0, k is the pseudo-first-order rate
constant, and t is the time. The slopes, -k, of linear plots
of ln(Ct) versus t were determined by linear regression. The
half-lives (11/2) were calculated from
ti/2 = 0.693/k (5)
Each hydrolysis reaction was run in triplicate and was
followed for a minimum of three half-lives. The correlation
coefficients were >0.999.
Skin Penetration Studies
Diffusion cell experiments were performed to measure the
transdermal delivery of 5FU and the 5FU prodrugs. Franz-type

32
diffusion cells from Crown Glass in Somerville, NJ with
4.9 cm2 donor surface areas and 20 mL receptor phase volumes
were used for this purpose. The full-thickness skins were
obtained from female hairless mice (SKH-hr-1) from Temple
University Skin and Cancer Hospital.
The mice were killed by cervical dislocation, their
skins were removed immediately by blunt dissection, and
dorsal sections were mounted in the diffusion cells. The
dermal sides of the skins were placed in contact with
receptor phase which contained 0.05 M phosphate buffer
(pH=7.1, 1=0.12) with 0.11% formaldehyde as a preservative.
The effectiveness of formaldehyde for this purpose has
recently been documented.89 The receptor phases were stirred
continuously and kept at constant temperature (32 °C) by a
circulating water bath. A preapplication period of 48 hours
was established to uniformly condition the skins and to
remove water-soluble UV-absorbing materials. The receptor
phases were changed three times during this period, and
control experiments from earlier studies have shown that this
procedure effectively removes those materials.97 The
epidermal sides of the skins were exposed to the air and were
left untreated during this period.
After the preapplication period, 0.5 mL aliquots from
suspensions of the prodrugs in IPM were applied to the
epidermal sides of the skins. The IPM suspensions were
stirred at 22+1 °C for 48 hours prior to application to
ensure that saturation was attained. Total concentrations of

33
the IPM suspensions ranged from 0.3 M to 0.8 M with enough
excess solid present to maintain saturation for the duration
of the application period (see below). Each drug-vehicle
combination was run in triplicate.
Samples were taken from the receptor phases at 4, 8, 12,
21, 24, 27, 30, 33, 36, 45, and 48 hours after donor phase
application. The receptor phases were changed following
removal of each sample so that "sink" conditions were
maintained. Samples were analyzed for total 5FU species that
had diffused by UV spectroscopy (e=7.13xl03 at 266 nm) after
allowing at least 72 hours for complete prodrug hydrolysis.
Cumulative amounts of total 5FU species that diffused (Jlmol)
were plotted against time (h), and the slopes of the linear,
"steady-state" regions were calculated using linear
regression. The slopes, when divided by 4.9 (the area of the
donor surface in cm2), gave the "steady-state" fluxes
()lmol/cm2/h) . In a separate experiment, HPLC was used to
determine intact prodrug content in the receptor phases at
each sampling time. Mobile phase containing 18-50% methanol
in 0.025 M acetate buffer (pH=5.0) was used with the system
described earlier. Aliquots were removed and chromatographed
immediately after the samples were taken. Prodrug fluxes
were calculated in the same manner as fluxes for total 5FU.
Donor phases were changed every twelve hours and were
set aside for i-H NMR analysis. Stability of the prodrugs in
IPM was determined from the chemical shift of C6-H. In
dimethylsulfoxide-d6, the C6-H signal for 5FU appears at

34
8=7.73 ppm. For each of the 1-alkyloxycarbonyl derivatives,
the same signal in dimethylsulfoxide-d6 appears at 8>8.10 ppm.
Since this area of the spectrum is free from interference by
IPM absorbances, the two signals can be identified and
quantified if necessary.
Following removal of the donor phases after the 48-hour
application period, the epidermal sides of the skins were
washed three times with 5 mL portions of methanol to remove
all remnants of prodrug and vehicle from the skin surfaces.
This was accomplished quickly (<3 min) to minimize contact
time between the skins and methanol. The receptor phases
were changed again, and the dermal sides were kept in contact
with the fresh buffer for 23 hours while the epidermal sides
were again left exposed to the air. After this "leaching"
period, another sample was taken from each cell to measure
the skin accumulation of total 5FU species.
Second applications to the epidermal sides of the skins
were made after the "leaching" period with a standard drug-
vehicle suspension . Theophylline in propylene glycol
(0.4 M) was applied to assess the damage to the skins from
application of the initial drug-vehicle combinations.
Samples were taken at 1, 2, 3, 5, 7, 9, and 11 or 12 hours
after application. The samples were analyzed for
theophylline by UV spectroscopy (E=1.02xl04 at 271 nm) and
second application fluxes were determined as described above.

35
Results and Discussion
Synthesis and Structure Determination
The known l-alkyloxycarbonyl-5FU derivatives have
melting points25'30'93 and spectral properties (UV30 and 3H
NMR25) in agreement with those reported in the literature.
The structures of the novel compounds were assigned by
comparison of their spectral properties with those of the
known homologs. Elemental microanalyses were obtained for
the novel compounds and were within acceptable limits
(±0.4%).
Acylation on the N1- or N3-position can be distinguished
by UV and 1H NMR analysis. Anions of N3-substituted
derivatives undergo a substantial shift of their UVmax to
longer wavelength while anions of N1-substituted compounds do
not.30'98 This is reportedly due to the extended conjugation
possible for the N1-anion.20 Differences in 3H NMR spectra
are also well defined. In chloroform-d, the C6-H signal for
Ni-substituted derivatives is a sharp doublet indicating
coupling with C5-F. The same signal in chloroform-d for
N3-substituted derivatives appears as a broad singlet or
triplet-like doublet of doublets from additional coupling of
C6-H with N1-!!.20 When the substituents contain a carbonyl
group attached to the N1-position, as they do in the
1-alkyloxycarbonyl series, an anisotropic effect is observed

36
in which the C6-H signal is shifted downfield relative to 5FU
or the N3-substituted derivatives.25 For example, the C®-H
chemical shift for l-ethyloxycarbonyl-5FU (5=8.00) is 0.77
ppm downfield when compared with 3-ethyloxycarbonyl-5FU
(5=7.23) in chloroform-d.
Solubility
Solubility determinations are generally accomplished by
stirring excess solute in a solvent until saturation is
attained. The excess solid is removed and the saturated
solution is assayed for solute content. This approach is
suitable for stable solutes and for unstable solutes in
aprotic solvents, but another method is needed for measuring
aqueous solubilities of chemically unstable compounds.
An alternative to the direct method for determining
aqueous solubilities is the partitioning-based method. The
advantage of this method is that contact time between the
unstable compound and the aqueous phase can be minimized.
However, several points regarding this procedure require
clarification.
First, partition coefficients are concentration
dependent except when compounds with low associating
tendencies are present in dilute solutions (<10_1 M).36
Solubilities based on partition coefficients can only be
reported as estimates since activity coefficients become more
important at higher solute concentrations."

37
Second, the partitioning and phase separation times that
were used in the partitioning-based method for estimating
aqueous solubility were chosen empirically. The times had to
be sufficiently long to allow equilibrium distribution of the
compounds between the phases to occur and to allow subsequent
separation of the phases to occur without substantial
hydrolysis of the prodrugs. Longer times than those chosen
could have been used for these more stable 1-alkyloxycarbonyl
derivatives, but in order to validate the procedure for all
four series, shorter times that were more appropriate for the
less stable derivatives were used.
Finally, due to the experimental design and the lability
of the prodrugs, mutual saturation of the phases prior to
partitioning could not be accomplished. Since the ester
(IPM) that was used as the lipid phase in these experiments
is practically insoluble in water,94 changes in the phase
volumes during partitioning are probably insignificant. This
potential volume change is a common source of error when more
water-soluble lipid solvents are not presaturated with their
corresponding aqueous phases.19®
The conventional, direct method was used for determining
lipid solubilities. Since the 1-alkyloxycarbonyl derivatives
are the most chemically stable compounds of the four series,
both the direct and the partitioning-based methods were used
for determining their aqueous solubilities. Thus, the two
methods for determining aqueous solubility were compared
using the 1-alkyloxycarbonyl derivatives as a model series.

38
Lipid (Sjpm) and aqueous (S*q) solubilities for the
1-alkyloxycarbonyl derivatives are presented in Table 2-2
along with their melting points. Lipid solubilities are
greatly enhanced by making derivatives. Solubility values
range from over 40 times greater than 5FD for 1-methyloxy-
carbonyl-5FU (1) to more than 3000 times greater than 5FO for
l-hexyloxycarbonyl-5FO (5). Increases in lipid solubility
with increasing chain length are accompanied by decreases in
melting point. A change in that trend is observed for
l-octyloxycarbonyl-5FU (6), but this is not unexpected.
Since melting point and solubility depend in part on crystal
lattice energies,101 this result indicates that the crystal
structure is dominated by the 5FU nucleus for lower homologs
and by the aliphatic side chain for higher homologs.
Aqueous solubilities are reported for both the direct
and partitioning methods. The results show that aqueous
solubility peaks for l-ethyloxycarbonyl-5FU (2) and then
decreases. Compounds 1 and 2 have aqueous solubilities
greater than 5FÜ even though a hydrogen-bonding group (N1—H)
has been masked. Again, this is reflected in the lowered
melting points for the derivatives when compared to 5FU.
Aqueous solubilities determined by the partitioning-
based method underestimated the direct solubilities by 7% for
compounds 5 and 6, 8% for compound 1, 21% for compound 4, 22%
for compound 3, and 34% for compound 2. Relative aqueous
solubility among members of the series is the same with

39
Table 2-2. Melting points (MP), lipid solubilities (Sipm) , and
aqueous solubilities (Saq) for 1-alkyloxycarbonyl derivatives.
Compound
MP
(°C)
Sipm3
(mM)
Saq5
(mM)
Saqc
(mM)
SAQd
(mM)
5FU
280-2
0.049
96
-
85
1
158-60
2.1
120
Ill
124
2
127-8
13
263
174
34
3
124-6
15
55
43
-
4
97-8
34
29
23
26
5
66-7
153
5.4
5.0
5.8
6
97-8
36
0.14
0.13
aStandard
deviations
from the
mean were
within ±3%
for IPM
solubilities.
bStandard deviations from the mean were within ±8% for
aqueous solubilities determined by direct method.
cStandard deviations from the mean were within ±6% for
experimental values used to calculate partition coefficients
and estimated aqueous solubilities (±11% for compound 2).
^Literature values from references 30 and 93.

40
Table 2-3. Solubility ratios (SR), partition coefficients
(PC) , and hydrophobicity parameters (7t) for
1-alkyloxycarbonyl derivatives.
log (PC) -
Compound
SRa
log (SR)
7tb
PCC
log (PC)
7td
log(SR)
1
0.018
-1.75
0.019
-1.72
0.03
2
0.050
-1.30
0.45
0.075
-1.12
0.60
0.18
3
0.28
-0.56
0.74
0.36
-0.45
0.67
0.11
4
1.2
0.07
0.63
1.4
0.16
0.61
0.09
5
29
1.46
0.70
31
1.48
0.66
0.02
6
257
2.41
0.48
285
2.45
0.49
0.04
aSolubility ratio calculated from Sipm/Saq.
bAlog(SR) for compound and preceding compound.
Experimental partition coefficient (Cipm/Caq) .
dAlog(PC) for compound and preceding compound.

41
either procedure, and overall, agreement between the two
methods is good.
The direct aqueous solubility values for compounds 1, 4,
and 5 agree with the corresponding literature values included
in Table 2-2. However, the value for compound 2 is much
higher than the literature value. The basis for this
discrepancy is not clear.
In Table 2-3, the solubility ratios (SR) and
experimentally determined partition coefficients (PC) are
compared for the 1-alkyloxycarbonyl series. The values for
log (PC)-log(SR) indicate that the partition coefficients are
somewhat higher than the corresponding solubility ratios.
Generally, the more polar derivatives (log[PC]<0) show the
greatest difference with the exception of compound 1.
Yalkowsky et al.102 studied solubility ratios and
octanol-water partition coefficients for a broad range of
solutes. They concluded that self-association of polar
solutes in octanol increases the ability of octanol to
accommodate the solute which increases the partition
coefficient. Conversely, a nonpolar solute causes a decrease
in the partition coefficient by self-associating in the
aqueous phase.ll^2 The present results can be explained on the
same basis. The lower than expected log(PC)-log(SR) value
for compound 1 could be due to the low concentration in the
IPM phase during partitioning of this derivative. Since the
aqueous solubility of compound 1 is nearly 60 times its lipid
solubility, the concentration in the IPM phase is reduced

42
well below its already low concentration at saturation.
Therefore, the solute-solute interactions that lead to higher
partition coefficients would also be reduced.
The hydrophobicity parameters100 (it) in Table 2-3 were
calculated from the relationship:103
log (PC) n = log (PC) o+Itn (6)
where n is the number of methylene units in a homologous
series using both log(PC) and log(SR) values. Both
calculations yield an average It value equal to 0.60. Values
for Jl from the literature include 0.54 for silicone oil-water
and 0.66 for hexane-water101 indicating that 0.60 is a
reasonable value for the IPM-buffer partitioning system.
Hydrolysis Kinetics
Hydrolysis of l-methyloxycarbonyl-5FU (1) to 5FÜ in
0.05 M phosphate buffer (pH=7.1, 1=0.12) with and without
0.11% formaldehyde (3.6xl0-2 M) was followed by HPLC at 32 °C.
Disappearance of compound 1, indicated by ln(C), is plotted
versus time (min) in Figure 2-1. The linearity of the plots
suggests that hydrolysis of 1 follows first-order kinetics in
the presence and absence of formaldehyde. Pseudo-first-order
rate constants (k) and half-lives (ti/2) from the linear plots
are presented in Table 2-4.
Hydrolysis of compound 1 is clearly faster in the
presence of formaldehyde indicating that general base
catalysis by formaldehyde hydrate may be involved (see

43
O -10
c
-11 .
-12
200
q Plain Buffer
t Formaldehyde Buffer
T
400
Time (min)
600
800
Figure 2—1. Plots of ln(C) versus time (min) for hydrolysis
of l-methyloxycarbonyl-5FU in 0.05 M phosphate buffer
(pH=7.1, 1=0.12) with and without formaldehyde
at 32 °C.

44
Table 2-4. Pseudo-first-order rate constants (k) and half-
lives (ti/2) for hydrolysis of l-raethyloxycarbonyl-5FD
in 0.05 M phosphate buffer (pH=7.1, 1=0.12)
with and without formaldehyde at 32 °C.
Formaldehyde
Compound (M)
k(±SD)a
(min-1)
ti/2
(min)
1 0 3.33x10-3(0.04x10-3) 208
1 3.6xl0-2 3.78xl0'3(0.05xl0-3) 183
aMean ± standard deviation for n=3 values.

45
Chapter 3). Significant buffer catalysis by phosphate has
also been demonstrated for compounds in this series, and the
reader is referred to the work of Buur and Bundgaard30 for
complete pH-rate profiles and probable hydrolysis mechanisms.
In any case, release of the parent drug from compound 1
by chemical means is too slow for dermal delivery purposes.
Other members of the series would also appear to be poor
candidates, since they hydrolyze chemically over two times
slower than compound 1.30>93 In 80% human plasma, however,
hydrolysis rates are much faster (ti/2=2-3 min)30»93
suggesting enzyme catalysis of the 1-alkyloxycarbonyl series.
Since the skin is metabolically active, further study of
these compounds in diffusion cells was warranted.
Skin Penetration
Skin penetration data from the diffusion cells are
plotted as cumulative amount of total 5FU species that
diffused ((Imol) versus time (h) . In Figure 2-2, results for
l-methyloxycarbonyl-5FU (1), l-ethyloxycarbonyl-5FU (2), and
l-propyloxycarbonyl-5FU (3) are compared to 5FU itself. In
Figure 2-3, results for l-butyloxycarbonyl-5FU (4),
l-hexyloxycarbonyl-5FU (5), and l-octyloxycarbonyl-5FU (6)
are compared to 5FU. Error bars correspond to the standard
deviation from the mean for n=3 values.
Fluxes (J) , lag times (ti,) , and skin accumulation (SA)
values for each compound are reported in Table 2-5. Lag time

Cumulative Amount (nmol)
46
Time (h)
Figure 2-2. Plots of cumulative amount of total 5FU species
that diffused (nmol) versus time (h) for
compounds 1, 2, 3, and 5FU.

Cumulative Amount (pmol)
47
Figure 2-3. Plots of cumulative amount of total 5FU species
that diffused (Umol) versus time (h) for
compounds 4, 5, 6, and 5FU.

48
Table 2-5. Fluxes
(SA) values
(J), lag times (tl), and skin accumulation
for 1-alkyloxycarbonyl derivatives.
Compound
J(+SD) a
(|lmol/cm2/h)
Prodrug
S-Sb(ll h)c
(%)
tL
(h)
SA(±SD)a
Oimol)
5FU
0.24 (0.09)
-
13
3.7 (0.9)
1
2.6(0.6)
42(16)
14
8.3(0.1)
2
5.9(1.3)
90(75)
13
18(4)
3
2.3(0.2)
78(43)
11
5.0(1.4)
4
2.2(0.1)
73(32)
10
4.2(0.5)
5
1.5(0.1)
79(14)
6.2
11(0)
6
0.29(0.02)
-
8.7
3.2(0.5)
aMean ± standard deviation for n=3 values.
bPercent of total 5FU as intact prodrug during "steady-state"
phase in separate experiment (n=l).
cPercent of total 5F0 as intact prodrug from 11 h sample in
separate experiment (n—1).

49
Table 2-6. Second application fluxes (J) and lag times (ti,)
for 1-alkyloxycarbonyl derivatives.
Compound
Ja(±SD)b
(Hmol/cm2/h)
tL
5FU
1.2 (0.2)
1.2
1
1.9(0.1)
0.8
2
1.8(0.4)
0.9
3
1.7(0.3)
1.0
4
1.8(0.2)
1.0
5
1.8(0.1)
0.8
6
1.8(0.2)
0.9
aFlux of 0.4 M theopylline from propylene glycol.
bMean ± standard deviation for n=3 values.

50
refers to the intersection of the linear, or "steady-state,"
region of each graph with the time (x) axis, and it is the
time required for establishing a uniform concentration
gradient within the skin.65 The percent of total 5FU present
as intact prodrug in the receptor phase is also reported in
Table 2-5. These values were calculated from samples taken
during the "steady-state" phase and from an earlier sample
(11 h) in a separate experiment using HPLC analysis (n=l).
The improvement in skin penetration of 5FU from the
1-alkyloxycarbonyl derivatives is significant except for
l-octyloxycarbonyl-5FU (6). Increases in flux are generally
about one order of magnitude, and l-ethyloxycarbonyl-5FU (2)
with nearly a 25-fold improvement is easily the best
derivative.
The presence of large amounts of intact prodrugs in the
receptor phases is a matter of interest. The high
percentages indicate that the hydrolytic enzymes of the skin
are not effectively converting the prodrugs to 5FU. It is
interesting to note, however, that percentages calculated
from a sample removed prior to "steady-state" are much lower
than the "steady-state" values. It is possible that the
large amounts of diffusant present at "steady-state" are
simply too much for the enzymes to handle. Another
possibility is that the continuous changing of the receptor
phase with each sample eventually depletes the enzymatic
activity.

51
The trend in skin accumulation is similar to the trend
in flux with the exception of l-hexyloxycarbonyl-5FU (5).
The large skin accumulation value and short lag time for
compound 5 may indicate a high affinity for the lipid regions
of the skin but less affinity for the hydrated regions and
the receptor phase.
Second application fluxes and lag times are presented in
Table 2-6. Skin penetration by theophylline from propylene
glycol, the standard drug-vehicle combination, is
approximately one and one-half times higher for the skins
treated with the 1-alkyloxycarbonyl derivatives than for
those treated with 5FU, and it is consistent throughout the
series. Therefore, skin damage is greater with the prodrugs,
but the difference is small when compared with the general
improvement in delivery of 5FU from the prodrugs.
The stability of the prodrugs in the IPM formulations
was assessed by ^-H NMR analysis of the donor phases. After a
minimum of five days from the time the suspensions were
prepared until their ^-H NMR spectra were recorded, including
at least twelve hours during which the formulations were in
contact with the skins, the 1-alkyloxycarbonyl derivatives
were found to be intact with no evidence of 5FU formation.
Snmma ry
The 1-alkyloxycarbonyl derivatives of 5FU exhibited
decreased melting points and increased lipid solubilities

52
when compared to 5FU. Aqueous solubility reached a maximum
for l-ethyloxycarbonyl-5FU (2) and decreased from there with
increasing chain length. Skin penetration and skin
accumulation were also maximized for compound 2 suggesting
that both lipid and aqueous solubilities are important for
predicting transdermal and dermal delivery of these 5FU
prodrugs. The presence of high percentages of prodrugs in
the receptor phases, indicating insufficient release of the
parent drugs in the hairless mouse skin model, may limit the
potential of this series of prodrugs at least for dermal
delivery purposes. Finally, the partitioning-based method
for determining aqueous solubility appears to be a useful
method particularly for determining relative solubilities in
an homologous series.

CHAPTER 3
1-ACYL DERIVATIVES
Introduction
The 1-acyl derivatives of 5-fluorouracil (5FU) are
chemically unstable in aqueous solutions at all pH values.
In fact, their lability has been cited as a limitation to
their usefulness as drugs or prodrugs.26'42 If properly
formulated in an aprotic vehicle, however, this series of 5FU
derivatives may have potential for use as prodrugs for dermal
delivery.
Six straight-chain 1-acyl derivatives were selected for
study. The derivatives and their structures are shown in
Table 3-1.
Materials and Methods
Synthesis
Melting points (mp) were determined with a Thomas-Hoover
capillary melting point apparatus and are uncorrected.
Elemental microanalyses were obtained for all novel compounds
through Atlantic Microlab, Incorporated in Norcross, Georgia.
Proton nuclear magnetic resonance (iH NMR) spectra were
obtained at 90 MHz on a Varian EM-390 spectrometer. Chemical
53

54
Table 3-1. Structures of 1-acyl derivatives.
Compound
R
l-acetyl-5FU (7)
1-propiony1-5FÜ (8)
l-butyryl-5FU (9)
l-valeryl-5FU (10)
l-hexanoyl-5FU (11)
l-octanoyl-5FU (12)
-ch3
-CH2CH3
-(CH2)2CH3
-(CH2)3CH3
-(CH2)4CH3
-(CH2) 6ch3

55
shifts (8) are reported in parts per million (ppm) from the
internal standard, tetramethylsilane (TMS). Coupling
constants (J) are expressed in cycles per second (Hz).
Infrared (IR) spectra were recorded with a Perkin-Elmer 1420
spectrophotometer and absorbances are reported in cm*1.
Ultraviolet (UV) spectra were obtained with a Cary 210 or
Shimadzu UV-265 spectrophotometer. Maximum absorbances are
reported in nm along with the molar absorptivities (£) in
L/mol. Single-crystal X-ray analysis was obtained for
l-acetyl-5FU through Hoffmann-La Roche in Nutley, NJ.
l-Acyl-5-fluorouracil (general procedure)
To 0.66 g (0.01 mol) of 85% potassium hydroxide
dissolved in methanol (20-50 mL) was added 1.31 g of
5-fluorouracil (0.01 mol). The methanol suspension was
stirred for 30 minutes, and the methanol was evaporated under
reduced pressure. The potassium salt was suspended in
acetonitrile (25-50 mL) which was evaporated under reduced
pressure to remove residual methanol. The salt was
resuspended in acetonitrile (25-50 mL), and the suspension
was added dropwise over 15 to 30 minutes to a well stirred
acetonitrile (25 mL) solution in an ice bath containing 1.0
to 1.2 equivalents of the appropriate acid chloride. The
mixture was stirred at 0 °C for 60 minutes. The mixture was
filtered, and the residue was washed with acetonitrile
(25 mL). The combined acetonitrile solutions were evaporated
under reduced pressure, and the solid residue was

56
crystallized from an appropriate solvent or solvent
combination.
l-Acetyl-5-fluorouracil ill
Crystallization from dichloromethane gave 0.98 g of 7
(57%) : mp 129-30 °C (lit.27 mp 126-7 °C) ; IR (KBr) 1670, 1695,
1725, and 1770 cm*1 and 8.23 (d, J=7 Hz, 1H, C6-fl> ; UVmax (CH3CN) 261 nm
(E=1.125x104) .
l-Propionyl-5-fluorouracil (8)
Crystallization from dichloromethane/hexane gave 1.32 g
of 8 (71%); mp 130-1 °C (lit.27 mp 124-5 °C) ; IR (KBr) 1695,
1710, and 1740 cm'1 (C=0); 7H NMR (CDCI3) 8 1.25 (t, J=7 Hz,
3H, CH3), 3.14 (q, J=7 Hz, 2H, COCH2) / and 8.27 (d, J=7 Hz,
1H, C6-H) ; UVmax (CH3CN) 261 nm (6=1.141xl04) .
Anal. Calc, for C7H7FN2O3: C, 45.17; H, 3.79; N, 15.05.
Found: C, 45.26; H, 3.83; N, 14.97.
l-Butvrvl-5-fluorouracil (9)
Crystallization from dichlormethane/hexane gave 0.86 g
of 9 (43%) : mp 145-6 °C; IR (KBr) 1690, 1710, and 1740 cm'1
(C=0); XH NMR (CDCI3) 8 1.01 (t, J=7 Hz, 3H, CH3) , 1.77 (m,
2H, COCH2CH2), 3.09 (t, J=7 Hz, 2H, COCH2) , and 8.25 (d,
J=6 Hz, 1H, C6-fi) ; UV max (CH3CN) 261 nm (6=1.168xl04) .
Anal. Calc, for C8H9FN2O3: C, 48.00; H, 4.53; N, 14.00.
Found: C, 48.12; H, 4.58; N, 13.91.

57
l-Valeryl-5-fluorouraci1 (10)
Crystallization from dichloromethane/hexane gave 1.35 g
of 10 (63%): mp 120-1 °C; IR (KBr) 1695, 1715, and 1740 cm'1
(C=0); *H NMR (CDCI3) 5 0.95 (t, J=7 Hz, 3H, CÜ3), 1.3-1.8 (m
4H, COCH2CÜ2CH2) , 3.11 (t, J=7 Hz, 2H, COCR2) , and 8.24 (d,
J=7 Hz, 1H, C6-H) ; OVmax (CH3CN) 261 nm (6=1.175xl04) .
Anal. Calc, for C9H11FN2O3: C, 50.47; H, 5.18; N, 13.08.
Found: C, 50.52; H, 5.23; N, 13.03.
l-Hexanovl-5-f luorouracil Llll.
Crystallization from dichloromethane/hexane gave 1.71 g
of 11 (75%): mp 101-2 °C; IR (KBr) 1690, 1715, and 1745 cm'1
(C=0); 1H NMR (CDCI3) 8 0.92 (distd t, 3H, CH3), 1.2-1.9 (m,
6H, COCH2CE2Cii2Ca2) , 3.09 (t, J=7 Hz, 2H, COCH2) , and 8.24 (d
J=7 Hz, 1H, C6-a) ; UVmax (CH3CN) 261 nm (6=1.158xl04) .
Anal. Calc, for C10H13FN2O3: C, 52.63; H, 5.74; N, 12.27
Found: C, 52.69; H, 5.75; N, 12.27.
l-Octanoyl-5-fluorouracil (12)
Crystallization from dichloromethane/hexane gave 1.28 g
of 12 (50%): mp 83-4 °C; IR (KBr) 1685, 1710, and 1745 cm'1
(C=0); XH NMR (CDCI3) 8 0.90 (distd t, 3H, CH3) , 1.2-1.8 (m,
10H, COCH2CH2CH2CH2CÍÍ2CH2) , 3.10 (t, J=7 Hz, 2H, C0Cfl2>, and
8.23 (d, J=7 Hz, 1H, C6-£); UVmax (CH3CN) 261 nm
(6=1.155xl04) .
Anal. Calc, for Ci2Hi7FN203: C, 56.24; H, 6.69; N, 10.93
Found: C, 56.22; H, 6.73; N, 10.96.

58
Lipid solubility
Lipid solubilities were determined using isopropyl
myristate (IPM), a commercial vehicle used in cosmetics and
topical medicináis,94 as the lipid solvent. The use of IPM as
a model lipophilic vehicle in skin penetration studies is
well established.77'95
Three suspensions of each derivative were stirred at
22±1 °C for 48 hours. The suspensions were filtered through
0.45 [lm nylon filters, and the saturated solutions were
diluted in acetonitrile and analyzed by UV spectroscopy.
Solubilities were calculated using Beer's Law:
A = e-C-d (1)
where A is the absorbance, £ is the molar absorptivity, C is
the concentration in mol/L, and d is the path length of the
cuvette in cm. Molar absorptivities were predetermined in
triplicate in acetonitrile at 261 nm.
Aqueous Solubility
Because of the chemical instability of the 1-acyl
derivatives, direct measurement of aqueous solubilities for
these prodrugs was not attempted. A comparison of the direct
and partitioning-based methods for determining aqueous
solubility was presented in Chapter 2 for the 1-alkyloxy-
carbonyl derivatives.

