Citation
Integrating biotechnology and pharmaceutics

Material Information

Title:
Integrating biotechnology and pharmaceutics development of the biocompatible allograft as an orthopedic drug delivery system
Alternate title:
Development of the biocompatible allograft as an orthopedic drug delivery system
Creator:
Mills, Charles Randal
Publication Date:
Language:
English
Physical Description:
vii, 141 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
Antibiotics ( jstor )
Antimicrobials ( jstor )
Bones ( jstor )
Drug evaluation ( jstor )
Homologous transplantation ( jstor )
Infections ( jstor )
Lipids ( jstor )
Microbial load ( jstor )
Specimens ( jstor )
Tissue grafting ( jstor )
Bacterial Infections -- drug therapy ( mesh )
Biocompatible Materials ( mesh )
Bone Diseases -- drug therapy ( mesh )
Bone and Bones ( mesh )
Cefazolin -- therapeutic use ( mesh )
Department of Pharmaceutics thesis Ph.D ( mesh )
Dissertations, Academic -- College of Pharmacy -- Department of Pharmaceutics -- UF ( mesh )
Drug Delivery Systems ( mesh )
Gentamicins -- therapeutic use ( mesh )
Lipids ( mesh )
Research ( mesh )
Transplantation, Homologous ( mesh )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph.D.)--University of Florida, 1999.
Bibliography:
Bibliography: leaves 131-140.
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Charles Randal Mills.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
021237039 ( ALEPH )
51624669 ( OCLC )

Downloads

This item has the following downloads:


Full Text
INTEGRATING BIOTECHNOLOGY AND PHARMACEUTICS:
DEVELOPMENT OF THE BIOCOMPATIBLE ALLOGRAFT AS AN ORTHOPEDIC DRUG DELIVERY SYSTEM
By
CHARLES RANDAL MILLS
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 1999




To organ and tissue donors and their families, who in their time of sorrow give compassionately to relieve the suffering of others.




ACKNOWLEDGMENTS
I would like to extend my sincere gratitude to Dr. Gayle Brazeau for her
continuous support throughout my graduate education. I would also like to thank Dr. Hochhaus, Dr. Derendorf, and Dr. Duggan for their time and effort they put forth on my committee. I would like to thank my family for their support. Most importantly, I offer my sincerest appreciation to Anna Radloff for always believing in me.
I would like to thank Regeneration Technologies, Inc. for its financial support, without which this work would not have been possible. I would like to thank Michael Roberts for his efforts and insight on this and many other projects. Lastly, I would like to thank Jamie Grooms for serving as my mentor and for encouraging me to follow my vision.
Ill




TABLE OF CONTENTS
page
A CK N O W LED G M EN TS ................................................................................................ iii
A B STRA CT ....................................................................................................................... vi
1 AN INTRODUCTION TO THE ALLOGRAFT AS A DRUG
D ELIV ERY SY STEM .................................................................................................. 1
Proposed Solution..................................................................................................... 8
Barriers to D evelopm ent ............................................................................................ 15
Concentration.............................................................................................................. 23
A ccom plishing Tissue Sterilization ........................................................................... 26
Sum m ary ..................................................................................................................... 34
2 EVALUATION OF ALLOGRAFT COMPOSITION FOR FACTORS THAT
MITIGATE ANTIMICROBIAL CHEMOPROPHYLAXIS ..................................... 35
Introduction................................................................................................................. 35
M aterials and M ethods............................................................................................... 40
Results ......................................................................................................................... 50
D iscussion................................................................................................................... 53
Conclusions................................................................................................................. 58
3 EVALUATION OF CHEMOTHERAPEUTIC AGENTS FOR USE IN
A CORTICAL BONE ALLOGRAFT DRUG DELIVERY SYSTEM...................... 59
Introduction................................................................................................................. 59
D iscussion................................................................................................................... 70
Conclusions................................................................................................................. 74
4 OPTIMIZATION OF GENTAMICIN LOADING INTO CORTICAL BONE........... 76
Introduction................................................................................................................. 76
M aterials and M ethods............................................................................................... 86
Results ......................................................................................................................... 93
D iscussion................................................................................................................... 96
Conclusions............................................................................................................... 100
iv




