Retinyl palmitate isomerization

MISSING IMAGE

Material Information

Title:
Retinyl palmitate isomerization analysis and quantitation in fortified foods and model systems
Physical Description:
xi, 140 leaves : ill. ; 28 cm.
Language:
English
Creator:
Mulry, Mary C., 1957-
Publication Date:

Subjects

Subjects / Keywords:
Vitamin A   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1983.
Bibliography:
Includes bibliographical references (leaves 123-139).
Statement of Responsibility:
by Mary C. Mulry.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000419155
notis - ACG6858
oclc - 10956820
System ID:
AA00003842:00001

Full Text










RETINYL PALMITATE ISOMERIZATION: ANALYSIS
AND QUANTITATION IN FORTIFIED
FOODS AND MODEL SYSTEMS











By

MARY C. MULRY


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


1983



























To my mother and father, Blanche and Douglas Mulry, for their love
and support and for encouraging me to continue to grow.













ACKNOWLEDGEMENTS


The author expresses her deepest thanks to Dr. James R. Kirk, her

major advisor, for his patience, support and invaluable guidance

throughout the course of this study, and during the writing of this

dissertation. Special thanks are expressed to Dr. Ronald H. Schmidt

whose encouraging words, frank discussion and friendship were much

appreciated.

The author wishes to acknowledge the advice and support of

her other committee members Dr. Esam Ahmed, Dr. John Cornell,

Dr. John Dorsey and Dr. Howard Appledorf (deceased).

Special appreciation goes to Cindy Swartz for typing this

manuscript, and to Walter Jones for the graphics work. In addition, the

author wishes to thank the faculty of the Food Science and Human

Nutrition Department for their technical help and encouragement,

especially Dr. Rachel Shireman and Dr. Peter Adams. The assistance of

the staff and secretaries is also acknowledged.

Finally, special thanks and appreciation are extended to

Carla Stubbs, for her continual friendship, love and support. The

comraderie of the many graduate students and special friends who made

graduate school enjoyable and memorable is greatly appreciated and will

not be forgotten.














TABLE OF CONTENTS

PAGE

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

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

LIST OF FIGURES................................................. vii

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

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

LITERATURE REVIEW.................. .............................. 3

Vitamin A -- History........................................ 3
Vitamin A -- Dietary Sources............................... 4
Vitamin A -- Units and Activity............................. 5
Vitamin A -- Biological Functions........................... 5
Vitamin A -- Requirements and Recommended Allowances........ 6
Vitamin A -- Basic Chemistry................................ 7
Methods of Determination............. .. ................ 12
Vitamin A Degradation Reactions............................ 32
Stability of Vitamin A in Processed Foods and
Pharmaceuticals ................................ ........ 38
Kinetics of Vitamin A Degradation......................... 40

EXPERIMENTAL PROCEDURES......................... ............ 45

Preparation of Standard Vitamin A Compounds................ 45
HPLC Apparatus .................... ....................... 47
Semi-Preparative HPLC Conditions ......................... 48
Analytical HPLC Separation of Retinyl Palmitate Isomers..... 48
Analytical HPLC Separation of Retinol Isomers............... 48
The Effect of Various Extraction Solvents on the
Isomerization of Retinyl Palmitate....................... 49
Extraction Procedures for Vitamin A Concentrates and
Food Products. ...... ... .............................. 50
Quantitation of Vitamin A Isomers......................... 53
Composition of the Model Systems........................... 55
Thermal Processing of the Model Systems..................... 55
Oxygen Determination in the Headspace of Reaction Vials..... 55
Conjugated Diene Determination............................ 56
Retinyl Palmitate Isomer Determination..................... 56
HPLC Column Reactivation Procedures......................... 57
Data Analysis................................. ........... 57









PAGE


RESULTS....................................................... 58

Semi-preparative HPLC Separation of Retinyl Esters.......... 58
Analytical HPLC Separation of Retinyl Palmitate
Isomers.................................................. 58
The Effect of Various Extraction Solvents on the
Isomerization of Retinyl Palmitate....................... 62
Determination of Vitamin A Isomers in Food Products
and Vitamin Concentrates by HPLC......................... 64
Retinyl Palmitate Isomerization During Heating of the
Coconut Oil Model System................................. 77
Retinyl Palmitate Isomerization During Heating of the
Coconut Oil Model System Containing Methyl Linoleate..... 91

DISCUSSION....................................................... 98

Semi-preparative HPLC Separation of Retinyl Esters.......... 98
The Effect of Various Extraction Solvents on the
Isomerization of Retinyl Palmitate....................... 99
Determination of Vitamin A Isomers in Food Products
and Vitamin Concentrates by HPLC......................... 102
Retinyl Palmitate Isomerization During Heating of the
Coconut Oil Model System.............................. 109
Retinyl Palmitate Isomerization During Heating of the
Coconut Oil Model System Containing Methyl Linoleate..... 114

SUMMARY AND CONCLUSIONS...................... ................. 118

FUTURE AREAS OF RESEARCH........................... ...... .. 122

REFERENCES.............................................. ....... 123

BIOGRAPHICAL SKETCH.............................................. 140













LIST OF TABLES


TABLE PAGE

I. Effect of Various Chemical Changes on the Biological
Activity of Vitamin A...................................... 15

II. Biopotencies of Retinyl Acetate Isomers.................... 16

III. Absorption Properties of Retinol Isomers in Hexane......... 19

IV. Integrated Kinetic Equations for Simple Reactions.......... 42

V. Determination of Retinyl Palmitate Isomers in
Vitamin A Concentrates by HPLC............................. 68

VI. Determination of Retinyl Palmitate Isomers in Food
Products by HPLC............... ............................. 69

VII. Comparison of Two Methods for the Determination of
Vitamin A Isomers in Fortified Breakfast Cereals
by HPLC .......................... ....... ..... ............ 76

VIII. Antioxidant Levels in Retinyl Palmitate Concentrates
(1.8 x 10 IU/g).............................. ....... 79

IX. Concentration of Retinyl Palmitate Isomers and
Prediction Equations for the Coconut Oil Model
System Fortified with Retinyl Palmitate Type P1.8
(15 Pg/mL) Heated at 1210C Under low Oxygen Tension........ 83

X. Concentration of Retinyl Palmitate Isomers and
Prediction Equations for the Coconut Oil Model
System Fortified with Retinyl Palmitate Type P1.8
(30 ug/mL) Heated at 1210C Under low Oxygen Tension........ 84

XI. Concentration of Retinyl Palmitate Isomers and
Prediction Equations for the Coconut Oil Model System
Fortified with Retinyl Palmitate Type P1.8/BH
(15 ug/mL) Heated at 1210C Under low Oxygen Tension........ 85

XII. Concentration of Retinyl Palmitate Isomers and
Prediction Equations for the Coconut Oil Model System
Fortified with Retinyl Palmitate Type P1.8/BH
(30 ug/mL) Heated at 1210C Under low Oxygen Tension........ 86








XIII. Slopes of the Linear Prediction Equations for the
Degradation and Isomerization of Retinyl Palmitate-
Fortified Coconut Oil Heated at 1210C Under low
Oxygen Tension for 4 Hours............................... 90

XIV. Headspace Oxygen Concentrations and Conjugated Diene
Levels in Reaction Vials Containing a Coconut Oil-
Methyl Linoleate Model System Fortified with Retinyl
Palmitate Heated at 1210C Under low Oxygen Tension
for 4 Hours................................... ... ....... 92


XV. Concentration of Retinyl Palmitate Isomers in the
Coconut Oil Model System Containing Methyl Linoleate
Fortified with Retinyl Palmitate Type P1.8 and
Type P1.8/BH (15 ug/mL) Heated at 1210C Under low
Oxygen Tension for 4 Hours................................. 94

XVI. Concentrations of Retinyl Palmitate Isomers in the
Coconut Oil Model System Containing Methyl Linoleate
Fortified with Retinyl Palmitate Type P1.8 and
Type P.18/BH (30 pg/mL) Heated at 1210C Under low
Oxygen Tension for 4 Hours................................. 95














LIST OF FIGURES


FIGURE PAGE

1 Nomenclature of vitamin A compounds...................... 8

2 Common carotenoid compounds exhibiting
vitamin A activity................. .................. 9

3 Structural formulas of nonhindered geometric
isomers of retinol........................... ......... 11

4 Vitamin A degradation products............................ 36

5 Preparation of pure vitamin A reference compounds......... 46

6 Semi-preparative HPLC chromatograms of crude
13-cis retinyl palmitate....................... 59

7 Semi-preparative HPLC chromatograms of crude all-trans
retinyl oleate......................................... 60

8 HPLC chromatogram of a test mixture of 9-cis, 13-cis,
and all-trans retinyl palmitate isomers................ 61

9 HPLC chromatograms of retinyl palmitate in CHoC12(A),
hexane (B) and CHC13(C) subjected to gold fluorescent
laboratory light for 2 hr., 4 hr. and 6.5 hr.......... 63

10 HPLC chromatograms of retinyl palmitate in hexane,
CHC13 or CH2C12 kept in the absence of light (A)
or in the presence of gold fluorescent light (B)
for 3.5 hr............................................. 66

11 Analytical HPLC chromatograms of hexane extracts
from ready-to-eat breakfast cereal (A) and UHT-
processed milk (B)..................................... 71

12 Analytical HPLC chromatograms showing the use of
retinyl oleate in a test mixture (A) and in
hexane extracts from margarine (B)..................... 72

13 Analytical HPLC chromatograms of a test mixture
of retinol isomers (A) and of an extract from
breakfast cereal (B).................................. 75


viii








PAGE


14 Reverse phase HPLC chromatograms of a test
mixture of BHA and BHT (A) and of an
acetonitrile extract of retinyl palmitate
Type P1.8/BH (B)............ ....................... 78

15 Gas chromatograms of a sample of ambient air
(A) and a headspace sample from a reaction
vial containing retinyl palmitate-fortified
coconut oil (B) ...................................... 80

16 Degradation and isomerization of retinyl palmitate
Type P1.8 in the coconut oil model system
heated at 1210C under low oxygen tension............... 82

17 HPLC chromatograms of hexane extracts from the
coconut oil model system fortified with
retinyl palmitate Type P1.8/BH (15 ug/mL)
heated at 1210C for 0 hr., 1 hr. and 4 hr............. 88

18 HPLC chromatogram of hexane extracts of the
coconut oil model system containing methyl
lineate fortified with retinyl palmitate
Type P1.8/BH (approximately 30 pg/mL) heated
at 1210C for 0 hr., 1 hr. and 4 hr................... 97








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



RETINYL PALMITATE ISOMERIZATION: ANALYSIS
AND QUANTITATION IN FORTIFIED
FOODS AND MODEL SYSTEMS


By

Mary C. Mulry

August 1983


Chairman: James R. Kirk
Cochairman: Ronald H. Schmidt
Major Department: Food Science and Human Nutrition

Vitamin A, a fat-soluble vitamin, is required in the diet by all

animals, including humans and is often added to foods as retinyl

palmitate. All-trans retinyl palmitate has been shown to undergo

physical transformation to form cis-isomers, which possess lower

biological activity. The presence of these isomers in fortified foods

and their formation during thermal processing have not been well

characterized. This study was designed to improve the analytical

technique for the determination of retinyl palmitate isomers in

fortified foods and to determine the effects of antioxidants,

fortification level (15 ug/mL and 30 ug/mL) and lipid composition,

including the presence or absence of an unsaturated fat, on the rates of

isomerization of retinyl palmitate in a coconut oil model system heated

at 1210C.








The 9-cis, 13-cis, and all-trans retinyl palmitate isomers,

synthesized and purified by semi-preparative high performance liquid

chromatography (HPLC), were separated on a silica analytical HPLC column

using fluorescence detection.

Retinyl palmitate isomers were determined in vitamin concentrates

and fortified foods by a direct hexane/isopropanol extraction technique

and compared to official AOAC saponification procedures, followed by

HPLC quantitation of retinol isomers. Chlorinated solvents were shown

to cause isomerization of the retinyl palmitate during direct

extraction. Significant quantities (3 35%) of 13-cis isomer were

found in the vitamin concentrates and fortified foods analyzed, but only

limited amounts of the 9-cis isomer (2 5%) were found in two food

products.

Retinyl palmitate isomerization was shown to occur in the coconut

oil model system at 1210C under low oxygen (1.5 2.4%) tension.

Retinyl palmitate degraded and isomerized in a curvilinear fashion,

forming a equilibrium mixture of the three isomers after 4 hours at

1210C. The rates of isomerization and degradation were higher (p <

0.01) at the 30 Pg/mL fortification level, indicating a concentration

effect on the stability of retinyl palmitate.

The addition of an unsaturated lipid, such as methyl linoleate, to

the coconut oil model system decreased the isomerization and degradation

of retinyl palmitate. Oxidation of methyl linoleate was shown to occur

and presumably the unsaturated lipid oxidation exhibited a sparing

effect on the retinyl palmitate stability.














INTRODUCTION


Vitamin A, a fat-soluble vitamin, is required in the diet by all

animals, including humans, for growth and maintenance of tissues in the

eye, bone, epithelia and reproductive organs. Dietary surveys, such as

the Ten State Survey (Guthrie, 1976) and the First Health and Nutrition

Examination Survey (Public Health Service, 1971) have shown that certain

age-groups are consuming intakes below the Recommended Dietary Allowance

(RDA) for vitamin A. Pennington (1976) listed vitamin A as one of seven

index nutrients to assess the nutritional adequacy of diets in the

United States. Therefore, it is important to assess the destruction of

vitamin A or its biological activity which may occur in foods.

Vitamin A is often added to foods in the form of retinyl esters.

The stability of added retinyl palmitate may be affected by heat, light

or the presence of acids which may cause degradation or isomerization to

cis-isomers, resulting in lowered biological activity. The presence of

cis-isomers has not been well characterized in fortified food products,

due to limitations in analytical methodology.

Thermal processing has been shown to cause isomerization of

carotenoids to compounds with lower biological activity, but this

reaction has not been studied with respect to fortification of retinyl

palmitate in foods. The objective of this study was to develop adequate

analytical methodology for the determination of retinyl palmitate

isomers in fortified foods using high performance liquid

chromatography. The content of 9-cis, 13-cis and all-trans retinyl








palmitate was determined in several fortified food products and vitamin

A concentrates. Application of this analytical HPLC method to determine

retinyl palmitate in a coconut oil model food system was performed to

assess the effect of heating on the rate of isomerization and

degradation of retinyl palmitate. The parameters investigated included

the effect of fortification level, antioxidants and lipid composition on

the rates of isomerization and degradation of retinyl palmitate.

The results from these studies will (1) improve the existing

analytical methodology for the determination of retinyl palmitate in

fortified foods and (2) provide information regarding the effect of

thermal processing on the rates of isomerization and degradation of

retinyl palmitate in fortified foods and (3) identify specific degrada-

tion parameters which are primarily responsible for the isomerization of

retinyl esters for further study.