59
Partition Coefficipnts
The partitioning-based method for determining aqueous
solubility utilized the saturated IPM solutions from the
lipid solubility study. For most compounds, equal volumes
(1 mL) of saturated IPM solution and 0.05 M acetate buffer
(pH=4.0) were used. The use of equal or near-equal phase
volumes is known to facilitate rapid equilibrium.96 The two
phases were mixed thoroughly for ten seconds and allowed to
separate for 60 seconds. A preliminary study with 1-acetyl-
5FD (7) showed that there was virtually no difference in
partition coefficient (PC) values when partitioning was
carried out for 10, 20, or 30 seconds (PC=0.183+0.006, a
standard deviation of only 3%). The IPM layers were diluted
in acetonitrile and analyzed by UV spectroscopy. The IPM-
buffer partition coefficients were calculated as follows:
PC = Aafter/ (Abefore“^after) 'Vaq/^IPM (2)
where Aafter is the absorbance from the IPM layer after
partitioning, Abefore is the absorbance from the IPM layer
before partitioning, Vaq is the volume of the aqueous phase,
and Vipm is the volume of the IPM phase. Estimated aqueous
solubilities (Saq) were calculated from the IPM solubility
(Sipm) and the partition coefficient:
Saq = Sjpm/PC (3)
Partitioning was carried out in triplicate for a fixed volume
ratio for each derivative. For those compounds with large

60
differences in solubility in one phase relative to the other,
volume ratios (IPM:buffer) other than 1:1 were necessary, but
the ratio never exceeded 10:1 or 1:10.
Hydrolysis Kinetics
Hydrolysis rates have previously been reported for one
member of this homologous series.26 In the present study,
hydrolysis rates were determined at 32 °C for all six 1-acyl
derivatives in 0.05 M phosphate buffer (pH=7.1, 1=0.12) and
for l-acetyl-5FU (7) in the same buffer with 0.11%
formaldehyde (3.6xl0~2 M). The rate in the presence of
formaldehyde was determined for comparison with the rate in
plain buffer since formaldehyde was used as a preservative in
the diffusion cell experiments described in the following
section. Two other concentrations of formaldehyde (1.8xl0_1 M
and 3.6xl0_1 M) were studied to assess the catalytic role of
formaldehyde in the hydrolysis of compound 7.
The hydrolyses were followed by ÃœV spectroscopy at
266 nm where the absorbance decrease accompanying hydrolysis
of the 1-acyl derivatives was maximized. Hydrolysis was
initiated by adding 60 to 75 |XL of stock solutions of the
derivatives in acetonitrile to 3 mL of buffer prewarmed to
32 °C in a thermostated quartz cuvette to give final
concentrations of l-2xl0~4 M. Absorbances were recorded at
appropriate intervals and pseudo-first-order rate constants

61
were determined from the expression:
In (At-Aoo) = In (A0-A.0)-kt (4)
where At is the absorbance at some time=t, A» is the
absorbance at t=°°, A0 is the absorbance at t=0, k is the
pseudo-first-order rate constant, and t is the time. The
hydrolyses were sufficiently fast to allow experimental
determination of Aoo. The slopes, -k, of linear plots of
In(At-A«) versus time were determined by linear regression.
The half-lives (11/2) were calculated from
ti/2 = 0.693/k (5)
Each hydrolysis reaction was run in triplicate and was
followed for a minimum of three half-lives. The correlation
coefficients were >0.999.
Skin Penetration Studies
Diffusion cell experiments were performed to measure the
transdermal delivery of 5FU and the 5FU prodrugs. Franz-type
diffusion cells from Crown Glass in Somerville, NJ with 4.9
cm2 donor surface area and 20 mL receptor phase volume were
used for this purpose. The full-thickness skins were
obtained from female hairless mice (SKH-hr-1) from Temple
University Skin and Cancer Hospital.
The mice were killed by cervical dislocation, their
skins were removed immediately by blunt dissection, and
dorsal sections were mounted in the diffusion cells. The
dermal sides of the skins were placed in contact with

62
receptor phase which contained 0.05 M phosphate buffer
(pH=7.1, 1=0.12) with 0.11% formaldehyde as a preservative.
The effectiveness of formaldehyde for this purpose has
recently been documented.89 The receptor phases were stirred
continuously and kept at constant temperature (32 °C) by a
circulating water bath. A preapplication period of 48 hours
was established to uniformly condition the skins and to
remove water-soluble UV-absorbing materials. The receptor
phases were changed three times during this period, and
control experiments from earlier studies have shown that this
procedure effectively removes those materials.97 The
epidermal sides of the skins were exposed to the air and were
left untreated during this period.
After the preapplication period, 0.5 mL aliquots from
suspensions of the prodrugs in IPM were applied to the
epidermal sides of the skins. The IPM suspensions were
stirred at 22±1 °C for 48 hours prior to application to
ensure that saturation was attained. Total concentrations of
the IPM suspensions ranged from 0.6 M to 1.0 M with enough
excess solid present to maintain saturation for the duration
of the application period (see below). Each drug-vehicle
combination was run in triplicate.
Samples were taken from the receptor phases at 4, 8, 12,
21, 24, 27, 30, 33, 36, 45, and 48 hours after donor phase
application. The receptor phases were changed following
removal of each sample so that "sink" conditions were
maintained. Samples were analyzed for total 5FU species that

63
had diffused by UV spectroscopy (6=7.13x103 at 266 nm) after
allowing at least 24 hours for complete prodrug hydrolysis.
Cumulative amounts of total 5FU species that diffused (Umol)
were plotted against time (h), and the slopes of the linear,
"steady-state" regions were calculated using linear
regression. The slopes, when divided by 4.9 (the area of the
donor surface in cm2), gave the "steady-state" fluxes
(|imol/cm2/h) . Because of the rapid chemical hydrolysis of
the 1-acyl derivatives, no attempt was made to analyze the
receptor phases for prodrug content.
Donor phases were changed every twelve hours and were
set aside for XH NMR analysis. Stability of the prodrugs in
IPM was determined from the chemical shift of C®-H. In
dimethylsulfoxide-d6, the C6-H signal for 5FÜ appears at
8=7.73 ppm. For each of the 1-acyl derivatives, the same
signal in dimethylsulfoxide-d6 appears at 5>8.20 ppm. Since
this area of the spectrum is free from interference by IPM
absorbances, the two signals can be identified and quantified
if necessary.
Following removal of the donor phases after the 48-hour
application period, the epidermal sides of the skins were
washed three times with 5 mL portions of methanol to remove
all remnants of prodrug and vehicle from the skin surfaces.
This was accomplished quickly (<3 min) to minimize contact
time between the skins and methanol. The receptor phases
were changed again, and the dermal sides were kept in contact
with the fresh buffer for 23 hours while the epidermal sides

64
were again left exposed to the air. After this "leaching"
period, another sample was taken from each cell to measure
the skin accumulation of total 5FU species.
Second applications to the epidermal sides of the skins
were made after the "leaching" period with a standard drug-
vehicle suspension . Theophylline in propylene glycol
(0.4 M) was applied to assess the damage to the skins from
application of the initial drug-vehicle combinations.
Samples were taken at 1, 2, 3, 5, 7, 9, and 11 or 12 hours
after application. The samples were analyzed for
theophylline by UV spectroscopy (e=1.02xl04 at 271 nm) and
second application fluxes were determined as described above.
Results and Discussion
Synthesis and Structure Determination
The known l-acyl-5FU derivatives have melting points27
and 7H NMR spectra25 in agreement with those reported in the
literature. The structures of novel compounds were assigned
by comparison of their NMR spectra with those of the known
homologs. Elemental microanalyses were obtained for the
novel compounds and were within acceptable limits (±0.4%).
The differences in spectral properties between N1- and
N3-acylated derivatives were discussed in detail in
Chapter 2. Ultraviolet (UV) spectra were not obtained for
the 1-acyl derivatives under basic conditions because of

65
their instability. However, differences in 1E NMR spectra
are very clear. In chloroform-d, the C6-H signal for
l-acetyl-5FU is a sharp doublet at 5=8.23, while the same
signal for 3-acetyl-5FU is a broad singlet at 5=7.23, a
difference of 1.00 ppm.
Results from the single-crystal X-ray analysis show that
the ^-assignment for l-acetyl-5FU is correct. The unit cell
was found to contain two independent molecules (Figure 3-1
and Figure 3-2). Both conformations show that the carbonyl
group from the 1-acyl group is positioned cis to the C6-H.
The circulating K electrons of carbonyl bonds are known to
deshield ortho protons in structurally similar compounds such
as acetophenone.105 Thus, the X-ray results support the
explanation that an anisotropic effect is responsible for the
downfield shift of C6-H in 1H NMR spectra of N1-acylated
derivatives.
Solubility
Lipid (Sipm) and aqueous (Saq) solubilities for the
1-acyl derivatives are presented in Table 3-2 along with
their melting points. In general, melting points decrease
and lipid solubilities increase with increasing chain length.
An exception to both of these trends is seen for 1-butyryl-
5FU (9). The basis for this variance from the observed
trends is not clear.

66
Figure 3-1. X-ray structure of l-acetyl-5FU (unprimed).
Source: Hoffman-La Roche, Nutley, NJ.

67
Figure 3-2. X-ray structure of l-acetyl-5FU (primed).
Source: Hoffmann-La Roche, Nutley, NJ.

68
Table 3-2. Melting points (MP), lipid solubilities (Sipm) , and
aqueous solubilities (Saq) for 1-acyl derivatives.
Compound
MP
(°C)
Sipm3
(mM)
SAQb
(mM)
saqc
(mM)
5FU
280-2
0.049
96
-
7
129-30
22
_
119
8
130-1
36
-
48
9
145-6
17
-
6.5
10
120-1
39
-
3.5
11
101-2
112
-
3.0
12
83-4
111
0.15
aStandard
deviations from
the mean
were within
±5% for IPM
solubilities.
bSolubility determined by direct method.
cStandard deviations from the mean were within +5% for
experimental values used to calculate partition coefficients
and estimated aqueous solubilities.

69
Table 3-3. Partition coefficients (PC) and hydrophobicity
parameters (JE) for 1-acyl derivatives.
Compound
PCa
log (PC)
7Cb
7
0.19
-0.73
_
8
0.76
-0.12
0.61
9
2.7
0.43
0.55
10
11
1.05
0.62
11
38
1.58
0.53
12
759
2.88
0.65
Experimental
partition
coefficient (Cipm/Caq) .
bAlog(PC) for
compound
and preceding compound.

70
Aqueous solubilities were determined by the
partitioning-based method for which a theoretical discussion
was presented in Chapter 2. For the 1-acyl series, aqueous
solubility decreases with increasing chain length, and only
l-acetyl-5FU (7) is more water soluble than 5FU.
In Table 3-3, partition coefficients (PC) and
hydrophobicity parameters (Jl) are listed for the 1-acyl
derivatives. The average m value is 0.60, the same value
that was calculated for the 1-alkyloxycarbonyl series.
Individual values for the 1-acyl derivatives are all within
0.07 of the average. These results further support use of
the partitioning-based method for determining relative
aqueous solubility in a homologous series.
Hydrolysis Kinetics
Hydrolysis of the 1-acyl derivatives was studied in
0.05 M phosphate buffer (pH=7.1, 1=0.12) at 32 °C. Pseudo-
first-order rate constants (k) and half-lives (ti/2) are
presented in Table 3-4. Half-lives are relatively consistent
throughout the series ranging from 3.1 minutes for
l-propionyl-5FU (8) to 4.8 minutes for l-acetyl-5FU (7).
Hydrolysis of compound 7 was also studied in the same
buffer with increasing concentrations of formaldehyde added
to assess the catalytic role of formaldehyde hydrate. These
results are shown in Table 3.5. It is clear that the
hydrolysis rate for compound 7 is faster in the presence of

71
Table 3-4. Pseudo-first-order rate constants (k) and half-
lives (11/2) for hydrolysis of 1-acyl derivatives in
0.05 M phosphate buffer Compound
k(±SD)a ti/2
(min-1) (min)
7
0.143(0.002)
4.8
8
0.222(0.004)
3.1
9
0.163(0.003)
4.3
10
0.169(0.006)
4.1
11
0.173(0.003)
4.0
12
0.183(0.003)
3.8
aMean ± standard deviation for n=3 values.

72
Table 3-5. Pseudo-first-order rate constants (k) and half-
lives (ti/2) for hydrolysis of l-acetyl-5FU in 0.05 M
phosphate buffer (pH=7.1, 1=0.12) with and without
formaldehyde at 32 °C.
Compound
Formaldehyde
(M)
k(±SD)
(min-1)
tl/2
(min)
7
0
0.143(0.002)a
4.8
7
3.6xl0'2
0.161(0.008)b
4.3
7
1.8X10'1
0.219(0.004)a
3.2
7
3.6xl0_1
0.279(0.006)a
2.5
aMean ± standard deviation for n=3 values.
bMean ± standard deviation for n=5 values.

k (min-1)
73
Formaldehyde (M)
Figure 3-3. Plot of pseudo-first-order rate constant (k)
versus formaldehyde concentration (M) for hydrolysis of
l-acetyl-5FU in 0.05 M phosphate buffer
(pH=7.1, 1=0.12) at 32 °C.

74
formaldehyde and the rate is concentration dependent. In
Figure 3-3, the pseudo-first-order rate constants are plotted
against formaldehyde concentration. The slope of the linear
plot gives the catalytic rate constant (kcat=0.375) for
formaldehyde catalysis of compound 7. Hydrolysis rates for
the 1-acyl derivatives have previously been shown to be
independent of pH at acidic pH values.26 Therefore,
formaldehyde catalysis is likely due to general base rather
than general acid catalysis.
Skin penetration data from the diffusion cells are
plotted as cumulative amount of total 5F0 species that
diffused (Umol) versus time (h). In Figure 3-4, results for
l-acetyl-5FU (7), l-propionyl-5FU (8), and l-butyryl-5FU (9)
are compared to 5FU itself. In Figure 3-5, results for
l-valeryl-5FD (10), l-hexanoyl-5FU (11), and l-octanoyl-5FU
(12) are compared to 5FU. Error bars correspond to the
standard deviation from the mean for n=3 values.
Fluxes (J), lag times (ti,) , and skin accumulation (SA)
values for each compound are reported in Table 3-6. Lag time
refers to the intersection of the linear, or "steady-state,"
region of each graph with the time (x) axis, and it is the
time required for establishing a uniform concentration
gradient within the skin.65 The percentages of total 5FD

Cumulative Amount (|imol)
75
Time (h)
Figure 3-4. Plots of cumulative amount of total 5FU species
that diffused (Jlmol) versus time (h) for
compounds 7, 8, 9, and 5FU.

Cumulative Amount (nmol)
76
Time (h)
Figure 3-5. Plots of cumulative amount of total 5FU species
that diffused (p.mol) versus time (h) for
compounds 10, 11, 12, and 5FU.

77
Table 3-6. Fluxes (J) , lag times (ti,) , and skin accumulation
(SA) values for 1-acyl derivatives.
Compound
J(±SD)a
(Hmol/cm2/h)
tL
SA (±SD) a
(Umol)
5FÜ
0.24 (0.09)
13
3.7(0.9)
7
9.3(0.3)
10
68(10)
8
4.3 (0.1)
13
69(10)
9
1.3(0.2)
12
8.2(2.7)
10
1.0(0.1)
9.0
16(4)
11
1.1(0.0)
5.6
11(3)
12
0.60(0.01)
6.5
12(3)
aMean ± standard deviation for n=3 values.

78
Table 3-7. Second application fluxes (J) and lag times (tx,)
for 1-acyl derivatives.
Compound
Ja(±SD)t>
(|lmol/cm2/h)
tL(h)
5FU
1.2(0.2)
1.2
7
1.6(0.0)
0.6
8
1.2(0.2)
0.1
9
1.0 (0.0)
0.6
10
0.80 (0.03)
0.8
11
0.47 (0.02)
0.2
12
0.72 (0.11)
0.4
aFlux of 0.4 M theopylline from propylene glycol.
bMean ± standard deviation for n=3 values.

79
present as prodrug are presumed to be zero since the half-
lives of these prodrugs under diffusion cell conditions are
only three to five minutes.
The improvement in skin penetration of 5FU from the
1-acyl derivatives is substantial. The best compound,
l-acetyl-5FU (7), shows an increase in flux of nearly 40
times when compared to 5F0. As chain length increases, the
fluxes decrease, but even l-octanoyl-5FU (12) improves
transdermal delivery of 5FU by two and one-half times.
Skin accumulation values are also much higher for the
1-acyl derivatives than for 5FU. They show a decrease with
increasing chain length, but the value for l-butyryl-5FU (9)
is lower than expected. As noted earlier, the lipid
solubility of compound 9 was also less than expected when
compared to the compounds around it in the 1-acyl series.
Therefore, the low skin accumulation value for compound 9 may
be due to its low affinity for the lipid regions of the skin.
Overall, skin accumulation is higher for this series than for
the 1-alkyloxycarbonyl series while skin penetration data for
the two series are similar. Rapid hydrolysis of the more
lipid-soluble 1-acyl derivatives to highly polar 5FU as they
partition into the skin may indicate that only 5FU is
diffusing through the remaining lipid regions of the skin.
This may effectively "lock in" large amounts of 5FU leading
to high skin accumulation values.
Second application fluxes and lag times are reported in
Table 3-7. Skin penetration by theophylline from propylene

80
glycol, the standard drug-vehicle combination, after
treatment with the 1-acyl derivatives is similar to that
following treatment with 5FU. In fact, the longer-chain
derivatives in this series actually appear to have a
protective effect on the skin since they cause less damage
than 5FU itself.
The stability of the prodrugs in the IPM formulations
was assessed by NMR analysis of the donor phases. After a
minimum of five days from the time the suspensions were
prepared until their 1H NMR spectra were recorded, including
at least twelve hours during which the formulations were in
contact with the skins, the 1-acyl derivatives were found to
be intact with no evidence of 5FU formation.
Snmma ry
The 1-acyl derivatives of 5FU exhibited decreased
melting points and increased lipid solubilities when compared
to 5FU. Aqueous solubility reached a maximum for 1-acetyl-
5FU (7) and decreased from there with increasing chain
length. Skin penetration was also maximized for compound 7
while skin accumulation values were highest and essentially
the same for compound 7 and l-propionyl-5FU (8). As was also
the case for the 1-alkyloxycarbonyl series, this demonstrates
that both lipid and aqueous solubilities are important for
predicting transdermal and dermal delivery of 5FU prodrugs.
The 1-acyl derivatives may be better candidates for dermal

81
delivery purposes than the 1-alkyloxycarbonyl derivatives
because of their rapid hydrolysis and higher skin
accumulation values. The partitioning-based method for
determining aqueous solubilities of chemically unstable
prodrugs was used to estimate the aqueous solubilities of the
1-acyl derivatives since this could not be accomplished by
other methods.

CHAPTER 4
1,3-BIS-ACYL DERIVATIVES
Tnt rodnot i on
The 1,3-bis-acyl derivatives of 5FU are double prodrugs.
The acyl groups at the N^-position are rapidly hydrolyzed
leaving the 3-acyl derivatives. The 3-acyl derivatives will
be discussed in Chapter 5. The rate of Ni-deacylation for
the 1,3-bis-acyl derivatives is even faster than hydrolysis
of the 1-acyl derivatives. This can best be explained by
comparing pKa values for the 3-acyl derivatives and 5FU. The
3-acyl derivatives have reported pKa values of 7.1 to 7.226
while the pKa for the first ionization of 5FU is 8.0. Thus,
the leaving group potential of the N3-acyl anions is greater
than the 5FO anion, and the rates of N1-deacylation are
faster for the 1,3-bis-acyl derivatives.
The 1,3-bis-acyl derivatives were chosen as prodrug
candidates in order to study a series in which both hydrogen¬
bonding groups were masked. It was anticipated that these
compounds would be more lipid soluble and less water soluble
than the other series and would be useful for comparison with
the other series.
82

83
Table 4-1. Structures of 1,3-bis-acyl derivatives.
Compound Ri,R2
1.3-bis-acetyl-5FU (13) -CH3
1.3-bis-propionyl-5FO (14) -CH2CH3
1.3-bis-butyryl-5FU (15) -(CH2)2CH3
1.3-bis-valeryl-5Fa (16) -(CH2)3CH3

84
Four straight-chain 1,3-bis-acyl derivatives were
selected for study. The derivatives and their structures are
shown in Table 4-1.
Materials and Methods
Melting points (mp) were determined with a Thomas-Hoover
capillary melting point apparatus and are uncorrected.
Elemental microanalyses were obtained for all novel compounds
through Atlantic Microlab, Incorporated in Norcross, Georgia.
Proton nuclear magnetic resonance (1H NMR) spectra were
obtained at 90 MHz on a Varian EM-390 spectrometer. Chemical
shifts (5) are reported in parts per million (ppm) from the
internal standard, tetramethylsilane (TMS). Coupling
constants (J) are expressed in cycles per second (Hz).
Infrared (IR) spectra were recorded with a Perkin-Elmer 1420
spectrophotometer and absorbances are reported in cm"1.
Ultraviolet (UV) spectra were obtained with a Cary 210 or
Shimadzu UV-265 spectrophotometer. Maximum absorbances are
reported in nm along with the molar absorptivities (E) in
L/mol. Single-crystal X-ray analysis was obtained for
1.3-bis-acetyl-5FU through Hoffmann-La Roche in Nutley, NJ.
1.3-bis-Acvl-5-fluorouracil (general procedure)
To 1.31 g (0.01 mol) of 5FU suspended in acetonitrile
(20 mL) was added 1.01 g of triethylamine (0.01 mol) in

85
acetonitrile (5 mL). The mixture was stirred continuously at
0 °C while 0.011 mol of the appropriate acid chloride in
acetonitrile (5 mL) was added dropwise over 5-10 minutes.
The above sequence was repeated until 0.03 mol of
triethylamine and 0.033 mol of acid chloride were added, then
the mixture was stirred for an additional 30 minutes at 0 °C.
Alternate addition of base and acylating agent was found to
increase yield and decrease formation of colored side
products that were difficult to remove. The mixture was
filtered, and the residue was washed with acetonitrile
(25 mL). The combined acetonitrile solutions were evaporated
under reduced pressure, and the solid residue was
crystallized from an appropriate solvent or solvent
combination.
1.3-bis-Acetyl-5-f luorourac.il (13)
Crystallization from ether gave 1.69 g of 13 (79%): mp
112-3 °C (lit.25 mp 111-3 °C) ; IR (KBr) 1680, 1695, 1740,
1750, and 1795 cm"1 (C=0); *H NMR (CDC13) 5 2.58 (s, 3H,
3-CH3), 2.72 (s, 3H, I-CH3), and 8.23 (d, J=7 Hz, 1H, C6-fi);
UVmax (CH3CN) 262 nm (E=9.75xl03) .
1.3-bis-Propionvl-5-fluorouracil (14)
Crystallization from ether gave 1.82 g of 14 (75%): mp
100-1 °C; IR (KBr) 1690, 1725, and 1795 cm'1 (CDCI3) 8 1.25 (t, J=7 Hz, 3H, 3-CH3), 1.28 (t, J=7 Hz, 3H,
I-CH3), 2.85 (q, J=7 Hz, 2H, 3-COCH2), 3.11 (q, J=7 Hz, 2H,

86
1-C0CÜ2), and 8.25 (d, J=7 Hz, 1H, C6-ü) ; UVmax (CH3CN) 262 nm
(6=9.65xl03) .
Anal. Calc, for C10H11FN2O4: C, 49.59; H, 4.58; N, 11.57.
Found: C, 49.44; H, 4.59; N, 11.53.
1.3-bis-Butyryl-5-fluorouracil (JJLL
Extraction of the residue with hot low-boiling petroleum
ether gave a cloudy solution which cleared upon cooling to
0 °C and produced a yellow resinous sediment. The mixture
was allowed to warm to room temperature, and the clear
solution was decanted from the sediment. The solution was
cooled again to 0 °C and crystallization gave 1.70 g of 15
(63%) : mp 48-9 °C (lit.26 mp 47.5-48.5 °C) ; IR (KBr) 1685,
1710, 1735, and 1795 cm'1 (C=0); iH NMR (CDCI3) 8 1.00 (t,
J=7 Hz, 3H, 3-CH3), 1.03 (t, J=7 Hz, 3H, I-CH3) , 1.6-1.9 (m,
4H, 3-COCH2CÜ2 and I-COCH2CIÍ2) , 2.80 (t, J=7 Hz, 2H, 3-COCÜ2) ,
3.07 (t, J=7 Hz, 2H, I-COCH2) , and 8.23 (d, J=6 Hz, 1H, C6-H);
UV max (CH3CN) 262 nm (e=l. 051xl04) .
1.3-bis-Valery1-5-fluorouracil (16)
Extraction of the residue with hot low-boiling petroleum
ether gave a cloudy solution which cleared upon cooling to
0 °C and produced a brown resinous sediment. The mixture was
allowed to warm to room temperature, and the clear solution
was decanted from the sediment. The solution was cooled
again to 0 °C and crystallization gave 2.24 g of 16 (75%) :
mp 47-8 °C; IR (KBr) 1685, 1710, 1735, and 1795 cm'1 (C=0);
1H NMR (CDCI3) 5 0.95 (t, J=7 Hz, 6H, 3-CH3 and I-CH3) ,

87
1.3-1.8 J=7 Hz, 2H, 3-COCÜ2)» 3.08 (t, J=7 Hz, 2H, I-COCH2) , and 8.22
(d, J=7 Hz, 1H, C6-H); UVmax (CH3CN) 262 nm (e=l.071xl04).
Anal. Calc, for C14H19FN2O4: C, 56.37; H, 6.42; N, 9.39.
Found: C, 56.29; H, 6.48; N, 9.33.
Lipid solubility
Lipid solubilities were determined using isopropyl
myristate (IPM), a commercial vehicle used in cosmetics and
topical medicináis,94 as the lipid solvent. The use of IPM as
a model lipophilic vehicle in skin penetration studies is
well established.77, 95
Three suspensions of each derivative were stirred at
22±1 °C for 48 hours. The suspensions were filtered through
0.45 Hm nylon filters, and the saturated solutions were
diluted in acetonitrile and analyzed by UV spectroscopy.
Solubilities were calculated using Beer's Law:
A = e-C-d (1)
where A is the absorbance, £ is the molar absorptivity, C is
the concentration in mol/L,and d is the path length of the
cuvette in cm. Molar absorptivities were predetermined in
triplicate in acetonitrile at 262 nm.
Aqueous .Solubility
Because of the chemical instability of the 1,3-bis-acyl
derivatives, direct measurement of aqueous solubilities for

88
these prodrugs was not attempted. A comparison of the direct
and partitioning-based methods for determining aqueous
solubility was presented in Chapter 2 for the 1-alkyloxy-
carbonyl derivatives.
Partition Coefficients
The partitioning-based method for determining aqueous
solubility utilized the saturated IPM solutions from the
lipid solubility study. For most compounds, equal volumes
(1 mL) of saturated IPM solution and 0.05 M acetate buffer
(pH=4.0) were used. The use of equal or near-equal phase
volumes is known to facilitate rapid equilibrium.96 The two
phases were mixed thoroughly for ten seconds and allowed to
separate for 60 seconds. A preliminary study showed that
there was virtually no difference in partition coefficient
(PC) values when partitioning was carried out for 10, 20, or
30 seconds (see Chapter 3). The IPM layers were diluted in
acetonitrile and analyzed by UV spectroscopy. The IPM-buffer
partition coefficients were calculated as follows:
PC = Aafter/ (Abefore-Aafter) ‘^aq/Vipm (2)
where Aafter is the absorbance from the IPM layer after
partitioning, Atefore is the absorbance from the IPM layer
before partitioning, Vaq is the volume of the aqueous phase,
and Vipm is the volume of the IPM phase. Estimated aqueous
solubilities (SAq) were calculated from the IPM solubility

89
(Sipm) and the partition coefficient:
Saq = Sipm/PC (3)
Partitioning was carried out in triplicate for a fixed volume
ratio for each derivative. For those compounds with large
differences in solubility in one phase relative to the other,
volume ratios (IPMrbuffer) other than 1:1 were necessary, but
the ratio never exceeded 10:1 or 1:10.
Hydrolysis Kinetics
Hydrolysis rates have previously been reported for
several members of this homologous series.26 In the present
study, hydrolysis rates for the conversion of 1,3-bis-acetyl-
5FU (13) to 3-acetyl-5FU were determined at 32 °C in 0.05 M
phosphate buffer (pH=7.1, 1=0.12) and in the same buffer with
0.11% formaldehyde (3.6xl0-2 M) . The rate in the presence of
formaldehyde was determined for comparison with the rate in
plain buffer since formaldehyde was used as a preservative in
the diffusion cell experiments described in the following
section.
The hydrolyses were followed by UV spectroscopy at
266 nm where the absorbance decrease accompanying conversion
of the 1,3-bis-acyl derivatives to 3-acyl derivatives was
maximized. Hydrolysis was initiated by adding 60 |1L of a
stock solution of the derivative in acetonitrile to 3 mL of
buffer prewarmed to 32 °C in a thermostated quartz cuvette to
give final concentrations of ~1.8xl0“4 M. Absorbances were

90
recorded at appropriate intervals and pseudo-first-order rate
constants were determined from the expression:
ln(At-Aoo) = ln(A0-A»)-kt (4)
where At is the absorbance at some time=t, hoo is the
absorbance at t=°°, Aq is the absorbance at t=0, k is the
pseudo-first-order rate constant, and t is the time. The
hydrolyses were sufficiently fast and the 3-acyl products
were stable enough to allow experimental determination of A».
The slopes, -k, of linear plots of ln(At-A«,) versus time were
determined by linear regression. The half-lives (ti/2) were
calculated from
tl/2 = 0.693/k (5)
Each hydrolysis reaction was run in triplicate and was
followed for a minimum of three half-lives. The correlation
coefficients were >0.999.
Skin Penetration Studies
Diffusion cell experiments were performed to measure the
transdermal delivery of 5F0 and the 5FU prodrugs. Franz-type
diffusion cells from Crown Glass in Somerville, NJ with 4.9
cm2 donor surface area and 20 mL receptor phase volume were
used for this purpose. The full-thickness skins were
obtained from female hairless mice (SKH-hr-1) from Temple
University Skin and Cancer Hospital.
The mice were killed by cervical dislocation, their
skins were removed immediately by blunt dissection, and

91
dorsal sections were mounted in the diffusion cells. The
dermal sides of the skins were placed in contact with
receptor phase which contained 0.05 M phosphate buffer
(pH=7.1, 1=0.12) with 0.11% formaldehyde as a preservative.
The effectiveness of formaldehyde for this purpose has
recently been documented.89 The receptor phases were stirred
continuously and kept at constant temperature (32 °C) by a
circulating water bath. A preapplication period of 48 hours
was established to uniformly condition the skins and to
remove water-soluble UV-absorbing materials. The receptor
phases were changed three times during this period, and
control experiments from earlier studies have shown that this
procedure effectively removes those materials.97 The
epidermal sides of the skins were exposed to the air and were
left untreated during this period.
After the preapplication period, 0.5 mL aliquots from
suspensions of the prodrugs in IPM were applied to the
epidermal sides of the skins. The IPM suspensions were
stirred at 22±1 °C for 48 hours prior to application to
ensure that saturation was attained. Total concentrations of
the IPM suspensions ranged from 0.5 M to 2.0 M with enough
excess solid present to maintain saturation for the duration
of the application period (see below). Each drug-vehicle
combination was run in triplicate.
Samples were taken from the receptor phases at 4, 8, 12,
21, 24, 27, 30, 33, 36, 45, and 48 hours after donor phase
application. The receptor phases were changed following

92
removal of each sample so that "sink" conditions were
maintained. Samples were analyzed for total 5FU species that
had diffused by UV spectroscopy (£-7.13xl03 at 266 nm) after
allowing at least 72 hours for complete prodrug hydrolysis.
Cumulative amounts of total 5F0 species that diffused (Jlmol)
were plotted against time (h), and the slopes of the linear,
"steady-state" regions were calculated using linear
regression. The slopes, when divided by 4.9 (the area of the
donor surface in cm2) , gave the "steady-state" fluxes
(Hmol/cm2/h). Because of the rapid chemical hydrolysis of
the 1,3-bis-acyl derivatives, no attempt was made to analyze
the receptor phases for prodrug content.
Donor phases were changed every twelve hours and were
set aside for XH NMR analysis. Stability of the prodrugs in
IPM was determined from the chemical shift of C6-H. In
dimethylsulfoxide-d6, the C6-H signal for 5FU appears at
5=7.73 ppm, and for the 3-acyl derivatives, it appears at
5=7.90. For each of the 1,3-bis-acyl derivatives, the same
signal in dimethylsulfoxide-dg appears at 5>8.40 ppm. Since
this area of the spectrum is free from interference by IPM
absorbances, the three signals can be identified and
quantified if necessary.
Following removal of the donor phases after the 48-hour
application period, the epidermal sides of the skins were
washed three times with 5 mL portions of methanol to remove
all remnants of prodrug and vehicle from the skin surfaces.
This was accomplished quickly (<3 min) to minimize contact

93
time between the skins and methanol. The receptor phases
were changed again, and the dermal sides were kept in contact
with the fresh buffer for 23 hours while the epidermal sides
were again left exposed to the air. After this "leaching"
period, another sample was taken from each cell to measure
the skin accumulation of total 5FU species.
Second applications to the epidermal sides of the skins
were made after the "leaching" period with a standard drug-
vehicle suspension . Theophylline in propylene glycol
(0.4 M) was applied to assess the damage to the skins from
application of the initial drug-vehicle combinations.
Samples were taken at 1, 2, 3, 5, 7, 9, and 11 or 12 hours
after application. The samples were analyzed for
theophylline by UV spectroscopy (£=1.02xl04 at 271 nm) and
second application fluxes were determined as described above.
Results and Discussion
Synthesis and Structure Determination
The known 1,3-bis-acyl-5FO derivatives have melting
points25-26 and 1H NMR spectra25 in agreement with those
reported in the literature. The structures of novel
compounds were assigned by comparison of their !h NMR spectra
with those of the known homologs. Elemental microanalyses
were obtained for the novel compounds and were within
acceptable limits (±0.4%).