5 IN VITRO PHARMACOKINETIC MODELING OF GENTAMICIN
RELEASE FROM CORTICAL BONE ALLOGRAFTS.......................................... 101
Introduction............................................................................................................... 101
M aterials and M ethods............................................................................................. 106
Results....................................................................................................................... 109
Discussion................................................................................................................. 117
Conclusions............................................................................................................... 122
6 CONCLUSIONS AND A DISCUSSION ON AREAS OF POTENTIAL APPLICATION AND FUTURE W ORK .................................................................. 123
Conclusions............................................................................................................... 123
Future W ork.............................................................................................................. 129
REFERENCES ............................................................................................................... 131
BIOGRAPHICAL SKETCH ......................................................................................... 141
V




Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy
INTEGRATING BIOTECHNOLOGY AND PHARMACEUTICS:
DEVELOPMENT OF THE BIOCOMPATIBLE ALLOGRAFT AS AN ORTHOPEDIC DRUG DELIVERY SYSTEM
By
Charles Randal Mills
December 1999
Chairman: Gayle Brazeau
Major Department: Pharmaceutics
The motivation behind the development of an allograft based drug delivery
system (ABDDS) is to treat or prevent orthopaedic infections using a biocompatible local delivery system. Currently, the majority of local antimicrobial therapy is accomplished with the polymethylmethacrylate (PMMA) gentamicin bead. However, an additional surgical procedure is generally required to remove the device after the drug has been delivered. An ABDDS would have the major advantage over PMMA in that allografts are bioincorporable, eliminating the need for a second surgical procedure to remove the device.
The influence of allograft architecture and composition on the effectiveness of
cefazolin is evaluated. The results indicate that cefazolin effectiveness may be hampered by architectural features, however, residual lipids may positively influence cefazolin effectiveness. It is hypothesized that this is the result of endogenous lipids blocking the
vi




uptake of contaminations into the allograft, making the bacteria accessible to higher concentration of the antibiotic.
By changing the route of delivery from systemic to local, it is hypothesized that the effectiveness of antimicrobial therapy can be enhanced. Gentamicin was selected for evaluation with an allograft based drug delivery system because of its appropriate antimicrobial spectrum, low incidence of provoking hypersensitivity reactions, and potency. A novel drug loading procedure was evaluated for its ability to impregnate cortical bone segments with this drug. This set of experiments demonstrated that gentamicin could be loaded into cortical bone without altering its activity or lowering the strength of the bone.
Lastly, the release profile of gentamicin from cortical bone was evaluated. An in vitro model was used to establish elution kinetics. The data suggest that the release profile was consistent with that predicted by a bi-exponential, diffusional based model. This data, however, also demonstrats that the release profile is rapid and that modification of the device to attain a more sustained release is likely needed.
Together, this work establishes the foundation for the further development of an allograft based drug delivery system. It is hoped that this research will ultimately provide surgeons and their patients with a superior alternative to synthetic antibiotic impregnated cements for the treatment or prevention of orthopaedic infections.
vii




CHAPTER 1
AN INTRODUCTION TO THE ALLOGRAFT AS A DRUG DELIVERY SYSTEM
Osteomyelitis and surrounding soft tissue infections are a significant concern for orthopaedic surgeons. Infection can arise from an acute traumatic injury such as a puncture wound, following invasive surgery, or be secondary to a predisposing condition such as sickle cell anemia or diabetes. Of particular concern is the infection rate with some types of open fractures. Current treatment for this type of wound typically involves the use of radical debridement of the wound site, prophylactic intravenous antibiotic, and open-wound irrigation and suction (16, 34, 53, 103, 122). Even with this meticulous care, the risk of subsequent infection for certain types of fractures is unacceptability high at nearly 50% (82).
The use of orthopaedic implants derived from human donor bone (allografts) also carries a significant risk of infection. Of the 500,000 surgical procedures using allograft that are performed each year, approximately 10% result in the development of an iatrogenic infection (31, 68, 80, 100). The severity of the infection can range from subclinical, only being identified by culture, to severe, requiring graft removal and lifesaving antimicrobial chemotherapy. The majority of such infections are mild and can be resolved with prolonged antibiotic therapy, usually lasting approximately 6 weeks. It is estimated that the additional charges for medication, laboratory testing, extended hospital
1