LITERATURE REVIEW


Vitamin A -- History

Knowledge of vitamin A had its beginnings around 1500 B.C. Night

blindness was the first disease attributed to a deficiency of

vitamin A. The prescribed cure for this affliction was a topical

application of liver or liver extracts on the eyes (Wolf, 1980). Drying

of the corneal pigment epithelium caused by severe lack of vitamin A in

the diet, termed xerophthalmia, was recognized early in the 19th century

as a dietary deficiency disease (Moore, 1957). Vitamin A deficiency

also affects growth; however, it was not until the advent of animal

experimentation that this effect was attributed to a lack of vitamin A

in the diet. Lunin (1881) found that mice were unable to survive on a

diet of pure casein, fats, sucrose and water, even when the diet was

supplemented with minerals. The growth rate of mice was normal when fed

a diet of dried milk and water; therefore, he concluded that dried milk

contained an "essential" nutrient.

This essential nutrient was termed "fat-soluble A" and its addition

to a basal diet by McCollum and Davis (1913, 1914, 1915) and Osborne and

Mendel (1913, 1915) restored the normal growth of deficient rats.

Drummond (1920) suggested that this nutrient be renamed vitamin A. In

subsequent experiments, vitamin A was then shown to maintain growth as

well as to prevent xerophthalmia and night blindness (McCollum and

Simmonds, 1917; Fredericia and Holm, 1925).









Steenbock (1919) demonstrated that there was a compound in plants

called provitaminn A" which could be converted to vitamin A in

animals. This pigment (B-carotene), which is found in animal and

vegetable fats, restored growth when added to the basal diet of the

rat. The role of 8-carotene as a vitamin A precursor was made more

clear after Karrer and coworkers (1930, 1931) determined the chemical

structure of B-carotene and retinol. Five years later, Holms and Corbet

(1937) were able to crystallize vitamin A from fish liver as the

monomethanolate derivative. Pure crystalline vitamin A and several

esters were prepared by Baxter and Robeson (1942). The synthesis of

retinol was achieved five years later (Arens and van Dorp, 1946; Isler

et al., 1947) as was the isolation and crystallization of the 13-cis

isomer of retinol (Robeson and Baxter, 1947). Shortly thereafter, 8-

carotene was synthesized by Karrer and Eugster (1950). A detailed

account of the discovery of vitamin A has been written by Moore (1957).


Vitamin A -- Dietary Sources

The term vitamin A refers to all compounds having vitamin-A-like

biological activity. Preformed retinol and retinyl esters occur only in

animal food products (mainly liver, butter, eggs and milk) or fortified

foods (cereals, skimmed milk, margarine, etc.) The provitamin A

carotenoids (a-, 0-, y-carotenes and cryptoxanthin) are found primarily

in plant food products (pigmented vegetables, green leafy vegetables and

fruits), and to a lesser extent in animal fats and organ tissues.

Vitamin A2 (3,4-dehydroretinol) has vitamin A activity and is present in

fresh water fish oils and fish livers, and to a lesser extent in the

oils and livers of marine species.









Vitamin A -- Units and Activity

Until recently, vitamin A activity in foods was expressed as

International Units (IU) with 1 IU being equivalent to 0.3 ug retinol,

0.344 ug retinyl acetate, 0.550 pg retinyl palmitate or 0.6 pg

8-carotene. These equivalencies were derived from studies using rats

and are assumed to be the same for humans. The poorer utilization of

provitamin A as compared to retinol and retinyl esters created confusion

in calculating the vitamin A value of diets. For this reason, the Food

and Nutrition Board's Committee on Dietary Allowances decided to

discontinue the use of the term IU and express the activity of vitamin A

as the equivalent weight of retinol or retinol equivalents (RE). One RE

is equal to 1 ug retinol, 6 pg B-carotene or 12 ug of other provitamin A

carotenoids.


Vitamin A -- Biological Functions

The biological functions of vitamin A can be broadly grouped into

five major areas based on the vitamin A deficiency effects in mammals.

Vitamin A deficiency adversely affects (1) vision, (2) bone growth, (3)

reproduction, (4) epithelial growth and (5) overall growth. For each of

these areas, there are well-defined deficiency symptoms in experimental

animals and man (Wolf, 1980).

The effect of vitamin A deficiency on vision is most often

demonstrated by the occurrence of night blindness, which is the

inability to see in dim light. This is caused by a decrease in the

amount of rhodopsin, the light-sensitive pigment in the eye. Vitamin A

deficiency also causes two major abnormalities in bone and tissue

growth. First is the laying down of excess bone in areas where bone

tissue is not normally present and second, the decreased absorption of









previously formed bone resulting in bone thickening (Barnicot and Datta,

1972). In the young animal, the first sign of vitamin A deficiency is

cessation of growth. In adult animals, reproductive capacity of both

the male and female is particularly sensitive to vitamin A deficiency.

Vitamin A deficiency is marked by atrophy of the testes and cessation of

spermatogenesis in the male; abnormal placental and fetal development

occur in the pregnant female. The effects of vitamin A deficiency on

epithelial growth and differentiation are manifested by an increase in

squamous keratinizing cells and a decrease in mucous secreting cells

throughout the animal body.

Vitiman A may be toxic if ingested at excessively high levels.

Vitamin A toxicity has been known since the 16th century when it was

realized that eating polar bear liver (3,900 5,400 RE/g) was toxic,

and caused drowsiness, irritability, headaches and vomiting.

Hypervitaminosis A induced in experimental animals results in bone

lesions, cartilage destruction, congenital malformations and hemorrhages

in the spleen, bladder and pectoral muscles (Wolf, 1980).


Vitamin A -- Requirements and Recommended Allowances

Many experiments have been conducted to determine the vitamin A

requirement in humans (Rodriquez and Irwin, 1972). Two major studies,

the Medical Research Council experiment (Hume and Krebs, 1979) and a

joint study by the University of Iowa and the United States Army

(Sauberlich et al., 1974) involved the largest number of subjects and

are in good agreement as to the basic requirement for vitamin A in

humans. Both of the studies indicate that 500 600 RE (500 -

600 ug) of retinol or twice as much B-carotene is needed by adults to

maintain an adequate blood concentration and to prevent deficiency









symptoms. Intakes greater than this are needed to produce liver

storage.

The recommended dietary allowance (RDA) for the adult male by the

Committee on Dietary Allowances (1980) is 1000 RE. Adult females are

able to maintain similar liver stores and adequate blood levels with

lower intakes than males; therefore the RDA for females is 800 RE. The

RDA for pregnant women is 1000 RE and 1200 RE for lactating females.

The RDA for children and adolescents is based on body size, and the

allowance for infants is based on the vitamin A content of human milk.


Vitamin A -- Basic Chemistry

The term "vitamin A" is normally used to refer to any substance or

mixture of substances possessing vitamin-A-like biological activity.

The parent compound, all-trans retinol, is an unsaturated primary

alcohol with the empirical formula C20H300 (MW = 286.46). Vitamin A

compounds (Figure 1) are numbered according to an official system

adopted by the International Union of Pure and Applied Chemistry

(1960). Other biologically active retinoid compounds include retinol,

found in the retina of the eye, retinoic acid, a polar retinol

metabolite, and the naturally occurring retinyl esters. Vitamin A2

(3,4-dehydroretinol) has the same structure as retinol, with an

additional double bond in the B-ionone ring.

The provitamins of retinol occurring naturally in plants

are a-, B- and y-carotene (Figure 2). The biological activity of the

carotenoids varies considerably, and those containing at least one

unsubstituted conjugated trimethyl cyclohexene ring (B-ionone) are the

most active. Oxygenated carotenoid pigments are less active.




















8 10 12 14


CH3
18


R-Group

- CH2OH

- CHO

- COOH

- CH20CH3

- CH20CO (CH2)14CH3


Compound Name

Retinol

Retinal

Retinoic acid

Retinyl acetate

Retinyl palmitate


Figure 1. Nomenclature of vitamin A compounds


17 16
CH, CH3


3 R
/ I15























a-carotene


B-carotene


y-carotene







Figure 2. Common carotenoid compounds exhibiting
vitamin A activity









The four double bonds present in the side-chain of retinol may give

rise to cis-trans isomerism, with sixteen isomers theoretically

possible. Pauling (1939) noted that substituents other than hydrogen

(e.g., methyl groups) in the 1,4-position of a cis-double bond result in

steric hindrance and tend to favor the isomerization of retinol to the

all-trans configuration. The 11-cis-isomer of retinol, important in

vision, is considered a "hindered" isomer due to spatial crowding

between the C-10 hydrogen and C-20 methyl group in the molecule. Four

non-hindered isomers (Figure 3), all-trans, 13-cis, 9-cis, and 9-cis,

13-cis, are believed to be formed in food products (Ames, 1966).

Other isomers of vitamin A were at one time thought to be labile

and nonexistent in synthetic mixtures (Zechmeister, 1962). With the

advent of improved analytical techniques, Denny and coworkers (1981)

have identified thirteen of the sixteen possible isomers of retinal,

including two tri-cis isomers.

Retinol (m.p. 62-630 C) is soluble in fats and in appropriate

organic solvents, especially ethanol. Retinyl palmitate (m.p. 26-28 OC)

is more soluble in nonpolar organic solvents (e.g., hexane). Both

compounds are insoluble in water, but may be dispersed in aqueous

systems by emulsification or by encapsulation in a protein matrix

(Knobloch and Cerna-Heyrovska, 1979).










17 16


7 91 II


10 12 14


al 1-trans


CH2OH


13-cis


9-cis


CH2OH


9-cis,13-cis



Figure 3. Structural Formulas of the nonhindered geometric
isomers of retinol









Retinol is highly unstable in the presence of oxygen, although

retinol can be converted to retinal with the use of highly selective

mild oxidizing agents such as manganous dioxide (Ball et al., 1948).

Retinyl esters are more stable against oxidation than retinol. The

reduction of retinol can be readily accomplished with the reducing

agents, lithium aluminum hydride or sodium borohydride. In the absence

of oxygen, retinol is stable to alkali, but is unstable in acidic

environments, resulting in dehydration and formation of rearrangement

products.

Ultraviolet light causes isomerization and degradation of retinoid

compounds in solution. An intense green fluorescence is observed with

ultraviolet light-irradiated solutions of retinol and retinyl esters.

Under more intense light, other transformations take place such as

dimerization of retinyl esters (Mousseron-Cadet, 1970). Due to the

properties described above, particular care must be taken to prevent

oxidation and/or isomerization of retinoid compounds by working in an

inert atmosphere and/or in the presence of antioxidants under low

intensity light.


Methods of Determination

Bioassay of Vitamin A Compounds

Methods. Although physiochemical methods of vitamin A analysis

have largely replaced biological assays for the quantitative determina-

tion of vitamin A in feed, food and pharmaceutical products, the

definitive assay for total vitamin A activity is based on its biological

activity (Ames, 1965). The biological activity of metabolites,

precursors, isomers and synthetic analogs of retinoid compounds may be

reliably determined only by biological assays. Biological assays are









especially useful when evaluating the effect of diet composition and the

variation within animal species with regard to vitamin utilization

(Ullrey, 1972).

Bioassays for vitamin A may be grouped into three categories based

on the physiological responses to the administration of vitamin A.

Methods are based on the reversal of deficiency symptoms in vitamin A-

depleted animals, the measurement of vitamin A tissue levels in vivo or

miscellaneous responses to vitamin A administration such as inducing

hypervitaminosis A or the in vitro opsin assay. An example of an assay

based on the reversal of deficiency symptoms is the rat growth curative

bioassay. Experimental animals are fed a vitamin A-deficient diet until

growth ceases. Graded levels of a vitamin A reference standard and the

test compound are then fed and growth responses are recorded. Growth

response is plotted versus the logarithm of the dose, and the slopes of

the lines for different test compounds are compared to determine

biopotency of the test material. Coefficients of variation found in

this assay are commonly 15-20%.

Measurement of tissue levels in vivo usually involves determination

of the concentration of vitamin A in the liver, which is the principal

storage organ. Liver storage assays require higher concentrations of

the vitamin A compound in the test diet fed than in the rat growth

bioassay, but the relative standard deviations can be as low as 5%.

Bioassays based on other responses relative to vitamin A administration

(e.g., plasma vitamin A levels) are less commonly used; however, working

details for these bioassays may be found in several excellent reviews

(Harris, 1960; Ames, 1965; Bliss and Roels, 1967).









Biopotencies of Vitamin A Compounds. All-trans retinyl acetate is

considered the parent vitamin A compound in biological assays, and other

compounds are evaluated for biopotency relative to the response elicited

by all-trans retinyl acetate. Table I shows the effect of chemical

changes on the biological activity of vitamin A (Roels, 1967). It has

been observed that most chemical changes in vitamin A compounds result

in a severe decrease in biological activity. Of special interest is the

trans-cis isomerization reaction of vitamin A compounds and its effect

on biological activity. Ames (1966) summarized the problems in

assessing vitamin A activity if cis-isomers are present. Data in Table

II summarize the biological activity of cis-isomers of retinyl

acetate. The 13-cis-retinyl acetate has a relative biopotency of 75% of

the all-trans; introduction of a cis double bond at the 9- or 11-

position decreases the potency to 24% or less.

The presence of vitamin A isomers, metabolites and degradation

products in foods and feeds has resulted in the use of bioassay proce-

dures as a comparison technique to physiochemical assay techniques.

Newton et al. (1979) used a chick liver storage assay to evaluate

molasses-based liquid feed supplements for cattle. Detection of a 10%

loss in vitamin A activity was estimated to require a minimum of 27

chicks per treatment. The biopotency of stablized vitamin A in

fortified flours held under accelerated storage conditions was evaluated

using rat growth, liver storage and serum levels as indicators of total

vitamin A bioactivity (Liu and Parrish, 1979). The stability of vitamin

A in commercial poultry feeds was evaluated by assessing liver storage

in chicks fed vitamin A standards in the experimental rations (Nir et

al., 1981). Pelleting of poultry feeds did not have an adverse effect

on the biological activity and availability of the vitamin A.












Table I. Effect of Various Chemical Changes on
the Biological Activity of Vitamin Aa


Process

Oxidation

Cis-Isomerism

Ether formation

Dehydrogenation

Loss of oxygen

Ketone formation

Demethylation

Addition of CH2

Dehydration

Condensation

Oxidation

Hydrogenation


Approximate Activity
Product all-trans retinyl acetate = 100

Aldehyde 91-100

Cis Isomers 15-75

Phenyl or methyl ethers 10-100

Vitamin A2 30

Axerophthene 10

C21-ketone 10
Norvitamin A 3

Homovitamin A 1.5

Anhydrovitamin A 0.4

Kitol 0

Epoxide 0

Dihydrovitamin A 0


aAmes (1965)













Table II. Biopotencies of Retinyl Acetate Isomersa


Isomers

all-trans

13-cis

9-cis

9-cis, 13-cis

11-cis

11-cis, 13-cis


Biopotency (RE/g)

872,000

657,000

529,000

206,000

175,000

128,000


% Relative Activities

100

75

21

24

24

15


aAmes (1966)









In recent years, interest in vitamin A compounds has been shown

related to the effectiveness of retinoid compounds in the treatment of

skin diseases, prevention of carcinogensis and wound healing (Wolf,

1980). Because large dosages of vitamin A are toxic, synthetic analogs

of vitamin A were synthesized to evaluate their biological activity with

respect to retinol and retinyl esters. Sporn et al. (1975) made an

extensive study of the biological activity of a series of retinoid

compounds using an in vitro system rather than traditional in vivo

techniques. This study was designed to account for variability in

absorption, storage, transport and toxicity of the retinoid compounds.

The in vitro system consists of tracheal organ cultures from vitamin A-

deficient hamsters which had shown signs of epithelial keratinization.