94
Structural analysis of the 1,3-bis-acyl derivatives was
helpful in resolving the apparent discrepancy between acidity
and reactivity at the two attachment sites. In Chapter 1, it
was noted that the monoanion of 5FU is actually a mixture of
N1- and N3-anions,30 and that a comparison of ionization
constants for 5FU derivatives identically substituted at the
N1- and N3-positions suggests that the N3-position is more
acidic.22'30 This is supported by spectral studies in aqueous
media (IR,306 gv,106-7 and ÍH NMR106) which showed that for the
5FU monoanion, the N3-anion is the dominant form. Formation
of the monoanion, however, leads exclusively to N1-acylated
products (Chapter 2 and Chapter 3). In addition, the
1,3-bis-acyl derivatives undergo rapid N1-deacylation during
hydrolysis.26 Ordinarily, this would imply that the N1-anion
is more stable than the N3-anion and that the the Ni-position
is more acidic.
The structure of 1,3-bis-acetyl-5FU (13) from single
crystal X-ray analysis is shown in Figure 4-1. The carbonyl
bond in the Ni-acetyl group is positioned cis to the C6-H bond
as it is in l-acetyl-5FU (7). As expected, the chemical
shifts of C6-H in chloroform-d show the same anisotropic
effect for compounds 7 and 13. The N3-acetyl group, however,
is shown to be nearly perpendicular to the plane of the 5FU
ring. Apparently, it is being forced out of plane by the
carbonyl groups at the C2- and expositions.

95
Figure 4-1. X-ray structure of 1,3-bis-acetyl-5FU.
Source: Hoffman-La Roche, Nutley, NJ.

96
The steric constraints at the N3-position may explain
the lower than expected reactivity at this site. The C2- and
C4-carbonyl groups make acylation more difficult by
restricting the orientations available to the relatively
bulky acylating agents as they interact with the 5F0 anion.
Likewise, slower hydrolysis rates are due to partial negative
charges on oxygen at C2=0 and C4=0 which impede the approach
of hydroxide ions to the N3-acyl carbonyl group when it is
perpendicular to the plane of the 5FU ring. Thus, although
the N3-anion is a better leaving group than the N3-anion,
steric hindrance prevents more rapid hydrolysis of the
N3 -acyl group than the Ni-acyl group in the 1,3-bis-acyl
series.
Infrared (IR) spectra for the 1,3-bis-acyl and 3-acyl
derivatives are also consistent with the conformation
depicted in Figure 4-1. The carbonyl stretching frequency
for the N3-acyl group is 25 to 50 cm'1 higher than those
associated with the 5FU ring or N1-acyl groups. The
perpendicular orientation of the N3-acyl group apparently
prevents electron delocalization from the ring to the
carbonyl bond which means that the inductive effect of the
ring predominates, and the carbonyl stretching frequency is
increased.105

97
â– Sol nhi 1 i ty
Lipid (Sipm) and aqueous (Saq) solubilities for the
1,3-bis-acyl derivatives are presented in Table 4-2 along
with their melting points. The downward trend in melting
point and upward trend in lipid solubility with increasing
chain length is pronounced in these diacylated compounds.
Aqueous solubilities were only determined for the first
two derivatives, and both are at least an order of magnitude
less than 5FU. The longer chain derivatives exhibit such
high lipid solubilites compared to their expected aqueous
solubilities that partitioning could not be done without
using unreasonable IPM-Buffer ratios.
Hydrolysis Kinptir^
Hydrolysis of 1,3-bis-acetyl-5FU (13) to 3-acetyl-5FU
was followed by UV spectroscopy in 0.05 M phosphate buffer
(pH=7.1, 1=0.12) at 32 °C with and without 0.11% formaldehyde
(3.6xl0-2 M) . Disappearance of compound 13, indicated by
ln(At-A«), is plotted versus time (min) in Figure 4-2. The
linearity of the plots suggests that hydrolysis of 13 follows
first-order kinetics in the presence and absence of
formaldehyde. Pseudo-first-order rate constants (k) and
half-lives (ti/2) from the linear plots are presented in
Table 4-3. The much greater stability of the product

98
Table 4-2. Melting points (MP), lipid solubilities (Sjpm) / and
aqueous solubilities (Saq) for 1,3-bis-acyl derivatives.
Compound
MP
(°C)
sIPMa
(mM)
Saq5
(mM)
Saqc
(mM)
5FU
280-2
0.049
96
-
13
112-3
26
-
9.0
14
100-1
72
-
3.3
15
48-9
625
-
-
16
47-8
1180
“
~
aStandard
deviation from
the means
were within ±4%
for IPM
solubilities.
bSolubility determined by direct method.
cStandard deviations from the mean were within ±4% for
experimental values used to calculate partition coefficients
and estimated aqueous solubilities.

In (At-A'
99
Plain Buffer
Formaldehyde Buffer
Time (min)
Figure 4-2. Plots of ln(At-Ao=) versus time (min) for
hydrolysis of 1,3-bis-acetyl-5FU in 0.05 M
phosphate buffer (pH=7.1, 1=0.12) with and
without formaldehyde at 32 °C.

100
Table 4-3. Pseudo-first-order rate constants (k) and half'
lives (11/2) for hydrolysis of 1,3-bis-acetyl-5FU in
0.05 M phosphate buffer (pH=7.1, 1=0.12) with and
without formaldehyde at 32 °C.
Compound
Formaldehyde k(±SD)a ti/2
(M) (min-1) (min)
13
13
0 0.802(0.010) 0.86
3.6xl0-2 0.868(0.039) 0.80
aMean ± standard deviation for n=3 values.

101
3-acyl derivatives compared to the 1,3-bis-acyl derivatives
allows the rates of hydrolysis for the N1-acyl groups to be
followed for three to four half-lives before detectable
hydrolysis of the N3-acyl groups is observed. The hydrolysis
of the 3-acyl derivatives will be discussed in Chapter 5.
Hydrolysis of compound 13 is clearly faster in the
presence of formaldehyde indicating that general base
catalysis by formaldehyde hydrate may be involved in this
series as well (see Chapter 3). Complete pH-rate profiles
for selected acyl derivatives of 5FU can be found in the work
of Buur and Bundgaard.26
Skin Penetration
Skin penetration data from the diffusion cells are
plotted as cumulative amount of total 5FU species that
diffused (|lmol) versus time (h) . In Figure 4-3, results for
1,3-bis-acetyl-5FU (13) and 1,3-bis-propionyl-5FD (14) are
compared to 5FU itself. In Figure 4-4, results for 1,3-bis-
butyryl-5FU (15) and 1,3-bis-valeryl-5FU (16) are compared to
5FU. Error bars correspond to the standard deviation from
the mean for n=3 values.
Fluxes (J), lag times (tx,) , and skin accumulation (SA)
values for each compound are reported in Table 4-4. The
highest flux obtained for the 1,3-bis-acyl series is for
1,3-bis-acetyl-5FU (13) with nearly an order of magnitude
improvement over 5FU. Overall, the 1,3-bis-acyl derivatives

Cumulative Amount (nmol)
102
Time (h)
Figure 4-3. Plots of cumulative amount of total 5FU species
that diffused ((imol) versus time (h) for
compounds 13, 14, and 5FU.

Cumulative Amount (|imol)
103
Time (h)
Figure 4-4. Plots of cumulative amount of total 5FU species
that diffused (|Imol) versus time (h) for
compounds 15, 16, and 5FU.

Table 4-4. Fluxes (J) , lag times (ti), and skin accumulation
(SA) values for 1,3-bis-acyl derivatives.
Compound
J(±SD)a
(|i.mol/cm2/h)
tL
(h)
SA(±SD)a
(Umol)
5FU
0.24 (0.09)
13
3.7(0.9)
13
2.2(0.5)
11
10(2)
14
0.69(0.06)
6.2
4.5(1.5)
15
0.98(0.06)
-2.3
12(2)
16
0.95(0.05)
-2.8
8.8(1.8)
aMean ± standard deviation for n=3 values.

105
Table 4-5. Second application fluxes (J) and lag times (tl)
for 1,3-bis-acyl derivatives.
Ja(+SD)b
Compound (|lmol/cm2/h) tt,(h)
5FU
1.2(0.2)
1.2
13
1.6(0.1)
0.7
14
1.6(0.3)
0.7
15
1.1(0.1)
0.9
16
0.87 (0.14)
0.8
aFlux of 0.4 M theopylline from propylene glycol.
bMean ± standard deviation for n=3 values.

106
are much less effective at delivering 5FU transdermally than
the 1-alkyloxycarbonyl and 1-acyl series. The results for
this highly lipophilic, but poorly aqueous-soluble series
demonstrate the importance of biphasic solubility for
maximizing skin penetration.
The negative lag times observed for 1,3-bis-butyryl-5FU
(15) and 1,3-bis-valeryl-5FU (16) may have been caused by a
change in the physical state of the donor phase. A large
amount of solid was required in formulating the donor phases
for compounds 15 and 16 due to their high solubility in IPM.
After a few hours of skin contact at 32 °C, it became
difficult to determine if excess solid from these low-melting
derivatives was still present in the donor phases since their
appearance was more like a gel than a suspension.
Skin accumulation values are greater for the 1,3-bis-
acyl series than for 5FU, but the improvement is small. The
best derivatives, compounds 13, 15, and 16, have skin
accumulation values less than three times higher than 5FU.
Second application fluxes and lag times are reported in
Table 4-5. Skin penetration by theophylline from propylene
glycol, the standard drug-vehicle combination, after
treatment with the 1,3-bis-acyl derivatives is similar to
that following treatment with 5FU.
The stability of the prodrugs in the IPM formulations
was assessed by 1H NMR analysis of the donor phases. After a
minimum of five days from the time the suspensions were
prepared until their *H NMR spectra were recorded, including

107
at least twelve hours during which the formulations were in
contact with the skins, the 1,3-bis-acyl derivatives were
found to be intact with no evidence of formation of the
corresponding 3-acyl derivatives or 5FU.
â– Snmma ry
The 1,3-bis-acyl derivatives of 5FU exhibited decreased
melting points and greatly increased lipid solubilities when
compared to 5FÜ. Aqueous solubility was at best an order of
magnitude less than 5FU and was highest for 1,3-bis-acetyl-
5FU (13). Flux was also maximized for compound 13 which
again shows the importance of aqueous solubility for
optimizing skin penetration within a homologous series of
more lipid-soluble prodrugs. Biphasic solubility data
appears to be useful for series to series comparisons as
well. The highly lipophilic 1,3-bis-acyl derivatives were
less effective at delivering 5FU transdermally than the less
lipid-soluble, but more aqueous-soluble series such as the
1-alkyloxycarbonyl and 1-acyl series. The partitioning-based
method for determining aqueous solubilities of chemically
unstable prodrugs was used to estimate the aqueous
solubilities of two of the 1,3-bis-acyl derivatives since
this could not be accomplished by other methods. The
difference in the stabilities of the N1- and N3-acyl groups of
the 1,3-bis-acyl derivatives has been rationalized based on
an X-ray crystallography study which showed that the N3-acyl

108
group is oriented perpendicular to the plane of the 5FU ring
and is sterically and electronically hindered from
nucleophilic attack.

CHAPTER 5
3-ACYL DERIVATIVES
Introduction
The 3-acyl derivatives of 5FU are more stable chemically
than either the 1-acyl or 1,3-bis-acyl derivatives.26
Enzymatic hydrolysis proceeds at a faster rate than chemical
hydrolysis, and this difference is more pronounced for the
short-chain derivatives of the 3-acyl series26 some of which
may be useful as prodrugs for dermal delivery.
The 3-acyl derivatives were chosen for evaluation
because of their chemical stability and for comparison with
the 1-acyl series. It was known that N1- and N3-substituted
derivatives have different physical-chemical properties,22'29"
30 and it was likely that they would exhibit differences in
skin penetration as well.
Four straight chain 3-acyl derivatives were selected for
study. The derivatives and their structures are presented in
Table 5-1.
109

110
Table 5-1. Structures of 3-acyl derivatives.
R
H
Compound R
3-acetyl-5FU (17) -CH3
3-propionyl-5FU (18) -CH2CH3
3-butyryl-5FU (19) -(CH2)2CH3
3-valeryl-5FU (20) -(CH2)3CH3

Ill
Materials and Methods
Synthesis
Melting points (mp) were determined with a Thomas-Hoover
capillary melting point apparatus and are uncorrected.
Elemental microanalyses were obtained for all novel compounds
through Atlantic Microlab, Incorporated in Norcross, Georgia.
Proton nuclear magnetic resonance (if! NMR) spectra were
obtained at 90 MHz on a Varian EM-390 spectrometer. Chemical
shifts (8) are reported in parts per million (ppm) from the
internal standard, tetramethylsilane (TMS). Coupling
constants (J) are expressed in cycles per second (Hz).
Infrared (IR) spectra were recorded with a Perkin-Elmer 1420
spectrophotometer and absorbances are reported in cm"1.
Ultraviolet (UV) spectra were obtained with a Cary 210 or
Shimadzu UV-265 spectrophotometer. Maximum absorbances are
reported in nm along with the molar absorptivities (6) in
L/mol. Thin layer chromatography (TLC) was run on Brinkman
Polygram Sil G/UV254 0.25 mm silica gel.
3-Acvl-5-fluorouracil (general procedure)
To a solution of an appropriate 1,3-bis-acyl derivative
(0.01 mol) in ether (60-650 mL) at 0 °C was added 0.73 g of
tert-butylamine (0.01 mol) in ether (25 mL) dropwise over ten
minutes with stirring. The solution was stirred for a total
of 30 minutes at 0 °C. The ether solution was evaporated

112
under reduced pressure at room temperature until the volume
was reduced to 25-50 mL. The solution was cooled again to
0 °C until crystallization occurred.
3-Acety1-5-fluorouracil (17)
Crystallization gave 1.22 g of 17 (71%) : mp 115-7 °C dec
(lit.25 mp 114-7 °C) ; IR (KBr) 1650, 1685, 1720, and 1805 cm"1
(C=0); 1H NMR (CDCI3) 8 2.58 (s, 3H, CÜ3) and 7.23 (bs, 1H,
C6-a) ; UVmax (CH3CN) 267 nm (e=6.57xl03) .
3-Propionyl-5-fluorouracil (18)
Crystallization gave 0.89 g of 18 (48%) : mp 102-3 °C dec
(lit.25 mp 99-102 °C); IR (KBr) 1655, 1680, 1730, and
1815 cm-1 (C=0); NMR (CDCI3) 8 1.27 (t, J=7 Hz, 3H, CH3),
2.86 (q, J=7 Hz, 2H, C0Cfi2) > and 7.26 (bs, 1H, C6-E) ; UV^x
(CH3CN) 267 nm (e=6.66xl03) .
3-Butvrvl-5-fluorouracil (19)
Crystallization gave 1.22 g of 19 (61%); mp 111-2 °C
(lit.26 mp 132-4 °C) ; IR (KBr) 1660, 1730, and 1810 cm'1
(C=0); XH NMR (CDCI3) 8 1.02 (t, J=7 Hz, 3H, CH3), 1.6-2.0
(m, 2H, COCH2CE2) , 2.82 (t, J=7 Hz, 2H, COCÍI2) , and 7.26
(bs, 1H, C6-H) ; UV max (CH3CN) 267 nm (e=6.55xl03) .
Anal. Calc, for C8H9FN203: C, 48.00; H, 4.53; N, 14.00.
Found: C, 48.08; H, 4.55; N, 13.93.
3-Valervl-5-fluorouracil (20)
Crystallization gave 1.54 g of 20 (72%) : mp 110-1 °C
dec; IR (KBr) 1665, 1730, and 1815 cm-1 (C=0) ; J-H NMR (CDCI3)
8 0.97 (t, J=7 Hz, 3H, Cfl3) , 1.3-1.9 (m, 4H, COCH2CH2CH2) ,

113
2.84 (t, J=7 Hz, 2H, COCH2) , and 7.23 (bs, 1H, C6-H); UVmax
(CH3CN) 267 nm (e=6.68xl03) .
Anal. Calc, for C9H11FN2O3: C, 50.47; H, 5.18; N, 13.08.
Found: C, 50.73; H, 5.24; N, 12.95.
T.i ni d snl nbi 1 i t~ y
Lipid solubilities were determined using isopropyl
myristate (IPM), a commercial vehicle used in cosmetics and
topical medicináis,94 as the lipid solvent. The use of IPM as
a model lipophilic vehicle in skin penetration studies is
well established.77, 95
Three suspensions of each derivative were stirred at
22±1 °C for 48 hours. The suspensions were filtered through
0.45 |lm nylon filters, and the saturated solutions were
diluted in acetonitrile and analyzed by UV spectroscopy.
Solubilities were calculated using Beer's Law:
A = e-C-d (1)
where A is the absorbance, E is the molar absorptivity, C is
the concentration in mol/L,and d is the path length of the
cuvette in cm. Molar absorptivities were predetermined in
triplicate in acetonitrile at 267 nm.
Aqnerm.g; Solubility
For direct measurement of aqueous solubilities, two or
more suspensions of each derivative were vigorously stirred
in 0.05 M acetate buffer (pH=4.0) at 22±1 °C for 60 minutes.

114
The suspensions were filtered through 0.45 (im nylon filters,
and the saturated solutions were diluted in acetonitrile and
analyzed by UV spectroscopy. Solubilities were calculated
using Beer's Law as previously described.
Partition Coefficients
The partitioning-based method for determining aqueous
solubility utilized the saturated IPM solutions from the
lipid solubility study. For most compounds, equal volumes
(1 mL) of saturated IPM solution and 0.05 M acetate buffer
(pH=4.0) were used. The use of equal or near-equal phase
volumes is known to facilitate rapid equilibrium.96 The two
phases were mixed thoroughly for ten seconds and allowed to
separate for 60 seconds. A preliminary study showed that
there was virtually no difference in partition coefficient
(PC) values when partitioning was carried out for 10, 20, or
30 seconds (see Chapter 3). The IPM layers were diluted in
acetonitrile and analyzed by UV spectroscopy. The IPM-buffer
partition coefficients were calculated as follows:
PC = Aafter/ Í^before“Aafter) '^aq/Vipm (2)
where Aafter is the absorbance from the IPM layer after
partitioning, Abefore is the absorbance from the IPM layer
before partitioning, V^g is the volume of the aqueous phase,
and Vjpm is the volume of the IPM phase. Estimated aqueous
solubilities (S*q) were calculated from the IPM solubility

115
(Sipm) and the partition coefficient:
Saq = Sipm/PC (3)
Partitioning was carried out in triplicate for a fixed volume
ratio for each derivative. For those compounds with large
differences in solubility in one phase relative to the other,
volume ratios (IPM:buffer) other than 1:1 were necessary, but
the ratio never exceeded 10:1 or 1:10.
Hydrolysis Kinetics
Hydrolysis rates have previously been reported for
several members of this homologous series.2^ In the present
study, hydrolysis rates were determined at 32 °C for
3-acetyl-5FU (17) in 0.05 M phosphate buffer (pH=7.1, 1=0.12)
and in the same buffer with 0.11% formaldehyde (3.6xl0~2 M).
The rate in the presence of formaldehyde was determined for
comparison with the rate in plain buffer since formaldehyde
was used as a preservative in the diffusion cell experiments
described in the following section. Two other concentrations
of formaldehyde were studied (3.6x10”^ M and 3.6xl0“4 M) when
it was discovered that products in addition to 5FU were being
formed during hydrolysis of compound 17.
The hydrolyses were followed by high performance liquid
chromatography (HPLC). The HPLC system consisted of a
Beckman model 110A pump with a model 153 UV detector, a
Rheodyne 7125 injector with a 20 |im loop, and a Hewlett-
Packard 3392A integrator. The column was a Lichrosorb RP-8

116
10 (Ira reversed-phase column, 250 mm x 4.6 mra (inside
diameter). The mobile phase contained 10% methanol and 90%
0.025 M acetate buffer (pH=5.0) and was run at 1.0 mL/min.
The column effluent was monitored at 254 nm, and quantitation
was based on peak areas. Standards chromatographed under the
same conditions were used for calibration.
Hydrolysis was initiated by adding 0.4 mL of a stock
solution of compound 17 in acetonitrile to 25 mL of buffer
prewarmed to 32 °C in a constant temperature water bath to
give final concentrations of ~1.8xl0-4 M. Aliquots were
removed at appropriate intervals and chromatographed
immediately. Pseudo-first-order rate constants were
determined from the expression:
ln(Ct) = In (C0) -kt (4)
where Ct is the concentration at some time=t, C0 is the
concentration at t=0, k is the pseudo-first-order rate
constant, and t is the time. The slopes, -k, of linear plots
of ln(Ct) versus t were determined by linear regression. The
half-lives (ti/2) were calculated from
ti/2 = 0.693/k (5)
The hydrolyses of compounds 17 and 18 (3-propionyl-5FU)
in buffer alone were followed by UV spectroscopy at 300 nm
where the absorbance decrease accompanying hydrolysis was
maximized. Hydrolysis was initiated by adding 60 |1L of a
stock solution of the derivative in acetonitrile to 3 mL of
buffer prewarmed to 32 °C in a thermostatted quartz cuvette
to give final concentrations of ~1.8xl0"4 M. Absorbances were

117
recorded at appropriate intervals and pseudo-first-order rate
constants were determined from the expression:
In (At“fioo) = lnfAo-A^J-kt (6)
where At is the absorbance at some time=t, A«, is the
absorbance at t=°°, A0 is the absorbance at t=0, k is the
pseudo-first-order rate constant, and t is the time. The
hydrolyses were sufficiently fast to allow experimental
determination of A«. The slopes, -k, of linear plots of
ln(At~A«,) versus time were determined by linear regression.
The half-lives (ti/2) were calculated from equation (5).
The hydrolysis reactions were run in duplicate or
triplicate and were followed for a minimum of three half-
lives. The correlation coefficients for all reported rate
constants were >0.999 except where noted.
Skin Penetration Studies
Diffusion cell experiments were performed to measure the
transdermal delivery of 5FU and the 5FU prodrugs. Franz-type
diffusion cells from Crown Glass in Somerville, NJ with 4.9
cm2 donor surface area and 20 mL receptor phase volume were
used for this purpose. The full-thickness skins were
obtained from female hairless mice (SKH-hr-1) from Temple
University Skin and Cancer Hospital.
The mice were killed by cervical dislocation, their
skins were removed immediately by blunt dissection, and
dorsal sections were mounted in the diffusion cells. The

118
dermal sides of the skins were placed in contact with
receptor phase which contained 0.05 M phosphate buffer
(pH=7.1, 1=0.12) with 0.11% formaldehyde as a preservative.
The effectiveness of formaldehyde for this purpose has
recently been documented.89 The receptor phases were stirred
continuously and kept at constant temperature (32 °C) by a
circulating water bath. A preapplication period of 48 hours
was established to uniformly condition the skins and to
remove water-soluble UV-absorbing materials. The receptor
phases were changed three times during this period, and
control experiments from earlier studies have shown that this
procedure effectively removes those materials.97 The
epidermal sides of the skins were exposed to the air and were
left untreated during this period.
After the preapplication period, 0.5 mL aliquots from
suspensions of the prodrugs in IPM were applied to the
epidermal sides of the skins. The IPM suspensions were
stirred at 22±1 °C for 48 hours prior to application to
ensure that saturation was attained. Total concentrations of
the IPM suspensions ranged from 0.4 M to 0.6 M with enough
excess solid present to maintain saturation for the duration
of the application period (see below). Each drug-vehicle
combination was run in triplicate.
Samples were taken from the receptor phases at 4, 8, 12,
21, 24, 27, 30, 33, 36, 45, and 48 hours after donor phase
application. The receptor phases were changed following
removal of each sample so that "sink" conditions were

119
maintained. Samples were analyzed for total 5FU species that
had diffused by UV spectroscopy (£=7.13x103 at 266 nm) after
allowing at least 72 hours for complete prodrug hydrolysis.
Cumulative amounts of total 5FU species that diffused (nmol)
were plotted against time (h) , and the slopes of the linear,
"steady-state" regions were calculated using linear
regression. The slopes, when divided by 4.9 (the area of the
donor surface in cm2) , gave the "steady-state" fluxes
(Hmol/cm2/h). In a separate experiment, HPLC was used to
determine intact prodrug content in the receptor phases at
each sampling time. Mobile phase containing 18-50% methanol
in 0.025 M acetate buffer (pH=5.0) was used with the system
described earlier. Aliquots were removed and chromatographed
immediately after the samples were taken. Prodrug fluxes
were calculated in the same manner as fluxes for total 5FU.
Donor phases were changed every twelve hours and were
set aside for XH NMR analysis. Stability of the prodrugs in
IPM was determined from the chemical shift of C6-H. In
dimethylsulfoxide-d6, the C6-H signal for 5FU appears at
8=7.73 ppm. For each of the 3-acyl derivatives, the same
signal in dimethylsulfoxide-d6 appears at 5=7.90. Since this
area of the spectrum is free from interference by IPM
absorbances, the two signals can be identified and quantified
if necessary.
Following removal of the donor phases after the 48-hour
application period, the epidermal sides of the skins were
washed three times with 5 mL portions of methanol to remove

120
all remnants of prodrug and vehicle from the skin surfaces.
This was accomplished quickly (<3 min) to minimize contact
time between the skins and methanol. The receptor phases
were changed again, and the dermal sides were kept in contact
with the fresh buffer for 23 hours while the epidermal sides
were again left exposed to the air. After this "leaching"
period, another sample was taken from each cell to measure
the skin accumulation of total 5FU species.
Second applications to the epidermal sides of the skins
were made after the "leaching" period with a standard drug-
vehicle suspension . Theophylline in propylene glycol
(0.4 M) was applied to assess the damage to the skins from
application of the initial drug-vehicle combinations.
Samples were taken at 1, 2, 3, 5, 7, 9, and 11 or 12 hours
after application. The samples were analyzed for
theophylline by UV spectroscopy (e-1.02xl04 at 271 nm) and
second application fluxes were determined as described above.
Results and Discussion
Synthesis and Structure Determination
The known 3-acyl-5FU derivatives have melting points25-26
and iH NMR spectra25 in agreement with those reported in the
literature. The structures of novel compounds were assigned
by comparison of their i-H NMR spectra with those of the known