2
stay, and follow-up examinations increase the cost of treatment by 20% (25, 42, 56, 88). Increased recovery time also leads to the potential for lost income for the patient.
In addition to the costs associated with iatrogenic allograft infection, there is increased morbidity and mortality (107). The use of extended antibiotic therapy, in sufficient concentrations to eradicate the infection, can lead to toxicity or hypersensitivity reactions (34). For example, high levels of gentamicin can lead to both renal toxicity and ototoxicity (28, 43, 66, 106, 113). Also, there is a substantial risk to the patient associated with any additional surgical procedure especially those requiring general anesthesia (44, 69, 89, 115).
Physician concern about infection often leads to the use of sub-optimal
alternatives to allografts. These consist of materials that lack the ability to incorporate, such as metallic implants (31). Long-term, this may lead to a poorer surgical outcome than would have been realized if a tissue graft was utilized. Other, more radical procedures, such as limb salvage for patients with osteosarcoma, are deferred for limb amputation because of this increased risk (41). In the case of neoplasm, the concern over infection is sometimes heightened further by adjuvant chemotherapy that may have immunosuppressive side effects.
Current Antimicrobial Treatments
The rationale of chemoprophylaxis is to attain an effective concentration of
antibiotic at the time of wound contamination. During this time the antibiotic reduces the quantity of introduced organisms to a number that can be cleared by the body's immune system without developing purulence. Four parameters need to be considered to achieve a




3
desired therapeutic end result including timing, route of administration, duration of treatment, and selection of antibiotic.
Early studies by Burke demonstrated the importance of timing in the effectiveness of chemoprophylaxis (24). He administered a single dose of penicillin at various times before and after the inoculation of penicillin-sensitive Staphylococcus aureus in simulated surgical wounds of guinea pigs. Pre- or perioperative administration of antibiotic resulted in lesions histologically identical to lesions induced by attenuated controls. A three hour postponement in the administration of antibiotic resulted in lesions similar to those in animals not receiving antibiotics, thus study establishing the critical dependence of prophylactic efficacy on the timing of drug administration.
For effective antimicrobial prophylaxis, adequate drug concentrations must be present in the tissues at the onset and throughout the operative procedure (23). The majority of surgical protocols call for the initial dose to be administered parenterally immediately prior to the operation (71). The optimal situation is to have peak concentrations (Cmax) well above the minimum inhibitory concentrations for the relevant organisms at the time of wound contamination and maintain these therapeutic concentrations throughout the procedure. For this reason, the half-life and time to Cmax for the chosen drug must be factored into the decisions of when and how much antibiotic to administer during these procedures.
A single dose of antibiotics before or during the procedure is recommended for prophylaxis in most surgical procedures (71, 102, 103, 108, 122). The logic behind a single dose regime is that contamination ends after the wound is closed. Studies comparing single-dose with multi-dose regimens have demonstrated no difference in




4
rates of postoperative infection (93). However, these studies did not specifically investigate allograft surgery and it was recognized by those authors that for some surgical procedures, the number of doses of antimicrobials required for optimal prophylaxis has not been precisely defined.
There are several considerations when selecting the appropriate antibiotic for
prophylaxis. The optimal prophylactic antibiotic should (1) be effective against relevant microorganisms; (2) attain sufficient local tissue concentrations; (3) result in minimal side effects; (4) be cost effective, and (5) not be likely to select for virulent organisms. Since Staphylococcus aureus, a Gram-positive cocci, is isolated from 35-55% of orthopaedic wound infections, it would clearly need to fall within the spectrum of an appropriately selected antibiotic (26). In addition, polymicrobial infection is also common with a reported incidence of 50% in allograft wound infections (108).
Cefazolin (Ancef6) is the most commonly used antibiotic for prophylaxis in orthopaedic surgery (71). It is a first generation cephalosporin effective against gram positive organisms including most species of staphylococci. It also has a wide range of effectiveness against gram negative organisms. Of the first generation cephalosporins, it has the longest half-life at approximately 1.8 hours. Typically, 1-2 grams is given within 30 minutes prior to the incision and a second dose is administered if the lasts longer than
3 hours (71, 122).
The two main side effects associated with cefazolin are allergic reactions and
antibiotic-associated colitis (AAC) (19). AAC is rarely a problem when the drug is given as a single intravenous prophylactic dose. High serum concentrations of cefazolin can