The dose required to reverse keratinization of the tracheas in 50% of

the cultures was defined as the ED50. Very small concentrations of

retinoids could be used to obtain adequate dose-response curves.

Sietsema and DeLuca (1982) have described an improved vaginal smear

assay to evaluate the biological activity of vitamin A compounds. This

assay has been highly correlated with the tracheal organ culture assay,

but is simpler and less expensive to perform. A dose-response curve was

developed using as little as 10 ug of retinoic acid. These assays have

been used to evaluate the vitamin A activity of metabolites of synthetic

analogs of vitamin A.


Physiochemical Methods of Vitamin A Assay

In spite of the improvement in biological assays, routine use of

biological methods for the evaluation of vitamin A activity in foods is

expensive and impractical. Determination of vitamin A by physiochemical

procedures is more rapid and precise. The most common analytical









procedures used for the quantitative analysis of vitamin A have been

reviewed by Kofler and Rubin (1960), Freed (1966), Roels and Mahadevan

(1967), Drujan (1971), Hubbard et al. (1970), Hashmi (1973), Parrish

(1977), and Knobloch and Cerna-Heyrovska (1979).

Spectrophotometric methods. The five conjugated double bonds in

the all-trans retinol molecule give a broad ultraviolet absorption

spectrum with an absorption maximum near 325 nm. The cis-isomers of

retinol have absorption maxima slightly shifted from the all-trans

retinol and have different molar absorptivities (Table III) (Groenendijk

et al., 1980). Retinyl acetate, retinyl palmitate and other long-chain

esters of retinol have been found to have the same molar absorptivities

as retinol (Ross, 1981).

Ultraviolet absorbance methods. Vitamin A may be determined by

direct reading of the absorbance in the 325-330 nm region. Lipids,

sterols and vitamins E and D absorb to some extent in the same general

region as vitamin A, and depending on the quantities of these compounds

present, may interfere with the quantitation of vitamin A. If

absorption of interfering compounds causes difficulties, these

substances may be removed by saponification and column chromography of

the extract. Morris and Stubbs (1946) reported that a formula could be

used to correct for extraneous absorption due to interfering

substances. The correction is based on the assumption that when

plotted, extraneous absorption is a straight line across the portion of

the absorption curve being examined. Readings are taken at 310 nm and

325 nm followed by the appropriate mathematical correction. This

correction is approximately 25% for the 13-cis isomer, and therefore

correctly estimates it, but overestimates the concentrations of other












Table III. Absorption Properties of Retinol Isomers in Hexanea


Isomer

all-trans

11-cis

9-cis

13-cis

9-cis, 13-cis


max absorption maximum
max
e = molar absorptivity

aGroenendijk et al. (1980)


x (nm)
max
325

318

322

328

324


e(M cm )

51800

34300

42200

48500

39500








vitamin A isomers. Spectrophotometric absorption is the official method

of the Association of Official Analytical Chemists (AOAC) (1980) for
\
vitamin A in margarine, and it is used for the quantitation of vitamin A

in pharmaceuticals as described by the U.S. Pharmacopeia (USP) (1970).

Vitamin A is also determined spectrophotometrically by dissolving

retinol in anhydrous benzene and subsequently dehydrating it with

p-toluenesulfonic acid to anhydroretinol. The absorbance is read in the

390-400 nm region, calculating the difference in absorption before and

after dehydration (Budowski and Bondi, 1957). Moisture interferes with

this analysis, and attempts to apply this method to foods have resulted

in somewhat low values (Parrish, 1977). Determination of individual

retinol isomers is not possible using this method.

Visible absorbance methods. The most common methods currently in

use for vitamin A determination are the colorimetric methods. The

antimony trichloride assay (Carr-Price method) which results in the

formation of a blue color is the most widely accepted method, and is the

current AOAC (1980) method for feeds, vitamin premixes and foods. This

procedure is based on the reaction of vitamin A with SbC13 in a

chloroform solution to form a colored complex with an absorbance maximum

at 620 nm. A pathway has been suggested for the reaction based on

spectroscopic studies involving retinylic and anhydroretinylic cations

(Blatz and Estrada, 1972). The disadvantages of this method include

rapid fading of the blue color, corrosiveness of the reagent,

sensitivity to traces of moisture and the production of atypical or

interfering colors due to the presence of sterols, carotenoids and other

materials. The AOAC (1980) procedure advises the addition of acetic

anhydride to overcome problems associated with the presence of









moisture. Other studies (Subramanyam and Parrish, 1976; Kamangar and

Fawri, 1978; Grys, 1980; Bayfield and Cole, 1980) have compared the

Carr-Price method with the use of other Lewis acids, tricholoracetic or

trifluoroacetic, in dichloromethane or dichloroethane. However, these

reagents are also corrosive, toxic and subject to interference by other

compounds.

The use of glycerol dichlorohydrin (1,3-dichloro-2-propanol) as a

colorimetric reagent has been advocated as a replacement for SbCl3.

Among its cited advantages are color stability, nonsensitivity to

moisture and noncorrosiveness of the reagent. Some problems have been

noted with the use of glycerol dichlorohydrin, the most important being

the low molar absorptivity of the chromophore which reduces the

sensitivity of the assay. In addition, preparation of active reagent

has proven to be difficult (Blake and Moran, 1976). More care in

reagent preparation was taken by Oliver (1980) and the method was used

to determine vitamin A and carotene in serum. The determination of

vitamin A in foods has not been performed using the glycerol

dichlorohydrin method.

All of the reagents used to quantitate vitamin A by chemical

methods react equally with all of the retinol isomers and may seriously

overestimate the biological activity of a mixture of the isomers. This

factor, along with safety problems previously stated has led researchers

to search for better analytical methods.

A colorimetric method which does distinguish retinol isomers is the

reaction of vitamin A with maleic anhydride. All-trans and 9-cis

retinol react completely with maleic anhydride in 16 hours to form a

Diels-Alder complex which fails to react subsequently with SbCl3 to









produce a blue complex. The maleic value (MV), or the percent of SbCl3-

reacting retinol remaining after treatment with maleic anhydride, is 0

and 100 for all-trans and 9-cis vitamin A and for the 13-cis and 9-cis,

13-cis-isomers, respectively. Through the application of a cubic

regression equation, an estimation of the biological activity is

obtained as described by Ames and Lehman (1960). The original method

for biopotency as determined by the maleic anhydride procedure required

a minimum vitamin A content of 1250 IU (375 RE) and a sixteen hour

incubation time, which is not applicable to foods with a low vitamin A

content. A modification by Parrish (1974 a,b) requiring a lower vitamin

A content and a one hour incubation time yielded similar values to the

more lengthy assay, but it did not give good results when evaluated in a

collaborative study.

Fluorometric methods. The strong fluorescent property of vitamin A

was first utilized by Sobotka et al. (1943) for the determination of

vitamin A in fish liver oils. The fluorescent properties of vitamin A

compounds have been reviewed by Kahan (1971). The corrected excitation

spectra of retinyl acetate and retinol correspond to their absorption

spectra with a maximum at 325-328 nm. The corrected emission spectra

consist of a single band with a maximum between 475 and 510 nm. The

absorption and emission spectra of the all-trans, 9-cis, 13-cis and

11-cis retinols were studied by Thomson (1969), but no useful

differences could be discerned. The fluorescence intensity of all-trans

retinol and all-trans retinyl acetate in different solvents varies

widely, with slight shifts in excitation and emission maxima. The

fluorescence intensity of retinol and retinyl acetate is linear with

respect to concentration in cyclohexane in the range 0.003-6pM;








considerable quenching occurs at concentrations exceeding 30pM (Kahan,

1970). The quantum efficiencies of retinol and retinyl acetate are

reported to be the same when dissolved in the same solvent (Kahan,

1967).

Fluorometric procedures for the determination of vitamin A were

developed early, but interference by carotenes, vitamin D and

phytofluene hindered their acceptance (Roels and Manadeven, 1967).

Better instrumentation, with the option to select specific excitation

and emission wavelengths, has increased the specificity of fluorometric

methods. The sensitivity of fluorometric methods is markedly better

than traditional colorimetric methods, with reported sensivities

of O.Olg to lug/ml (Parrish, 1977).

Vitamin A was determined fluorometrically by Thompson et al.,

(1972) Senyk et al., (1975), and later automated by Thompson and Madere

(1978) for analysis of milk products. Erdman et al. (1973) reported a

fluorometric assay for vitamin A in foods which employed a correction

for phytofluene, but removal of carotenes by column chromatography was

required. An improved fluorometric assay for vitamin A in rat serum by

fluorescence has recently been reported (Hansen and Warwick, 1978; Wu et

al., 1981).

Interferences associated with fluorometric methods include reagent

impurities, fluorescing lipids, chemical residues on glassware and

contamination from rubber stoppers (Davidson, 1979). Fluorometric

methods are also unable to distinguish isomeric forms of vitamin A.

Other methods. Other methods for determining vitamin A and

identifying isomeric forms include electrochemical, infrared

spectrophotometry, nuclear magnetic resonance, X-ray diffraction and









mass spectrometry. Most of these methods have not been applied to

foods, feeds and pharmaceuticals, but they are important in determining

molecular structure and identifying isomeric forms of vitamin A

compounds.

Kuta (1964) investigated the polarographic behavior of fat-soluble

vitamins, demonstrating that a minimum of three double bonds were needed

to produce a reduction wave. The 13-cis isomer of retinol was shown to

be reduced somewhat more readily and could be distinguished from the

all-trans isomer using this method. An electrochemical method for

determining vitamin A in margarine and pharmaceuticals has been

published (Atuma et al., 1975); however, neither of these methods has

gained wide acceptance as a vitamin A assay.

The infrared absorption spectra of four isomers of retinol were

determined by Robeson et al. (1955) and no quantitative differences were

found; however, significant differences were found in the infrared

spectra of isomeric aldehydes of vitamin A. Brown et al. (1959)

analyzed fish liver oils for four unhindered isomers using infrared

analysis by first oxidizing the retinol to retinal. The presence of the

9-cis, the 13-cis and the 9-cis,13-cis isomers was confirmed by this

technique.

Nuclear magnetic resonance (NMR) has been applied to the analysis

of vitamin A compounds. Proton nuclear magnetic resonance spectra have

been determined for the isomeric retinols (Kofler and Rubin, 1960) and

isomeric retinals (Patel, 1969). Tsukida et al. (1972) have determined

the isomeric composition of vitamin A in a mixture using a proton shift

reagent for analysis by NMR. Recently, 13 C-NMR spectra have been

generated for several cis-trans isomers of vitamin A compounds (Englert,

1975; Hanafusa et al., 1980).









X-ray powder diagrams of retinol isomers have been applied to

crystalline preparations and show distinct differences in the lattice

structure between isomers. This is a useful technique for

identification of isomers only.

The application of mass spectrometry to vitamin A and synthetic

analogues has been achieved (Lin et al., 1970; Elliot and Waller, 1972;

Reid et al., 1973). Mass spectra can differentiate easily between

different retinoid compounds, but the fragmentation patterns for vitamin

A isomers are very similar.


Chromatographic Methods

Chromatographic methods have proven very useful in the analysis of

vitamin A in foods because interfering components in the complex sample

extracts are separated from the various forms of vitamin A prior to

quantitation by UV-visible, colorimetric, fluorometric or other

techniques. Chromatography has become a component of almost all vitamin

A assays of food products.

Column chromatography. Conventional column chromatography with

alumina or silica as adsorbents is often used for cleanup procedures for

vitamin A extracts (AOAC, 1980). Barnholdt (1956) and separated cis-

isomers as a group from all-trans vitamin A using chromatography on

coarse alumina. A qualitative separation of several geometric isomers

of retinol (Zile and DeLuca, 1968) was achieved using a silicic acid

column with a gradient elution technique. Lipids and other fat-soluble

vitamins were also separated from vitamin A using this method. Strong

(1976) used a reverse phase partition chromatography procedure to

separate vitamins A, D and E in a multi-vitamin preparation. Gel

permeation chromatography has been used to separate vitamin A compounds









from lipids and other fat-soluble vitamins (Holsova and Blattna,

1976).

The use of larger conventional chromatography columns are rarely

used in routine assays because the procedures are long, difficult to

standardize and result in the oxidation and isomerization of vitamin A

while on the column. Column chromatography, utilizing small columns, is

used primarily as a cleanup step in combination with other efficient

chromatographic techniques at the present time.

Thin-layer chromatography. Thin-layer chromatography (TLC) is a

common procedure for the separation of vitamin A and related compounds

(Varma et al., 1964; John et al., 1965; Kahan, 1967). TLC has been used

to separate antioxidants and fat-soluble vitamins in pharmaceutical

products and feeds (Johnson and Vickers, 1973). Fung et al. (1978) were

able to separate several vitamin A compounds using TLC. Colorimetric

spray reagents or fluorescence techniques were used to visualize the

spots, the spots were eluted from the plate and vitamin A compounds were

quantitated by UV spectrophotometry.

Separation of isomeric forms of vitamin A using TLC is much more

difficult than the separation of vitamin A from other compounds. Von

Planta et al. (1962) obtained a qualitative, but incomplete separation

of the cis-isomers of retinol on silica gel G. Futterman and Rollins

(1973) obtained a separation of the mono-cis isomers of retinol, but

failed to separate the 11-cis and 13-cis isomers completely. Futterman

et al. (1979) used unidimensional multiple TLC to separate isomeric

retinals. The plates were developed three times, with hexane/diisopropyl

ether (90/10) drying the plate 1 minute between runs. Dobrucki (1979)

developed a derivatization procedure to obtain a separation of the









all-trans, 9-cis and 13-cis retinal isomers as their 2,4-

dinitrophenylhydrazone derivatives using benzene/chloroform/ethyl

acetate (30/3/1) as the developing solvent.

These TLC procedures have not gained wide acceptance for

quantitative determination of vitamin A compounds because losses due to

oxidation result in poor recoveries of the vitamin A compounds. These

methods are generally not as sensitive as newer analytical methods;

therefore, TLC is more widely used for qualitative estimations,

determination of homogeneity and separation from other vitamins and

lipids (Groenendijk et al., 1980).

Gas-liquid chromatography. There have been several reviews

concerning the use of gas-liquid chromatography (GLC) for the

determination of vitamin A, but the high temperatures and stationary

phases required for separation tend to degrade vitamin A (Dunagin and

Olson, 1964; Sheppard et al., 1972; Vecchi et al., 1973).

Derivatization techniques have been developed to improve GC techniques

for quantitation of vitamin A, but none of these methods have been

applied to the determination of vitamin A in foods.

High performance liquid chromatography. A relatively new and

rapidly developing technique in food analysis is high performance liquid

chromatography (HPLC). A comprehensive treatise on the subject has been

written by Snyder and Kirkland (1979). This method has shown great

promise for the determination of total vitamin A and/or fat-soluble

vitamins, as well as resolving the isomeric forms of retinoid

compounds. The use of HPLC offers several advantages over conventional

chromatographic methods of vitamin A analysis, including rapid separa-

tion, excellent resolution, quantitative recovery and elimination of








interference by artifacts and degradation products. HPLC can be

applied to the separation of vitamin A compounds with the wide range of

polarities (Roberts et al., 1978; McCormick et al., 1978; Halley and

Nelson, 1979b; Taylor and Ikawa, 1980; McCormick et al., 1980; McClean

et al., 1982; Kurokawa and DeLuca, 1982).