121
homologs. Elemental microanalyses were obtained for the
novel compounds and were within acceptable limits (±0.4%).
Differences in the spectral properties between N1- and
N3-acylated derivatives were discussed in detail in Chapter 2
and Chapter 3. The 3-acyl derivatives are synthesized by
t^-deacylation of the 1,3-bis-acyl derivatives, and ^-H NMR and
IR spectra confirm that it is the N3-acyl groups that are
present in the products.
The melting points of the 3-acyl derivatives are rate
dependent and exhibit an incomplete melt followed by
resolidification and a complete melt at 40 to 100 °C above
the initial thermal event suggesting thermal decomposition of
the prodrugs. The literature melting points for 3-propionyl-
5FU (18) are 99-102 °C25 and 113-4 °C.26 In the present
study, the melting point for compound 18 was 102-3 °C when
the temperature was raised by 2 °C per minute, but if the
rate was increased to 5 °C per minute, the observed melting
point was 111-3 °C.
The thermal decomposition of 3-acetyl-5FU (17) was
studied by dissolving 50 mg of compound 17 in 10 mL of ethyl
acetate in a beaker and heating gently to evaporate the
solvent. The temperature was increased slowly until the
residual solid had melted completely. The beaker was then
removed from the heat to allow resolidification to occur.
The decomposition was followed by TLC on silica gel
using ethyl acetate as the eluent. The TLC showed no
decomposition until after the solvent had completely

122
evaporated at which time gradual disappearance of compound 17
(Rf=0.43) began with formation of both l-acetyl-5FU (Rf=0.62)
and 1,3-bis-acetyl-5FU (Rf=0.66) as well as 5FU (Rf=0.30).
Opon resolidification of the melt, its 1H NMR spectrum showed
that compound 17 had been converted to l-acetyl-5FU (~70%) ,
5FU (-30%) and a trace of 1,3-bis-acetyl-5FU. Percentages
were calculated from integration of the corresponding C6-H
signals.
Similar studies showed that melting 1,3-bis-acetyl-5FO
alone produced l-acetyl-5FU (-40%) and 1,3-bis-acetyl-5FU
(-60%) while melting equivalent amounts of 1,3-bis-acetyl-5FU
and 5FU produced l-acetyl-5FU (-75%), 5FU (-20%), and 1,3-
bis-acetyl-5FU (-5%).
Since rate and duration of heating were not controlled
in these experiments, conclusions regarding the relative
amounts of decomposition products formed can not be drawn.
However, the data do suggest that thermal N3-deacylation
occurs from both 3-acetyl-5FU and 1,3-bis-acetyl-5FU, and
overall, there is a net loss of acetyl groups. Some of the
N3-acetyl groups (or even N1-acetyl groups) could be lost to
ketene formation as shown in Figure 5-1. There is also a net
gain of N1-acetyl groups when 3-acetyl-5FU is heated alone or
a mixture of 1,3-bis-acetyl-5FO and 5FU is heated. This
indicates that a rearrangement has occurred. The TLC and
3H NMR data suggest that the rearrangement is to some degree
intermolecular since some 1,3-bis-acetyl-5FU is produced when
3-acetyl-5FU is heated alone, but an intramolecular

123
H
Figure 5-1. Possible scheme for thermal decomposition of
3-acetyl-5FU to 5FU.

124
Figure 5-2. Possible scheme for thermal intramolecular
rearrangement for 3-acetyl-5FU to l-acetyl-5FU.

125
rearrangement from N3- to N1-acetyl is possible in this case
as well. In Figure 5-2, an intramolecular rearrangement
scheme is presented in which a transient 02-acyl intermediate
is proposed.
The existence of intramolecular 1,3 acyl migrations from
both N to 0 and 0 to N has been documented.109 Of the several
possible mechanisms for the 0 to N thermal rearrangement, one
that is preferred involves rate-determining nucleophilic
attack by the nitrogen lone pair on the carbonyl carbon with
partial C-0 bond cleavage to relieve the strain associated
with formation of a four-membered ring.110 This synchronous
mechanism is depicted in Figure 5-2 for both the N to 0 and
0 to N rearrangements. The perpendicular orientation of the
N3-acetyl group to the 5F0 ring would facilitate a
rearrangement of this type. All steps are shown as
reversible since acyl or aroyl rearrangements are potentially
reversible whether they are intramolecular or
intermolecular.111'3 It is worth noting that similar melting
behavior to that seen for the 3-acyl derivatives in the
present study was also observed for the 0 to N benzoyl
migration in 6-phenanthridinone.113 Further discussion
regarding the proposed 02-acyl intermediate will be presented
in the hydrolysis kinetics section.

126
Solubility
Lipid (Sipm) and aqueous (Saq) solubilities for the
3-acyl derivatives are presented in Table 5-2 along with
their melting points. Lipid solubilities are greatly
enhanced by making these derivatives, and solubility
generally increases with increasing chain length. A change
in this trend is observed for 3-valeryl-5FU (20). A similar
downturn in lipid solubility was also seen for l-butyryl-5FU
in Chapter 3.
Because of the relative stability of the 3-acyl
derivatives,26 aqueous solubilities were determined by both
the direct and partitioning-based methods. The results show
that aqueous solubility is highest for 3-propionyl-5FU (18)
and then decreases. Both compound 18 and 3-acetyl-5FU (17)
exhibit aqueous solubilities greater than 5FU.
Aqueous solubilities determined by the partitioning-
based method underestimated the direct solubilities by 32%
for compound 18, 37% for compound 17, 57% for compound 19,
and overestimated the direct solubility by 10% for
compound 20. Relative aqueous solubility among members of
the series is the same with either procedure, but agreement
between the two methods is not as good as that seen for the
1-alkyloxycarbonyl derivatives.

127
Table 5-2. Melting points (MP), lipid solubilities (Sipm) , and
aqueous solubilities (Saq) for 3-acyl derivatives.
Compound
MP
(°C)
SlPM3
(mM)
SAQb
(mM)
SAQC
(mM)
SAQd
(mM)
5FU
280-2
0.049
96
-
85
17
115-7
4.3
166
105
249
18
102-3
14
198
135
190
19
111-2
22
53
23
-
20
110-1
9.2
5.0
5.5
aStandard
deviations
from the
mean were
within ±3%
for IPM
solubilities (±10% for compound 19).
bStandard deviations from the mean were within ±8% for
aqueous solubilities determined by direct method
cStandard deviations from the mean were within ±5% for
experimental values used to calculate partition coefficients
and estimated aqueous solubilities (±17% for compound 19).
dLiterature values from reference 26.

128
Table 5-3. Solubility ratios (SR), partition coefficients
(PC) , and hydrophobicity parameters (Jl) for
3-acyl derivatives.
log (PC) -
Compound
SRa
log(SR)
7lb
PCC
log (PC)
7ld
log(SR)
17
0.02 6
-1.59
0.041
-1.39
0.20
18
0.072
-1.14
0.45
0.11
-0.97
0.42
0.17
19
0.42
-0.38
0.76
0.98
-0.01
0.96
0.37
20
1.9
0.27
0.65
1.7
0.22
0.23
-0.05
aSolubility ratio calculated from Sipm/Saq.
bAlog(SR) for compound and previous compound.
Experimental partition coefficient (Cipm/Caq)
dAlog(PC) for compound and previous compound.

129
The direct aqueous solubility value for compound 18
agrees with the corresponding literature value included in
Table 5-2. However, the value for compound 17 is lower than
the literature value. The basis for this discrepancy is not
clear.
In Table 5-3, the solubility ratios (SR) and
experimentally determined partition coefficients (PC) are
compared for the 3-acyl series. The log(PC)-log(SR) values
indicate that the partition coefficients are generally larger
than the corresponding solubility ratios. As discussed in
Chapter 2, this is common for polar solutes (log[PC]<0) that
self-associate in the organic phase. The small negative
log (PC)-log(SR) value for compound 20 also fits the expected
trend, but the large positive value for 3-butyryl-5FU (19)
remains unexplained.
This deviation from the expected trend can also be seen
in the hydrophobicity parameter (ic) data. From the
experimental partition coefficients, the Alog(PC) value is
0.96 for the comparison between compounds 18 and 19 and 0.23
for the comparison between compounds 19 and 20, whereas the
expected average Jl value is around 0.60. On the other hand,
the Alog(PC) value for the comparison between compounds 18
and 20 is 0.60 per methylene unit. This suggests that the
butyryl derivative in this series exhibits unusual solubility
and partitioning properties compared to other members of the
series.

In (At-A.
130
Time (min)
Figure 5-3. Plot of ln(At-A«.) versus time (min) for hydrolysis
of 3-acetyl-5FU in 0.05 M phosphate buffer
(pH=7.1, 1=0.12) at 32 °C (n=3).

131
Time (min)
Figure 5-4. Plot of In (At-Aoo) versus time (min) for hydrolysis
of 3-propionyl-5FU in 0.05 M phosphate buffer
(pH-7.1, 1-0.12) at 32 °C (n=2).

132
-9.0
O
c
-10.0 .
-11.0
1 00
â–¡ Plain Buffer
t 3.6e-4 M Formaldehyde
I 3.6e-3 M Formaldehyde
3.6e-2 M Formaldehyde
♦
0 0
♦ □
O
I
200
time (min)
300
400
Figure 5-5. Plots of ln(C) versus time (min) for hydrolysis
of 3-acetyl-5FU in 0.05 M phosphate buffer (pH=7.1, 1=0.12)
with (n=2) and without (n=3) formaldehyde at 32 °C.

133
time (min)
Figure 5-6. Plots of ln(C) versus time (min) for hydrolysis
of 3-acetyl-5FU in 0.05 M phosphate buffer (pH=7.1, 1=0.12)
at 32 °C using actual concentration (Ct) and concentration
corrected for secondary degradation (Coorr) •

134
Table 5-4. Pseudo-first-order rate constants (k) and half-
lives (ti/2) for hydrolysis of 3-acyl derivatives in
0.05 M phosphate buffer (pH=7.1, 1=0.12) with and
without formaldehyde at 32 °C.
Compound
(Assay)
Phase
Formaldehyde
(M)
k(±SD)a
(min-1)
tl/2
(min)
17 (UV)
initial
0
7.01x10-3 (0.10x10-3)
99
17 (UV)
terminal
0
3.91x10-3(0.23x10-3)
177
18 (UV)
initial
0
6.20x10-3(0.14x10-3)
112
18(UV)
terminal
0
2.31x10-3(0.01x10-3)
300
17(HPLC)
initial
0
6.83x10-3(0.31xl0-3)b
101
17(HPLC)
terminal
0
3.92x10-3(0.32x10-3)
177
17 (HPLC)
initial
3.6xl0~4
7.20x10-3(0.17x10-3)
96
17(HPLC)
terminal
3.6XIO-4
5.07x10-3(0.08x10-3)
137
17(HPLC)
all
3.6x10-3
7.19x10-3(0.14x10-3)
96
17(HPLC)
all
3.6xl0-2
7.00x10-3(0.28x10-3)
99
aMean ± standard deviation for n=2-3 values (see Figures).
Correlation coefficient <0.999 (=0.998).

135
Table 5-5. Reaction products formed during hydrolysis of
3-acetyl-5FU in 0.05 M phosphate buffer (pH=7.1, 1-0.12)
with formaldehyde at 32 °C.
Reactant (M)
3-Acetyl-5FU Formaldehyde
Product (%)
5FU l-AOM-5FUa3-AOM-5FUb
1.8xl0-4
3.6xl0~4
91
4.7
4.9
1.8xl0"4
3.6xl0-3
65
18
17
1.8x10-4
3.6xl0-2
57
22
21
al-Acetyloxymethyl-5F0.
b3-Acetyloxymethyl-5FU.

136
ch2o
o
ch2o
HO—H
0=
I
H3C
Figure 5-7. Possible scheme for reaction of 3-acetyl-5FU with
formaldehyde to form l-acetyloxymethyl-5FU and
3-acetyloxymethyl-5FD.

137
Hydrolysis Kinetics
Hydrolysis of 3-acetyl-5FU (17) to 5FU was followed by
HPLC in 0.05 M phosphate buffer (pH=7.1, 1=0.12) at 32 °C
with and without formaldehyde and the hydrolyses of compounds
17 and 18 (3-propionyl-5FU) were followed by UV spectroscopy
in the buffer alone. Disappearance of compounds 17 and 18 in
the UV studies, indicated by In (At-Aoo) , is plotted versus time
(min) in Figure 5-3 and Figure 5-4 respectively. The plots
appear to be biexponential with a short initial linear phase
followed by curvature to an extended linear terminal phase.
In Figure 5-5, disappearance of compound 17 as ln(C)
from the HPLC studies is plotted versus time (rain). The same
biexponential behavior is apparent for plain buffer and to a
lesser extent for the lowest formaldehyde concentration.
However, the plots for the two higher formaldehyde
concentrations are linear throughout the course of the
hydrolysis reactions.
Biexponential plots of this type suggest that the 3-acyl
derivatives undergo rapid reversible reactions (initial
phase) followed by slower hydrolysis reactions in which one
or more of the compounds in equilibrium are converted to 5FU
(terminal phase). The pseudo-first-order plots, obtained
when formaldehyde is added to the buffer, indicate that
formaldehyde must be trapping the product or products of the
initial reaction thereby preventing the reverse reaction.114*5

138
In Figure 5-6, the HPLC plot for compound 17 is shown
along with a corrected concentration plot from the
equation:114-5
In (Ccorr) = In (Ct) +kt (7)
where CCorr is the concentration of compound 17 after
correcting for terminal phase hydrolysis and k is the
observed rate constant for the linear terminal phase. The
flattened appearance of the In(Ccorr) versus time plot
supports the assertion that a reversible reaction is involved
as the initial step in the hydrolysis of compound 17.114-5
In Table 5-4, rate constants (k) and half-lives (tj./2>
from the preceding plots are presented. The OV and HPLC data
for compound 17 are consistent, and the initial-phase rate
constants are essentially the same regardless of the amount
of formaldehyde added to the buffer. The linearity of the
plots with the higher formaldehyde concentrations is
indicated by a single rate constant.
Evidence that formaldehyde is trapping an intermediate
is presented in Table 5-5. Although an intermediate was
never observed during the HPLC studies, two products were
identified that suggest an 02-acylated intermediate. The
formation of l-acetyloxymethyl-5FU and 3-acetyloxymethyl-5FU
in equal concentrations was confirmed by comparison with
authentic samples.1 Identical retention times from the HPLC
chromatograms were observed for the authentic samples and the
1 Authentic samples were provided by Dr. Hans Bundgaard of
the Royal Danish School of Pharmacy, Copenhagen, Denmark.

139
reaction products from the hydrolysis studies, and 1H NMR
spectra obtained from a scaled-up hydrolysis reaction were
consistent with the assigned structures based on the
positions of the C6-H and N-CH2-0 absorptions.23 In addition,
l-acetyloxymethyl-5FU was isolated from the reaction mixture,
and its melting point (124-5 °C) was consistent with the
literature-reported value (122-3 °C).22
A possible scheme for formation of 1- and 3-acetyloxy-
methyl-5FU is presented in Figure 5-7. The even distribution
of the two products at all formaldehyde concentrations
(Table 5-5) suggests an intramolecular reaction with an
02-acyl intermediate.
Skin Penetration
Skin penetration data from the diffusion cells are
plotted as cumulative amount of total 5FU species that
diffused ((Imol) versus time (h) . In Figure 5-8, results for
3-acetyl-5FU (17) and 3-propionyl-5FU (18) are compared to
5FU itself. In Figure 5-9, results for 3-butyryl-5FU (19)
and 3-valeryl-5FU (20) are compared to 5FU. Error bars
correspond to the standard deviation from the mean for n=3
values.
Values for cumulative amount of total 5FU species that
diffused and skin accumulation of total 5FU species were
obtained by UV analysis of the receptor phases. Since the UV
spectra showed lower UVmax/UVmin ratios than expected for 5FU

Cumulative Amount (pmol)
140
Time (h)
Figure 5-8. Plots of cumulative amount of total 5FO species
that diffused (nmol) versus time (h) for
compounds 17, 18, and 5FU.

(loiurl) lunouiv aAjjeiniuno
141
Time (h)
Figure 5-9. Plots of cumulative amount of total 5FU species
that diffused (|imol) versus time (h) for
compounds 19, 20, and 5FU.

142
Table 5-6. Fluxes (J) , lag times (ti,) , and skin accumulation
(SA) values for 3-acyl derivatives.
Compound
J(±SD)a
(|lmol/cm2/h)
Prodrug
S-Sb(ll h)c
(%)
tL
(h)
SA(±SD)a
(^.mol)
5FU
0.24(0.09)
-
13
3.7(0.9)
17
4.4(0.5)
24(19)
19
16(3)
18
5.2(0.7)
49 (43)
15
15(2)
19
2.2(0.3)
62(52)
9.8
7.1(1.6)
20
0.55 (0.07)
-
6.2
3.3(1.1)
aMean ± standard deviation for n=3 values.
bPercent of total 5FÜ as intact prodrug during "steady-state"
phase in separate experiment (n=l).
°Percent of total 5FD as intact prodrug from 11 h sample in
separate experiment (n=l).

143
Table 5-7. Second application fluxes (J) and lag times (ti,)
for 3-acyl derivatives.
Compound
Ja(±SD)b
4lmol/cm2/h)
tL
(h)
5FU
1.2(0.2)
1.2
17
1.6(0.1)
0.5
18
1.8(0.2)
1.2
19
1.1(0.1)
1.0
20
1.1(0.2)
1.3
aFlux of 0.4 M theopylline from propylene glycol.
bMean ± standard deviation for n=3 values.

144
alone and hydrolysis studies showed that products other than
5F0 were formed during hydrolysis of the 3-acyl derivatives
in formaldehyde buffer, it was necessary to compare the molar
absorptivity (£) at 266 nm for 5FU with e values at 266 nm for
hydrolyzed 3-acyl and 1,3-bis-acyl derivatives in solutions
containing receptor phase. Solutions containing known
concentrations of each of the 3-acyl and 1,3-bis-acyl
derivatives were incubated for 24 hours at 32 °C at which
time HPLC analysis showed that only 5FU and the two
formaldehyde reaction products were present. Subsequent
analysis by UV spectroscopy showed that the multiple
component solutions had e values that were within ±5% of the
£ value for 5FU alone at 266 nm. This verified that the UV
assay for determining cumulative amount of total 5FU species
that diffused and skin accumulation of total 5FU species is
valid.
Fluxes (J), lag times (ti.) , and skin accumulation (SA)
values for each compound are reported in Table 5-6. Lag time
refers to the intersection of the linear, or "steady-state,"
region of each graph with the time (x) axis, and it is the
time required for establishing a uniform concentration
gradient within the skin.65 The percentages of total 5FU
present as intact prodrugs in the receptor phases are also
reported in Table 5-6. These values were calculated from
samples taken during the "steady-state" phase and from an
earlier sample (11 h) in a separate experiment using HPLC
analysis (n=l).

145
All of the 3-acyl derivatives improved the skin
penetration of 5F0. Increases in flux ranged from more than
two times (3-valeryl-5FU [20]) to more than 20 times
(3-propionyl-5FU [18]) the flux of 5FU itself.
As was the case with the 1-alkyloxycarbonyl derivatives,
large amounts of intact 3-acyl prodrugs were detected in the
receptor phases. Unlike the 1-alkyloxycarbonyl derivatives,
however, the differences between the "steady-state"
percentages and the percentages calculated from samples taken
prior to "steady-state" are small. This suggests that
hydrolytic enzymes play a relatively minor role in the
hydrolysis of the 3-acyl derivatives when compared with the
1-alkyloxycarbonyl derivatives. This is consistent with the
known hydrolytic behavior of the 3-acyl series in human
plasma26 and is probably due to the steric and electronic
effects described earlier. It is important to note that the
intact prodrug percentages reported in Table 5-6 represent
the lower limits of prodrug content. Formaldehyde reaction
products (1- and 3-acyloxymethyl derivatives) were also
present in the receptor phases at the time of analysis which
indicates that more than the observed amounts of 3-acyl
derivatives had actually diffused through the skin intact.
Skin accumulation values are highest for compounds 17
and 18 and are over four times higher than the value for 5FU.
Overall, skin accumulation is much lower from the 3-acyl
derivatives than the 1-acyl derivatives which supports the

146
earlier assertion that rapid hydrolysis of 5FU prodrugs leads
to higher skin levels of 5FU.
Second application fluxes and lag times are reported in
Table 5-7. Skin penetration by theophylline from propylene
glycol, the standard drug-vehicle combination, after
treatment with the 3-acyl derivatives is similar to that
following treatment with 5FU.
The stability of the prodrugs in the IPM formulations
was assessed by 1H NMR analysis of the donor phases. After a
minimum of five days from the time the suspensions were
prepared until their 7-H NMR spectra were recorded, including
at least twelve hours during which the formulations were in
contact with the skins, the 3-acyl derivatives were found to
be intact with no evidence of 5FU formation.
.gumma ry
The 3-acyl derivatives of 5FU exhibited increased lipid
solubilities when compared to 5FU. Aqueous solubility
reached a maximum for 3-propionyl-5FU (18) and then decreased
with increasing chain length. Skin penetration and skin
accumulation were also maximized for compound 18 suggesting
that both lipid and aqueous solubilities are important for
predicting transdermal and dermal delivery of these 5FD
prodrugs. The presence of high percentages of prodrugs in
the receptor phases, indicating insufficient release of the
parent drugs in the hairless mouse skin model, may limit the

147
potential of this series of prodrugs at least for dermal
delivery purposes. The partitioning-based method for
determining aqueous solubility was effective for determining
the relative solubilities in the 3-acyl series, but overall,
the agreement with the direct method for determining aqueous
solubility was not as good as it was for the 1-alkyloxy-
carbonyl series. An 02-acyl intermediate was proposed to
explain the thermal decomposition and rearrangement of the
3-acyl derivatives and the unusual hydrolytic behavior of the
series including the formation of acyloxymethyl derivatives
of 5FÜ.

CHAPTER 6
SUMMARY AND CONCLUSIONS
Fluorouracil (5FU) is a commercially available
antineoplastic agent that has been used topically for
treatment of actinic keratoses, superficial basal cell
carcinomas, psoriasis, and other precancerous conditions,
malignant and benign tumors, and dermatoses. As a typical
polar, heterocyclic compound, however, 5FU has poor
solubility and skin penetration properties, and topical
treatment with 5FU is often ineffective.
In an attempt to improve the topical delivery of 5FU,
four homologous series of bioreversible derivatives
(prodrugs) of 5FU were synthesized and characterized, and
their ability to penetrate through and accumulate in the skin
was evaluated. The four series included six 1-alkyloxy-
carbonyl, six 1-acyl, four 1,3-bis-acyl, and four 3-acyl
derivatives.
Solubilities and melting points for the four series of
prodrugs showed the expected trends. Melting points
generally decreased and lipid solubilities generally
increased with increasing chain length, and in fact, lipid
solubilities were a minimum of 40 times greater for the
derivatives than for 5FU. Aqueous solubilities were
maximized for the first or second member of each series and
then decreased with increasing chain length. Five
148

14 9
derivatives had aqueous solubilities greater than 5FU even
though an hydrogen-bonding N-H group was masked in each. The
partitioning-based method for determining aqueous
solubilities was shown to give reliable values for relative
solubility in a homologous series, and overall, the values
correlated well with the conventionally determined aqueous
solubilities. It is noteworthy that this is the first known
successful attempt to obtain aqueous solubilities for such
chemically unstable prodrugs.
Hydrolysis of the 1-alkyloxycarbonyl, 1-acyl, and
1,3-bis-acyl (N1-acyl group only) derivatives followed
pseudo-first-order kinetics in 0.05 M phosphate buffer
(pH=7.1, 1=0.12) with and without formaldehyde which was used
as a preservative in the diffusion cell experiments.
Hydrolysis was faster in the presence of formaldehyde which
apparently functions as a general base catalyst. The N^-acyl
groups were much more rapidly hydrolyzed than the N3-acyl
groups even though the N3-anion is a better leaving group
than the N1-anion. The X-ray crystal structure of 1,3-bis-
acetyl-5FU showed that the N3-acetyl group was oriented
perpendicular to the plane of the 5FU ring, and the carbonyl
carbon was sterically and electronically hindered from attack
by hydroxide ion. It was suggested that the unusual
orientation of the N3-acyl group caused the slower than
expected hydrolysis rates for the 3-acyl derivatives compared
to the 1-acyl derivatives.

150
Hydrolysis of the 3-acyl derivatives was biexponential.
A reversible reaction involving an O-acyl intermediate was
proposed for the initial phase followed by terminal phase
hydrolysis of the 3-acyl derivative, an O-acyl intermediate,
or both to 5FU. Formation of nearly equal amounts of 1- and
3-acetyloxymethyl-5FU during hydrolysis of 3-acetyl-5FU in
buffer with formaldehyde suggested that the intermediate may
be the 02-acetyl derivative.
An 02-acyl intermediate was also proposed to explain the
thermal decomposition and rearrangement of the 3-acyl
derivatives to form N1-acylated compounds. Ketene formation
was proposed to account for the formation of 5FU from the
3-acyl derivatives (and possibly the 1-acyl derivatives)
during their thermal decomposition.
Skin penetration was 1.2 to nearly 40 times greater for
the prodrugs than for 5FU itself. The highest flux was
recorded for l-acetyl-5F0,whereas 3-propionyl-5FU,
l-ethyloxycarbonyl-5FU, and 1,3-bis-acetyl-5FU exhibited the
highest rates of delivery of 5FU for their respective series.
Without exception, the highest fluxes were obtained for the
most aqueous soluble derivative in each series. The skin
penetration results demonstrate the importance of biphasic
solubility for achieving optimal transdermal delivery of 5FU
from a homologous series of prodrugs.
Intact prodrug content in the receptor phases
constituted over 40% of total 5FÜ species that had diffused
for the 1-alkyloxycarbonyl and the 3-acyl derivatives with

151
the exception of 3-acetyl-5FU which was much lower. Since
the 1,3-bis-acyl derivatives hydrolyzed to give the
corresponding 3-acyl derivatives almost immediately, they
should behave in a similar manner to the 3-acyl derivatives.
Testing of the 3-acyl, 1,3-bis-acyl, and 1-alkyloxycarbonyl
series in other skin models should be carried out before they
are dismissed as too stable for use as prodrugs for dermal as
opposed to transdermal delivery.
Skin accumulation of total 5FU species was highest for
the 1-acyl derivatives particularly for l-acetyl-5F0 and
l-propionyl-5FU. These two prodrugs had skin accumulation
values more than 18 times the value for 5FU and nearly four
times more than the next best derivative. It was suggested
that rapid hydrolysis of the 1-acyl derivatives upon
partitioning into the skin effectively "locks in" large
amounts of 5FU since it is the highly polar 5FU molecule that
must diffuse through the remaining lipid regions of the skin.
Second application studies using the standard drug-
vehicle combination showed that skin damage was at most one
and one-half times greater for the prodrugs than for 5FU, a
small increase when compared to the much greater improvement
in skin penetration and skin accumulation that was achieved.
The stability of the prodrugs in the IPM formulations
was documented and no decomposition was observed even after a
twelve-hour period during which the formulations were in
contact with the skins. This shows that even chemically

152
unstable derivatives can potentially be formulated for
topical delivery.
Overall, l-acetyl-5FU was the most effective prodrug for
delivering 5FU through the skin (transdermal) and retaining
5FU in the skin (dermal). The combination of high flux,
rapid hydrolysis, high skin accumulation, and stability in
formulation makes l-acetyl-5FU the best candidate for
improving the dermal or transdermal delivery of 5F0.