5
cause seizures and renal function should be considered before a multiple dose regime is initiated (16).
Limitations of Current Practice
The obvious limitation of the current strategies in preventing post-operative
complications is the persistence of an unacceptably high rate of infection. Although it is recognized that there is a chance of infection any time natural barriers to infection are compromised, a goal for all surgical procedures should be to achieve infection rates no higher than those found with the cleanest procedures. In with some types of wounds this is currently not being approached. Specifically, there is nearly a 50% rate of infection following the reduction of some open fractures (82). Allograft use is associated with a higher infection rate compared with similar synthetic prosthetics. One reason for the ineffectiveness of chemoprophylaxis on allograft tissue is the physical difference between synthetic and tissue implants. Current recommendations for prophylactic antibiotic administration in orthopedic procedures do not differentiate between those procedures that utilize allograft tissue from those that involve synthetic implants. The differences between the two types of can be hypothesized to cause prophylactic chemotherapy to be less effective with allografts thereby resulting in a higher incidence of infection.
There are several important features of allograft cortical bone that are dissimilar to metallic implants (Table 1-1). An examination of these specific features reveals that it may not be appropriate to consider allografts and synthetics equivalent for the purpose of chemoprophylaxis. An appreciation of the complexity of cortical bone is requisite to understand how these differences influence the effectiveness of systemic therapy.




6
The porosity of allograft material makes it possible for contaminants that are
absorbed into the matrix during graft reconstitution to not be immediately available to the antibiotic. Although studies have demonstrated that antibiotic concentrations reach prophylactic levels within bone tissue using the standard prophylactic protocol, this is perhaps an incorrect assumption in the graft itself (14, 16, 27, 99, 121). In fact it seems highly unlikely that this is possible considering the half-life of cefazolin and that only a single dose is typically administered. Although the recipient's bone is usually well vascularized, no vascular connections are made to the implanted bone by the surgeon. This leaves diffusion as the only mechanism for delivering antibiotic into the newly grafted material. Metallic implants, in contrast, are solid and non-porous. Therefore, metallic implants would not have the ability to harbor organisms in a protective manner from systemically administered antibiotics. Table 1-1. Compositional and architectural features that are unique to allografts and the hypothesized association with bacterial infection.
Feature Hypothesized Association with Bacterial Infection
Porosity Bacteria sequestered deep within the internal matrix of the tissue will be protected from optimal concentrations of antibiotic and as well as the recipients immune system. Bacteria sequestered within complex surface features of the graft, such as threads, will Architecture be protected from optimal concentrations of antibiotic and as well as the recipients immune system.
S Protection of microorganisms from hydrophilic compounds such as cefazolin and Endogenous Lipids gentamicin.
Igentamicmn.
Similar to porosity, the surface features of allografts appear to have the ability to shelter microorganisms. When examined at close proximity it is evident that the surface of bone is irregular, containing peaks and valleys that increase surface area. It is possible that bacteria can reside in these valleys, protected from optimal antibiotic concentrations and phagocytosis by neutrophils and macrophages. Metallic implants are usually