One of the most notable advantages of HPLC over other analytical

techniques is the ability to separate and quantitate isomeric forms of

vitamin A. Picomole quantities of vitamin A compounds can be detected

by analytical HPLC and isolation of micromole quantities have been

achieved by preparative HPLC (Paanakker and Groenendijk, 1979;

Groenendijk et al., 1980). Vecchi et al. (1973) performed a

qualitative separation of retinyl acetate isomers by means of liquid-

liquid chromatography, using B,B'-oxydiproprionitrile as the stationary

phase. This early work led to qualitative and quantitative separation

of isomeric retinals (Rotmans and Kropf, 1975; Tsukida, et al., 1977a;

Pilkiewicz et al., 1977; Maeda et al., 1978) and retinols (Bridges,

1976; Tsukida et al., 1977b). Paanakker and Groenendijk (1979)

described analytical and preparative separations of the mono-cis isomers

of retinal, retinol and retinyl palmitate using the same adsorption

column and varying the mixture of hexane/dioxane (0.1 5% dioxane) as

the mobile phases. A programmed-gradient has been used (Bridges et al.,

1980) to separate isomeric retinals, retinols and retinyl esters in a

single HPLC run. HPLC is also a good analytical technique for the

separation of retinyl esters of varying chain length (deRuyter and

deLeenheer, 1979; Ross, 1981) and their isomers (Alvarez et al., 1981;

Steurle, 1981).









Early HPLC methods for the determination of vitamin A in vitamin

concentrates, foods and feeds were concerned with the development of a

single method for the determination of various forms of vitamin A,

provitamins A and other fat-soluble vitamins. Simultaneous determina-

tion of vitamin A with other fat-soluble vitamins was accomplished by

Williams et al., (1972). Separation and quantitation of fat-soluble

vitamins using HPLC has been applied to pharmaceuticals (Dolan et al,

1978; Macleod and Wiggins, 1978; Eriksson et al., 1978; Barnett and

Frick, 1979; Burns and McKay, 1980; Santoro et al., 1982), biological

materials (Bieri et al., 1979; deLeenheer et al., 1979) and

feeds or foods (Cohen and LaPointe, 1978; Soderhjelm and Andersson,

1978; Widicus and Kirk, 1979; Henderson and McClean, 1979; Barnett et

al., 1980; Elton-Bott and Stacey, 1981; Landen, 1980, 1982). In all of

these methods, interfering materials were separated from vitamin A,

specificity was increased and the potential for an increased number of

assays over conventional methods was realized.

Methods for the determination of vitamin A and B-carotene in food

products by HPLC are usually designed to assess only the total vitamin A

activity present in foods. These methods may be divided into two

groups--those which determine retinol after saponification of the

retinyl esters and those which determine the retinyl esters directly

after extraction. Both types of extraction methods may be followed by

normal or reverse-phase HPLC.

Early methods for analysis of vitamin A in foods using HPLC favored

the determination of retinol in the unsaponifiable fraction of the

sample. Tamegai et al. (1976) were able to quantitate vitamin A as

retinol in pharmaceutical preparations by reverse-phase HPLC. Total








vitamin A in the nonsaponifiable fraction of food composites was

determined (Head and Gibbs, 1977) using a gradient elution technique.

Retinol was determined in cereal products, infant formula and dehydrated

whole egg using an isocratic adsorption HPLC method (Dennison and Kirk,

1977). Subsequently, reverse phase HPLC methods for quantitation of

retinol have been applied to margarine, milk and infant formula

(Thompson and Maxwell, 1977; Iguchi et al., 1979; Bohman et al., 1982),

vitamin and mineral supplements (Ranfft and Ruckemann, 1978), pizza

(Kamel and Bueno, 1980) and cheese (Bui-Nguyen and Blanc, 1980). All of

these methods used absorbance detection at or near the absorption

maximum of 325 nm for determination of retinol.

Only two HPLC methods have been published which separate and

quantitate retinol isomers in the unsaponifiable extract of foods.

Egberg et al. (1977) analyzed a variety of food products for all-trans

and 13-cis retinol after saponification. Both a normal phase and

reverse-phase HPLC separation were described with the reverse-phase

procedure being favored because it was easier to perform. Detection of

the retinols was either by absorbance or fluorescence spectroscopy.

Comparison of this reverse-phase method with the AOAC procedure showed

good agreement when the 13-cis isomer was summed with the all-trans

isomer. The average recovery of vitamin A added during the extraction

was 94.6 6.6% with a relative standard deviation of 3.9%. More

recently, Stancher and Zonta (1982a) described a normal phase HPLC

procedure to separate five cis-trans isomers of retinol. Application of

this method to the analysis of cheese (Stancher and Zonta, 1982b)

resulted in the determination of the 13-cis and all-trans retinol in

these products. Other retinol isomers were present in the









chromatograms, but were not adequately resolved from the all-trans

isomer to allow quantitation. The retinols were detected at 340 nm and

8-carotene as monitored at 450 nm. Recoveries of retinol and

B-carotene were somewhat low, at 75.7% and 79.6% respectively.

Other investigators using HPLC for quantitation of vitamin A have

found a considerable savings in analysis time and have reduced losses of

vitamin A due to oxidation by omitting the saponification step during

sample preparation. Retinol is much less stable to oxidation than

retinyl esters; therefore, methods using direct extraction result in

higher, more precise recoveries of vitamin A as compared to methods

which involved saponification of the sample. Another advantage of this

extraction technique is the ability to separate naturally occurring

retinol from retinyl esters in biological materials (Elton-Bott and

Stacey, 1981; Stowe, 1982).

Normal phase adsorption HPLC has become the method of choice for

vitamin A analysis following direct extraction of retinyl esters, since

silica columns can withstand the relatively high triglyceride load of

the sample extract as compared to reverse-phase columns. Direct

extraction of retinyl palmitate from margarine with heptane followed by

passage through a cleanup column of Na2SO4 was used by Aitzenmuller

et al. (1979). Other direct extraction techniques have been reported

for the determination of retinyl palmitate in cereal products (Widicus

and Kirk, 1979), margarine, milk, skimmed milk (Thompson et al., 1980)

and fortified milk products (Woollard and Woollard, 1981).

Landen and Eitenmiller (1979) have taken a different approach to

the determination of retinyl palmitate in high fat products such as

margarine. Their use of a nonaqueous reverse-phase HPLC procedure








necessitated the use of a high performance gel permeation (HP-GPC)

prefractionation step to prevent buildup of triglycerides on the

reverse-phase column. Landen has applied this procedure to fortified

breakfast cereals (1980) and infant formulas (1982). Using this

reverse-phase technique, recoveries and precision are very good (> 95%

and < 3%, respectively); however, the prefractionation step of the

HP-GPC procedure requires approximately 20 minutes with an additional 25

minutes per sample for the analytical HPLC, resulting in a method that

takes two to four times longer for the HPLC analyses than other direct

extraction methods.

Quantitation of cis-isomers in vitamin A extracts has not been

determined in foods until recently. Thompson et al. (1980) reported the

separation of the 13-cis isomer from the all-trans, but the individual

peak areas were combined to determine the retinyl palmitate present. An

HPLC method to separate the 13-cis, 9-cis and all-trans retinyl

palmitate on a silica column using a mixture of hexane/methyl t-butyl

ether (MTBE) was developed and applied to food products (Mulry et al.,

1980). Van Antwerp and LePore (1982) used a silica column and a

hexane/MTBE mobile phase to resolve four isomers in liquid multivitamin

preparations; however, quantitation of vitamin A was based on the

addition of individual peak areas, even though the 9-cis isomers

accounted for 15% of the total peak area in the degraded samples.


Vitamin A Degradation Reactions

The degradation of all-trans retinol and retinyl esters to

compounds with decreased biological activity is known to occur by three

primary mechanisms: isomerization, photodimerization and oxidation.

Investigation of the extent of these reactions and elucidation of the









reaction products have often been hampered by the lack of adequate

analytical methods. A clearer understanding of these reactions has

occurred in recent years due to the development of analytical techniques

such as HPLC, 13 C-NMR and mass spectrometry.

Isomerization of vitamin A compounds occurs in the presence of

heat, light, acid and iodine (Zechmeister, 1962; Schwieter and Isler,

1967). Thermal isomerization of vitamin A compounds has been observed

primarily with the hindered isomers--11-cis retinal rearranges to the

all-trans isomer, and the 11-cis, 13-cis retinal results in the

formation of the 13-cis retinal after 3 hours at 70C. No apparent

isomerization was observed when all-trans retinyl acetate was refluxed

under nitrogen for one hour in the absence of light (Tsukida et al.,

1972). Isomerization of vitamin A compounds at temperatures above 1000C

has not been addressed in the research literature.

Isomerization of retinoid compounds is most often caused by

exposure to light with or without the addition of catalytic

concentrations of iodine. Tuskida et al. (1972) reported the

productions of 30% 13-cis retinyl acetate in a solution of all-trans

retinyl acetate following the addition of 0.1% iodine to the hexane

solution after incubation for 2 hours in the dark. Irradiation of the

same solution for 30 minutes produced a ratio of all-trans: 13-cis:

9-cis: 9-cis, 13-cis of 49: 26: 13: 12. The 13-cis isomer of retinyl

acetate subjected to the same protocol resulted in an almost identical

ratio of 46: 28: 12: 14. It appears that an equilibrium mixture is

formed when a hexane/iodine solution of a vitamin A isomer is

irradiated. Substituting a more polar solvent (e.g., ethanol,

acetonitrile) for hexane in these studies results in the appearance of








the 11-cis and 7-cis isomers (Tuskida et al., 1978); therefore, the

solution environment has an effect on the mixture of isomers produced by

irradiation. Recent studies by Mulry et al. (1983) showed that

extraction with chloroform caused isomerization of vitamin A, presumably

as a result of light-induced free radical formation.

Vitamin A compounds are extremely sensitive to the presence of

acids, which cause double bond rearrangement and dehydration followed by

cis-trans isomerization. Products of these reactions include

retrovitamin A, anhydrovitamin A and isoanhydrovitamin A (Figure 4), as

well as cis-isomers.

Kahan (1967) and Mousseron-Cadet (1971) have reviewed investiga-

tions of prolonged photodecomposition of vitamin A compounds and

reported the presence of anhydrovitamin A, epoxy vitamin A compounds

and the production of asymmetric dimers. Only preliminary research

regarding the physiochemical characteristics (Tsukida et al., 1971)

and quantitation of these compounds has appeared in the literature.

The oxidation of vitamin A compounds has received somewhat more

attention than photodecomposition. The same factors shown to affect

conventional lipid oxidation have been shown to affect vitamin A

stability, namely temperature, light, oxygen, metal ions, moisture and

antioxidants (Labuza, 1971).

The first investigation of vitamin A oxidation demonstrated that

the coupled autoxidations of retinyl acetate with methyl linoleate and

of retinyl acetate alone were very similar with regards to oxygen uptake

and spectral changes. Retinyl acetate was virtually destroyed before

10% of the methyl linoleate was oxidized, demonstrating the extreme

liability of retinyl acetate to the presence of free radicals from the








oxidation of unsaturated lipids (Holman, 1950). Further studies on the

oxidation of retinyl palmitate in paraffin (Budowski and Bondi, 1960)

demonstrated that dilute solutions of retinyl palmitate were more stable

than concentrated solutions. This so-called "dilution effect" was also

observed by Debodard et al. (1951) for the stability of vitamin A in

oils. Other findings in the same study showed that the addition of

unsaturated fat resulted in higher rates of autoxidation, with the reac-

tion rate increasing as a function of increasing unsaturation. Addition

of antioxidants to the system with or without acid oxidation synergists

(e.g., citric acid) decreased the rate of autoxidation as expected.

Other studies investigating vitmain A stability were performed by

Anmo and coworkers (Anmo et al., 1966; Anmo et al., 1968; Anmo et al.,

1972; Takashima et al., 1979) who investigated the effect of heat and

water content on vitamin A degradation reactions. Vitamin A alcohol

remained unchanged when heated at 500C or 1000C in alcoholic solutions.

The stability of retinol in 60-100% ethanol solutions at 350C, 500C, and

1000C decreased gradually as the water content increased. Degradation

products of vitamin A alcohol were anhydrovitamin A, vitamin A ethyl

ether (Figure 4), vitamin A aldehyde, retinoic acid (Figure 3) and 2

unknown products. Heating aqueous ethanolic solutions of retinyl

acetate also showed increasing liability of retinol as the water content

was increased. Separation and identification of reaction products by

HPLC, NMR, UV, IR and mass spectrometry demonstrated the production of

anhydrovitamin A, vitamin A ethyl ether and 4-ethoxy anhydrovitamin A

(Figure 4).

Parizkova and Blattna (1980) reported the presence of fourteen

oxidation products of retinyl acetate using preparative TLC. The Rf
























R-group

- CH2CH20H

- CH2CH20CH3

- CH = CH2

- CH CH3

OCH3

- CH2 CH20CH2CH3


R





Compound

retrovitamin A alcohol

retrovitamin A acetate

anhydrovitamin A

isoanhydrovitamin A



retrovitamin A ethyl ether


.CH3


5
4 r

OCH2CH3





4-ethoxy anhydrovitamin A


Figure 4. Vitamin A degradation products









values and UV absorbance characteristics were reported. Suyama et al.

(1983) identified nineteen volatile oxidation products from the thermal

oxidation of retinyl palmitate at 100-110C and 1500C using GC/MS

analysis of the steam-distilled products.

Wahlgren (1979) showed an increase in the destruction of retinyl

palmitate in liquid paraffin stored at 400C as compared to 250C. Under

anaerobic conditions retinyl palmitate showed no significant degradation

when stored in paraffin for four years at room temperature while

considerable decomposition occurred when the solution was saturated with

oxygen. Finkel'shtein et al. (1978, 1980) probed the reaction mechanism

of retinyl acetate oxidation in solid films and the kinetics of retinyl

acetate oxidation in a chlorobenzene solution at 450C. They showed that

dialkyl peroxides, carbonyl compounds and hydroxyl compounds are formed

in the oxidized amorphous films of retinyl acetate. Rate constants were

determined for the chain propagation and termination steps in the

oxidation of retinyl acetate in chlorobenzene. The rates of oxidation

inhibition by 3-t-butyl methoxyphenol, hydroquinone and 2,6-di-t-butyl-

4-methylphenol were determined.

Metal-catalyzed destruction of vitamin A has been shown to occur.

Ogata et al. (1970) demonstrated the dependence of retinol autoxidation

in paraffin on the concentration of cobalt stearate. The effect of

increasing levels of alkali and alkaline earth salts on the autoxidation

of retinol in aqueous dispersions was investigated by Fisher et al.

(1972). The catalytic efficiency of the univalent cations was lower

than divalent cations at concentrations of less than 10 mM. The

oxidation of all-trans retinyl acetate in 1% aqueous and pure acetic

acid is retarded by copper acetate; however, this antioxidant effect of








copper acetate was less for retinyl acetate than for eleostearic

acids. Cobalt acetate was shown to be pro-oxidant in these studies

(Pekkarinen, 1973).