LIST OF REFERENCES
1. Diasio, R. B.; Harris, B. E. Clin. Pharmacokin. 1989,
1£, 215.
2. Kovach, J. S.; Beart, R. W. Invest. New Drugs 1989, 7,
13.
3. Bennett, D. R., Ed. Drug Evaluations Annual 1991;
American Medical Association: Milwaukee, WI, 1990;
p. 1772.
4. McEvoy, G. K., Ed. AHFS Drug Information 90: American
Society of Hospital Pharmacists: Bethesda, MD, 1990;
p. 513.
5.
Dillaha,
C. J.;
Jansen
, G. T.;
Honeycutt,
w.
M. ;
Bradford,
A. C.
Arch.
Dermatol
^ 1963, M,
247 .
6.
Dillaha,
C. J.;
Jansen
, G . T .;
Honeycutt,
w.
M. ;
Holt
G. A. Arch. Dermatol. 1965, 12., 410.
7. Simmonds, W. L. Cutis 1973, 12., 615.
8. Simmonds, W. L. Cutis 1976, 18., 298.
9. Klein, E.; Stoll, H. L.; Milgrom, H.; Traenkle, H. L.;
Case, R. W.; Laor, Y.; Helm, F.; Nadel, R. S. J. Invest
Dermatol. 1965, li, 489.
10.
Stoll,
H.
L.; Klein, E. i
Invest. Dermatol. 1969, 52
304.
11.
Tsuji,
T. ;
: Sugai, T. Arch.
Dermatol. 1972. 105, 208.
12.
L junggren,
. B.; Moller, H.
Arch.
Dermatol. 1972,
263.
13.
Pearlman,
D. L.; Youngberg, B.;
Engelhard. C. J. Am.
Acad.
Dermatol. 1986, 11,
1247 .
14 .
Goette
, D
. K. J. Am. Acad.
Dermatol. 1981, i, 633.
15.
Robinson,
T. A.; Kligman,
A. M.
Br. J. Dermatol. 1975
12. 703.
153

154
16. Duschinsky, R.; Pleven, E.; Heidelberger, C. J. Am.
Chem. S no. 1957, 22, 4559.
17. Heidelberger, C.: Chaudhuri, N. K.; Danneberg, P.;
Mooren, D.; Griesbach, L.; Duschinsky, R.; Schnitzer,
R. J.; Pleven, E.; Scheiner, J. Nature 1957, 179. 663.
18 .
Stella, V.
, J.
; Himmelstein, K. J.
J,
Med,
. Chem. 1980.
22, 1275.
19.
Bansal, P.
, C.
; Pitman, I. H.; Tam,
. J
. N.
S.; Mertes, M
Kaminski,
J.
J. J. Pharm. Scl^ 1981,
22,
850 .
20.
Ozaki, S.;
Watanabe, Y.; Hoshiko,
T.;
: Nagase, T.;
Ogasawara, T.; Furukawa, H.; Uemura, A.; Ishikawa, K.;
Mori, H.; Hoshi, A.; ligo, M.; Tokuzen, R. Chem. Pharm.
Bull. 1986, 21, 150.
21. Mollgaard, B.; Hoelgaard, A.; Bundgaard, H. Tnt. J.
Pharm. 1982, 12., 153.
22. Buur, A.; Bundgaard, H.; Falch, E. Int. J. Pharm. 1985,
21, 43.
23. Ozaki, S.; Watanabe, Y.; Hoshiko, T.; Mizuno, H.;
Ishikawa, K.; Mori, H. Chem. Pharm. Bull. 1984, ¿2, 733
24. Ahmad, S.; Ozaki, S.; Nagase, T.; ligo, M.; Tokuzen, R.
Hoshi, A. Chem. Pharm. Bull. 1987, 15., 4137.
25. Kametani, T.; Kigasawa, K.; Hiiragi, M.; Wakisaka, K.;
Haga, S.; Nagamatsu, Y.; Sugi, H.; Fukawa, K.; Irino,
0.; Yamamoto, T.; Nishimura, N.; Taguchi, A. J. Med.
Chem. 1980, 22, 1324.
26
Buur,
A.;
Bundgaard,
H. Int. J. Pharm. 1984, 21,
349.
27 .
Masao,
T.
Chem. Lett.
1975, 129.
28.
Buur,
A.;
Bundgaard,
H. Arch. Pharm.
Chem.. Sci.
Ed.
1986,
ii,
99.
29.
Buur,
A.;
Bundgaard,
H. Arch. Pharm.
Chem.. Sel.
Ed.
1984,
12,
37.
30.
Buur,
A. ;
Bundgaard,
H. J. Pharm. Scj
^ 1986, 22,
522.
31.
Nagase
i, T
.; Seike, K.
; Shiraishi, K.;
Yamada, Y.
; Ozak
S. Chem.
Lett. 1988,
1381.
32.
Nagase
!, T
.; Shiraishi
, K.; Yamada, Y.
; Ozaki, S.
Heterocvclesâ–  1988, 22, 1155.

155
33. Buur, A.; Bundgaard, H.; Falch, E. Acta Pharm. Suec.
1986, 21, 205.
34. Buur, A.; Bundgaard, H. Int. J. Pharm. 1985, 21, 209.
35. Ozaki, S.; Ike, Y.; Mizuno, H.; Ishikawa, K.; Mori, H.
Bull. Chero. Soc. Jap. 1977, ¿2, 2406.
36. Sasaki, H.; Takahashi, T.; Nakamura, J.; Konishi, R.;
Shibasaki, J. J. Pharm. Sci. 1986, 75â–  676.
37. Sasaki, H.; Takahashi, T.; Mori, Y.; Nakamura, J.;
Shibasaki, J. Int. J. Pharm. 1990, 12, 1.
38. Sloan, K. B.; Koch, S. A. M.; Siver, K. G. Tnt. ,T.
Pharm. 1984, 21, 251.
39. Sloan, K. B.; Sherertz, E. F.; McTiernan, R. G. Int. J
Pharm. 1988, 11, 87.
40. Kametani, T.; Kígasawa, K.; Hiiragi, M.; Wakisaka, K.;
Nakazato, k.; Ichikawa, K.; Fukawa, K.; Irino, 0.;
Nishimura, N.; Okada, T. J. Med. Chem. 1982, 21, 1219.
41. Kamata, S.; Haga, N.; Matsui, T.; Nagata, W. Chem.
Pharm. Bull. 1985, 11, 3160.
42. Hoshiko, T.; Ozaki, S.; Watanabe, Y.; Ogasawara, T.;
Yamauchi, S.; Fujiwara, K.; Hoshi, A.; ligo, M. Chem.
Pharm.
Bull. 1985,
11,
2832 .
43.
Ozaki,
S.; Nagase,
T.;
Tamai, H.;
: Mori, H.;
1 Hoshi, A
ligo, M. Chem, Pharm. Bull. 1987, H, 3894.
44. Ozaki, S.; Ohnishi, J.; Watanabe, Y.; Nohda, T.; Nagase
T.; Akiyama, T.; Uehara, N.; Hoshi, A. Polymer. I. 1989
21, 955.
45. Ouchi, T.; Fujino, A.; Tanaka, K.; Banba, T. J.
Controlled Release 1990, 12, 143.
46. Kingsbury, W. D.; Boehm, J. C.; Mehta, R. J.; Grappel,
S. F.; Gilvarg, C. J. Med. Chem. 1984, 22, 1447.
47. Hiller, S. A.; Zhuk, R. A.; Lidaks, M. Y. Dokl. Akad.
Nauk. SSSR 1967, 122, 332.
48. Hoshi, A.; ligo, M.,- Nakamura, A.; Yoshida, M.;
Kuretani, K. Gann 1976, 22, 725.

156
49. Ajmera, S.; Bapat, A. R.; Stephanian, E.; Danenberg, P.
V. J, Med. Chem. 1988, £L, 1094.
50. Pinedo, H. M.; Peters, G. F. J. J. Clin. Oncol. 1988, £
1653.
51. Douglas, K. T. Med. Res. Rev. 1987, 2, 441.
52. Odland, G. F. In Biochemistry and Physiology of the
Skin: Goldsmith, L. A., Ed.; Oxford: New York, 1983;
Vol. 1, p. 3.
53. Scheuplein, R. J.; Bronaugh, R. L. In Biochemistry and
Physiology of the Skin: Goldsmith, L. A., Ed.; Oxford:
New York, 1983; Vol. 2, p. 1255.
54. Flynn, G. L.; Stewart, B. Drug Dev. Res. 1988, 12, 169.
55. Blank, I. H. In Percutaneous Absorption: Bronaugh, R.
L., Maibach, H. I., Eds.; Marcel Dekker: New York, 1985
p. 97 .
56. Elias, P. M. Arch. Dermatol. Res. 1981, 270. 95.
57. Elias, P. M.; Grayson, S.; Lampe, M. A.; Williams, M.
L.; Brown, B. E. In Stratum Corneum: Marks, R., Plewig,
G., Eds.; Springer-Verlag: New York, 1983; p. 53.
58.
Flynn, G.
Rhodes, C
p. 263.
L.
. T
In Modern Pharmaceutics: Banker, G. S.
., Eds.; Marcel Dekker: New York, 1990;
59.
Barry, B.
W.
J. Controlled Release
1991, 1£, 237.
60.
Idson, B.
Pharm. Sci. 1975, £1,
901.
61.
Hadgraft,
J.
J. Controlled Release
1991, 1£, 221.
62 .
Potts, R.
0.;
' Golden, G. M.; Francoeur, M. L.; Mak,
V. H. W.; Guy, R. H. J, Controlled Release 1991, 1£,
249.
63. Boddé, H. E.; van den Brink, I.; Koerten, H. K.;
de Haan, F. H. N. J. Controlled Release 1991, 1£, 227.
64. Elias, P. M. J. Controlled Release 1991, 12, 199.
65. Martin, A.; Swarbrick, J.; Cammarata, A. Phy.si cal
Pharmacy; Lea and Febiger: Philadelphia, 1983; p. 399.
66.Scheuplein, R. J. J, Invest. Dermatol. 1976, £7, 31.

157
67 .
Mak, V. H
1991, £,
. w.,
1064
: Potts, R. 0.;
Guy, R
. H.
Pharm. Res.
68.
Pikal, M.
J.;
Shah, S. Pharm,
, Res.
1990,
, 7, 222.
69.
Sloan, K.
B.;
Koch, S. A. M.;
: Siver
, K.
G.; Flowers
F. P. J. Invest. Dermatol. 1986, £7, 244.
70. Sloan, K. B.; Siver, K. G.; Koch, S. A. M. J. Pharm.
Sci. 1986, 11, 744.
71. Sherertz, E. F.; Sloan, K. B.; McTiernan, R. G. .T.
Invest. Dermatol. 1987, SI, 147.
72. Waranis, R. P.; Siver, K. G.; Sloan, K. B. Int. J.
Pharm. 1987, 211.
73. Stella, V. In Prodrnas as Novel Drug Delivery Svst-pmg.
ACS Symposium Series 14: Higuchi, T., Stella, V., Eds.;
American Chemical Society: Washington, 1975; p. 1.
74. Stella, V. J.; Mikkelson, T. J.; Pipkin, J. D. In Drug
Delivery Systems. Characteristics and Biomedical
Applications: Juliano, R. L., Ed.; Oxford Onversity
Press: New York, 1980; p. 112.
75. Michaels, A. S.; Chandrasekaran, S. K.; Shaw, J. E.
AIChE J. 1975, 21, 985.
76. Yalkowsky, S. H.; Flynn, G. L. J. Pharm. Sen . 1973, 62.
210.
77. Sloan, K. B. Advanced Drug Delivery Rev. 1989, 2, 67.
78. Pitman, I. H. Med. Res. Rev. 1981, 1, 189.
79. Sherertz, E. F.; Sloan, K. B.; McTeirnan, R. G. Arch.
Dermatol. Res. 1990, 282. 463.
80. Sloan, K. B.; Sherertz, E. F.; McTiernan, R. G. Arch.
Dermatol. Res. 1990, 282. 484.
81. Pannatier, A.; Jenner. P.; Testa, B.; Etter, J. C. nrna
Metab. Rev. 1978, S, 319.
82. Kao, J.; Carver, M. P. Drug Mefah. Rev. 1990, 22, 363.
83. Itoh, T.; Magavi, R.; Casady, R. I.; Nishihata, T.;
Rytting, J. H. Pharm. Res. 1990, 1, 1302.

158
84. Skelly, J. P.; Shah, U. P.; Maibach, H. I.; Guy, R. H.;
Wester, R. C.; Flynn, G.; Yacobi, A. Pharm. Res. 1987,
Í, 265.
85. Lambert, W. J.; Higuehi, W. I.; Knutson, K.; Krill,
S. L. J. Pharm. Sci. 1989, 7£, 925.
86. Kai, T.; Mak, V. H. W.; Potts, R. 0.; Guy, R. H. 2L.
Controlled Release 1990, 12, 103.
87. Bond, J. R.; Barry, B. W. J. Invest. Dermatol. 1988, 90
486.
88. Hinz, R. S.; Hodson, C. D.; Lorence, C. R.; Guy, R. H.
J. Invest. Dermatol. 1989, 22, 87.
89. Sloan, K. B.; Beall, H. D.; Weimar, W. R.; Villanueva,
R. Int. J. Pharm. 1991, 22, 97.
90. Scheuplein, R. J.; Ross, L. W. J. Invest. Dermatol.
1974, £2., 353.
91. Rudy, B. C.; Senkowski, B. Z. In Analytical Profiles of
Drug Substances: Florey, K., Ed.; Academic Press:
New York, 1973; Vol. 2, p. 222.
92. Sloan, K. B.; Selk, S.; Haslam, J.; Caldwell, L.;
Shaffer, R. J, Pharm. Sci. 1984, 22, 1734.
93. Buur, A.; Bundgaard, H. Int. J. Pharm. 1987, 2£, 41.
94. Merck Index; Windholz, M., Ed.; Merck: Rahway, NJ, 1976
p. 5069.
95. Surber, C.; Wilhelm, K-P.; Hori, M.; Maibach, H. I.;
Guy, R. H. Pharm. Res. 1990, 1, 1320.
96. Hansch, C. In Strategy of Drug Design: A Guide to
Biological Activity: Purcell, W. P., Bass, G. E.,
Clayton, J. M. Eds.; John Wiley and Sons: New York,
1973; p. 126.
97.
Saab, A.
J. Pharm.
N.; Sloan, K. B.; Beall,
Sci. 1990. 79. 1099.
H . D . ;
Villanueva, R.
98.
Gacek, M.
; Undheim. K. Acta Chem.
Scand.
1979,
fi22, 515
99.
Smith, R.
N.; Hansch, C.; Ames, M.
. M. jL.
Pharm.
. Sci.
1975, M,
599.

159
100. Hansch, C.; Leo, A. In Substituent Constants for
Corre!ation Analysis in Chemistry and Riolngy: John
Wiley and Sons: New York, 1979; p. 13.
101. Yalkowsky, S. H.; Flynn, G. L.; Slunick, T. G. J. Pharm
Sci. 1972, £1, 852.
102.
Yalkowsky,
J. Pharm.
. s.
Sci
H.; Valvani, S. C.;
1983, 72, 866.
Roseman, T.
J.
103.
Flynn, G.
L.;
Yalkowsky, S. H. J.
Pharm. Sci.
1972
838.
104. Chem. Abstr. 1963, £fi, 1477a.
105. Silverstein, R. M.; Bassler, G. C.; Morrill, T. C. In
Spent rnmet ri n Trient i f i rat i nn of Organic Compounds: John
Wiley and Sons: New York, 1981; p. 95, 181.
106. Wierzchowski, K. L.; Litonska, E.; Shugar, D. J. Am.
Chem. Soc. 1965, £7, 4621.
107. Wempen, I.; Fox, J. J. J. Am. Chem. Soc. 1964, £_£, 2474.
108. Stolarski, R.; Remin, M.; Shugar, D. 7,. Natnrfnrsnh.
1977, £2fi, 894.
109. Curtin, D. Y.; Miller, L. L. J. Am. Chem. Soc. 1967, &£,
637 .
110. McCarthy, D. G.; Hegarty, A. F. J. Chem. Soc. Perkin II
1977, 1085.
111. Minkin, V. I.; Olekhnovich, L. P.; Zhdanov, Y. A. A or..
Chem. Res. 1981, 14, 210.
112. Fleming, I.; Philippides, D. J. Chem. Soc. 1C) 1970,
2426.
113. Curtin, D. Y.; Engelmann, J. H. Tetrahedron Letters
1968, 3911.
114. Prankerd, R. J.; Stella, V. J. Int. J. Pharm. 1989, £2,
71.
115.Prankerd, R. J.; Walters, J. M.; Parnés, J. H. Int. J
Pharm. in press.

BIOGRAPHICAL SKETCH
Howard D. Beall was born July 6, 1954, in Salt Lake
City, Utah. The Beall family returned to Montana before he
was two years old, and he spent his school years in several
Montana cities.
He graduated with honors from the University of Montana
in 1977 with a Bachelor of Science degree in pharmacy. He
practiced pharmacy in various settings in Montana and Oregon
before moving to Florida and earning a Master of Science
degree in pharmacy from the University of Florida in 1983.
He entered the doctoral program in the Department of
Medicinal Chemistry at the University of Florida in 1987.
He is a member of Rho Chi National Honor Society and the
American Association of Pharmaceutical Scientists. He
received the Merck Award for academic achievement in 1977 and
was a fellow of the American Foundation for Pharmaceutical
Education from 1990-91.
He married Donna Goyette in 1984, and they have a son,
Michael, born in 1989, and a daughter, Allison, born in 1991.
160

I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
a dissertation for the degree of Do^pfr of Pbilqsophy.
Kenneth B. Sloan, Chair
Associate Professor of
Medicinal Chemistry
X certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
Margaret 0. James
Professor of Medicinal
Chemistry
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor,of Philosophy.
¿=£L
Koppatfeí'V. Rao
Professor of Medicinal
Chemistry
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequat
a dissertation for the degree of
Assistant Professor of
Pharmaceutics
sand quality,
ililosqphy.

I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
jjhn^A. Zoltewicz ^0
Professor of Chemistry
This dissertation was submitted to the Graduate Faculty
of the College of Pharmacy and to the Graduate School and was
accepted as partial fulfillment of the requirements for the
degree of Doctor of Philosophy.
December, 1991
%/•
Dean, College of Pharmacy'-
Dean, Graduate School

UNIVERSITY OF FLORIDA
ii m in urn mi' ; : t -
3 1262 08554 6553



84
Four straight-chain 1,3-bis-acyl derivatives were
selected for study. The derivatives and their structures are
shown in Table 4-1.
Materials and Methods
Melting points (mp) were determined with a Thomas-Hoover
capillary melting point apparatus and are uncorrected.
Elemental microanalyses were obtained for all novel compounds
through Atlantic Microlab, Incorporated in Norcross, Georgia.
Proton nuclear magnetic resonance (1H NMR) spectra were
obtained at 90 MHz on a Varian EM-390 spectrometer. Chemical
shifts (5) are reported in parts per million (ppm) from the
internal standard, tetramethylsilane (TMS). Coupling
constants (J) are expressed in cycles per second (Hz).
Infrared (IR) spectra were recorded with a Perkin-Elmer 1420
spectrophotometer and absorbances are reported in cm"1.
Ultraviolet (UV) spectra were obtained with a Cary 210 or
Shimadzu UV-265 spectrophotometer. Maximum absorbances are
reported in nm along with the molar absorptivities (E) in
L/mol. Single-crystal X-ray analysis was obtained for
1.3-bis-acetyl-5FU through Hoffmann-La Roche in Nutley, NJ.
1.3-bis-Acvl-5-fluorouracil (general procedure)
To 1.31 g (0.01 mol) of 5FU suspended in acetonitrile
(20 mL) was added 1.01 g of triethylamine (0.01 mol) in


40
Table 2-3. Solubility ratios (SR), partition coefficients
(PC) and hydrophobicity parameters (7t) for
1-alkyloxycarbonyl derivatives.
log (PC) -
Compound
SRa
log (SR)
7tb
PCC
log (PC)
7td
log(SR)
1
0.018
-1.75
0.019
-1.72
0.03
2
0.050
-1.30
0.45
0.075
-1.12
0.60
0.18
3
0.28
-0.56
0.74
0.36
-0.45
0.67
0.11
4
1.2
0.07
0.63
1.4
0.16
0.61
0.09
5
29
1.46
0.70
31
1.48
0.66
0.02
6
257
2.41
0.48
285
2.45
0.49
0.04
aSolubility ratio calculated from Sipm/Saq.
bAlog(SR) for compound and preceding compound.
Experimental partition coefficient (Cipm/Caq) .
dAlog(PC) for compound and preceding compound.


30
solubilities (Saq) were calculated from the IPM solubility
(Sipm) and the partition coefficient:
Saq = Sipm/PC (3)
Partitioning was carried out in triplicate for a fixed volume
ratio for each derivative. For those compounds with large
differences in solubility in one phase relative to the other,
volume ratios (IPM:buffer) other than 1:1 were necessary, but
the ratio never exceeded 10:1 or 1:10.
Hydrolysis Kinetics
Hydrolysis rates have previously been reported for
several members of this homologous series.30,93 In the
present study, hydrolysis rates were determined at 32 C for
l-methyloxycarbonyl-5FU (1) in 0.05 M phosphate buffer
(pH=7.1, 1=0.12) and in the same buffer with 0.11%
formaldehyde (3.6xl0-2 M). The rate in the presence of
formaldehyde was determined for comparison with the rate in
plain buffer since formaldehyde was used as a preservative in
the diffusion cell experiments described in the following
section.
The hydrolyses were followed by high performance liquid
chromatography (HPLC). The HPLC system consisted of a
Beckman model 110A pump with a model 153 V detector, a
Rheodyne 7125 injector with a 20 (im loop, and a Hewlett-
Packard 3392A integrator. The column was a Lichrosorb RP-8
10 )tm reversed-phase column, 250 mm x 4.6 mm (inside


35
Results and Discussion
Synthesis and Structure Determination
The known l-alkyloxycarbonyl-5FU derivatives have
melting points25'30'93 and spectral properties (UV30 and 3H
NMR25) in agreement with those reported in the literature.
The structures of the novel compounds were assigned by
comparison of their spectral properties with those of the
known homologs. Elemental microanalyses were obtained for
the novel compounds and were within acceptable limits
(0.4%).
Acylation on the N1- or N3-position can be distinguished
by UV and 1H NMR analysis. Anions of N3-substituted
derivatives undergo a substantial shift of their UVmax to
longer wavelength while anions of ^-substituted compounds do
not.30'98 This is reportedly due to the extended conjugation
possible for the N1-anion.20 Differences in 3H NMR spectra
are also well defined. In chloroform-d, the C6-H signal for
Nl-substituted derivatives is a sharp doublet indicating
coupling with C5-F. The same signal in chloroform-d for
N3-substituted derivatives appears as a broad singlet or
triplet-like doublet of doublets from additional coupling of
C6-H with N1-!!.20 When the substituents contain a carbonyl
group attached to the N1-position, as they do in the
1-alkyloxycarbonyl series, an anisotropic effect is observed


CHAPTER 4
1,3-BIS-ACYL DERIVATIVES
Tnt rodnot i on
The 1,3-bis-acyl derivatives of 5FU are double prodrugs.
The acyl groups at the N^-position are rapidly hydrolyzed
leaving the 3-acyl derivatives. The 3-acyl derivatives will
be discussed in Chapter 5. The rate of Nl-deacylation for
the 1,3-bis-acyl derivatives is even faster than hydrolysis
of the 1-acyl derivatives. This can best be explained by
comparing pKa values for the 3-acyl derivatives and 5FU. The
3-acyl derivatives have reported pKa values of 7.1 to 7.226
while the pKa for the first ionization of 5FU is 8.0. Thus,
the leaving group potential of the N3-acyl anions is greater
than the 5FO anion, and the rates of N1-deacylation are
faster for the 1,3-bis-acyl derivatives.
The 1,3-bis-acyl derivatives were chosen as prodrug
candidates in order to study a series in which both hydrogen
bonding groups were masked. It was anticipated that these
compounds would be more lipid soluble and less water soluble
than the other series and would be useful for comparison with
the other series.
82


3
alkyloxycarbonyl,25'29-30 alkyloxycarbonyloxyalkyl,31-33 alkyl-
carbamoyl, 34-37 aminoalkyl (Mannich base),38-39 phthalid-
yl, 40-41 anc derivatives containing sulfur rather than oxygen
in some of the above functional groups.42-43 Polymer44-45 and
peptide46 derivatives have also been proposed as sources of
5FU in vivo. Currently, at least two bioreversible
derivatives of 5FU, 1-(tetrahydrofuran-2-yl)-5-fluorouracil47
and l-hexylcarbamoyl-5-fluorouracil,48 have been marketed in
Japan.
The above is just a small sample of the synthetic
literature involving 5FU. In addition, a large number of
studies have been devoted to synthesizing analogs of FUR and
FdUR containing an N1-glycosidic linkage. FdUR, or
floxuridine, is available in the United States, but is only
approved for intraarterial infusion.4 It appears to offer no
advantage over intravenous 5FU, and this therapy is both
hazardous and expensive.3 The nucleosides and their analogs
are converted to 5FU in vivo by a pyrimidine phosphorylase.
Since phosphorylase activity is higher in tumor tissue, it is
believed that development of compounds that are good
substrates for phosphorylase may lead to greater tumor
specificity and decreased toxicity.49
Fluorouracil Metabolism
Fluorouracil (5FU) undergoes extensive catabolic and
anabolic metabolism. Approximately 15% of a single


66
Figure 3-1. X-ray structure of l-acetyl-5FU (unprimed).
Source: Hoffman-La Roche, Nutley, NJ.


11
Since Ci is essentially the concentration in the first
layer(s) of skin, Ci will now be referred to as Cs, so
J = D-Cs/h (4)
While Cs is not usually a known quantity, it can be
represented instead by the product of the penetrant
concentration in the donor vehicle (Cv) and the
membrane-vehicle partition coefficient (Km = Cs/Cv), and the
expression now becomes
J = D-Km-Cv/h (5)
This expanded form of Fick's first law is often cited and has
been verified by data from numerous skin penetration
studies.66
Another useful measure of skin penetration is the
permeability coefficient (P). This is simply the
concentration-normalized flux which is expressed:
P = J/Cv = DKm/h (6)
Permeability coefficients are often used when comparing a
single penetrant in a series of vehicles or when studying an
homologous series of penetrants.
Enhancement of Skin Penetration
The driving force for skin penetration is the membrane-
vehicle partition coefficient (Km). If a penetrant is
delivered in a saturated solution in the presence of excess
solid, then the chemical potential, or escaping tendency, is
maximized, and the actual concentration in the vehicle (Cv)


16
compounds in water, no enzymatic activation was necessary to
release the parent compound. In a later report,79 one of
these prodrugs was compared with a number of 5FU formulations
including four commercially available creams and solutions.
The prodrug outperformed the formulations by a minimum of
four times in terms of 5FU delivered through hairless mouse
skin.
Sasaki et al.37 studied the delivery of 1-alkylcarbamoyl
derivatives of 5F through rat skin. All three derivatives
(butyl, hexyl, and octyl) were more effective in delivering
5F than 5FU itself. The lipid solubilities of the three
derivatives were comparable with each other and much higher
than 5FU. The aqueous solubilities, while less than 5FU,
showed an order of magnitude decrease between each derivative
beginning with the butyl derivative. Interestingly, the best
performing compound was the least lipid-soluble and most
water-soluble derivative, the butyl derivative.
Dermal versus Transdermal. .Delivery
It is important to make a distinction between dermal and
transdermal delivery. Most in vitro skin penetration testing
is done with excised skin to which a drug in a formulation is
applied on the donor side and samples are removed from the
receptor side. These experiments give information on
transdermal rates of delivery.


67
Figure 3-2. X-ray structure of l-acetyl-5FU (primed).
Source: Hoffmann-La Roche, Nutley, NJ.


Cumulative Amount (nmol)
46
Time (h)
Figure 2-2. Plots of cumulative amount of total 5FU species
that diffused (nmol) versus time (h) for
compounds 1, 2, 3, and 5FU.


In (At-A'
99
Plain Buffer
Formaldehyde Buffer
Time (min)
Figure 4-2. Plots of ln(At-Ao=) versus time (min) for
hydrolysis of 1,3-bis-acetyl-5FU in 0.05 M
phosphate buffer (pH=7.1, 1=0.12) with and
without formaldehyde at 32 C.


18
rather than chemical activation, then an appropriate
cutaneous enzyme must be present.
The ability of the skin to metabolize drugs and other
foreign compounds is well known.81-82 phase I reactions
(oxidation, reduction, and hydrolysis) and phase II
conjugations are known to occur. Drug-metabolizing enzymes
are distributed in all layers of the skin and the appendages.
Of particular interest in prodrug chemistry is the
presence of nonspecific esterases in the skin.82 The
predictable metabolism of esters by these hydrolytic enzymes
makes this functional group a popular choice for prodrug
synthesis.
Hairless Mouse Skin as a Model Membrane
A number of animal skins have been suggested as model
membranes for skin penetration studies. The two most common
models for in vitro diffusion studies are human and hairless
mouse skin. While the advantages of human skin are obvious,
there are also disadvantages. Human skin can be difficult to
obtain and store,83 it can be expensive,83 and it is known to
have high barrier variability.83-84 Factors such as age,
diet, and disease state may not be well controlled with human
skin.85 Hairless mouse skin, on the other hand, is easily
obtained and prepared, and other factors can ordinarily be
controlled.


90
recorded at appropriate intervals and pseudo-first-order rate
constants were determined from the expression:
ln(At-Aoo) = ln(A0-A)-kt (4)
where At is the absorbance at some time=t, hoo is the
absorbance at t=, Aq is the absorbance at t=0, k is the
pseudo-first-order rate constant, and t is the time. The
hydrolyses were sufficiently fast and the 3-acyl products
were stable enough to allow experimental determination of A.
The slopes, -k, of linear plots of ln(At-A,) versus time were
determined by linear regression. The half-lives (ti/2) were
calculated from
tl/2 = 0.693/k (5)
Each hydrolysis reaction was run in triplicate and was
followed for a minimum of three half-lives. The correlation
coefficients were >0.999.
Skin Penetration Studies
Diffusion cell experiments were performed to measure the
transdermal delivery of 5FU and the 5FU prodrugs. Franz-type
diffusion cells from Crown Glass in Somerville, NJ with 4.9
cm2 donor surface area and 20 mL receptor phase volume were
used for this purpose. The full-thickness skins were
obtained from female hairless mice (SKH-hr-1) from Temple
University Skin and Cancer Hospital.
The mice were killed by cervical dislocation, their
skins were removed immediately by blunt dissection, and


partitioning-based method for determining aqueous solubility
gave reliable values for relative solubility in each
homologous series, and overall, the values correlated well
with conventionally determined solubilities. This was the
first known successful attempt to obtain aqueous solubilities
for chemically unstable prodrugs (e.g., 1-acyl and 1,3-bis-
acyl derivatives).
Hydrolysis rates for N^-substituents followed pseudo-
first-order kinetics, but plots for N3-acyl hydrolysis were
biexponential. An 02-acyl intermediate was proposed to
account for the unusual hydrolysis and thermal decomposition
of the N3-acyl group. X-ray crystal analysis showed that the
N3-acyl group was oriented perpendicular to the 5FU ring and
was sterically and electronically hindered from nucleophilic
attack.
Skin penetration was 1.2 to nearly 40 times greater for
the prodrugs than for 5FU. The highest flux was recorded for
l-acetyl-5FD, whereas 3-propionyl-5FU, 1-ethyloxycarbonyl-
5F, and 1,3-bis-acetyl-5FU exhibited the highest rates of
delivery for their respective series. These derivatives were
also the most aqueous soluble members of each series. This
demonstrates the importance of biphasic solubility for
achieving optimal transdermal delivery of 5FU from a
homologous series of prodrugs.
Skin accumulation was highest for l-acetyl-5FU and
l-propionyl-5FD <>18 times more than 5FU) Rapid hydrolysis
of the 1-acyl series upon partitioning into the skin may
xii


135
Table 5-5. Reaction products formed during hydrolysis of
3-acetyl-5FU in 0.05 M phosphate buffer (pH=7.1, 1-0.12)
with formaldehyde at 32 C.
Reactant (M)
3-Acetyl-5FU Formaldehyde
Product (%)
5FU l-AOM-5FUa3-AOM-5FUb
1.8xl0-4
3.6xl0~4
91
4.7
4.9
1.8xl0"4
3.6xl0-3
65
18
17
1.8x10-4
3.6xl0-2
57
22
21
al-Acetyloxymethyl-5F0.
b3-Acetyloxymethyl-5FU.