7
polished for a smooth surface, decreasing surface area and providing better presentation of a potential contaminant to the antibiotic (1, 67).
A third unique feature of allografts is the presence of residual lipids. The amount of fat that remains on a graft is a function of the tissue type and of the extent of cleaning to which the graft was exposed. Cancellous or trabecular bone has a fat content of 7090% w/w, whereas the fat content of cortical bone is lower, 6-9% w/w. Fat content is not necessary for proper graft function and is removed to the extent possible during graft preparation (15). The amount of residual fat that remains in allografts is highly variable and depends upon the processing facility's methods and graft type. Despite efforts to remove fat, most bone grafts still carry a significant amount after processing. Bacteria can potentially partition into fat reservoirs carried on the allograft and remain protected from the antibiotic, particularly if only single dose is given. This would be augmented by the relative insolubility of water-soluble cephalosporins in lipids (27). Even if bacteria do not preferentially partition into fat, organisms that become surrounded by lipids carried on the graft would be protected for this reason. The Concept of Local Therapy
Orthopaedic implants containing an antibiotic for either therapeutic or
prophylactic delivery have been used since the 1970s (61). During that time there have been significant improvements to both the devices and their application. Currently, the majority of local antimicrobial therapy used in orthopaedic surgery is accomplished with the polymethylmethacrylate (PMMA) gentamicin bead (11, 72). This system provides many advantages over the more conventional systemic antimicrobial therapy. With local delivery, systemic toxicity is avoided because serum drug concentrations are 10-100 times lower (33, 34, 53). Because the drug reservoir is located at the site requiring




8
therapy, high tissue concentrations are achieved only in the location they are needed. Organisms reported to be resistant to a drug at systemically attainable plasma concentrations may be sensitive to the drug at concentrations found at the wound site via local delivery (61). For example, if a bacterial strain was resistant to an antibiotic because it was able to product an enzyme that inactivated the drug, the addition of more drug would eventually saturate the enzyme allowing the accumulation of the antibiotic. This is in contrast to the scenario that accompanies intravenous therapy where drug concentrations in poorly perfused wound tissue are often much lower than plasma levels
(53).
However, targeted delivery systems are not without limitations. During
preparation of the standard bead, dry powders of both PMMA and an antibiotic are mixed with water to form a cement. This reaction is very exothermic resulting in beads reaching temperatures in excess of 1000C during the drying process (32). For reasons pertaining to the stability of the drug, exposure to such high temperatures places significant constraints on the antibiotic that may be used. Additionally, PMMA beads serve no function other than being a non-resorbable carrier for the antibiotic. Because of this, additional surgery is generally required to remove the device after the drug has been delivered, adding the expense and risk of a second procedure (61).
Proposed Solution
As shown in Figures 1-l a-c, an allograft based drug delivery system would have the major advantage over PMMA in that allografts are bioincorporable, meaning they will eventually integrate into and become part of the recipient's own tissues (15, 17, 22, 50). This eliminates the need for a second surgical procedure to remove the device, as




9
Figure 1-la. Immediate postoperative radiograph of an extensive defect of the tibial diaphysis. The oval-shaped, dark area at the center is a void in the structure of the bone itself. This void is filled with a combination of demineralized bone matrix and cancellous chips (non-radioevident).




10
Figure 1-lb. Three week postoperative radiograph of the tibial diaphysis shown in 1-1 a. The size and extent of the void has decreased noticeably as new bone formation has occurred.




Figure 1-10. Six week postoperative radiograph of the tibial diaphysis shown in 1-1 a-b. Remodeling has completely filled the void with new bone restoring the tibia to a structural unit.




12
Figure 1-2. Postoperative radiograph of a threaded cortical dowel fusing L4 and L5 (top); An example of a threaded cortical dowel allograft (bottom).




13
with the PMMA bead. In addition, allografts are indicated in orthopaedic procedures for
reasons other than drug delivery. Allografts are frequently used to lend structural support
or promote new bone formation (Figure 1-2). These added advantages do not come at the
expense of previously mentioned benefits associated with local drug delivery systems.
Table 1-2 summarizes and compares the advantages of an allograft based delivery system
and the PMMA system over systemic drug delivery.
Table 1-2. A comparison of the advantages and disadvantages associated with the PMMA and allograft based drug delivery systems.
PMMA Based System Allograft Based System
Advantages Disadvantages Advantages Disadvantages
Locally high drug High temperatures Locally high drug Potential for disease concentrations (>100C) are required concentrations transmission for formulation
Decreased risk of Second surgery is Decreased risk of Restrictions on the systemic side effects required for removal systemic side effects availability of donated following drug tissue expenditure
Allows for primary No functional purposes Allows for primary Difficulty in drug wound closure other than drug delivery wound closure loading and material homogeneity
Avoids compliance Avoids compliance issues issues Incorporates into the
recipients tissue
eliminating the need for
a second surgical
procedure for removing
the device
Provides structural
support at the graft site
Promotes new bone
formation
Indications for Use
There are several possible indications for an allograft based drug delivery system,
raging from simple graft preservation to therapeutics. The following list briefly describes
four major uses of antibiotics with allografts.