Stability of Vitamin A in Processed Foods and Pharmaceuticals

The stability of vitamin A incorporated into food products is

difficult to predict since it may be affected by pH, moisture content,

water activity, package type, storage conditions and product composition

(DeRitter, 1976). Vitamin A is marketed in a number of forms for

fortification or pharmaceutical use (Klaui, 1974). Liquid forms are

usually dissolved in a vegetable oil or specifically prepared

emulsions. Dry products include powders, granulates, microspheres and

beadlets processed by methods involving absorption, granulation, spray

congealing, encapsulation in gelatin and chemical complexation. Both

liquid and dry products are made to be water- or oil-dispersible.

Few data have been published on the stability of vitamin A in

fortified foods. In general, these data for stability of vitamin A are

proprietary information of food and pharmaceutical manufacturing

companies. Limited data, available in the literature, are sketchy and

difficult to use for prediction of stability due to the variability of

processing conditions for various food products. Compilation of the

available vitamin A stability data for many food products may be found

in several excellent reviews (Bauernfeind and Cort, 1974; Paden et al.,

1979; Bauernfeind, 1980; Klaui and Bauernfeind, 1981).

In general, vitamin A has been found to be stable in most fortified

foods. Hartman and Dryden (1974) reported that procedures such as

pasteurization, sterilization, spray and roller drying or evaporation

caused little loss of vitamin A in milk products. However, prolonged









heating of milk, butter or butterfat at high temperatures in the

presence of air has resulted in a considerable decrease in vitamin A

activity.

Vitamin A is readily destroyed by y-irradiation, and thus is

present in lower amounts in irradiated food products than in

conventionally processed foods (Kung et al., 1953; Bhushan and Kumta,

1977; Diehl, 1979). In these studies, it was found that the losses can

be mediated by storage at low temperatures, exclusion of air or

complexation of the vitamin A with starch, sugars and albumin.

The stability of vitamin A in an extruded cornmeal product was

studied by Lee et al. (1978). Extrusion at 130C caused losses of

vitamin A from 0 to 48%. The extent of vitamin A degradation was

dependent upon initial concentration of vitamin A, retention time in the

extruder barrel and the form of the vitamin. In this extruded product,

retinyl acetate was the most stable fortification form, followed closely

by retinyl palmitate. Retinol was found to be the least stable.

Little information is available regarding the effect of retort

canning or pouch processing of fortified food on the stability of

vitamin A, while trans- to cis isomerization of carotenoids is well-

known (Sweeney and Marsh, 1971; Panalaks and Murray, 1970; Lee and

Ammerman, 1974; Ogunlesi and Lee, 1978). As previously discussed,

isomerization of retinoids can result in a severe decrease in biological

activity of carotenoids and vitamin A present.

Parrish (1977) stated that isomerization of vitamin A in food

products was not a major problem. This is based on his previous work

(Parrish and Aguilar, 1971), which showed vitamin A to be stable in

fortified foods. More recently, Liu and Parrish (1979) determined that









there was little decrease in biopotency of vitamin A in fortified flour

after storage at elevated temperatures. However, if isomerization is

not a problem, then it is difficult to explain the recent reports of the

presence of significant levels of cis-isomers in food products. Egberg

et al. (1977) reported 11.4% to 34.8% of the total vitamin A in the food

products analyzed as the 13-cis isomer. The concentration of 13-cis

retinyl palmitate in Italian cheeses was shown to be 12% to 29.5% of the

total vitamin A present (Stancher and Zonta, 1982b). Formation of the

9-cis, the 13-cis, and the 9-cis,13-cis isomers in pharmaceutical

preparations has been confirmed by several studies (Lehman et al., 1960;

Ames and Lehman, 1960; DeRitter, 1976; Ames 1966; Van Antwerp and

LePore, 1982). Recently, Thompson et al. (1980) have separated the 13-

cis isomer from the all-trans in fortified milk products.

More studies are needed to quantitate vitamin A isomers in

fortified foods, and what effects, if any, processing and storage have

on isomerization of vitamin A compounds in fortified food products.

These data would be used to allow a more accurate assessment of the

total biological activity of vitamin A in a processed food product.

Kinetics of Vitamin A Degradation

In recent years, the kinetics of nutrient losses upon heating,

dehydration and storage have been determined. Nutrient losses during

thermal processing are not only dependent upon time and temperature

relationships but also upon pH, oxidation-reduction potential, media

composition and the presence of heavy metals, moisture, and oxygen

(Lund, 1975). To optimize nutrient retention in processes in which

environmental factors and/or composition are controlled variables, it is

necessary to know the relationship between nutrient destruction rate and

these factors.








Kinetic Evaluation of Reactions

A brief review of the terminology of chemical kinetics will be

presented here. A more indepth review may be found in Moore and Pearson

(1981).

A general overall rate equation can be written as



Rate of reaction = constant [reactant(s)]n


where n is the order of the reaction. The integrated kinetic equations

for chemical reactions are listed in Table IV.

Kinetic data are generated by analyzing samples for reactant or

product concentration as a function of time. The reaction order is

determined by choosing the kinetic equation (Table IV) which best fits

the data or by use of graphic or integration methods.


Thermodynamic Activation Parameters

The most common and generally valid assumption is that the

temperature dependence of reaction rate will follow the Arrhenius

equation:



k = Ae Ea/RT


where k is the reaction rate, A is the Arrhenius pre-exponential

constant, Ea is the activation energy (kcal mol-1), R is the gas

constant (1.987 cal K-1 mol-1), and T is the temperature (K).









Table IV. Integrated Kinetic Equations for Simple Reactions


Reaction Order


Kinetic Equation


Zero


First


Pseudo-firsta


C = Co -kt


In C/Co = -kt


In C/Co = -kt

B (C)
n C = -k (8 Co)t


Second


B or C denote concentration of reactant(s) at time t

Bo or Co denote centration of reactants at time = 0

t = time of sampling

k = rate constant
apseudo first order assumes that the reaction is a second order
reaction, but the concentration of the second reactant is
excessive








Transition state theory states that Ea is the energy required to form

the activated complex from the reactants. The Ea may vary with

concentration and other composition factors or when the reaction

mechanism varies with temperature.

Other thermodynamic activation parameters which may be calculated

from kinetic data include AGI (Gibbs free energy of activation), AHt

(enthalpy of activation), and AST (entropy of activation). Eyring

(1935) described the absolute reaction rate theory as

b T ASl -AHI
k = h e T /RT


where K is the first order rate constant at temperature T, Kb is the

Boltzman constant (1.38 x 10-16 erg K-1), T is the reaction temperature

(K), h is Planck's constant (6.63 x 10-27 erg sec), R is the gas
constant (1.987 cal mol-1 K-1) and AH = Ea RT. The Gibbs free energy

of activation may be calculated from


AGt = AH- TASt




Kinetics of Vitamin A Losses

There are few reports in the literature which give a kinetic

evaluation of vitamin A degradation reactions in foods and

pharmaceuticals. Tardif (1965) determined that vitamin A degraded in a

pseudo-first order fashion in multivitamin tablets. These results were

confirmed by Slater et al. (1979), but they found errors in room

temperature shelf-life predictions from the kinetic data generated at









higher temperatures. Analysis of Slater's kinetic for vitamin A

degradation data (Wilkinson, et al., 1981) gave an activation energy

(Ea) of 118 KJ/mol (28.1 Kcal/mol). Wilkinson et al. (1981) determined

the first-order rate constants for the degradation of total vitamin A in

beef liver puree heated between 102.90C and 126.70C. The Ea calculated

from the reaction was 112 9 KJ/mol (26.9 2.1 Kcal/mol). They

further investigated the loss of vitamin A in beef liver puree as a

function of product composition (moisture, fat, protein), pH and copper

(Wilkinson et al., 1982) and found that moisture and copper were the

most significant factors affecting the rate of vitamin A destruction.

The Ea's determined for various treatments ranged from 36-122 KJ/mol.

An increase in moisture content increased Ea, whereas copper addition

decreased Ea. These data suggest that vitamin A destruction follows

pseudo-first order kinetics and are similar to those found by Ogata et

al. (1970) which showed an increased thermal destruction rate of retinol

in liquid paraffin with increasing concentrations of cobalt ion and

oxygen partial pressure.

There is a need for further study of the kinetics of vitamin A

degradation and isomerization in simple systems where composition

variables are controlled. The effects of heating on vitamin A

isomerization need to be clearly defined. These data may be used to

optimize retention of the biological activity of vitamin A present in

processed food products. The following studies of retinyl palmitate

degradation and isomerization in fortified food products and model

systems will provide useful information regarding vitamin A stability.












EXPERIMENTAL PROCEDURES

Preparation of Standard Vitamin A Compounds

The use of authentic reference compounds for vitamin A analysis of

foods using HPLC is essential for unambiguous identification and

quantitation of vitamin A compounds. Only a few vitamin A compounds and

isomers (i.e., all-trans and 13-cis retinal, all-trans retinol, all-

trans retinyl acetate and all-trans retinyl palmitate) are available

commercially. The 9-cis isomers of retinol and retinyl acetate were

provided by Hoffmann La Roche, Inc., for these studies. Commercially

available vitamin A compounds were subjected to further purification.

Methods used to prepare authentic vitamin A reference compounds are

summarized in Figure 5 (Groenendijk et al., 1980). All solvents used

are reagent-grade, unless stated otherwise. The following procedures

were carried out under gold fluorescent laboratory light (Sylvania,

F40GO) to minimize light-induced degradation and isomerization of

vitamin A compounds. Retinal was reduced to retinol by treatment with

NaBH4 in ethanol (Hubbard et al., 1971). This procedure was used to

prepare 9-cis, 13-cis and all-trans retinol from the respective isomeric

retinals. No detectable isomerization was observed during the

reduction.

Following their reductions, palmitoyl chloride was then reacted

with 9-cis retinol and 13-cis retinol in methylene chloride (CH2C12)

using triethylamine as a catalyst (Bridges et al., 1980) to form 9-cis

and 13-cis retinyl palmitate. In the same manner, all-trans retinol was

reacted with oleoyl chloride to produce all-trans retinyl oleate.

Excess reagents and undesirable reaction products were separated
















retinal

NaBH4


retinol

acyl KOH
chloride

retinyl ester

semi-preparative
HPLC


pure retinyl ester

KOH



pure retinol


Figure 5. Preparation of pure vitamin A reference compounds









from the retinyl esters by preparative TLC on poured plates (2 mm,

Silica 60, PF 254). The plates were developed with cyclohexane/toluene/

ethyl acetate (5/3/2 v/v/v). The band corresponding to the retinyl

ester was located under UV light (254 nm), scraped from the plate and

the retinyl ester was removed from the silica gel by extraction with

acetone. An alternative to the preparative TLC clean-up step was to

chromatograph the reaction mixture on a Biosil A (Bio-Rad Laboratories)

packed column (1.5 cm x 8 cm) weakened with 5% H20. The retinyl esters

were eluted with 5% methyl-t-butyl ether (MTBE) in hexane. Column

chromatography resulted in less oxidation of the retinyl esters during

purification than did TLC.

The crude retinyl esters and commercially available all-trans

retinyl palmitate were purified by semi-preparative HPLC according to

the following procedures:


HPLC Apparatus

An Altex Model 312 (Altex Scientific, Inc.) liquid chromatograph

equipped with dual Model 110A pumps and Altex Model 312 and Rheodyne

Model 7125 loop injectors were used for these studies. Eluting

components were detected with either an Altex Model 153 absorbance

detector or an Aminco FluoroMonitor (American Instrument Company;

Corning 7-51 excitation filter, 365 nm max.; Wratten 8 emission filter

500 nm, sharp cut; 70 uL flowcell; G.E. 4T4-BL lamp). The detector

signals were recorded with a Linear Instruments Model 250 single pen

recorder and/or a Fisher Recordall Series 5000 double pen recorder. The

chart speed used was 0.5 cm/min.









Semi-Preparative HPLC Conditions

The semi-preparative HPLC column used was a LiChrosorb Si-60-5 (9

mm x 250 mm) column (Chrompack U.S.A., Inc.). The mobile phase used was

0.5% MTBE in hexane (HPLC grade, Fisher Scientific) at a flowrate of

1.0 2.5 mL/min. Detection of retinyl esters was by UV absorbance

(100 1L flowcell, 254 nm filter). Fractions corresponding to the

retinyl ester isomer of interest were collected, pooled and subjected to

analytical HPLC using the following procedure to check the purity of the

retinyl esters.


Analytical HPLC Separation of Retinyl Palmitate Isomers

The analytical HPLC column used was a Supelcosil LC-Si (5 Wm, 4.6

mm x 150 mm), protected by a guard column (4.6 mm x 50 mm) tap-filled

with Pelliguard LC-Si (40 un) packing (Supelco, Inc.). The mobile phase

used to separate the retinyl palmitate isomers was 0.09% MTBE (HPLC

grade, Fisher Scientific) in hexane (HPLC grade, Fisher Scientific)

containing 0.005% triethylamine pumped at a flowrate of 0.9 2.0

mL/min. The retinyl palmitate isomers were monitored by fluorescence

(365 nm excitation, 500 nm emission). Purified vitamin A standards were

stored at -200 C in hexane under N2.


Analytical HPLC Separation of Retinol Isomers

Retinol isomers were prepared from the purified retinyl palmitate

isomers by saponification with alcoholic KOH and subsequent extraction

into hexane. Isomerization of the vitamin A compounds was negligible

during these procedures. A Zorbax (Dupont) CN column (4.6 mm x 250 mm)

was used to separate the retinol isomers and was maintained at 450C in a

water bath. Hexane/MTBE/isopropanol (95.5/0.4/0.1 v/v/v) was used as a









mobile phase and was pumped at a flowrate of 1.0 mL/min. Detection of

retinol isomers was by fluorescence (365 nm excitation, 500 nm

emission).

The Effect of Various Extraction Solvents on the Isomerization
of Retinyl Palmitate

The possible effects of light and chlorinated solvents on isomeri-

zation of retinyl palmitate were of particular interest because

preliminary data showed isomerization of retinyl palmitate when chloro-

form was used during the extraction of retinyl palmitate from polar

solvents. The effect of reaction products and/or impurities in the

solvent systems was also evaluated to determine their role, if any, in

causing the isomerization of retinyl palmitate during extraction.

Purification and Stabilization of CHC13

Reagent-grade chloroform (CHC13) (bp. 59-620 C) was redistilled.

Absolute ethanol (0.5%) or 50 ppm amylene (2-methyl-2-butene) was added

and the CHC13 was stored over molecular sieve (Type 13X).

Effect of Solvents on Isomerization

Five milliliters of a stock all-trans retinyl palmitate in hexane

were added to each of three separate 25 mL volumetric flasks and made to

volume with either hexane, stabilized CHC13 or CH2C12. The solutions

were subjected to gold fluorescent laboratory light (Sylvania, F40GO),

and 5 mL aliquots were removed at 2 hr., 4 hr. and 6.5 hr. for

qualitative analysis of the solutions for the presence of cis-isomers.

The solvent was evaporated from each aliquot under N2 at 350C and the

samples were redissolved in hexane. Twenty microliters of the hexane

extract were injected into the HPLC to separate retinyl palmitate

isomers following the previously described procedure for the analytical

HPLC separation of retinyl palmitate isomers.









Effect of Light on Isomerization

The protocol described above was followed except a duplicate set of

samples was prepared. One set of samples was subjected to gold

fluorescent laboratory light for a fixed time of 3.5 hr. and the

duplicate samples were stored in the dark for the same time period.