119
maintained. Samples were analyzed for total 5FU species that
had diffused by UV spectroscopy (£=7.13x103 at 266 nm) after
allowing at least 72 hours for complete prodrug hydrolysis.
Cumulative amounts of total 5FU species that diffused (nmol)
were plotted against time (h) and the slopes of the linear,
"steady-state" regions were calculated using linear
regression. The slopes, when divided by 4.9 (the area of the
donor surface in cm2) gave the "steady-state" fluxes
(Hmol/cm2/h). In a separate experiment, HPLC was used to
determine intact prodrug content in the receptor phases at
each sampling time. Mobile phase containing 18-50% methanol
in 0.025 M acetate buffer (pH=5.0) was used with the system
described earlier. Aliquots were removed and chromatographed
immediately after the samples were taken. Prodrug fluxes
were calculated in the same manner as fluxes for total 5FU.
Donor phases were changed every twelve hours and were
set aside for XH NMR analysis. Stability of the prodrugs in
IPM was determined from the chemical shift of C6-H. In
dimethylsulfoxide-d6, the C6-H signal for 5FU appears at
8=7.73 ppm. For each of the 3-acyl derivatives, the same
signal in dimethylsulfoxide-d6 appears at 5=7.90. Since this
area of the spectrum is free from interference by IPM
absorbances, the two signals can be identified and quantified
if necessary.
Following removal of the donor phases after the 48-hour
application period, the epidermal sides of the skins were
washed three times with 5 mL portions of methanol to remove


127
Table 5-2. Melting points (MP), lipid solubilities (Sipm) and
aqueous solubilities (Saq) for 3-acyl derivatives.
Compound
MP
(C)
SlPM3
(mM)
SAQb
(mM)
SAQC
(mM)
SAQd
(mM)
5FU
280-2
0.049
96
-
85
17
115-7
4.3
166
105
249
18
102-3
14
198
135
190
19
111-2
22
53
23
-
20
110-1
9.2
5.0
5.5
aStandard
deviations
from the
mean were
within 3%
for IPM
solubilities (10% for compound 19).
bStandard deviations from the mean were within 8% for
aqueous solubilities determined by direct method
cStandard deviations from the mean were within 5% for
experimental values used to calculate partition coefficients
and estimated aqueous solubilities (17% for compound 19).
^Literature values from reference 26.


I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
a dissertation for the degree of Do^pfr of Pbilqsophy.
Kenneth B. Sloan, Chair
Associate Professor of
Medicinal Chemistry
X certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
Margaret 0. James
Professor of Medicinal
Chemistry
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor,of Philosophy.
=£L
KoppaJfe'V. Rao
Professor of Medicinal
Chemistry
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequat
a dissertation for the degree of
Assistant Professor of
Pharmaceutics
sand quality,
ililosqphy.


93
time between the skins and methanol. The receptor phases
were changed again, and the dermal sides were kept in contact
with the fresh buffer for 23 hours while the epidermal sides
were again left exposed to the air. After this "leaching"
period, another sample was taken from each cell to measure
the skin accumulation of total 5FU species.
Second applications to the epidermal sides of the skins
were made after the "leaching" period with a standard drug-
vehicle suspension Theophylline in propylene glycol
(0.4 M) was applied to assess the damage to the skins from
application of the initial drug-vehicle combinations.
Samples were taken at 1, 2, 3, 5, 7, 9, and 11 or 12 hours
after application. The samples were analyzed for
theophylline by UV spectroscopy (£=1.02xl04 at 271 nm) and
second application fluxes were determined as described above.
Results and Discussion
Synthesis and Structure Determination
The known 1,3-bis-acyl-5FO derivatives have melting
points25-26 and 1H NMR spectra25 in agreement with those
reported in the literature. The structures of novel
compounds were assigned by comparison of their !h NMR spectra
with those of the known homologs. Elemental microanalyses
were obtained for the novel compounds and were within
acceptable limits (0.4%).


145
All of the 3-acyl derivatives improved the skin
penetration of 5F0. Increases in flux ranged from more than
two times (3-valeryl-5FU [20]) to more than 20 times
(3-propionyl-5FU [18]) the flux of 5FU itself.
As was the case with the 1-alkyloxycarbonyl derivatives,
large amounts of intact 3-acyl prodrugs were detected in the
receptor phases. Unlike the 1-alkyloxycarbonyl derivatives,
however, the differences between the "steady-state"
percentages and the percentages calculated from samples taken
prior to "steady-state" are small. This suggests that
hydrolytic enzymes play a relatively minor role in the
hydrolysis of the 3-acyl derivatives when compared with the
1-alkyloxycarbonyl derivatives. This is consistent with the
known hydrolytic behavior of the 3-acyl series in human
plasma26 and is probably due to the steric and electronic
effects described earlier. It is important to note that the
intact prodrug percentages reported in Table 5-6 represent
the lower limits of prodrug content. Formaldehyde reaction
products (1- and 3-acyloxymethyl derivatives) were also
present in the receptor phases at the time of analysis which
indicates that more than the observed amounts of 3-acyl
derivatives had actually diffused through the skin intact.
Skin accumulation values are highest for compounds 17
and 18 and are over four times higher than the value for 5FU.
Overall, skin accumulation is much lower from the 3-acyl
derivatives than the 1-acyl derivatives which supports the


71
Table 3-4. Pseudo-first-order rate constants (k) and half-
lives (11/2) for hydrolysis of 1-acyl derivatives in
0.05 M phosphate buffer Compound
k(SD)a ti/2
(min-1) (min)
7
0.143(0.002)
4.8
8
0.222(0.004)
3.1
9
0.163(0.003)
4.3
10
0.169(0.006)
4.1
11
0.173(0.003)
4.0
12
0.183(0.003)
3.8
aMean standard deviation for n=3 values.


39
Table 2-2. Melting points (MP), lipid solubilities (Sjpm) and
aqueous solubilities (Saq) for 1-alkyloxycarbonyl derivatives.
Compound
MP
(C)
Sipm3
(mM)
Saq5
(mM)
Saqc
(mM)
SAQd
(mM)
5FU
280-2
0.049
96
-
85
1
158-60
2.1
120
Ill
124
2
127-8
13
263
174
34
3
124-6
15
55
43
-
4
97-8
34
29
23
26
5
66-7
153
5.4
5.0
5.8
6
97-8
36
0.14
0.13
aStandard
deviations
from the
mean were
within 3%
for IPM
solubilities.
bStandard deviations from the mean were within 8% for
aqueous solubilities determined by direct method.
Standard deviations from the mean were within 6% for
experimental values used to calculate partition coefficients
and estimated aqueous solubilities (11% for compound 2).
^Literature values from references 30 and 93.


14 9
derivatives had aqueous solubilities greater than 5FU even
though an hydrogen-bonding N-H group was masked in each. The
partitioning-based method for determining aqueous
solubilities was shown to give reliable values for relative
solubility in a homologous series, and overall, the values
correlated well with the conventionally determined aqueous
solubilities. It is noteworthy that this is the first known
successful attempt to obtain aqueous solubilities for such
chemically unstable prodrugs.
Hydrolysis of the 1-alkyloxycarbonyl, 1-acyl, and
1,3-bis-acyl (N1-acyl group only) derivatives followed
pseudo-first-order kinetics in 0.05 M phosphate buffer
(pH=7.1, 1=0.12) with and without formaldehyde which was used
as a preservative in the diffusion cell experiments.
Hydrolysis was faster in the presence of formaldehyde which
apparently functions as a general base catalyst. The N^-acyl
groups were much more rapidly hydrolyzed than the N3-acyl
groups even though the N3-anion is a better leaving group
than the N1-anion. The X-ray crystal structure of 1,3-bis-
acetyl-5FU showed that the N3-acetyl group was oriented
perpendicular to the plane of the 5FU ring, and the carbonyl
carbon was sterically and electronically hindered from attack
by hydroxide ion. It was suggested that the unusual
orientation of the N3-acyl group caused the slower than
expected hydrolysis rates for the 3-acyl derivatives compared
to the 1-acyl derivatives.


49
Table 2-6. Second application fluxes (J) and lag times (ti,)
for 1-alkyloxycarbonyl derivatives.
Compound
Ja(SD)b
(Hmol/cm2/h)
tL
5FU
1.2 (0.2)
1.2
1
1.9(0.1)
0.8
2
1.8(0.4)
0.9
3
1.7(0.3)
1.0
4
1.8(0.2)
1.0
5
1.8(0.1)
0.8
6
1.8(0.2)
0.9
aFlux of 0.4 M theopylline from propylene glycol.
bMean standard deviation for n=3 values.


120
all remnants of prodrug and vehicle from the skin surfaces.
This was accomplished quickly (<3 min) to minimize contact
time between the skins and methanol. The receptor phases
were changed again, and the dermal sides were kept in contact
with the fresh buffer for 23 hours while the epidermal sides
were again left exposed to the air. After this "leaching"
period, another sample was taken from each cell to measure
the skin accumulation of total 5FU species.
Second applications to the epidermal sides of the skins
were made after the "leaching" period with a standard drug-
vehicle suspension Theophylline in propylene glycol
(0.4 M) was applied to assess the damage to the skins from
application of the initial drug-vehicle combinations.
Samples were taken at 1, 2, 3, 5, 7, 9, and 11 or 12 hours
after application. The samples were analyzed for
theophylline by UV spectroscopy (e-1.02xl04 at 271 nm) and
second application fluxes were determined as described above.
Results and Discussion
Synthesis and Structure Determination
The known 3-acyl-5FU derivatives have melting points25-26
and iH NMR spectra25 in agreement with those reported in the
literature. The structures of novel compounds were assigned
by comparison of their iH NMR spectra with those of the known


28
(m, 12H, OCH2CH2CH2CI2CI2CH2CII2) 4-27 (t/ J=6 Hz, 2H, 0C2) ,
and 8.15 (d, J=7 Hz, 1H, C6-H) ; UVmax (CH3CN) 254 nm
(e=1.009xl04) .
Anal. Calc, for C13H19FN2O4: C, 54.53; H, 6.69; N, 9.79.
Found: C, 54.46; H, 6.73; N, 9.77.
Lipid Solubility
Lipid solubilities were determined using isopropyl
myristate (IPM), a commercial vehicle used in cosmetics and
topical medicinis,94 as the lipid solvent. The use of IPM as
a model lipophilic vehicle in skin penetration studies is
well established.77'95
Three suspensions of each derivative were stirred at
221 C for 48 hours. The suspensions were filtered through
0.45 |lm nylon filters, and the saturated solutions were
diluted in acetonitrile and analyzed by UV spectroscopy.
Solubilities were calculated using Beer's Law:
A = e-C-d (1)
where A is the absorbance, £ is the molar absorptivity, C is
the concentration in mol/L, and d is the path length of the
cuvette in cm. Molar absorptivities were predetermined in
triplicate in acetonitrile at 254 nm.
Aqueous Solubility
For direct measurement of aqueous solubilities, three
suspensions of each derivative were vigorously stirred in


88
these prodrugs was not attempted. A comparison of the direct
and partitioning-based methods for determining aqueous
solubility was presented in Chapter 2 for the 1-alkyloxy-
carbonyl derivatives.
Partition Coefficients
The partitioning-based method for determining aqueous
solubility utilized the saturated IPM solutions from the
lipid solubility study. For most compounds, equal volumes
(1 mL) of saturated IPM solution and 0.05 M acetate buffer
(pH=4.0) were used. The use of equal or near-equal phase
volumes is known to facilitate rapid equilibrium.96 The two
phases were mixed thoroughly for ten seconds and allowed to
separate for 60 seconds. A preliminary study showed that
there was virtually no difference in partition coefficient
(PC) values when partitioning was carried out for 10, 20, or
30 seconds (see Chapter 3). The IPM layers were diluted in
acetonitrile and analyzed by UV spectroscopy. The IPM-buffer
partition coefficients were calculated as follows:
PC = Aafter/ (Afceforg-Aafter) ^aq/Vipm (2)
where Aafter is the absorbance from the IPM layer after
partitioning, Atefore is the absorbance from the IPM layer
before partitioning, Vaq is the volume of the aqueous phase,
and Vipm is the volume of the IPM phase. Estimated aqueous
solubilities (SAq) were calculated from the IPM solubility


Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
BIOREVERSIBLE DERIVATIVES OF 5-FLU0R0URACIL (5FU) :
IMPROVING DERMAL AND TRANSDERMAL DELIVERY
WITH PRODRUGS
By
Howard D. Beall
December, 1991
Chairman: Kenneth B. Sloan
Major Department: Medicinal Chemistry
Fluorouracil (5FU) is an antineoplastic agent used
topically for treatment of actinic keratoses, superficial
basal cell carcinomas, psoriasis, and other skin conditions,
but treatment is often ineffective due to its poor skin
penetration properties. Four homologous series of
bioreversible derivatives (prodrugs) of 5FU were synthesized
characterized, and their ability to penetrate through
(transdermal) and accumulate in (dermal) the skin was
evaluated. Six 1-alkyloxycarbonyl, six 1-acyl, four 1,3-bis
acyl, and four 3-acyl derivatives were studied.
Lipid solubility was a minimum of 40 times greater for
the derivatives than for 5FU, and aqueous solubility was
actually higher than 5FU for five derivatives. The
xi


158
84. Skelly, J. P.; Shah, U. P.; Maibach, H. I.; Guy, R. H.;
Wester, R. C.; Flynn, G.; Yacobi, A. Pharm. Res. 1987,
, 265.
85. Lambert, W. J.; Higuehi, W. I.; Knutson, K.; Krill,
S. L. J. Pharm. Sci. 1989, 7£, 925.
86. Kai, T.; Mak, V. H. W.; Potts, R. 0.; Guy, R. H. ¡L.
Controlled Release 1990, 12, 103.
87. Bond, J. R.; Barry, B. W. J. Invest. Dermatol. 1988, 90
486.
88. Hinz, R. S.; Hodson, C. D.; Lorence, C. R.; Guy, R. H.
J. Invest. Dermatol. 1989, 22, 87.
89. Sloan, K. B.; Beall, H. D.; Weimar, W. R.; Villanueva,
R. Int. J. Pharm. 1991, 22, 97.
90. Scheuplein, R. J.; Ross, L. W. J. Invest. Dermatol.
1974, £2., 353.
91. Rudy, B. C.; Senkowski, B. Z. In Analytical Profiles of
Drug Substances: Florey, K., Ed.; Academic Press:
New York, 1973; Vol. 2, p. 222.
92. Sloan, K. B.; Selk, S.; Haslam, J.; Caldwell, L.;
Shaffer, R. J, Pharm. Sci. 1984, 22, 1734.
93. Buur, A.; Bundgaard, H. Int. J. Pharm. 1987, 2£, 41.
94. Merck Index; Windholz, M., Ed.; Merck: Rahway, NJ, 1976
p. 5069.
95. Surber, C.; Wilhelm, K-P.; Hori, M.; Maibach, H. I.;
Guy, R. H. Pharm. Res. 1990, 1, 1320.
96. Hansch, C. In Strategy of Drug Design: A Guide to
Biological Activity: Purcell, W. P., Bass, G. E.,
Clayton, J. M. Eds.; John Wiley and Sons: New York,
1973; p. 126.
97.
Saab, A.
J. Pharm.
N.; Sloan, K. B.; Beall,
Sci. 1990. 79. 1099.
H D ;
Villanueva, R.
98.
Gacek, M.
; Undheim. K. Acta Chem.
Scand.
1979,
fi22, 515
99.
Smith, R.
N.; Hansch, C.; Ames, M.
. M. jL.
Pharm.
. Sci.
1975, M,
599.


55
shifts (8) are reported in parts per million (ppm) from the
internal standard, tetramethylsilane (TMS). Coupling
constants (J) are expressed in cycles per second (Hz).
Infrared (IR) spectra were recorded with a Perkin-Elmer 1420
spectrophotometer and absorbances are reported in cm*1.
Ultraviolet (UV) spectra were obtained with a Cary 210 or
Shimadzu UV-265 spectrophotometer. Maximum absorbances are
reported in nm along with the molar absorptivities (£) in
L/mol. Single-crystal X-ray analysis was obtained for
l-acetyl-5FU through Hoffmann-La Roche in Nutley, NJ.
l-Acyl-5-fluorouracil (general procedure)
To 0.66 g (0.01 mol) of 85% potassium hydroxide
dissolved in methanol (20-50 mL) was added 1.31 g of
5-fluorouracil (0.01 mol). The methanol suspension was
stirred for 30 minutes, and the methanol was evaporated under
reduced pressure. The potassium salt was suspended in
acetonitrile (25-50 mL) which was evaporated under reduced
pressure to remove residual methanol. The salt was
resuspended in acetonitrile (25-50 mL), and the suspension
was added dropwise over 15 to 30 minutes to a well stirred
acetonitrile (25 mL) solution in an ice bath containing 1.0
to 1.2 equivalents of the appropriate acid chloride. The
mixture was stirred at 0 C for 60 minutes. The mixture was
filtered, and the residue was washed with acetonitrile
(25 mL). The combined acetonitrile solutions were evaporated
under reduced pressure, and the solid residue was


33
the IPM suspensions ranged from 0.3 M to 0.8 M with enough
excess solid present to maintain saturation for the duration
of the application period (see below). Each drug-vehicle
combination was run in triplicate.
Samples were taken from the receptor phases at 4, 8, 12,
21, 24, 27, 30, 33, 36, 45, and 48 hours after donor phase
application. The receptor phases were changed following
removal of each sample so that "sink" conditions were
maintained. Samples were analyzed for total 5FU species that
had diffused by UV spectroscopy (e=7.13xl03 at 266 nm) after
allowing at least 72 hours for complete prodrug hydrolysis.
Cumulative amounts of total 5FU species that diffused (Jlmol)
were plotted against time (h), and the slopes of the linear,
"steady-state" regions were calculated using linear
regression. The slopes, when divided by 4.9 (the area of the
donor surface in cm2), gave the "steady-state" fluxes
(Hmol/cm2/h). In a separate experiment, HPLC was used to
determine intact prodrug content in the receptor phases at
each sampling time. Mobile phase containing 18-50% methanol
in 0.025 M acetate buffer (pH=5.0) was used with the system
described earlier. Aliquots were removed and chromatographed
immediately after the samples were taken. Prodrug fluxes
were calculated in the same manner as fluxes for total 5FU.
Donor phases were changed every twelve hours and were
set aside for i-H NMR analysis. Stability of the prodrugs in
IPM was determined from the chemical shift of C6-H. In
dimethylsulfoxide-d6, the C6-H signal for 5FU appears at


Mechanisms of Transdermal Penetration
For most substances, the stratum corneum provides the
rate-limiting barrier to skin penetration.53-55 The stratum
corneum is 75-85% protein (dry weight),54 mostly intracellular
keratin, while the intercellular space is a lipid-enriched
region. During the outward migration of the keratinocytes
through the epidermis, lamellar bodies are synthesized, which
contain lipids, polysaccharides, and hydrolytic enzymes. As
the granular cells prepare to enter the stratum corneum, the
lamellar bodies move to the cell periphery and empty their
contents into the spaces between the cells.56 The corneocytes
and intercellular lipids are arranged in a brick and mortar
fashion,57 and the result is a very dense (1.4 g/cm3 in the
dry state),58 highly efficient moisture barrier.
In order for a substance to penetrate the stratum
corneum, it must either (1) cross through the densely packed
corneocytes, (2) traverse through the intercellular region,
thereby avoiding the keratinized cells, or (3) bypass the
stratum corneum completely by diffusing through shunt
pathways such as sweat ducts and hair follicles.53 The shunt
pathways are not considered to be significant, especially
when steady state diffusion has been attained,59 and these
pathways are generally disregarded in discussions of
penetration mechanisms.


38
Lipid (Sjpm) and aqueous (S*q) solubilities for the
1-alkyloxycarbonyl derivatives are presented in Table 2-2
along with their melting points. Lipid solubilities are
greatly enhanced by making derivatives. Solubility values
range from over 40 times greater than 5FD for 1-methyloxy-
carbonyl-5FU (1) to more than 3000 times greater than 5FO for
l-hexyloxycarbonyl-5FO (5). Increases in lipid solubility
with increasing chain length are accompanied by decreases in
melting point. A change in that trend is observed for
l-octyloxycarbonyl-5FU (6), but this is not unexpected.
Since melting point and solubility depend in part on crystal
lattice energies,101 this result indicates that the crystal
structure is dominated by the 5FU nucleus for lower homologs
and by the aliphatic side chain for higher homologs.
Aqueous solubilities are reported for both the direct
and partitioning methods. The results show that aqueous
solubility peaks for l-ethyloxycarbonyl-5FU (2) and then
decreases. Compounds 1 and 2 have aqueous solubilities
greater than 5F even though a hydrogen-bonding group (N1-H)
has been masked. Again, this is reflected in the lowered
melting points for the derivatives when compared to 5FU.
Aqueous solubilities determined by the partitioning-
based method underestimated the direct solubilities by 7% for
compounds 5 and 6, 8% for compound 1, 21% for compound 4, 22%
for compound 3, and 34% for compound 2. Relative aqueous
solubility among members of the series is the same with


139
reaction products from the hydrolysis studies, and 1H NMR
spectra obtained from a scaled-up hydrolysis reaction were
consistent with the assigned structures based on the
positions of the C6-H and N-CH2-0 absorptions.23 In addition,
l-acetyloxymethyl-5FU was isolated from the reaction mixture,
and its melting point (124-5 C) was consistent with the
literature-reported value (122-3 C).22
A possible scheme for formation of 1- and 3-acetyloxy-
methyl-5FU is presented in Figure 5-7. The even distribution
of the two products at all formaldehyde concentrations
(Table 5-5) suggests an intramolecular reaction with an
02-acyl intermediate.
Skin Penetration
Skin penetration data from the diffusion cells are
plotted as cumulative amount of total 5FU species that
diffused ((Imol) versus time (h) In Figure 5-8, results for
3-acetyl-5FU (17) and 3-propionyl-5FU (18) are compared to
5FU itself. In Figure 5-9, results for 3-butyryl-5FU (19)
and 3-valeryl-5FU (20) are compared to 5FU. Error bars
correspond to the standard deviation from the mean for n=3
values.
Values for cumulative amount of total 5FU species that
diffused and skin accumulation of total 5FU species were
obtained by UV analysis of the receptor phases. Since the UV
spectra showed lower UVmax/UVmin ratios than expected for 5FU


142
Table 5-6. Fluxes (J) lag times (ti,) and skin accumulation
(SA) values for 3-acyl derivatives.
Compound
J(SD)a
(|lmol/cm2/h)
Prodrug
S-Sb(ll h)c
(%)
tL
(h)
SA(SD)a
(^.mol)
5FU
0.24(0.09)
-
13
3.7(0.9)
17
4.4(0.5)
24(19)
19
16(3)
18
5.2(0.7)
49 (43)
15
15(2)
19
2.2(0.3)
62(52)
9.8
7.1(1.6)
20
0.55 (0.07)
-
6.2
3.3(1.1)
aMean standard deviation for n=3 values.
bPercent of total 5F as intact prodrug during "steady-state"
phase in separate experiment (n=l).
Percent of total 5FD as intact prodrug from 11 h sample in
separate experiment (n=l).


44
Table 2-4. Pseudo-first-order rate constants (k) and half-
lives (ti/2) for hydrolysis of l-raethyloxycarbonyl-5FD
in 0.05 M phosphate buffer (pH=7.1, 1=0.12)
with and without formaldehyde at 32 C.
Formaldehyde
Compound (M)
k(SD)a
(min-1)
ti/2
(min)
1 0 3.33x10-3(0.04x10-3) 208
1 3.6xl0-2 3.78xl0'3(0.05xl0-3) 183
aMean standard deviation for n=3 values.


69
Table 3-3. Partition coefficients (PC) and hydrophobicity
parameters (JE) for 1-acyl derivatives.
Compound
PCa
log (PC)
7Cb
7
0.19
-0.73
_
8
0.76
-0.12
0.61
9
2.7
0.43
0.55
10
11
1.05
0.62
11
38
1.58
0.53
12
759
2.88
0.65
Experimental
partition
coefficient (Cipm/Caq) .
bAlog(PC) for
compound
and preceding compound.


129
The direct aqueous solubility value for compound 18
agrees with the corresponding literature value included in
Table 5-2. However, the value for compound 17 is lower than
the literature value. The basis for this discrepancy is not
clear.
In Table 5-3, the solubility ratios (SR) and
experimentally determined partition coefficients (PC) are
compared for the 3-acyl series. The log(PC)-log(SR) values
indicate that the partition coefficients are generally larger
than the corresponding solubility ratios. As discussed in
Chapter 2, this is common for polar solutes (log[PC]<0) that
self-associate in the organic phase. The small negative
log(PC)-log(SR) value for compound 20 also fits the expected
trend, but the large positive value for 3-butyryl-5FU (19)
remains unexplained.
This deviation from the expected trend can also be seen
in the hydrophobicity parameter (ft) data. From the
experimental partition coefficients, the Alog(PC) value is
0.96 for the comparison between compounds 18 and 19 and 0.23
for the comparison between compounds 19 and 20, whereas the
expected average Jt value is around 0.60. On the other hand,
the Alog(PC) value for the comparison between compounds 18
and 20 is 0.60 per methylene unit. This suggests that the
butyryl derivative in this series exhibits unusual solubility
and partitioning properties compared to other members of the
series.


Table 4-4. Fluxes (J) lag times (ti), and skin accumulation
(SA) values for 1,3-bis-acyl derivatives.
Compound
J(SD)a
(|i.mol/cm2/h)
tL
(h)
SA(SD)a
(Umol)
5FU
0.24 (0.09)
13
3.7(0.9)
13
2.2(0.5)
11
10(2)
14
0.69(0.06)
6.2
4.5(1.5)
15
0.98(0.06)
-2.3
12(2)
16
0.95(0.05)
-2.8
8.8(1.8)
aMean standard deviation for n=3 values.


7
layer is only about 0.5 Urn in thickness. The outer layers
are continuously desquamated and replaced from below. The
entire transit time from basal layer to desquamation is 26 to
42 days. The cells of the epidermis are attached to each
other by desmosomes, which degrade just prior to
desquamation.52
Below the epidermis lies the dermis which constitutes
the bulk of the approximately 2 mm thickness of human skin.
The two layers are anchored to the basal lamina by various
fibrils and microfibrils. The attachment is enhanced by the
interlocking nature of the junction.52
The dermis consists of two regions, the papillary dermis
and reticular dermis. The papillary dermis is the thinner,
outermost region that is molded against the ridges and
grooves of the epidermis. It contains small, loosely
distributed fibrils and encloses the microcirculatory blood
and lymph vessels. In contrast, the collagen bundles and
elastin fibers of the reticular dermis are more densely
packed, and this region is relatively acellular and
avascular. Collagen is the major component of the dermis,
and it gives the skin its tensile strength. The dermis also
contains nerves, excretory and secretory glands, and hair
follicles.52


137
Hydrolysis Kinetics
Hydrolysis of 3-acetyl-5FU (17) to 5FU was followed by
HPLC in 0.05 M phosphate buffer (pH=7.1, 1=0.12) at 32 C
with and without formaldehyde and the hydrolyses of compounds
17 and 18 (3-propionyl-5FU) were followed by UV spectroscopy
in the buffer alone. Disappearance of compounds 17 and 18 in
the UV studies, indicated by In (At-Aoo) is plotted versus time
(min) in Figure 5-3 and Figure 5-4 respectively. The plots
appear to be biexponential with a short initial linear phase
followed by curvature to an extended linear terminal phase.
In Figure 5-5, disappearance of compound 17 as ln(C)
from the HPLC studies is plotted versus time (rain). The same
biexponential behavior is apparent for plain buffer and to a
lesser extent for the lowest formaldehyde concentration.
However, the plots for the two higher formaldehyde
concentrations are linear throughout the course of the
hydrolysis reactions.
Biexponential plots of this type suggest that the 3-acyl
derivatives undergo rapid reversible reactions (initial
phase) followed by slower hydrolysis reactions in which one
or more of the compounds in equilibrium are converted to 5FU
(terminal phase). The pseudo-first-order plots, obtained
when formaldehyde is added to the buffer, indicate that
formaldehyde must be trapping the product or products of the
initial reaction thereby preventing the reverse reaction.114*5


138
In Figure 5-6, the HPLC plot for compound 17 is shown
along with a corrected concentration plot from the
equation:114-5
In (Ccorr) = ln(Ct)+kt (7)
where CCorr is the concentration of compound 17 after
correcting for terminal phase hydrolysis and k is the
observed rate constant for the linear terminal phase. The
flattened appearance of the In(Ccorr) versus time plot
supports the assertion that a reversible reaction is involved
as the initial step in the hydrolysis of compound 17.114-5
In Table 5-4, rate constants (k) and half-lives (tj./2>
from the preceding plots are presented. The OV and HPLC data
for compound 17 are consistent, and the initial-phase rate
constants are essentially the same regardless of the amount
of formaldehyde added to the buffer. The linearity of the
plots with the higher formaldehyde concentrations is
indicated by a single rate constant.
Evidence that formaldehyde is trapping an intermediate
is presented in Table 5-5. Although an intermediate was
never observed during the HPLC studies, two products were
identified that suggest an 02-acylated intermediate. The
formation of l-acetyloxymethyl-5FU and 3-acetyloxymethyl-5FU
in equal concentrations was confirmed by comparison with
authentic samples.1 Identical retention times from the HPLC
chromatograms were observed for the authentic samples and the
1 Authentic samples were provided by Dr. Hans Bundgaard of
the Royal Danish School of Pharmacy, Copenhagen, Denmark.