14
1. Graft preservation. Allografts, being aseptically harvested and processed, are
frequently treated with a solution containing one or more antibiotics to prevent
incidental contamination during recovery or processing leading to an infection in the
recipient.
2. Simple prophylaxis. An assumably sterile allograft could be loaded with sufficient
antibiotic to prevent infection due to contamination introduced during surgery. The allograft is independently indicated in the surgery and the antibiotic loaded into the tissue simply replaces or improves upon systemic perioperative chemoprophylaxis.
3. Directed prophylaxis. An allograft containing an antibiotic would be used
specifically for the prevention of infection in a wound where one would likely occur if no prophylaxis were administered. This setting would be analogous to the use of the PMMA- gentamicin beads in preventing infection of an open fracture. The tissue
based delivery system would have the two fold advantage of not requiring a second
surgery for removal and lending structural and healing support to the fracture site.
4. Therapeutics. Tissue based delivery systems could be used to treat an existing
condition such as chronic osteomyelitis. Current treatment for certain types of
osteomyelitis includes a primary operation for wound debridement and placement of
PMMA-gentamicin beads followed by a second procedure for bead removal and placement of a bone allograft. Tissue based delivery systems could eliminate the
need for the second operation by delivering a therapeutic dose of antibiotic carried
within the prescribed allograft, which would then be followed by normal graft
incorporation.




15
Barriers to Development
The use of allograft bone as a carrier for antimicrobial agents has previously been proposed, however limitations with tissue preparation and drug loading have prevented the development of a tissue based delivery system(73). The major barrier in developing an allograft based drug delivery system is the potential for disease transmission that is associated with any biological material.
Potential for Disease Transmission
Annually, over 500,000 allografts are surgically implanted and the vast majority are done without serious complication (8, 15, 17, 31, 59, 111, 116). Viral transmission, predominantly HIV and HCV, is a rare event (1 in 1 million for HIV) (15, 59, 111). However, because these diseases are incurable as of now, their statistically small risk of transmission remains a significant concern in the operations of tissue banks.
There are two cases known to date, where grafts from HIV infected individuals have resulted in disease transmission. The first incident was in 1984 when a femoral head removed from one patient during hip arthroplasty was subsequently used for spinal fusion in a second patient (7). This bone segment was neither tested for HIV (no test was licensed for HIV in 1984) nor processed by a tissue bank. In addition, the donor was not screened for symptoms or high risk behaviors associated with HIV, both of which were present (7, 36, 37).
The second incident occurred in 1985. In this case, a young man involved in a robbery was fatally shot. Four solid organs were procured and all four recipients seroconverted prior to their death. In addition, 46 tissue grafts were produced, however only three transmissions have been reported. Each of the three grafts implicated with




16
transmission were large, non-purged grafts with largely intact marrow reservoirs. These grafts were preserved via freezing and had no secondary sterilization. Interestingly, four grafts of this type were implanted, indicating one was a non-transmission (a non-purged femoral head). The other grafts, all of which were purged to some extent and freezedried, did not result in disease transmission (7).
In this second transmission, the only available test for HIV was the HTLV-III antibody test that is far less sensitive than HIV tests used today (36, 39, 54, 63). At the time of implantation, the donor tested negative with this assay. However, subsequent retesting of banked white blood cell preparations by polymerase chain reaction (PCR), were positive for HIV proviral DNA. Based on this evidence, it is estimated that the donor had contracted HIV approximately 3 weeks prior to donation and was at that time, in the time period from when a person is infected to when the infection is detectable.
When taken together, these two incidents indicate that HIV can be transmitted
through implantation of infected allografts. Also evident is the potential reduction of this infectivity through tissue processing.
Limiting Risk by Controlling the Supply
To reduce the potential for disease transmission and increase the safety of the supply of tissues made available to implanting surgeons, the Food and Drug Administration (FDA) in 1993 mandated that blood from each donor be tested for the presence of antibodies to the human immunodeficiency virus type 1 and type 2 (HIV 1 and 2), hepatitis B virus (HBV) surface antigen, and antibodies to the hepatitis C virus (HCV) (35-38). Many of the recovery agencies and processing have elected to test for disease markers beyond those required by the FDA (Table 1-3).