Hexane, CH2C12, MTBE and stabilized CHC13 were used as the diluents.


Extraction Procedures for Vitamin A Concentrates and Food Products

Three vitamin A concentrates, retinyl palmitate Type P1.8 (Hoffman

La Roche, Inc.), retinyl palmitate in corn oil (106 IU/g) and gelatin-

stabilized retinyl palmitate (ICN Nutritional Biochemicals) were

extracted and subjected to analytical HPLC. Six food products,

purchased from a local market, were analyzed for retinyl palmitate

isomer content.

Vitamin A Concentrates with Gelatin-Stabilization

Samples containing 10-20 mg retinyl palmitate were weighed into a

10 mL volumetric flask and diluted to volume with hexane containing 0.1%

butylated hydroxytoluene (BHT). An aliquot (0.1 mL) was diluted to 50

mL with 0.1% BHT in hexane, and 20 pL were injected into the HPLC.

Recovery samples were prepared by adding purified all-trans retinyl

palmitate in hexane to the weighed vitamin A concentrates.

Gelatin-stabilized Vitamin A Concentrates

Samples (250,000 500,000 IU/g) equivalent to 10-20 mg retinyl

palmitate were weighed and placed in screw-capped culture tubes. Two

milliliters of H20 were added and the solution was heated at 650C for 15

minutes to dissolve the gelatin. Two milliliters of ethanol were added,

and the tubes were mixed briefly to obtain a homogeneous distribution of

the vitamin A in solution. Hexane containing 0.1% BHT (10 mL) was added









and the solution was mixed vigorously on a vortex mixer for 1 min. An

aliquot (0.1 mL) of the hexane layer was diluted to 25 mL with hexane

containing 0.1% BHT and 20 pL was injected into the HPLC. Recovery

samples were prepared by adding all-trans retinyl palmitate (ca 10 mg)

dissolved in ethanol to the gelatin-stabilized vitamin A concentrate

before extraction into hexane.

Milk and Infant Formulas

Samples were opened and allowed to equilibrate to room temperature

and mixed thoroughly with a glass rod. The hexane/ethanol extraction

procedure of Thompson et al. (1980) was used to extract retinyl

palmitate from milk and infant formula with the following

modifications. BHT was added to the hexane at a level of 0.1% and a 1

mL aliquot of the hexane extract was passed through a Biosil A cleanup

column (0.5 cm x 4 cm) weakened with 5% H20 and equilibrated with 1 ml

of 5% MTBE in hexane. The retinyl palmitate was eluted from the cleanup

column with 5% MTBE in hexane. The hexane/MTBE mixture was removed at

350C under N2, and the sample was redissolved in hexane before injection

into the HPLC to determine retinyl palmitate. All-trans retinyl

palmitate was added in ethanol to the milk or infant formula to prepare

recovery samples.

Margarine

Samples were warmed to 370C and mixed thoroughly. A sub-sample

(5 g 0.1 mg) was weighed into a 100 mL volumetric flask. A 10 mL

aliquot of retinyl oleate in hexane was added as an internal standard to

each sample. The contents were dissolved in hexane containing 0.1% BHT

and diluted to volume. The solution was allowed to stand until the

insoluble material had settled to the bottom and a clear solution was








obtained. A 1 mL aliquot of the sample extract was passed through the

Biosil A cleanup column before injection into the HPLC, as described for

milk and infant formulas, to determine retinyl palmitate. Recovery

samples were prepared by adding a known concentration of all-trans

retinyl palmitate in hexane to the sample extract before diluting to

volume.

Breakfast Cereal

Direct extraction of retinyl palmitate. Samples were finely ground

in a blender and 1.0 g + 0.1 mg was weighed into a screw-capped culture

tube (24 mm x 150 mm). Twenty milliliters of hexane containing 0.1%

BHT/isopropanol (3/2) were added and the samples were mixed vigorously

on a vortex mixer for 30 sec. and were allowed to stand for 3-5 min.

This mixing and standing was repeated twice. Eight milliliters

deionized H20 were added; the solutions were inverted several times and

allowed to stand to permit phase separation. The hexane layer was then

quantitatively transferred to a 25 mL volumetric flask. This procedure

was repeated with 10 mL of hexane containing 0.1% BHT, and the hexane

layers were combined and diluted to volume with 0.1 % BHT in hexane. A

1 mL aliquot of this solution was passed through a Biosil A cleanup

column, as described for margarine, and injected into the HPLC to

determine retinyl palmitate using the analytical HPLC procedure. All-

trans retinyl palmitate was added in ethanol to the weighed cereal

samples before extraction into hexane to prepare recovery samples.

Saponification to determine retinol. Samples were finely ground in

a blender and 1.0 g 0.1 mg was weighed into a screw-capped culture

tube (24 mm x 150 mm). Ten milliliters of 0.1% BHT in absolute ethanol

were added and the samples were mixed briefly. One milliliters of 50%









KOH was added to the sample, the solutions were again mixed and the

tubes were heated in a water bath at 750C for 40 minutes. The samples

was removed from the water bath, 10 mL of deionized H20 were added and

they were allowed to cool to room temperature. The samples were

filtered through coarse filter paper (Kleen test milk filter discs) onto

a ClinElut column (20 mL capacity; Analytichem International). Twenty

milliliters of 0.1% BHT in hexane were used to rinse the culture tube

and filter paper and were added to the ClinElut column. The hexane

extract was allowed to pass through the column and was collected into a

50 mL volumetric flask. Two additional 15 mL aliquots of 0.1% BHT in

hexane were passed through the column and the hexane fractions were

combined. Five milliliters of deionized H20 were added to 5 mL of the

hexane extract in a screw-capped culture tube. The solution was mixed

vigorously and centrifuged briefly at 2000 rpm to induce phase

separation. Fifty microliters of the hexane solution were injected into

the HPLC to determine retinol. Recovery solutions were prepared by

adding all-trans retinyl palmitate in ethanol to the cereal before

saponification of the sample.


Quantitation of Vitamin A Isomers

Serial dilutions of all-trans vitamin A standards (all-trans

retinyl palmitate, all-trans retinol) were prepared. Aliquots were

injected into the HPLC and peak heights were determined. The slope of

the standard curve was calculated using linear regression and sample

peak heights were compared to the standards. Retinyl palmitate content

in margarine was determined by using retinyl palmitate/retinyl oleate

response ratios.









Preliminary investigation of the relative response of the

fluorescence detector under the analytical HPLC conditions to varying

amounts of injected 9-cis retinyl palmitate and 13-cis retinyl palmitate

was performed. The sensitivity (response per unit weight) ratio of the

fluorescence detector for all-trans, 13-cis and 9-cis and retinyl

palmitate was determined to be 1.00, 1.93 and 1.47, respectively.

Similar response ratios were found for the retinol isomers.


Determination of Antioxidants in Concentrated Retinyl Palmitate

BHA and BHT were determined in concentrated retinyl palmitate (P1.8

and P1.8/BH; Hoffmann La Roche, Inc.) using the reverse-phase HPLC

procedure described by Pellerin et al. (1980).

Retinyl palmitate (0.5 g 0.1 mg) was weighed into a 50 mL screw-

capped test tube. Ten milliliters of hexane were added and the solution

was mixed on a vortex mixer to form a homogeneous solution. Twenty

milliliters of acetonitrile were added and the solution was mixed

thoroughly for 30 sec. on a vortex mixer. The phases were allowed to

separate and 1 mL of the acetonitrile phase was diluted to 10 mL with

acetonitrile. Fifty microliters of this solution were sampled for

injection into the HPLC. Duplicate recovery samples were prepared by

adding 10 mL of a mixture of BHA and BHT at a known concentration in

hexane to the weighed samples. These samples were treated as described

above.

The HPLC separation was carried out using a mobile phase of

acetonitrile/water (70/30 v/v) pumped at a flowrate of 2.0 mL/min

through an Altex Spherisorb ODS (5 0n; 4.6 mm x 250 mm) column at

ambient temperature. The eluent was monitored by absorbance detection

at 280 nm.









Composition of the Model Systems

The first model system used in these studies consisted of

hydrogenated coconut oil (Hydrol 92, Durkee Co.) fortified with retinyl

palmitate (Type P1.8, Hoffman La Roche, Inc.) with and without BHA and

BHT added as antioxidants. Two fortification levels of retinyl

palmitate (approximately 15 and 30 pg/mL oil) were evaluated for the

effect of heating on the extent of isomerization of retinyl palmitate.

Methyl linoleate (>99% pure, Sigma Chemical Co.) was added at a

level of 10% to the hydrogenated coconut oil in a separate study to

evaluate the effect of oxidation of an unsaturated fat on isomerization

of retinyl palmitate as a function of heating time. Retinyl palmitate

(with and without antioxidants added) was added at two different

fortification levels as described above.


Thermal Processing of the Model Systems

One mL of retinyl palmitate-fortified model system was placed in

3.5 mL amber vials sealed with open screw caps fitted with a

teflon/silicone septum. The headspace in the vials was flushed with

oxygen-free nitrogen (0.5 TM nitrogen, Airco) at a flowrate of 120

mL/min for 5 min. The vials were placed in an oil bath maintained at

1210 0.50C. Duplicate vials were removed from the oil bath at

appropriate time intervals, cooled in an ice/water bath and stored in

the dark at -200 C until analyzed for retinyl palmitate isomer content.


Oxygen Determination in the Headspace of Reaction Vials

The vials were warmed to 30 C and 1 mL of HPLC grade hexane was

injected through the septum; then 1 mL of headspace was sampled for

subsequent analysis by gas chromatography for oxygen content. A Varian








Aerograph Series 1520 B gas chromatograph equipped with a 6 ft. x 1/4"

00 stainless steel CTR column (Alltech, Inc.) was used. The carrier gas

used was helium at a flowrate of 50 mL/min. Detection of the headspace

gases was by thermal conductivity (145 mA). The injector, column and

detector were at ambient temperature. The responses of the samples from

the reaction vials were compared to that of ambient air (20.6 % oxygen).


Conjugated Diene Determination

The model system consisting of 10% methyl linoleate in coconut oil

was analyzed for conjugated diene content. One milliliter of HPLC grade

hexane was added to each vial containing 1 mL coconut oil and 1 mL

hexane (total volume = 3 mL) and the vials were mixed on a vortex mixer

for 30 sec. One hundred microliters of the HPLC grade hexane extract

were diluted to 10 mL. The UV absorbance of this solution was measured

at 233 nm with a Beckman Model 25 spectrophotometer.


Retinyl Palmitate Isomer Determination

Retinyl palmitate isomers were determined in the heated coconut

oil-based model systems as described previously for fortified foods.

One milliliter of 60% ethanol was added to the hexane extract; the

mixture was mixed briefly on a vortex mixer and was allowed to stand to

permit phase separation. One milliliter of the hexane extract was

passed through a Biosil A column and eluted with 5% MTBE in hexane as

previously described for the quantitation of retinyl palmitate in milk

and infant formula. Aliquots (20 jL) of the hexane extracts were

analyzed by the analytical HPLC method for retinyl palmitate isomer

content as previously described.








HPLC Column Reactivation Procedures

Reactivation of the silica HPLC columns was necessary when the

column efficiency had decreased and separation of retinyl ester isomers

was inadequate for quantitation. This decrease in column efficiency was

due to the absorption of polar compounds and/or water on the surface of

the column. Column efficiencies were restored using a reactivation

procedure which reacts 2,2-dimethoxypropane (DMP) with water on the

column in the presence of an acid catalyst (Bredeweg et al., 1979).

Using this procedure, the water is converted to acetone which is then

eluted with CH2C12. The following sequence of solvents was used to

reactivate the columns: 20 mL CH2C12, 20 mL CH2Cl2/acetic acid/DMP

(90/2/2 v/v/v) followed by 20 mL CH2C12 and subsequent equilibration

with the mobile phase. Regeneration of the reverse-phase bonded HPLC

columns was accomplished by pumping 30 mL of acetonitrile through the

column to remove any nonpolar components which were strongly retained

(Snyder and Kirkland, 1979).


Data Analysis

Data comparing the saponification technique with the direct

extraction technique were analyzed using "Student's" t-test (Snedecor

and Cochran, 1980). The degradation and isomerization rates of retinyl

palmitate were calculated by least squares linear and nonlinear

regression techniques (Draper and Smith, 1981). Slopes of the

calculated linear regression lines for the different treatments were

compared using analysis of variance techniques (Draper and Smith,

1981).















RESULTS


Semi-Preparative HPLC Purification of Retinyl Esters

Preparation of authentic reference retinyl ester isomers required

the use of semi-preparative HPLC for the purification of 9-cis, 13-cis,

all-trans retinyl palmitate and all-trans retinyl oleate. Isolation of

purified 13-cis retinyl palmitate is shown in Figure 6 and the

purification of retinyl oleate is shown in Figure 7.

Several injections were made for the preparation of each reference

compound since only 1.0-1.5 mg could be collected during each HPLC

run. Approximately 7-10 separate injections were collected and pooled

for tentative identification by UV absorbance spectral comparisons.

Compounds used as reference standards for later experiments were

determined to be greater than 99% pure by HPLC and UV methods.


Analytical HPLC Separation of Retinyl Palmitate Isomers

An efficient separation of three retinyl palmitate isomers was

obtained with an analytical silica HPLC column. The separation of a

test mixture of 9-cis, 13-cis and all-trans retinyl palmitate isomers is

shown in Figure 8. The three isomers were eluted in approximately 10.5

minutes.
















0.128 AUFS


13-cis RP


0 4 8 12 16 20 24 28 32 36

Time (min)



Figure 6. Typical semi-preparative HPLC chromatogram of
crude 13-cis retinyl palmitate.


100.


r
Ct
0
0
a)




a)












0.128 AUFS


a 70--
C AT RO



"6 50--
o
E50-
0

c 40--


30--


20--

10


0-|
0 4 8 12 16 20 24 28 32
Time (min)

Figure 7. Typical semi-preparative HPLC chromatogram
of crude all-trans retinyl oleate.


100








































I I I I I
0 2 4 6 8 10 12

Time (min)


HPLC chromatogram of a test mixture of three
retinyl palmitate isomers. 9 = 9-cis, 13 = 13-cis,
AT = all-trans retinyl palmitate.


40+


C
o
Q.
CO



-o


30+


20+


10+


Figure 8.








The Effect of Various Extraction Solvents on the
Isomerization of Retinyl Palmitate

The first experiment was performed to test various solvent

purification procedures such as the addition of ethanol and/or amylene

and storage over molecular sieves to remove degradation products in the

chlorinated solvents. Standard retinyl palmitate solutions used in this

study were subjected to gold fluorescent laboratory light to simulate

exposure to light encountered during routine extraction and analysis of

vitamin A. Gold fluorescent tubes are routinely used in most

laboratories where vitamin analyses are performed since it excludes

light of less than 500 nm wavelength.

The chromatograms A and C in Figure 9 indicate that exposure of

all-trans retinyl palmitate to gold fluorescent laboratory light for as

little as two hours produced a pronounced change in the isomeric nature

of the retinyl palmitate extracted in methylene chloride or

chloroform. Isomerization was minimal when hexane was used as a solvent

(Figure 9B).