152
unstable derivatives can potentially be formulated for
topical delivery.
Overall, l-acetyl-5FU was the most effective prodrug for
delivering 5FU through the skin (transdermal) and retaining
5FU in the skin (dermal). The combination of high flux,
rapid hydrolysis, high skin accumulation, and stability in
formulation makes l-acetyl-5FU the best candidate for
improving the dermal or transdermal delivery of 5F0.


79
present as prodrug are presumed to be zero since the half-
lives of these prodrugs under diffusion cell conditions are
only three to five minutes.
The improvement in skin penetration of 5FU from the
1-acyl derivatives is substantial. The best compound,
l-acetyl-5FU (7), shows an increase in flux of nearly 40
times when compared to 5F0. As chain length increases, the
fluxes decrease, but even l-octanoyl-5FU (12) improves
transdermal delivery of 5FU by two and one-half times.
Skin accumulation values are also much higher for the
1-acyl derivatives than for 5FU. They show a decrease with
increasing chain length, but the value for l-butyryl-5FU (9)
is lower than expected. As noted earlier, the lipid
solubility of compound 9 was also less than expected when
compared to the compounds around it in the 1-acyl series.
Therefore, the low skin accumulation value for compound 9 may
be due to its low affinity for the lipid regions of the skin.
Overall, skin accumulation is higher for this series than for
the 1-alkyloxycarbonyl series while skin penetration data for
the two series are similar. Rapid hydrolysis of the more
lipid-soluble 1-acyl derivatives to highly polar 5FU as they
partition into the skin may indicate that only 5FU is
diffusing through the remaining lipid regions of the skin.
This may effectively "lock in" large amounts of 5FU leading
to high skin accumulation values.
Second application fluxes and lag times are reported in
Table 3-7. Skin penetration by theophylline from propylene


97
Sol nhi 1 i ty
Lipid (Sipm) and aqueous (Saq) solubilities for the
1,3-bis-acyl derivatives are presented in Table 4-2 along
with their melting points. The downward trend in melting
point and upward trend in lipid solubility with increasing
chain length is pronounced in these diacylated compounds.
Aqueous solubilities were only determined for the first
two derivatives, and both are at least an order of magnitude
less than 5FU. The longer chain derivatives exhibit such
high lipid solubilites compared to their expected aqueous
solubilities that partitioning could not be done without
using unreasonable IPM-Buffer ratios.
Hydrolysis Kinptins
Hydrolysis of 1,3-bis-acetyl-5FU (13) to 3-acetyl-5FU
was followed by UV spectroscopy in 0.05 M phosphate buffer
(pH=7.1, 1=0.12) at 32 C with and without 0.11% formaldehyde
(3.6xl0-2 M) Disappearance of compound 13, indicated by
ln(At-A), is plotted versus time (min) in Figure 4-2. The
linearity of the plots suggests that hydrolysis of 13 follows
first-order kinetics in the presence and absence of
formaldehyde. Pseudo-first-order rate constants (k) and
half-lives (11/2) from the linear plots are presented in
Table 4-3. The much greater stability of the product


50
refers to the intersection of the linear, or "steady-state,"
region of each graph with the time (x) axis, and it is the
time required for establishing a uniform concentration
gradient within the skin.65 The percent of total 5FU present
as intact prodrug in the receptor phase is also reported in
Table 2-5. These values were calculated from samples taken
during the "steady-state" phase and from an earlier sample
(11 h) in a separate experiment using HPLC analysis (n=l).
The improvement in skin penetration of 5FU from the
1-alkyloxycarbonyl derivatives is significant except for
l-octyloxycarbonyl-5FU (6). Increases in flux are generally
about one order of magnitude, and l-ethyloxycarbonyl-5FU (2)
with nearly a 25-fold improvement is easily the best
derivative.
The presence of large amounts of intact prodrugs in the
receptor phases is a matter of interest. The high
percentages indicate that the hydrolytic enzymes of the skin
are not effectively converting the prodrugs to 5FU. It is
interesting to note, however, that percentages calculated
from a sample removed prior to "steady-state" are much lower
than the "steady-state" values. It is possible that the
large amounts of diffusant present at "steady-state" are
simply too much for the enzymes to handle. Another
possibility is that the continuous changing of the receptor
phase with each sample eventually depletes the enzymatic
activity.


17
Topical 5FU works on afflicted cells in the epidermal
region of the skin. This is considered dermal delivery.
Transdermal techniques provide valuable information for
dermally targeted drugs since they indicate how effectively
the drugs penetrate the barrier layer of the skin. Other
experiments can be done to augment the transdermal results
such as measuring accumulation of the drug in the skin.
The correlation between transdermal delivery rates and
epidermal uptake was studied by Sloan et al.79'60 Incorpora
tion of 3H-deoxyuridine into epidermal DNA of live hairless
mice was quantitated by scintillation counting following
application of various 5F0 formulations and a 5FU prodrug.
The prodrug, which had the highest in vitro transdermal flux,
was also the most effective at inhibiting epidermal DNA
synthesis in vivo. The correlation was also good among the
formulations.
Cutaneous Metabolism
When a prodrug is applied topically and targeted for a
dermal site, it is essential that the parent drug is released
before the prodrug leaves the epidermal region. If a prodrug
has a relatively short half-life under physiological
conditions, such as the aforementioned Mannich-base prodrugs
of 5FU, then release of the parent drug will probably occur
either prior to or during its transit through the viable
epidermis. However, if a compound depends on enzymatic


CHAPTER 1
INTRODUCTION
Fluorouracil
Fluorouracil (5FU) is one of the most widely used1 and
studied2 anticancer agents. It is used in the palliative
treatment of solid tumors and is most effective in
combination with other antineoplastic agents2 and in
conjunction with radiation therapy.2 It has been used to
treat carcinoma of the colon, rectum, breast, and stomach
and, with less effectiveness, carcinoma of the ovary, cervix,
urinary bladder, liver, and pancreas.4
Fluorouracil (5FU) is used as a single agent in topical
preparations and is effective in treating actinic (solar)
keratoses,5-8 superficial basal cell carcinomas,9-10 and
psoriasis.11'12 Topical 5FU has also been used to treat a
variety of other precancerous conditions, malignant and
benign tumors, and dermatoses.14 Commercial 5FU creams and
solutions may be adequate for treating lesions on the face,
but other areas, especially the forearms and hands, respond
poorly probably due to lack of penetration by 5FU.15 Some
conditions, such as psoriasis, respond to topical 5FU therapy
when the drug is applied under an occlusive dressing.12
1


24
Table 2-1. Structures of 1-alkyloxycarbonyl derivatives.
Compound
R
l-methyloxycarbonyl-5FU
l-ethyloxycarbonyl-5FU
l-propyloxycarbonyl-5FU
l-butyloxycarbonyl-5FU
l-hexyloxycarbonyl-5F0
l-octyloxycarbonyl-5F0
(1)
-ch3
(2)
-CH2CH3
(3)
-(CH2)2CH3
(4)
-(CH2)3CH3
(5)
-(ch2)5ch3
(6)
-(CH2)7CH3


95
Figure 4-1. X-ray structure of 1,3-bis-acetyl-5FU.
Source: Hoffman-La Roche, Nutley, NJ.


155
33. Buur, A.; Bundgaard, H.; Falch, E. Acta Pharm. Suec.
1986, 21, 205.
34. Buur, A.; Bundgaard, H. Int. J. Pharm. 1985, 21, 209.
35. Ozaki, S.; Ike, Y.; Mizuno, H.; Ishikawa, K.; Mori, H.
Bull. Chero. Soc. Jap. 1977, 2, 2406.
36. Sasaki, H.; Takahashi, T.; Nakamura, J.; Konishi, R.;
Shibasaki, J. J. Pharm. Sci. 1986, 75 676.
37. Sasaki, H.; Takahashi, T.; Mori, Y.; Nakamura, J.;
Shibasaki, J. Int. J. Pharm. 1990, 12, 1.
38. Sloan, K. B.; Koch, S. A. M.; Siver, K. G. Tnt. ,T.
Pharm. 1984, 21, 251.
39. Sloan, K. B.; Sherertz, E. F.; McTiernan, R. G. Int. J
Pharm. 1988, 11, 87.
40. Kametani, T.; Kgasawa, K.; Hiiragi, M.; Wakisaka, K.;
Nakazato, k.; Ichikawa, K.; Fukawa, K.; Irino, 0.;
Nishimura, N.; Okada, T. J. Med. Chem. 1982, 21, 1219.
41. Kamata, S.; Haga, N.; Matsui, T.; Nagata, W. Chem.
Pharm. Bull. 1985, 11, 3160.
42. Hoshiko, T.; Ozaki, S.; Watanabe, Y.; Ogasawara, T.;
Yamauchi, S.; Fujiwara, K.; Hoshi, A.; ligo, M. Chem.
Pharm.
Bull. 1985,
11,
2832 .
43.
Ozaki,
S.; Nagase,
T.;
Tamai, H.;
: Mori, H.;
1 Hoshi, A
ligo, M. Chem, Pharm. Bull. 1987, H, 3894.
44. Ozaki, S.; Ohnishi, J.; Watanabe, Y.; Nohda, T.; Nagase
T.; Akiyama, T.; Uehara, N.; Hoshi, A. Polymer. I. 1989
21, 955.
45. Ouchi, T.; Fujino, A.; Tanaka, K.; Banba, T. J.
Controlled Release 1990, 12, 143.
46. Kingsbury, W. D.; Boehm, J. C.; Mehta, R. J.; Grappel,
S. F.; Gilvarg, C. J. Med. Chem. 1984, 22, 1447.
47. Hiller, S. A.; Zhuk, R. A.; Lidaks, M. Y. Dokl. Akad.
Nauk. SSSR 1967, 122, 332.
48. Hoshi, A.; ligo, M.,- Nakamura, A.; Yoshida, M.;
Kuretani, K. Gann 1976, 22, 725.


41
either procedure, and overall, agreement between the two
methods is good.
The direct aqueous solubility values for compounds 1, 4,
and 5 agree with the corresponding literature values included
in Table 2-2. However, the value for compound 2 is much
higher than the literature value. The basis for this
discrepancy is not clear.
In Table 2-3, the solubility ratios (SR) and
experimentally determined partition coefficients (PC) are
compared for the 1-alkyloxycarbonyl series. The values for
log (PC)-log(SR) indicate that the partition coefficients are
somewhat higher than the corresponding solubility ratios.
Generally, the more polar derivatives (log[PC]<0) show the
greatest difference with the exception of compound 1.
Yalkowsky et al.102 studied solubility ratios and
octanol-water partition coefficients for a broad range of
solutes. They concluded that self-association of polar
solutes in octanol increases the ability of octanol to
accommodate the solute which increases the partition
coefficient. Conversely, a nonpolar solute causes a decrease
in the partition coefficient by self-associating in the
aqueous phase.1(^2 The present results can be explained on the
same basis. The lower than expected log(PC)-log(SR) value
for compound 1 could be due to the low concentration in the
IPM phase during partitioning of this derivative. Since the
aqueous solubility of compound 1 is nearly 60 times its lipid
solubility, the concentration in the IPM phase is reduced


77
Table 3-6. Fluxes (J) lag times (ti,) and skin accumulation
(SA) values for 1-acyl derivatives.
Compound
J(SD)a
(Hmol/cm2/h)
tL
SA(SD)a
(pimol)
5F
0.24 (0.09)
13
3.7(0.9)
7
9.3(0.3)
10
68(10)
8
4.3(0.1)
13
69(10)
9
1.3(0.2)
12
8.2(2.7)
10
1.0(0.1)
9.0
16(4)
11
1.1(0.0)
5.6
11(3)
12
0.60(0.01)
6.5
12(3)
aMean standard deviation for n=3 values.


Table 3-6. Fluxes (J), lag times (ti.), and skin
accumulation (SA) values for 1-acyl derivatives 77
Table 3-7. Second application fluxes (J) and lag times
(ti) for 1-acyl derivatives 78
Table 4-1. Structures of 1,3-bis-acyl derivatives 83
Table 4-2. Melting points (MP), lipid solubilities
(Sipm) and aqueous solubilities (Saq) for 1,3-bis-
acyl derivatives 98
Table 4-3. Pseudo-first-order rate constants (k) and
half-lives (tj./2> for hydrolysis of 1,3-bis-acetyl-5FU
in 0.05 M phosphate buffer (pH=7.1, 1=0.12) with and
without formaldehyde at 32 C 100
Table 4-4. Fluxes (J), lag times (tl), and skin
accumulation (SA) values for 1,3-bis-acyl
derivatives 104
Table 4-5. Second application fluxes (J) and lag times
(tl) for 1,3-bis-acyl derivatives 105
Table 5-1. Structures of 3-acyl derivatives 110
Table 5-2. Melting points (MP), lipid solubilities
(Sipm) / and aqueous solubilities (Saq) for 3-acyl
derivatives 127
Table 5-3. Solubility ratios (SR), partition
coefficients (PC) and hydrophobicity parameters (7C)
for 3-acyl derivatives 128
Table 5-4. Pseudo-first-order rate constants (k) and
half-lives (ti/2) for hydrolysis of 3-acyl derivatives
in 0.05 M phosphate buffer (pH=7.1, 1=0.12) with and
without formaldehyde at 32 C 134
Table 5-5. Reaction products formed during hydrolysis of
3-acetyl-5FU in 0.05 M phosphate buffer (pH=7.1,
1=0.12) with formaldehyde at 32 C 135
Table 5-6. Fluxes (J) lag times (tj,) and skin
accumulation (SA) values for 3-acyl derivatives 142
Table 5-7. Second application fluxes (J) and lag times
(tl) for 3-acyl derivatives 143
vii


143
Table 5-7. Second application fluxes (J) and lag times (ti,)
for 3-acyl derivatives.
Compound
Ja(SD)b
(Hmol/cm2/h)
tL
(h)
5FU
1.2(0.2)
1.2
17
1.6(0.1)
0.5
18
1.8(0.2)
1.2
19
1.1(0.1)
1.0
20
1.1(0.2)
1.3
aFlux of 0.4 M theopylline from propylene glycol.
bMean standard deviation for n=3 values.


126
Solubility
Lipid (Sipm) and aqueous (Saq) solubilities for the
3-acyl derivatives are presented in Table 5-2 along with
their melting points. Lipid solubilities are greatly
enhanced by making these derivatives, and solubility
generally increases with increasing chain length. A change
in this trend is observed for 3-valeryl-5FU (20). A similar
downturn in lipid solubility was also seen for l-butyryl-5FU
in Chapter 3.
Because of the relative stability of the 3-acyl
derivatives,26 aqueous solubilities were determined by both
the direct and partitioning-based methods. The results show
that aqueous solubility is highest for 3-propionyl-5FU (18)
and then decreases. Both compound 18 and 3-acetyl-5FU (17)
exhibit aqueous solubilities greater than 5FU.
Aqueous solubilities determined by the partitioning-
based method underestimated the direct solubilities by 32%
for compound 18, 37% for compound 17, 57% for compound 19,
and overestimated the direct solubility by 10% for
compound 20. Relative aqueous solubility among members of
the series is the same with either procedure, but agreement
between the two methods is not as good as that seen for the
1-alkyloxycarbonyl derivatives.


116
10 (Ira reversed-phase column, 250 mm x 4.6 mra (inside
diameter). The mobile phase contained 10% methanol and 90%
0.025 M acetate buffer (pH=5.0) and was run at 1.0 mL/min.
The column effluent was monitored at 254 nm, and quantitation
was based on peak areas. Standards chromatographed under the
same conditions were used for calibration.
Hydrolysis was initiated by adding 0.4 mL of a stock
solution of compound 17 in acetonitrile to 25 mL of buffer
prewarmed to 32 C in a constant temperature water bath to
give final concentrations of ~1.8xl0-4 M. Aliquots were
removed at appropriate intervals and chromatographed
immediately. Pseudo-first-order rate constants were
determined from the expression:
ln(Ct) = In (C0) -kt (4)
where Ct is the concentration at some time=t, C0 is the
concentration at t=0, k is the pseudo-first-order rate
constant, and t is the time. The slopes, -k, of linear plots
of ln(Ct) versus t were determined by linear regression. The
half-lives (ti/2) were calculated from
ti/2 = 0.693/k (5)
The hydrolyses of compounds 17 and 18 (3-propionyl-5FU)
in buffer alone were followed by UV spectroscopy at 300 nm
where the absorbance decrease accompanying hydrolysis was
maximized. Hydrolysis was initiated by adding 60 |1L of a
stock solution of the derivative in acetonitrile to 3 mL of
buffer prewarmed to 32 C in a thermostatted quartz cuvette
to give final concentrations of ~1.8xl0"4 M. Absorbances were


92
removal of each sample so that "sink" conditions were
maintained. Samples were analyzed for total 5FU species that
had diffused by UV spectroscopy (£-7.13xl03 at 266 nm) after
allowing at least 72 hours for complete prodrug hydrolysis.
Cumulative amounts of total 5F0 species that diffused (Jlmol)
were plotted against time (h), and the slopes of the linear,
"steady-state" regions were calculated using linear
regression. The slopes, when divided by 4.9 (the area of the
donor surface in cm2) gave the "steady-state" fluxes
(Hmol/cm2/h). Because of the rapid chemical hydrolysis of
the 1,3-bis-acyl derivatives, no attempt was made to analyze
the receptor phases for prodrug content.
Donor phases were changed every twelve hours and were
set aside for XH NMR analysis. Stability of the prodrugs in
IPM was determined from the chemical shift of C6-H. In
dimethylsulfoxide-d6, the C6-H signal for 5FU appears at
5=7.73 ppm, and for the 3-acyl derivatives, it appears at
5=7.90. For each of the 1,3-bis-acyl derivatives, the same
signal in dimethylsulfoxide-dg appears at 5>8.40 ppm. Since
this area of the spectrum is free from interference by IPM
absorbances, the three signals can be identified and
quantified if necessary.
Following removal of the donor phases after the 48-hour
application period, the epidermal sides of the skins were
washed three times with 5 mL portions of methanol to remove
all remnants of prodrug and vehicle from the skin surfaces.
This was accomplished quickly (<3 min) to minimize contact


13
coefficient (Km) decreases and vice versa. In a series of
papers,69-72 Sloan and coworkers suggested that a parabolic
relationship exists for log Km and log P when they are
plotted against vehicle polarity. Log Km and log P reach a
minimum where vehicle polarity is equal to penetrant polarity
or in other words, where penetrant solubility in the vehicle
is the greatest. While this relationship might be expected,
a more interesting finding was that in most instances fluxes
are also lower from those vehicles in which the penetrants
are most soluble. While a formulation approach may permit
some improvement in skin penetration values, the maximum
achievable levels appear to be limited.
The final method for enhancing skin penetration is the
prodrug approach. This is the only method in which a
substantial increase in Km can be realized, and according to
equation (5), this should translate into substantially
improved skin penetration. It was this knowledge that led to
adoption of the prodrug method for the current project.
Prodruas and Skin Penetration
The term, prodrug, was defined earlier in this chapter.
Prodrugs, bioreversible derivatives, or latentiated drugs are
simply compounds which undergo biotransformation before
producing their pharmacological effects.7^
Generally, prodrugs are designed to overcome some kind
of barrier to a drug's usefulness. These barriers can


k (min-1)
73
Formaldehyde (M)
Figure 3-3. Plot of pseudo-first-order rate constant (k)
versus formaldehyde concentration (M) for hydrolysis of
l-acetyl-5FU in 0.05 M phosphate buffer
(pH=7.1, 1=0.12) at 32 C.


BIOREVERSIBLE DERIVATIVES OF 5-FLUOROURACIL (5FU):
IMPROVING DERMAL AND TRANSDERMAL DELIVERY
WITH PRODRUGS
BY
HOWARD D. BEALL
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1991


89
(Sipm) and the partition coefficient:
Saq = Sipm/PC (3)
Partitioning was carried out in triplicate for a fixed volume
ratio for each derivative. For those compounds with large
differences in solubility in one phase relative to the other,
volume ratios (IPMrbuffer) other than 1:1 were necessary, but
the ratio never exceeded 10:1 or 1:10.
Hydrolysis Kinetics
Hydrolysis rates have previously been reported for
several members of this homologous series.26 In the present
study, hydrolysis rates for the conversion of 1,3-bis-acetyl-
5FU (13) to 3-acetyl-5FU were determined at 32 C in 0.05 M
phosphate buffer (pH=7.1, 1=0.12) and in the same buffer with
0.11% formaldehyde (3.6xl0-2 M) The rate in the presence of
formaldehyde was determined for comparison with the rate in
plain buffer since formaldehyde was used as a preservative in
the diffusion cell experiments described in the following
section.
The hydrolyses were followed by UV spectroscopy at
266 nm where the absorbance decrease accompanying conversion
of the 1,3-bis-acyl derivatives to 3-acyl derivatives was
maximized. Hydrolysis was initiated by adding 60 |1L of a
stock solution of the derivative in acetonitrile to 3 mL of
buffer prewarmed to 32 C in a thermostated quartz cuvette to
give final concentrations of ~1.8xl04 M. Absorbances were


94
Structural analysis of the 1,3-bis-acyl derivatives was
helpful in resolving the apparent discrepancy between acidity
and reactivity at the two attachment sites. In Chapter 1, it
was noted that the monoanion of 5FU is actually a mixture of
N1- and N3-anions,30 and that a comparison of ionization
constants for 5FU derivatives identically substituted at the
N1- and N3-positions suggests that the N3-position is more
acidic.22'30 This is supported by spectral studies in aqueous
media (IR,16 ov,106-7 and H NMR108) which showed that for the
5FU monoanion, the N3-anion is the dominant form. Formation
of the monoanion, however, leads exclusively to N1-acylated
products (Chapter 2 and Chapter 3). In addition, the
1,3-bis-acyl derivatives undergo rapid N1-deacylation during
hydrolysis.26 Ordinarily, this would imply that the N1-anion
is more stable than the N3-anion and that the the Ni-position
is more acidic.
The structure of 1,3-bis-acetyl-5FU (13) from single
crystal X-ray analysis is shown in Figure 4-1. The carbonyl
bond in the Ni-acetyl group is positioned cis to the C6-H bond
as it is in l-acetyl-5FU (7). As expected, the chemical
shifts of C6-H in chloroform-d show the same anisotropic
effect for compounds 7 and 13. The N3-acetyl group, however,
is shown to be nearly perpendicular to the plane of the 5FU
ring. Apparently, it is being forced out of plane by the
carbonyl groups at the C2- and expositions.


Figure 5-1. Possible scheme for thermal decomposition of
3-acetyl-5FU 123
Figure 5-2. Possible scheme for thermal intramolecular
rearrangement for 3-acetyl-5FU to l-acetyl-5FU 124
Figure 5-3. Plot of In (At-Aod versus time (min) for
hydrolysis of 3-acetyl-5FU in 0.05 M phosphate buffer
(pH-7.1, 1=0.12) at 32 C 130
Figure 5-4. Plot of ln(At-A) versus time (min) for
hydrolysis of 3-propionyl-5FU in 0.05 M phosphate
buffer (pH=7.1, 1=0.12) at 32 C 131
Figure 5-5. Plots of ln(C) versus time (min) for
hydrolysis of 3-acetyl-5FU in 0.05 M phosphate buffer
(pH=7.1, 1=0.12) with (n=2) and without (n=3)
formaldehyde at 32 C 132
Figure 5-6. Plots of ln(C) versus time (min) for
hydrolysis of 3-acetyl-5FU in 0.05 M phosphate buffer
(pH=7.1, 1=0.12) at 32 C using actual concentration
(Ct) and concentration corrected for secondary
degradation (CCorr) 133
Figure 5-7. Possible scheme for reaction of 3-acetyl-5FU
with formaldehyde to form l-acetyloxymethyl-5FU and
3-acetyloxymethyl-5FU 136
Figure 5-8. Plots of cumulative amount of total 5FU
species that diffused (nmol) versus time (h) for
compounds 17, 18, and 5FU 140
Figure 5-9. Plots of cumulative amount of total 5FU
species that diffused (|lmol) versus time (h) for
compounds 19, 20, and 5FU 141
ix


Cumulative Amount (nmol)
102
Time (h)
Figure 4-3. Plots of cumulative amount of total 5FU species
that diffused ((imol) versus time (h) for
compounds 13, 14, and 5FU.


87
1.3-1.8 J=7 Hz, 2H, 3-COC2) 3.08 (t, J=7 Hz, 2H, I-COCH2) and 8.22
(d, J=7 Hz, 1H, C6-fi); UVmax (CH3CN) 262 nm (e=l.071xl04).
Anal. Calc, for C14H19FN2O4: C, 56.37; H, 6.42; N, 9.39.
Found: C, 56.29; H, 6.48; N, 9.33.
Lipid solubility
Lipid solubilities were determined using isopropyl
myristate (IPM), a commercial vehicle used in cosmetics and
topical medicinis,94 as the lipid solvent. The use of IPM as
a model lipophilic vehicle in skin penetration studies is
well established.77, 95
Three suspensions of each derivative were stirred at
221 C for 48 hours. The suspensions were filtered through
0.45 Hm nylon filters, and the saturated solutions were
diluted in acetonitrile and analyzed by UV spectroscopy.
Solubilities were calculated using Beer's Law:
A = e-C-d (1)
where A is the absorbance, £ is the molar absorptivity, C is
the concentration in mol/L,and d is the path length of the
cuvette in cm. Molar absorptivities were predetermined in
triplicate in acetonitrile at 262 nm.
Aqueous .Solubility
Because of the chemical instability of the 1,3-bis-acyl
derivatives, direct measurement of aqueous solubilities for


156
49. Ajmera, S.; Bapat, A. R.; Stephanian, E.; Danenberg, P.
V. J, Med. Chem. 1988, L, 1094.
50. Pinedo, H. M.; Peters, G. F. J. J. Clin. Oncol. 1988, £
1653.
51. Douglas, K. T. Med. Res. Rev. 1987, 2, 441.
52. Odland, G. F. In Biochemistry and Physiology of the
Skin: Goldsmith, L. A., Ed.; Oxford: New York, 1983;
Vol. 1, p. 3.
53. Scheuplein, R. J.; Bronaugh, R. L. In Biochemistry and
Physiology of the Skin: Goldsmith, L. A., Ed.; Oxford:
New York, 1983; Vol. 2, p. 1255.
54. Flynn, G. L.; Stewart, B. Drug Dev. Res. 1988, 12, 169.
55. Blank, I. H. In Percutaneous Absorption: Bronaugh, R.
L., Maibach, H. I., Eds.; Marcel Dekker: New York, 1985
p. 97 .
56. Elias, P. M. Arch. Dermatol. Res. 1981, 270. 95.
57. Elias, P. M.; Grayson, S.; Lampe, M. A.; Williams, M.
L.; Brown, B. E. In Stratum Corneum: Marks, R., Plewig,
G., Eds.; Springer-Verlag: New York, 1983; p. 53.
58.
Flynn, G.
Rhodes, C
p. 263.
L.
. T
In Modern Pharmaceutics: Banker, G. S.
., Eds.; Marcel Dekker: New York, 1990;
59.
Barry, B.
W.
J. Controlled Release
1991, 1£, 237.
60.
Idson, B.
Pharm. Sci. 1975, £1,
901.
61.
Hadgraft,
J.
J. Controlled Release
1991, 1£, 221.
62 .
Potts, R.
0.;
' Golden, G. M.; Francoeur, M. L.; Mak,
V. H. W.; Guy, R. H. J, Controlled Release 1991, 1£,
249.
63. Bodd, H. E.; van den Brink, I.; Koerten, H. K.;
de Haan, F. H. N. J. Controlled Release 1991, 1£, 227.
64. Elias, P. M. J. Controlled Release 1991, 12, 199.
65. Martin, A.; Swarbrick, J.; Cammarata, A. Physical
Pharmacy; Lea and Febiger: Philadelphia, 1983; p. 399.
66.Scheuplein, R. J. J, Invest. Dermatol. 1976, £7, 31.


4
intravenous dose of 5FU is excreted unchanged in the urine
within six hours, and 90% of that is detected within the
first hour.4 Approximately 80% of the dose is metabolized by
the liver and extrahepatic tissues.1 The principal
catabolite of 5FU is a-fluoro-p-alanine (FBAL), which
accounts for over 95% of the catabolic products found in the
urine. Bile acid conjugates of FBAL represent the major
biliary catabolites of 5FU.1
A number of anabolic pathways have been characterized
for 5FU, and one or more of them may be responsible for the
activation and subsequent cytotoxicity of the drug. The most
direct activation pathway utilizes orotidine monophosphate
phosphoribosyl transferase (OMPT) to form 5-fluorouridine-5'-
monophosphate (FUMP) from 5FU and 5-phosphoribosyl-l-pyro-
phosphate. FUMP can also be generated in two steps with
uridine phosphorylase and uridine kinase. With two more
kinases 5-fluorouridine-5'-triphosphate (FUTP) is formed
which can be incorporated into ribonucleic acid (RNA) leading
to RNA dysfunction.1 Some cell lines appear to favor the
OMPT pathway, while others favor the uridine phosphorylase
pathway, and the FUTP formed from each is added to different
fractions of RNA.50
Some of the 5-fluorouridine-5'-diphosphate (FUDP) that
is generated above as an intermediate can become a substrate
for ribonucleotide reductase. The resulting deoxynucleotide,
5-fluoro-2'-deoxyuridine-5'-diphosphate (FdUDP), can form
5-fluoro-2'-deoxyuridine-5'-triphosphate (FdUTP) using


146
earlier assertion that rapid hydrolysis of 5FU prodrugs leads
to higher skin levels of 5FU.
Second application fluxes and lag times are reported in
Table 5-7. Skin penetration by theophylline from propylene
glycol, the standard drug-vehicle combination, after
treatment with the 3-acyl derivatives is similar to that
following treatment with 5FU.
The stability of the prodrugs in the IPM formulations
was assessed by 1H NMR analysis of the donor phases. After a
minimum of five days from the time the suspensions were
prepared until their ifi NMR spectra were recorded, including
at least twelve hours during which the formulations were in
contact with the skins, the 3-acyl derivatives were found to
be intact with no evidence of 5FU formation.
.gumma ry
The 3-acyl derivatives of 5FU exhibited increased lipid
solubilities when compared to 5FU. Aqueous solubility
reached a maximum for 3-propionyl-5FU (18) and then decreased
with increasing chain length. Skin penetration and skin
accumulation were also maximized for compound 18 suggesting
that both lipid and aqueous solubilities are important for
predicting transdermal and dermal delivery of these 5FD
prodrugs. The presence of high percentages of prodrugs in
the receptor phases, indicating insufficient release of the
parent drugs in the hairless mouse skin model, may limit the


63
had diffused by UV spectroscopy (6=7.13x103 at 266 nm) after
allowing at least 24 hours for complete prodrug hydrolysis.
Cumulative amounts of total 5FU species that diffused (Umol)
were plotted against time (h), and the slopes of the linear,
"steady-state" regions were calculated using linear
regression. The slopes, when divided by 4.9 (the area of the
donor surface in cm2), gave the "steady-state" fluxes
(|imol/cm2/h) Because of the rapid chemical hydrolysis of
the 1-acyl derivatives, no attempt was made to analyze the
receptor phases for prodrug content.
Donor phases were changed every twelve hours and were
set aside for XH NMR analysis. Stability of the prodrugs in
IPM was determined from the chemical shift of C-H. In
dimethylsulfoxide-d6, the C6-H signal for 5F appears at
8=7.73 ppm. For each of the 1-acyl derivatives, the same
signal in dimethylsulfoxide-d6 appears at 5>8.20 ppm. Since
this area of the spectrum is free from interference by IPM
absorbances, the two signals can be identified and quantified
if necessary.
Following removal of the donor phases after the 48-hour
application period, the epidermal sides of the skins were
washed three times with 5 mL portions of methanol to remove
all remnants of prodrug and vehicle from the skin surfaces.
This was accomplished quickly (<3 min) to minimize contact
time between the skins and methanol. The receptor phases
were changed again, and the dermal sides were kept in contact
with the fresh buffer for 23 hours while the epidermal sides


80
glycol, the standard drug-vehicle combination, after
treatment with the 1-acyl derivatives is similar to that
following treatment with 5FU. In fact, the longer-chain
derivatives in this series actually appear to have a
protective effect on the skin since they cause less damage
than 5FU itself.
The stability of the prodrugs in the IPM formulations
was assessed by NMR analysis of the donor phases. After a
minimum of five days from the time the suspensions were
prepared until their 1H NMR spectra were recorded, including
at least twelve hours during which the formulations were in
contact with the skins, the 1-acyl derivatives were found to
be intact with no evidence of 5FU formation.
Snmma ry
The 1-acyl derivatives of 5FU exhibited decreased
melting points and increased lipid solubilities when compared
to 5FU. Aqueous solubility reached a maximum for 1-acetyl-
5FU (7) and decreased from there with increasing chain
length. Skin penetration was also maximized for compound 7
while skin accumulation values were highest and essentially
the same for compound 7 and l-propionyl-5FU (8). As was also
the case for the 1-alkyloxycarbonyl series, this demonstrates
that both lipid and aqueous solubilities are important for
predicting transdermal and dermal delivery of 5FU prodrugs.
The 1-acyl derivatives may be better candidates for dermal


19
The importance of hairless mouse skin for in vitro
experimentation has been noted86 despite its generally greater
permeability when compared to human skin. It has been
suggested that this difference may actually be an asset in
that small changes in permeability will be amplified in the
mouse skin model.86
A major criticism of hairless mouse skin involves its
ability to withstand the effects of hydration,87-88 a
necessary condition for controlled in vitro diffusion
experiments. Permeability increases as a function of
hydration time have been attributed to breakdown of the
stratum corneum barrier in these studies. A recent report,89
however, suggested that the absence of an adequate
preservative in the receptor phase may actually be
responsible for breakdown of the skin. Skin penetration data
was collected for delivery of theophylline from a propylene
glycol vehicle following skin hydration periods ranging from
4 to 120 hours. It was found that increased theophylline
flux and loss of barrier function corresponded to the
presence of microbial growth in the receptor phase. When
microbial growth was completely inhibited, fluxes were
essentially constant for all hydration periods.89 Finally,
with regard to hydration, Scheuplein and Ross noted that
"even well-hydrated stratum corneum preferentially dissolves
lipid-soluble molecules, so that the selective permeability
of these molecules is preserved" (p. 353).90


61
were determined from the expression:
In (At-Aoo) = In (Aq-A)-kt (4)
where At is the absorbance at some time=t, A is the
absorbance at t=, A0 is the absorbance at t=0, k is the
pseudo-first-order rate constant, and t is the time. The
hydrolyses were sufficiently fast to allow experimental
determination of Aoo. The slopes, -k, of linear plots of
In(At-A) versus time were determined by linear regression.
The half-lives (11/2) were calculated from
ti/2 = 0.693/k (5)
Each hydrolysis reaction was run in triplicate and was
followed for a minimum of three half-lives. The correlation
coefficients were >0.999.
Skin Penetration Studies
Diffusion cell experiments were performed to measure the
transdermal delivery of 5FU and the 5FU prodrugs. Franz-type
diffusion cells from Crown Glass in Somerville, NJ with 4.9
cm2 donor surface area and 20 mL receptor phase volume were
used for this purpose. The full-thickness skins were
obtained from female hairless mice (SKH-hr-1) from Temple
University Skin and Cancer Hospital.
The mice were killed by cervical dislocation, their
skins were removed immediately by blunt dissection, and
dorsal sections were mounted in the diffusion cells. The
dermal sides of the skins were placed in contact with


147
potential of this series of prodrugs at least for dermal
delivery purposes. The partitioning-based method for
determining aqueous solubility was effective for determining
the relative solubilities in the 3-acyl series, but overall,
the agreement with the direct method for determining aqueous
solubility was not as good as it was for the 1-alkyloxy-
carbonyl series. An 02-acyl intermediate was proposed to
explain the thermal decomposition and rearrangement of the
3-acyl derivatives and the unusual hydrolytic behavior of the
series including the formation of acyloxymethyl derivatives
of 5F.


57
l-Valeryl-5-fluorouraci1 (10)
Crystallization from dichloromethane/hexane gave 1.35 g
of 10 (63%): mp 120-1 C; IR (KBr) 1695, 1715, and 1740 cm'1
(C=0); *H NMR (CDCI3) 5 0.95 (t, J=7 Hz, 3H, C3), 1.3-1.8 (m
4H, COCH2CH2CH2) 3.11 (t, J=7 Hz, 2H, COCR2) and 8.24 (d,
J7 Hz, 1H, C6-H) ; OVmax (CH3CN) 261 nm (6=1.175xl04).
Anal. Calc, for C9H11FN2O3: C, 50.47; H, 5.18; N, 13.08.
Found: C, 50.52; H, 5.23; N, 13.03.
l-Hexanovl-5-f luorouracil Llll.
Crystallization from dichloromethane/hexane gave 1.71 g
of 11 (75%): mp 101-2 C; IR (KBr) 1690, 1715, and 1745 cm'1
(C=0); 1H NMR (CDCI3) 8 0.92 (distd t, 3H, CH3) 1.2-1.9 (m,
6H, COCH2CE2CH2CH2) 3.09 (t, J=7 Hz, 2H, COCH2), and 8.24 (d
J=7 Hz, 1H, C6-H) ; UVmax (CH3CN) 261 nm (6=1.158xl04) .
Anal. Calc, for C10H13FN2O3: C, 52.63; H, 5.74; N, 12.27
Found: C, 52.69; H, 5.75; N, 12.27.
l-Octanoyl-5-fluorouracil (12)
Crystallization from dichloromethane/hexane gave 1.28 g
of 12 (50%): mp 83-4 C; IR (KBr) 1685, 1710, and 1745 cm'1
(C=0); XH NMR (CDCI3) 5 0.90 (distd t, 3H, CH3) 1.2-1.8 (m,
10H, COCH2CH2CH2CH2CB2CH2) 3.10 (t, J=7 Hz, 2H, C0Cfl2), and
8.23 (d, J=7 Hz, 1H, C6-£); UVmax (CH3CN) 261 nm
(6=1.155xl04) .
Anal. Calc, for Ci2Hi7FN203: C, 56.24; H, 6.69; N, 10.93
Found: C, 56.22; H, 6.73; N, 10.96.


115
(Sipm) and the partition coefficient:
Saq = Sipm/PC (3)
Partitioning was carried out in triplicate for a fixed volume
ratio for each derivative. For those compounds with large
differences in solubility in one phase relative to the other,
volume ratios (IPM:buffer) other than 1:1 were necessary, but
the ratio never exceeded 10:1 or 1:10.
Hydro!vsis Kinetics
Hydrolysis rates have previously been reported for
several members of this homologous series.2^ In the present
study, hydrolysis rates were determined at 32 C for
3-acetyl-5FU (17) in 0.05 M phosphate buffer (pH=7.1, 1=0.12)
and in the same buffer with 0.11% formaldehyde (3.6xl0~2 M) .
The rate in the presence of formaldehyde was determined for
comparison with the rate in plain buffer since formaldehyde
was used as a preservative in the diffusion cell experiments
described in the following section. Two other concentrations
of formaldehyde were studied (3.6x10^ M and 3.6xl04 M) when
it was discovered that products in addition to 5FU were being
formed during hydrolysis of compound 17.
The hydrolyses were followed by high performance liquid
chromatography (HPLC). The HPLC system consisted of a
Beckman model 110A pump with a model 153 UV detector, a
Rheodyne 7125 injector with a 20 Hm loop, and a Hewlett-
Packard 3392A integrator. The column was a Lichrosorb RP-8


Cumulative Amount (|imol)
75
Time (h)
Figure 3-4. Plots of cumulative amount of total 5FU species
that diffused (Jlmol) versus time (h) for
compounds 7, 8, 9, and 5FU.


65
their instability. However, differences in 1H NMR spectra
are very clear. In chloroform-d, the C6-H signal for
l-acetyl-5FU is a sharp doublet at 5=8.23, while the same
signal for 3-acetyl-5FU is a broad singlet at 5=7.23, a
difference of 1.00 ppm.
Results from the single-crystal X-ray analysis show that
the ^-assignment for l-acetyl-5FU is correct. The unit cell
was found to contain two independent molecules (Figure 3-1
and Figure 3-2). Both conformations show that the carbonyl
group from the 1-acyl group is positioned cis to the C6-H.
The circulating K electrons of carbonyl bonds are known to
deshield ortho protons in structurally similar compounds such
as acetophenone.105 Thus, the X-ray results support the
explanation that an anisotropic effect is responsible for the
downfield shift of C6-H in 1H NMR spectra of N1-acylated
derivatives.
Solubility
Lipid (Sipm) and aqueous (Saq) solubilities for the
1-acyl derivatives are presented in Table 3-2 along with
their melting points. In general, melting points decrease
and lipid solubilities increase with increasing chain length.
An exception to both of these trends is seen for 1-butyryl-
5FU (9). The basis for this variance from the observed
trends is not clear.


100
Table 4-3. Pseudo-first-order rate constants (k) and half'
lives (11/2) for hydrolysis of 1,3-bis-acetyl-5FU in
0.05 M phosphate buffer (pH=7.1, 1=0.12) with and
without formaldehyde at 32 C.
Compound
Formaldehyde k(SD)a ti/2
(M) (min-1) (min)
13
13
0 0.802(0.010) 0.86
3.6xl0-2 0.868(0.039) 0.80
aMean standard deviation for n=3 values.


15
obviously improving lipid solubility, also increases water
solubility from 3 mg/mL for uracil to 500 mg/mL for 1,3-di-
methyluracil even though the methyl group is hydrophobic
itself. Melting points are also decreased as the two
intermolecular hydrogen-bonding N-H sites are masked.78 Since
alkyl groups such as methyl are stable and therefore not
bioreversible, they are not candidates to function as prodrug
promoieties. However, many other derivatives, such as those
cited earlier, when linked to the N1- or N8-site of 5FU, do
qualify as promoieties and could potentially give 5FO the
improved solubility characteristics necessary for enhanced
skin penetration.
Of the many studies devoted to making bioreversible
derivatives of 5FU, only a few have been directed at
improving its topical delivery. Mollgaard et al.21 looked at
two 1-acyloxymethyl derivatives of 5FO. One of the compounds
delivered 5FU more readily than 5FU itself through excised
human skin. Both compounds showed greatly increased lipid
solubilities with only slightly reduced water solubilities.
Hydrolysis of these derivatives was attributed to cutaneous
metabolism by hydrolytic enzymes.
Three 1,3-bis-aminomethyl (Mannich base) derivatives of
5FU were prepared by Sloan and coworkers,38-39 and their
topical delivery was studied using hairless mouse skin.
Solubilities in lipid and aqueous phases were substantially
increased as were the rates of delivery through the skin of
5FU from these prodrugs. Due to the instability of these


54
Table 3-1. Structures of 1-acyl derivatives.
Compound
R
l-acetyl-5F0 (7)
1-propiony1-5F (8)
l-butyryl-5FU (9)
l-valeryl-5FU (10)
l-hexanoyl-5FU (11)
l-octanoyl-5FU (12)
-ch3
-CH2CH3
-(CH2)2CH3
-(CH2)3CH3
-(CH2)4CH3
-(CH2) 6ch3


121
homologs. Elemental microanalyses were obtained for the
novel compounds and were within acceptable limits (0.4%).
Differences in the spectral properties between N1- and
N3-acylated derivatives were discussed in detail in Chapter 2
and Chapter 3. The 3-acyl derivatives are synthesized by
t^-deacylation of the 1,3-bis-acyl derivatives, and ^-H NMR and
IR spectra confirm that it is the N3-acyl groups that are
present in the products.
The melting points of the 3-acyl derivatives are rate
dependent and exhibit an incomplete melt followed by
resolidification and a complete melt at 40 to 100 C above
the initial thermal event suggesting thermal decomposition of
the prodrugs. The literature melting points for 3-propionyl-
5FU (18) are 99-102 C25 and 113-4 C.26 In the present
study, the melting point for compound 18 was 102-3 C when
the temperature was raised by 2 C per minute, but if the
rate was increased to 5 C per minute, the observed melting
point was 111-3 C.
The thermal decomposition of 3-acetyl-5FU (17) was
studied by dissolving 50 mg of compound 17 in 10 mL of ethyl
acetate in a beaker and heating gently to evaporate the
solvent. The temperature was increased slowly until the
residual solid had melted completely. The beaker was then
removed from the heat to allow resolidification to occur.
The decomposition was followed by TLC on silica gel
using ethyl acetate as the eluent. The TLC showed no
decomposition until after the solvent had completely


78
Table 3-7. Second application fluxes (J) and lag times (tx,)
for 1-acyl derivatives.
Compound
Ja(SD)t>
(|lmol/cm2/h)
tL(h)
5FU
1.2(0.2)
1.2
7
1.6(0.0)
0.6
8
1.2(0.2)
0.1
9
1.0 (0.0)
0.6
10
0.80 (0.03)
0.8
11
0.47 (0.02)
0.2
12
0.72 (0.11)
0.4
aFlux of 0.4 M theopylline from propylene glycol.
bMean standard deviation for n=3 values.


27
1-Propvloxvcarbonv1-5-fluorouraci] L21
Crystallization from acetone/ether gave 1.37 g of 3
(64%): mp 124-6 C; IR (KBr) 1690, 1730, and 1755 cm'1 (C=0);
XH NMR [ (CD3) 2SO] 8 0.95 (t, J=7 Hz, 3H, CH3) 1.70 (m, 2H,
OCH2CH2), 4.23 (t, J=7 Hz, 2H, OC2) and 8.15 (d, J=7 Hz, 1H,
C6-H) ; UV max (CH3CN) 254 nm (E=l. OOlxlO4) .
Anal. Calc, for C8H9FN2O4: C, 44.45; H, 4.20; N, 12.96.
Found: C, 44.53; H, 4.23; N, 12.89.
1-Butyloxycarbony1-5-fluorouraci1 14)
Crystallization from dichloromethane/hexane gave 1.33 g
of 4 (58%): mp 97-8 C (lit.30 mp 102-4 C); IR (KBr) 1695,
1735, and 1765 cm-1 (C=0); XH NMR [(CD3)2SO] 6 0.91 (t,
J=7 Hz, 3H, C3), 1.3-1.8 (m, 4H, OCH2CH2CH2) 4.27 (t, J=6
Hz, 2H, OCH2), and 8.13 (d, J=7 Hz, 1H, C6-2); (CH3CN)
254 nm <£=9.93xl03) .
1-Hexyloxvcarbonyl-5-fluorouraci1 (5)
Crystallization from dichloromethane/hexane gave 1.27 g
of 5 (49%) : mp 66-7 C (lit.93 mp 68-9 C) ; IR (KBr) 1690,
1730, and 1750 cm-1 (C=0); 3H NMR [(CD3)2SO] 5 0.87 (distd t,
3H, CH3), 1.1-1.8 (m, 8H, OCH2CH2CH2CH2CH2) 4.26 (t, J=6 Hz,
2H, OCH2), and 8.13 (d, J=7 Hz, 1H, C6-H); UVmax (CH3CN) 254
nm (6=1.004xl04).
l-0ctvloxvcarbonvl-5-fluorouracil (6)
Crystallization from dichloromethane/hexane gave 1.80 g
of 6 (61%): mp 97-8 C; IR (KBr) 1690, 1730, and 1750 cm"l
(C=0); 3H NMR [(CD3)2SO] 5 0.87 (distd t, 3H, CH3), 1.2-1.9


31
diameter). The mobile phase contained 10% methanol and 90%
0.025 M acetate buffer (pH=5.0) and was run at 1.0 mL/min.
The column effluent was monitored at 254 nm, and quantitation
was based on peak areas. Standards chromatographed under the
same conditions were used for calibration.
Hydrolysis was initiated by adding 0.4 mL of a stock
solution of compound 1 in acetonitrile to 25 mL of buffer
prewarmed to 32 C in a constant temperature water bath to
give final concentrations of ~1.8xl0~4 M. Aliquots were
removed at appropriate intervals and chromatographed
immediately. Pseudo-first-order rate constants were
determined from the expression:
ln(Ct) = In (C0) -kt (4)
where Ct is the concentration at some time=t, C0 is the
concentration at t=0, k is the pseudo-first-order rate
constant, and t is the time. The slopes, -k, of linear plots
of ln(Ct) versus t were determined by linear regression. The
half-lives (11/2) were calculated from
ti/2 = 0.693/k (5)
Each hydrolysis reaction was run in triplicate and was
followed for a minimum of three half-lives. The correlation
coefficients were >0.999.
Skin Penetration Studies
Diffusion cell experiments were performed to measure the
transdermal delivery of 5FU and the 5FU prodrugs. Franz-type


96
The steric constraints at the N3-position may explain
the lower than expected reactivity at this site. The C2- and
C4-carbonyl groups make acylation more difficult by
restricting the orientations available to the relatively
bulky acylating agents as they interact with the 5F0 anion.
Likewise, slower hydrolysis rates are due to partial negative
charges on oxygen at C2=0 and C4=0 which impede the approach
of hydroxide ions to the N3-acyl carbonyl group when it is
perpendicular to the plane of the 5FD ring. Thus, although
the N3-anion is a better leaving group than the N3-anion,
steric hindrance prevents more rapid hydrolysis of the
N3 -acyl group than the Ni-acyl group in the 1,3-bis-acyl
series.
Infrared (IR) spectra for the 1,3-bis-acyl and 3-acyl
derivatives are also consistent with the conformation
depicted in Figure 4-1. The carbonyl stretching frequency
for the N3-acyl group is 25 to 50 cm'1 higher than those
associated with the 5FU ring or N1-acyl groups. The
perpendicular orientation of the N3-acyl group apparently
prevents electron delocalization from the ring to the
carbonyl bond which means that the inductive effect of the
ring predominates, and the carbonyl stretching frequency is
increased.105


157
67 .
Mak, V. H
1991, £,
. w.,
1064
: Potts, R. 0.;
Guy, R
. H.
Pharm. Res.
68.
Pikal, M.
J.;
Shah, S. Pharm,
, Res.
1990,
, 7, 222.
69.
Sloan, K.
B.;
Koch, S. A. M.;
: Siver
, K.
G.; Flowers
F. P. J. Invest. Dermatol. 1986, £7, 244.
70. Sloan, K. B.; Siver, K. G.; Koch, S. A. M. J. Pharm.
Sci. 1986, 22, 744.
71. Sherertz, E. F.; Sloan, K. B.; McTiernan, R. G. .T.
Invest. Dermatol. 1987, £2, 147.
72. Waranis, R. P.; Siver, K. G.; Sloan, K. B. Int. J.
Pharm. 1987, 22, 211.
73. Stella, V. In Prodruas as Novel Drug Delivery Svst-pms.
ACS Symposium Series 14: Higuchi, T., Stella, V., Eds.;
American Chemical Society: Washington, 1975; p. 1.
74. Stella, V. J.; Mikkelson, T. J.; Pipkin, J. D. In Drug
Delivery Systems. Characteristics and Biomedical
Applications: Juliano, R. L., Ed.; Oxford Onversity
Press: New York, 1980; p. 112.
75. Michaels, A. S.; Chandrasekaran, S. K.; Shaw, J. E.
AIChE J. 1975, 21, 985.
76. Yalkowsky, S. H.; Flynn, G. L. J. Pharm. Sen 1973, 62.
210.
77. Sloan, K. B. Advanced Drug Delivery Rev. 1989, 2, 67.
78. Pitman, I. H. Med. Res. Rev. 1981, 1, 189.
79. Sherertz, E. F.; Sloan, K. B.; McTeirnan, R. G. Arch.
Dermatol. Res. 1990, 282. 463.
80. Sloan, K. B.; Sherertz, E. F.; McTiernan, R. G. Arch.
Dermatol. Res. 1990, 282. 484.
81. Pannatier, A.; Jenner. P.; Testa, B.; Etter, J. C. nrna
Metab. Rev. 1978, 2, 319.
82. Kao, J.; Carver, M. P. Drug Metah. Rev. 1990, 22, 363.
83.Itoh, T.; Magavi, R.; Casady, R. I.; Nishihata, T.;
Rytting, J. H. Pharm. Res. 1990, 1, 1302.


Ill
Materials and Methods
Synthesis
Melting points (mp) were determined with a Thomas-Hoover
capillary melting point apparatus and are uncorrected.
Elemental microanalyses were obtained for all novel compounds
through Atlantic Microlab, Incorporated in Norcross, Georgia.
Proton nuclear magnetic resonance (if! NMR) spectra were
obtained at 90 MHz on a Varian EM-390 spectrometer. Chemical
shifts (8) are reported in parts per million (ppm) from the
internal standard, tetramethylsilane (TMS). Coupling
constants (J) are expressed in cycles per second (Hz).
Infrared (IR) spectra were recorded with a Perkin-Elmer 1420
spectrophotometer and absorbances are reported in cm"1.
Ultraviolet (UV) spectra were obtained with a Cary 210 or
Shimadzu UV-265 spectrophotometer. Maximum absorbances are
reported in nm along with the molar absorptivities (6) in
L/mol. Thin layer chromatography (TLC) was run on Brinkman
Polygram Sil G/UV254 0.25 mm silica gel.
3-Acvl-5-fluorouracil (general procedure)
To a solution of an appropriate 1,3-bis-acyl derivative
(0.01 mol) in ether (60-650 mL) at 0 C was added 0.73 g of
tert-butylamine (0.01 mol) in ether (25 mL) dropwise over ten
minutes with stirring. The solution was stirred for a total
of 30 minutes at 0 C. The ether solution was evaporated


14
include (1) premature metabolism prior to reaching the active
site, (2) too rapid absorption and distribution when
prolonged action is needed, (3) toxicity associated with
(a) local irritation or (b) distribution to tissues other
than the target site, (4) poor site specificity leading to
subtherapeutic levels at the target site, and (5) generally
poor physical chemical properties resulting in (a) solubility
problems in the dosage form or (b) poor absorption across
biological membranes such as the blood brain barrier,
gastrointestinal lining, or skin.74 Obviously, it is the
latter problem that is addressed in the current project.
The idea that improving skin penetration is a solubility
problem has been documented in this chapter, and the
importance of biphasic (lipid and aqueous) solubility is well
recognized.7S_7 In a recent review of prodrug approaches
for improving dermal delivery, numerous literature examples
are cited in which both solubility and skin penetration data
are presented.77 The evidence strongly suggests that:
although an increase in lipid solubility due to
transient masking of a polar functional group almost
always results in enhanced dermal delivery of the parent
drug, in order to optimize delivery, it is necessary to
use the members of the (homologous) series (of prodrugs)
that are more water soluble than the parent drug or that
are the more water-soluble member(s) of the series
(p. 68) ,77
Derivatization to improve solubility characteristics of
polar heterocycles with amide and/or imide functional groups
is relatively easy to accomplish. For example, successive
methylation of uracil at the N1- and N3-positions, while


85
acetonitrile (5 mL). The mixture was stirred continuously at
0 C while 0.011 mol of the appropriate acid chloride in
acetonitrile (5 mL) was added dropwise over 5-10 minutes.
The above sequence was repeated until 0.03 mol of
triethylamine and 0.033 mol of acid chloride were added, then
the mixture was stirred for an additional 30 minutes at 0 C.
Alternate addition of base and acylating agent was found to
increase yield and decrease formation of colored side
products that were difficult to remove. The mixture was
filtered, and the residue was washed with acetonitrile
(25 mL). The combined acetonitrile solutions were evaporated
under reduced pressure, and the solid residue was
crystallized from an appropriate solvent or solvent
combination.
1.3-bis-Acetyl-5-f luorourac.il (13)
Crystallization from ether gave 1.69 g of 13 (79%): mp
112-3 C (lit.25 mp 111-3 C) ; IR (KBr) 1680, 1695, 1740,
1750, and 1795 cm'1 (C=0); NMR (CDC13) 5 2.58 (s, 3H,
3-CH3), 2.72 (s, 3H, I-CH3), and 8.23 (d, J=7 Hz, 1H, C6-fi);
UVmax (CH3CN) 262 nm (E=9.75xl03) .
1.3-bis-Propionvl-5-fluorouracil (14)
Crystallization from ether gave 1.82 g of 14 (75%): mp
100-1 C; IR (KBr) 1690, 1725, and 1795 cm'1 (CDCI3) 8 1.25 (t, J=7 Hz, 3H, 3-CH3), 1.28 (t, J=7 Hz, 3H,
I-CH3), 2.85 (q, J=7 Hz, 2H, 3-COCH2), 3.11 (q, J=7 Hz, 2H,


51
The trend in skin accumulation is similar to the trend
in flux with the exception of l-hexyloxycarbonyl-5FU (5).
The large skin accumulation value and short lag time for
compound 5 may indicate a high affinity for the lipid regions
of the skin but less affinity for the hydrated regions and
the receptor phase.
Second application fluxes and lag times are presented in
Table 2-6. Skin penetration by theophylline from propylene
glycol, the standard drug-vehicle combination, is
approximately one and one-half times higher for the skins
treated with the 1-alkyloxycarbonyl derivatives than for
those treated with 5FU, and it is consistent throughout the
series. Therefore, skin damage is greater with the prodrugs,
but the difference is small when compared with the general
improvement in delivery of 5FU from the prodrugs.
The stability of the prodrugs in the IPM formulations
was assessed by ^-H NMR analysis of the donor phases. After a
minimum of five days from the time the suspensions were
prepared until their 1H NMR spectra were recorded, including
at least twelve hours during which the formulations were in
contact with the skins, the 1-alkyloxycarbonyl derivatives
were found to be intact with no evidence of 5FU formation.
Snmma ry
The 1-alkyloxycarbonyl derivatives of 5FU exhibited
decreased melting points and increased lipid solubilities


UNIVERSITY OF FLORIDA,
II in in mu mu''
3 1262 08554 6553


37
Second, the partitioning and phase separation times that
were used in the partitioning-based method for estimating
aqueous solubility were chosen empirically. The times had to
be sufficiently long to allow equilibrium distribution of the
compounds between the phases to occur and to allow subsequent
separation of the phases to occur without substantial
hydrolysis of the prodrugs. Longer times than those chosen
could have been used for these more stable 1-alkyloxycarbonyl
derivatives, but in order to validate the procedure for all
four series, shorter times that were more appropriate for the
less stable derivatives were used.
Finally, due to the experimental design and the lability
of the prodrugs, mutual saturation of the phases prior to
partitioning could not be accomplished. Since the ester
(IPM) that was used as the lipid phase in these experiments
is practically insoluble in water,94 changes in the phase
volumes during partitioning are probably insignificant. This
potential volume change is a common source of error when more
water-soluble lipid solvents are not presaturated with their
corresponding aqueous phases.19
The conventional, direct method was used for determining
lipid solubilities. Since the 1-alkyloxycarbonyl derivatives
are the most chemically stable compounds of the four series,
both the direct and the partitioning-based methods were used
for determining their aqueous solubilities. Thus, the two
methods for determining aqueous solubility were compared
using the 1-alkyloxycarbonyl derivatives as a model series.