Data in Figure 9A and C also indicate there was a noticeable

increase in the 9-cis retinyl palmitate peak and a corresponding

decrease in the all-trans retinyl palmitate concentration with increased

storage time in both chloroform and methylene chloride, as compared to

hexane. Identification of the peaks in the chromatograms as isomers of

retinyl palmitate was made by matching the appropriate retention time

with those of the standard mixture of three purified isomers under

identical HPLC conditions (Figure 8). Fractions corresponding to the

individual isomers were isolated under semi-preparative HPLC conditions

and the absorption maxima of the fractions corresponded to the

appropriate isomer (Table III) as tentatively identified. Treatment of





63





100-
AT AT AT
80-

AAT
AT0-
S A AT 8 2h 4h 6.5h (B)


2h 4h 6.5h (A)
40-
40-

20- 20-








80--


e AT
J 13
II I I i Il 1i I I :







8 40-

SAT


| 20-





0 2 4 6 8 0 2 4 6 80 24 6
Time (min)







Figure 9. HPLC chromatograms of retinyl palmitate in CHhCI (A),
ha ( ,adi aT














hexane (), and CHC (C) subjected to gold uoescent
laboratory light for 2 hr. 4 hr. and 6.5 hr. 9 = 9-cis
retinyl palmitate, 13 = 13-cis retinyl palmitate, AT
all-trans retinyl palmitate.-









each isolated fraction with dilute iodine produced an equilibrium

mixture of the three isomers plus a fourth peak, possibly the 9-cis

13-cis retinyl palmitate (Ames, 1966).

The isomerization process occurred only when the retinyl palmitate

is present in the chlorinated solvents. No further detectable

isomerization was observed if chlorinated solvents were stripped under

nitrogen and the retinyl palmitate is redissolved in hexane.


Effect of Light Exposure

The results of a study to investigate isomerization of vitamin A

under gold fluorescent lights suggest that isomerization of retinyl

palmitate is due to an interaction between light exposure and the

chlorinated solvents (Figure 108). The presence of light had little

effect on the isomerization in hexane (Figure 10B) or MTBE; however,

pronounced isomerization occurred in methylene chloride and

chloroform. Vitamin A solubilized in chlorinated solvents and stored in

the dark did not show isomerization of the retinyl palmitate during

sample preparation (Figure 10A).


Determination of Vitamin A Isomers in Food Products and
Vitamin Concentrates by HPLC


Direct Extraction Method

The limit of detection (LOD), defined as three times the measured

peak to peak noise level on the baseline nearest the analyte peak, for

all-trans retinyl palmitate was determined to be 0.4 ng (0.7 picomoles)

using the analytical HPLC method. The limit of quantitation (LOQ),

defined as ten times the peak to peak noise level (ACS Committee on

Environmental Improvement, 1980), for all-trans retinyl palmitate was






























Figure 10. Representative chromatograms of retinyl palmitate in hexane,
CHC13 or CH2C12 kept in the absence of light (A) or in the
presence of goTd fluorescent laboratory light (B) for 3.5 hr.













































80+


60+


404


20+


Hexane


S 2 4 6 8


CKCI3


CH, C12


9i


0 2 4 6 0 2 4 6
Time (min)


I I i I I I I I I I I i |


r ~ r









found to be 1.2 ng (2.3 picomoles) while the LOQ for 13-cis and 9-cis

retinyl palmitate are 2.4 ng and 1.75 ng, respectively.

Vitamin A Concentrates. Quantitation of retinyl palmitate isomers

in three types of vitamin A concentrates was performed. The levels of

13-cis and all-trans retinyl palmitate found in three vitamin A

concentrates are shown in Table V. The average values for total retinyl

palmitate ranged from 127.8 mg/g in the gelatin-stabilized concentrate

to 968 mg/g in the P1.8 (1.8 x 106 IU/g) concentrate. Average

recoveries in all three concentrates were greater than 95%. The

coefficients of variation (CV) for these analyses were 8.4% in retinyl

palmitate in corn oil, 3.3% in the P1.8 concentrate and 5.2% in the

gelatin-coated retinyl palmitate.

The 13-cis retinyl palmitate isomer was shown to be present in all

of the vitamin concentrates; however, the 9-cis isomer was not detected

in any of these products. The 13-cis isomer comprised 3.8% of the total

retinyl palmitate in the P1.8 concentrate, 7.3% of the gelatin-coated

concentrate and 12.2% of the retinyl palmitate in corn oil. Retinol

equivalents (RE) in the concentrates were calculated by and correcting

for the presence of the 13-cis isomer using a factor of 75% biological

activity of the all-trans isomer. When the activity of 13-cis isomer

was included in the calculation of RE all three concentrates had greater

than 91% of their label claim for retinyl palmitate (Table V).

Fortified Food Products. Retinyl palmitate was determined in six

food products and the results are shown in Table VI. The data indicate

appreciable quantities of the 13-cis isomer were present in all food

products, and the 9-cis isomer was present in the high sugar corn cereal

and the infant formula concentrate. In all other products, the 9-cis








68





) E
S- coi rco Q-









0 0 0



(0 CD0 C









t..



>) I cn a**


0

+1 cn



0
6--


U,



I N

- COj
-. 00
dl
10r


S CN *t
LM LA




+1 +4 +I


R> mn
L -D 4


m
in -


* Jj




0
C-








u +o 00 -
a c C.) -


0U -
V 01 0 I



'U 0 Co r- o
C CO <-- W LO
0 *- Q. -- 0N
Q -
0 r 0'' ~T


0

Lc


L

I













*r
0





*4-
in>




'I
*



" 0
*r-









I-)
0 a
C =


-C ')





0 0 -
**- +





L r-
0t0
0

OC 0 .

*f0 .0 *i-
.iu E .4.

-IV C L

II It II

. Q.
U: CC*
C I V
('V 0 0







































(r) CT
clei







43



0
o-
+1:


o -
-










o C-. 0

CD









0~' '.0 -4



+-1 +1 +1

co r- Co
O.. .
*. 0











-4 O ,-
ON












OD O










m r-- =
-4 t- C

ro 0 o
T O 0


ce) in ro)

CO
00
o C:) 0





c.. (Y Oci
* *
* CM4










c0 C-i C)
C- -4 0
-4


CM
i-m
* 0 0
O Z Z


oD m
o M
0 0
* *


00 C0
0) 0
L 0)
0) L




0 1 0 a
) C -* 4- 0)
.C 0 -
3O) 4.) = =
C. EC- Ec fo
Lo) (L tC LU
4LU 0 0 0) 0 0) 0
10 oi 4-.0) 4- u C
00) 3 '- C .1

CL C 0. 0- 0
O *I C C 0


(I,








C%
L





..)







0
4-4




















CM
VI








*-
Cn















a.)
+ ce





















U
*0 0




ac .
0 Ce




3 U U






(0 0
C- 4- D- L-

4U C 4.)






~C 0





a c e

0O ) L
0 C1 10





IC 4 -) 0





0) 0 I






O X )0
=0 -. 0 .0





0 D u )


C" CO m
* *











o 0 0C


U,


I
C
0 I










I1
l;ll~i


C,, .
-4









isomer was below the LOQ and therefore was listed as ND (not detected)

in Table VI. Representative HPLC chromatograms of hexane extracts from

ready-to-eat cereal and UHT-processed milk are shown in Figure 11.

For high fat products such as margarine, there were initial

problems with variations in retention times due to the high triglyceride

concentration in the extracts, resulting in variability in the vitamin A

data for margarine. This problem was addressed by the addition of

purified retinyl oleate as an internal standard prior to extraction. A

typical HPLC chromatogram showing the separation of a test mixture of

three mono-cis retinyl palmitate isomers and retinyl oleate is shown in

Figure 12A. A HPLC chromatogram of a hexane extract from margarine

containing retinyl oleate as an internal standard is shown in

Figure 128.

Recoveries of added retinyl palmitate were high for all food

products, exceeding 96 2% in all cases. The CV of these analyses

ranged from 0.3% for ready-to-feed infant formula to 8.1% for margarine.

The level of the 13-cis isomer was found to be 3.2% and 4.3% of the

total retinyl palmitate in margarine and the corn, oat, wheat and rice

ready-to-eat breakfast cereals, which is considered to be a low value.

High sugar corn breakfast cereal contained a somewhat higher value of

12.5%, as did thermally processed dairy products. Concentrated infant

formula, ready-to-feed infant formula and UHT-processed milk contained

20.6%, 24.0% and 35.4%, respectively.

The 9-cis isomer, which is biologically active, was present in two

food products at a quantifiable level. High sugar corn breakfast cereal

contained 2.1% 9-cis isomer and concentrated infant formula had 5.0%

9-cis isomer present.













50-


40t


(A)


30+


20+


10+


(B)


- i i I~ :- I I I I I


6 80 Ib 12


8 10 12


Time (min)





Figure 11. Analytical HPLC chromatograms of hexane extracts from ready-
to-eat breakfast cereal (A) and UHT- processed milk (8).














(A)


(B)


2 4 6 8 10 12 0 2 4 6 8 10

Time (min)


Figure 12.


Analytical HPLC chromatograms showing the use of retinyl
oleate as an internal standard in a test mixture (A)
and in hexane extracts from margarine (B).


u,
0
cl
0
CL


O
a)



Q)
a)









Calculation of the total RE/g was accomplished by taking into

account the differences in biological activity among the three

isomers. The product containing the largest percentage 13-cis isomer

(35.4%) was calculated to have 9% less biological activity present by

this method than would be found if the 13-cis isomer was summed with the

all-trans form and assumed to have 100% biological activity. This is

the current AOAC method for determining vitamin A content.


Direct Extraction vs. Saponification

The extraction of retinyl palmitate from margarine and dairy

products using mixtures of hexane and ethanol previously described was

compared with the standard saponification procedures used by AOAC and

many HPLC researchers (Thompson et al., 1980). These published

procedures were used in the preceding study; therefore, testing of

extraction efficiency by comparison to a saponification techniques was

not deemed to be necessary.

The direct extraction of breakfast cereal products using

hexane/isopropanol had not previously been reported in the literature

and therefore a comparison of direct extraction with saponification was

necessary to ascertain the efficiency of the extraction procedure. This

comparison required the development of an HPLC separation of retinol

isomers to permit comparison of the isomer profile between the direct

extraction and the saponification techniques.

The chromatogram in Figure 13A shows separation of a test mixture

of 9-cis, 13-cis and all-trans retinol on a CN-bonded phase HPLC column

using a mixture of hexane/MTBE/isopropanol, with fluorescence detection

at 450C. The LOD and LOQ for this HPLC method was 0.3 ng (1 picomole)

and 3 ng (10 picomoles), all-trans retinyl palmitate, respectively.








Two breakfast cereals, one fortified to 100% USRDA (United States

Recommended Dietary Allowance) for vitamin A and one fortified to 25%

USRDA, were extracted and analyzed by the two different methods.

Determination of vitamin A isomers by direct hexane/isopropanol

extraction followed by analysis on a silica HPLC column and by

saponification with subsequent analysis on a CN-bonded phase HPLC column

(Figure 13B) were compared. The results are shown in Table VII and

expressed as RE/g to aid in statistical analysis of the results. No

differences were observed between the two methods in either breakfast

cereal for the all-trans isomer content or total vitamin A present,

accounting for differences in biological activity. Significant (p<0.05)

differences were observed in the 13-cis isomer content between the

direct extraction procedure in both breakfast cereals. Larger

quantities of 13-cis retinyl palmitate were found in the saponified

samples. These data show a significant increase in 13-cis isomer

concentration due to saponification, but the absolute increase is quite

small, 0.7% and 1.3% for the 100% USRDA-fortified breakfast cereal and

25% USRDA-fortified breakfast cereal, respectively. Calculated

recoveries of added all-trans retinyl palmitate in both methods

exceeded 95%.


Retinyl Palmitate Isomerization During Heating of
the Coconut Oil Model System


The isomerization and degradation of retinyl palmitate in a coconut

oil model system subjected to heating at 1210C in amber reaction vials

under low oxygen tension were studied. Two types of retinyl palmitate

concentrate with and without BHA and BHT and two fortification levels

(approximately 15 pg and 30 ug/mL oil) were evaluated to determine the

effect of heating at 1210C on the isomerization of retinyl palmitate.
























JCO






--,-0




0 a-

o











0 .0
L. o
U- L








0 L




o -
00'







ea





.o e
0--0
C4-





*4-J LT
1- -, C


















II -






0--0


0 0 0 0 0 0 0
wu d) 4a l Ile "-
) D 0 '<







76


C LA


CCO

M 0

+1 +1


S 0 *


CO
*r- c

E 0


4o *-

> 0




0
4





cr ;




-.-- 0
*(- 4-
E m 0:*







- LA
4J %
CLLC




u- 01 C C-
*

00 U

4 C
L L 0 cc

0 41 41- 0 41
C- 4J I 0i 4--



4 u
00 O 0


C C4 C
CL a0
0 .- L0 0 0

aa CL
0 4- X a
S*r- L N 0

L 4) I CMJ 4 c-
4- 0 U -
0 0 *r-

0 *c- Q 4-

*r- 0
LL 0 U

o aE *- r-
EO I *

o .-I E
s0 o o

- 0 0o

V V

CU C C I- Q0. 0
r- 0 < -' -

.0
I- -0 <0 C C 0)
*r- *U- )- -
L E L 1.
o r- t a au u
*ei- *4-- 4 r
) 4) 0 4- 4- c -
L L *r- *t- 0.
-0 0 0 A
r0 4- 10 4- /l t

{L. L<: U r- 4-.
iO a 0)C I +j .4 o
o* a c u t: ed e c
1/) 4 0 0 L
4) ) +3 ) U U 0)
Cd I II .- *r- -D
10 a 0a 4- 4- E
0) 4-C 4-L i- *- *-- 3
.- O YC ( 0 C C C
0Q. 10 0 4) 0 O
Sa0 < o *.- *r- It
St -- vi In
m 0 CO co U C









Concentration of Antioxidants in the Retinyl Palmitate Concentrates

The concentrations of the antioxidants BHA and BHT in the retinyl

palmitate concentrates prior to heat treatment were determined by a

reverse-phase HPLC procedure (Pellerin et al., 1980). HPLC

chromatograms characteristic for the separation and quantitation of a

standard mixture of BHA and BHT from an acetonitrile extract of retinyl

palmitate Type P1.8/BH (1.8 x 106 IU/g with BHA and BHT added) are shown

in Figure 14. BHA eluted in approximately 2.3 minutes and BHT eluted in

7.0 minutes. No interfering peaks were noted with absorbance detection

at 280 nm. Data presented in Table VIII show that retinyl palmitate

Type P1.8/BH contained 11.0 mg/g and 15.7 mg/g BHA and BHT,

respectively. BHA and BHT were not detected in retinyl palmitate

Type P1.8.


Oxygen Levels in the Headspace of the Reaction Vials

Headspace oxygen levels in the reaction vials were monitored

throughout these studies by gas chromatography. Gas chromatograms from

a 1 mL sample of ambient air and the headspace gas from a reaction vial

containing retinyl palmitate-fortified coconut oil are shown in

Figure 15. Each analysis was completed within 3 minutes.

The reaction vials were flushed with oxygen-free nitrogen prior to

the heat treatment in order to decrease the oxygen tension in the

headspace. Data from a preliminary study indicated that the headspace

oxygen level in a reaction vial decreased from 20.6% in a curvilinear

fashion to an asymptotic level of 1.9 0.3% after 2.5 minutes of

purging with oxygen-free nitrogen. Therefore, a 5 minute flushing time

was chosen to minimize variability in headspace oxygen concentration as

a function of flushing time.










(A)


BHA


BHT


(3)

BHA


BHT


6 8


0 2 4 6 8 0 2 4
Time (min)


Figure 14.


Reverse-phase HPLC chromatograms of a test mixture of
BHA and BHT (A) and an acetonitrile extract of a retinyl
palmitate concentrate Type P1.8/BH. (B). BHT = butylated
hydroxytoluene and BHA = butylated hydroxyanisole.


(,

cn
r?

O
8.
c?










Table VIII. Antioxidant Levels in Rgtinyl Palmitate
Concentrates (1.8 x 10 IU/g)

Concentrate BHA mg/g % BHT mg/g %

P1.8/BH 110 1.10 157 1.57

P1.8 ND ND ND NO



aHoffman LaRoche, Inc. (Nutley, NJ)


ND = non detected





















































Time (min)


Figure 15. Gas chromatograms of a sample of
headspace sample from a reaction
palmitate-fortified coconut oil.
denoted by x16 and x2.


ambient air (A) and a
vial containing retinyl
Attenuation values are









Headspace oxygen levels were monitored by sampling duplicate

reaction vials as a function of heating time at 1210C up to 10.5 hours.

No change in headspace oxygen concentration (p > 0.05) was observed as a

function of the heat treatment for any of the four treatment combina-

tions. The average oxygen concentrations for the four treatment

combinations ranged from 1.3 2.4% with no discernible differences

between the oxygen concentration for each of the four treatment groups.


Isomerization and Degradation of Retinyl Palmitate

The concentrations of 9-cis, 13-cis, and all-trans retinyl

palmitate were monitored in a coconut oil model system as a function of

heating time. The individual isomers were summed to obtain the total

retinyl palmitate present. The biological activity of the retinyl

palmitate was calculated based on 75% and 21% biological activity for

the 13-cis and 9-cis isomers, respectively (Ames, 1966).

Data shown in Figure 16 depict the degradation and isomerization of

retinyl palmitate Type P1.8 (30 jg/mL) in coconut oil heated at 1210

C. The increase in the concentrations of 9-cis and 13-cis retinyl

palmitate as well as the decrease in the all-trans retinyl palmitate

concentration was curvilinear with heating time. This same curvilinear

function was noted for the destruction of retinyl palmitate in both Type

PI.8 and Type P1.8/BH retinyl palmitate preparations at both

15 jg and 30 ug/mL.

Isomerization and degradation data for each of the four treatments

over time (t) were fitted using a second-degree polynomial,

E(Y) = 80 + Bt + B2t2 and the results are shown in Tables IX XII.

All of the data fit a second degree polymonial satisfactorily except for

the degradation of retinyl palmitate in the Type P1.8/BH preparation

























0 O


I 1I I I
00r ~C


0 0


< C


< a


o 0 0


0 O 0


on 0o


0 o 0


a-


< Ca-


-oo-O I l I I IlI II I -
o C) co <0u co U
ro ro 0N ~

(iujW/r) uoi4D4U93uoO a !lu|IDd lAu!la8


4-)


Ur-
0 >
U.-
C u
e.- (
c,-

(D 0
CD

eo L.
_ CD
4J 4.
*- II I


(0 ,
*J -
E.




4- C *r



-> 0 U
oEo
C+
N 00
) t) -





NO U,


Lr
.o -
*v- o -"m










0 = u|
*I- 1-


g^ E~








Table IX.


Concentration of Retinyl Palmitate Isomers and Prediction
Equations for the Coconut Oil Model System Fortified with
Retinyl Palmitate Type P1.8 (15 ug/mL) Heated at 1210C Under
low Oxygen Tension.


Retinyl Palmitate (ug/mL oil)


Time (hr.)

0
0
2
2
3
3
4
4
6.5
6.5
8.5
10.5
10.5


13

0.56
0.69
1.37
1.63
1.80
2.06
1.89
1.89
1.97
1.93
2.06
2.14
2.40


9

0.28
0.34
0.34
0.45
0.68
0.68
0.56
0.56
0.68
0.79
0.68
0.62
0.79


AT

14.42
16.24
13.59
13.09
11.19
11.11
9.95
10.78
8.85
9.20
7.38
7.13
7.21


Total

15.26
17.27
15.30
15.18
13.67
13.84
12.40
13.23
11.10
11.92
10.11
9.89
10.40


BAa

14.90
16.83
14.69
14.41
12.68
12.79
11.48
12.31
10.08
10.81
9.06
8.87
9.18


Prediction equations (and standard error of coefficient estimates):


13
c


= 0.78 +
(0.14)


0.36t 0.022t2
(0.07) (0.005)


9 = 0.30 + 0.11t 0.006t2
c (0.06) (0.03) (0.002)

AT = 15.42 1.46t + 0.064t2
c (0.39) (0.18) (0.016)


Total
c


= 16.54 1.03t
(0.41) (0.19)


+ 0.039t2
(0.016)


BA = 16.11 1.20t + 0.049t2
c (0.39) (0.18) (0.016)


R2 = 0.8586


R2 = 0.7589


R2 = 0.9623


R2 = 0.9362


R2 = 0.9498


In Total = 2.78 -
2 0.927
R = 0.927


13 = 13-cis retinyl palmitate
9 = 9-cis retinyl palmitate
AT = all-trans retinyl palmitate
aBA = biological activity, calculated by 0.75 (13) + 0.21 (9) + AT


0.048t
(0.004)









Table X. Concentration of Retinyl Palmitate Isomers and Prediction
Equations for the Coconut Oil Model System Fortified with
Retinyl Palmitate Type P1.8 (30 ug/mL) Heated at 1210C Under
low Oxygen Tension.


Retinyl Palmitate (ug/mL oil)


Time (hr.)


0
0
1
1
2
2
3
3
4
4
6.5
6.5
8.5
10.5
10.5


1.34
1.61
3.23
3.40
3.23
3.36
3.67
3.31
4.75
4.75
5.29
5.29
6.09
6.18
6.41


9

0.29
0.35
0.77
0.71
0.88
0.94
1.12
0.77
1.30
1.18
1.30
1.35
1.47
1.53
1.47


AT

33.75
34.96
29.51
30.63
26.13
25.87
23.19
21.72
21.81
21.03
20.60
19.56
19.04
17.13
17.31


Total

35.39
36.93
33.50
34.74
30.24
30.18
27.98
25.80
27.85
26.95
27.18
26.20
26.60
24.85
25.18


BAa

34.82
36.24
32.09
33.33
28.74
28.59
26.18
24.37
25.64
24.84
24.83
23.81
23.92
22.09
22.42


Prediction equations (and standard error of coefficient estimates):


1.90 +
(0.23)


0.78t 0.034t2
(0.11) (0.010)


0.42 + 0.24t 0.013t2
(0.06) (0.03) (0.003)

33.33 3.73t + 0.22t2
(0.77) (0.39) (0.04)


Total =35.65 2.72t + 0.17t2
c (0.82) (0.41) (0.04)


= 34.84 3.10t
(0.80) (0.40)


+ 0.19t2
(0.04)


R2 = 0.9407


R2 = 0.9259


R2 = 0.9493


R2 = 0.8790


R2 = 0.9147


In Total = 3.50 0.03t
c (0.006)
R2 = 0.7004


13-cis retinyl palmitate
9-cis retinyl palmitate
all-trans retinyl palmitate
biological activity, calculated by 0.75 (13) + 0.21 (9) + AT


13 =
c

9 =
c

AT =
c









Table XI. Concentration of Retinyl Palmitate Isomers and Prediction
Equations for the Coconut Oil Model System Fortified with
Retinyl Palmitate Type P1.8/BH (15 ug/mL) Heated at 1210C Under
low Oxygen Tension.


Retinyl Palmitate (pg/mL oil)


Time (hr.)


0
0
1
1
2
2
3
3
4
6.5
6.5
8.5
10.5
10.5


1.10
1.39
1.72
1.92
2.66
2.45
2.90
2.78
2.90
2.94
2.70
2.94
2.66
2.78


9

0.43
0.32
0.54
0.65
0.70
0.70
0.65
0.75
0.75
0.75
1.08
0.97
1.34
0.75


AT Total


15.64
14.38
14.38
14.22
12.64
12.95
11.53
11.37
10.11
10.27
10.03
9.56
8.14
8.53


17.17
16.09
16.63
16.78
15.99
16.11
15.08
14.91
13.77
13.96
13.81
13.47
12.14
12.06


Prediction equations (and standard error of the coefficient estimates):


= 1.42 +
(0.14)


0.51t 0.038t2
(0.07) (0.007)


9 = 0.44 + 0.11t 0.005t2
c (0.09) (0.05) (0.004)


= 15.01 1.20t
(0.33) (0.17)


+ 0.057t2
(0.016)


R2 = 0.863


R2 = 0.684 n.s.


R2 = 0.945


Total = 16.87 0.58t + 0.013t2
c (0.28) (0.15) (0.013)

BA = 16.17 0.79t + 0.027t2
c (0.28) (0.15) (.014)


R2 = 0.924 n.s.


In Total
)


R2= 0.942


= 2.82 0.03t
(0.002)


R = 0.927


13 = 13-cis retinyl palmitate
9 = 9-cis retinyl palmitate
AT = all-trans retinyl palmitate
aBA = biological activity, calculated by 0.75 (13) + 0.21 (9) + AT
n.s. = not significant at p > 0.05


BAa

16.56
15.49
15.78
15.79
14.78
14.94
13.84
13.62
12.45
12.63
12.28
11.97
10.41
10.78









Table XII.


Concentration of Retinyl Palmitate Isomers and Prediction
Equations for the Coconut Oil Model System Fortified with
Retinyl Palmitate Type P1.8/BH (30 ug/mL) Heated at 1210C Under
low Oxygen Tension.


Retinyl Palmitate (ug/mL oil)


Time (hr.)


0
0
1
1
2
2
3
3
4
4
6.0
6.0
8.5
10.5
10.5


2.09
2.37
2.92
3.61
3.32
3.34
3.48
3.48
3.62
3.60
3.90
4.37
4.75
5.11
5.20


9

0.79
0.79
1.18
1.18
1.18
1.18
1.18
1.58
1.58
1.58
1.58
1.58
1.62
1.60
1.63


AT

29.81
29.35
27.95
27.48
24.22
22.83
20.50
18.63
17.70
17.68
16.30
15.37
14.50
14.04
13.89


Total

32.69
32.50
32.06
32.27
28.73
27.35
25.16
23.69
22.90
22.86
21.78
21.32
20.87
20.75
20.72


BAa

31.54
31.29
30.39
30.44
26.96
25.58
23.35
21.57
20.75
20.71
19.56
18.98
18.40
18.21
18.13


Prediction equations (and standard erros of coefficient estimates):


13 = 2.57 + 0.33t 0.008t2
c (0.16) (0.08) (0.007)


0.87 + 0.21t 0.013t2
(0.04) (0.00) (0.00078)


= 30.06 3.80t
(0.55) (0.27)


+ 0.22t2
(0.02)


Total = 33.42 3.20t + 0.19t2
c (0.63) (0.32) (0.03)


= 32.08 -
(0.62)


3.44t + 0.20t2
(0.31) (0.03)


R2 = 0.9617


R = 0.9264


R = 0.9600


R2 = 0.9506


R2 = 0.9592


In Total = 3.419 0.45t
c (0.007)
R2 = 0.7890


13-cis retinyl palmitate
9-cis retinyl palmitate
all-trans retinyl palmitate
biological activity, calculated by 0.75 (13) + 0.21 (9) + AT


9 -=
c

AT
c


BA
c


13
9
AT
aBA









(15 ug/mL) which is shown in Table XI. The 82 coefficient was not

significantly different from zero (p > 0.05) for the formation of the

9-cis isomer or for the loss of the total retinyl palmitate.

An alternative to using the second-degree polynomial E(Y) = 80 + 0it + 2t2

to model the degradation data would be to transform the data to natural

logarithms and fit E(lnY) = 80 + Blt. These equations for the degradation of

total retinyl palmitate are shown in Tables IX XII.

In all treatments, the concentration of all-trans retinyl palmitate

decreased as a function of heating time, while the 9-cis and 13-cis

isomer increased in concentration reaching an equilibrium condition

after 4 hours of heating (Figure 16). The 9-cis and 13-cis isomers

reached a stable concentration, while the all-trans retinyl palmitate

show a slight decrease in concentration.

The loss in biological activity, calculated by accounting for the

lower biological activity of the 13-cis and 9-cis isomers, closely

parallels the loss in the total retinyl palmitate (Figure 16). The

difference between the calculated biological activity and the total

retinyl palmitate present is 1.6 3.6% at zero time. As the retinyl

palmitate-fortified coconut oil is heated, the concentrations of 9-cis

and 13-cis retinyl palmitate increase (Tables IX XII). The difference

between the calculated biological activity and the total retinyl

palmitate concentration increases with time, reaching approximately 11%

after 10.5 hours heating.

High performance liquid chromatograms of the hexane extracts from

coconut oil fortified with retinyl palmitate Type P1.8/BH (15 ug/mL)

heated for 0 hrs., 1 hr. and 4 hr. are shown in Figure 17. The increase



































































0



o o
nr


asuodsa ljap0oo al


*.- .
4-

4.-)
L. *
0
4- L



.l)
>1 0
.4-i




-C-4

o
0

= 4-)

0 0)




E c
o



0

t-
4- Ln
C-J









4-) 00




e-
OQC





0r
EE
Ca)

X >*





0)
4-3

10 .-*
t- O
0




S.. -
C- *r-





31


C:




t


-O
I-

-0


C
Q
0,

rr)









in the concentration of 13-cis and corresponding decrease in the all-

trans retinyl palmitate levels are a quadratic function of heating

time. The concentration of 9-cis also increases, but at a slower rate

than does 13-cis retinyl palmitate. The small peak between the 13-cis

and 9-cis peak is believed to be the 9-cis,13-cis peak; however, it was

not positively identified. The small peak eluting before the 13-cis

peak was shown to be an impurity in the MTBE used in the extraction.

Effect of fortification level and antioxidants. The statistical

comparison of the effects of fortification level and the presence of

antioxidants on the rates of isomerization and degradation of retinyl

palmitate is difficult when comparing curvilinear lines. However, the

data in Figure 16 indicate that the initial loss at all-trans retinyl

palmitate concentration is linear as is the increase in the 9-cis and

13-cis retinyl palmitate concentrations with heating times up to 4

hours.

The degradation and isomerization data for retinyl palmitate-

fortified coconut oil heated for 4 hours at 1210C were fitted to the

linear expression, Y = 80 + t to allow statistical analysis of the

four treatments on the rates of isomerization as a function of heating

time. All of the slopes of the calculated linear regression lines were

significantly (p < 0.05) greater than 0.

The slopes of the linear prediction equations for the effect of

heating retinyl palmitate fortified coconut oil at 1210C under low

oxygen tension are shown in Table XIII. The rates of isomerization and

degradation of retinyl palmitate Type P1.8 and Type P1.8/BH under low

oxygen tension were significantly higher (p < 0.01) at the

30 pg/mL fortification level than at the 15 ug/mL fortification level.