17
The use of this complex battery of tests greatly reduces the window period of an infection (Figure 1-1). A donor in the window period has contracted the disease and is infectious, although will not test positive to markers for the disease. This type of result is often referred to as a false negative. As the disease progresses the donor will begin to test positive for the disease, a process termed seroconversion. Consequently, these tests narrow the window period and serve as a more reliable indicator of donor thus reducing the chance of yielding a false negative.
Table 1-3. An example serological profile run on potential tissue donors. Donors
testing positive to an one of these markers are excluded prior to graft release.
Test Name Marker Detected
HIV-DNA by PCR Human Immunodeficiency Virus Proviral DNA
HIV /V2 Ab Human Immunodeficiency Virus 1 and 2 Antibody
HBsAg Hepatitis B Surface Antigen
HBcAb Hepatitis B Virus Core Antibody
HCV Hepatitis C Virus Antibody
HTLV I & II Antibodies Human T-Cell Lymphotropic Virus Type I & II
RPR Antibodies to Treponemal palladium
In addition to serological testing, each potential donor is given a physical
examination and a comprehensive medical and social history is obtained to exclude donors with "high risk" factors for infectious diseases (15, 59, 111). Lastly, bacterial and fungal testing is done on processed grafts prior to their release for implantation.
The potential for transmission of HIV through allograft implantation was more likely during in the earlier stages of the epidemic, before reliable screening assays were available and the importance of excluding donors with high-risk symptoms and behaviors




18
10
4 6
M 8 .....".. HIV-DA
0-
0 2 4 6 8 10 Weeks After Exposure
Figure 1-3. Idealized serological profile following HIV infection. The solid horizontal line indicates the theoretical level of detection (sensitivity). The bracket denotes the window period for HIV when tested with the PCR assay. Note that the window period is longer with the HIV antibody test.




19
was realized. The two cases discussed earlier would likely not happen today. Both of the donors would have been excluded if PCR testing for HIV was available and performed. In addition, the first donor would have been excluded based on his medical and social history, which revealed past IV drug abuse and generalized lymphadenopathy. It is also speculated that the second donor, who tested negative for antibodies to HTLV-III would have tested positive for antibodies to HIV-1 with the current, more sensitive testing assays.
Controlling the Risk through Processing
Although donor screening has led to a dramatic increase in tissue safety, it has inherent limitations. For one, it is not practical nor possible to test for every potential pathogen. In addition, the emergence of new, yet unrecognized pathogens is a certainty due to increased resistance to antibiotics. For these reasons, it is prudent to incorporate steps in the production of a tissue product that address the possibility of a contaminated starting material or the introduction of an adventitious contaminant. Limitations to Bone Sterilization
Bone does not lend itself to sterilization due to a number of factors. First, the number of potential contaminants is high. Most materials that are sterilized, such as plastics or metals, do not have a substantial reservoir for viral contamination. Human tissue obviously does carry the potential for significant viral contamination. This is an important factor as viruses are highly resistant to some sterilization processes that are effective against bacteria such as irradiation (6, 45, 68, 91, 111). Another consideration is the delicate nature of the proteins carried within the bone. Bone morphogenic proteins or BMP's that enhance new bone formation are necessary for proper graft. It is recognized that several common sterilization processes alter or eliminate the beneficial




20
properties of these proteins (64, 110). Decreasing strength and sufficient penetration are the other considerations for bone sterilization. The biomechanical strength of tissue can be effected by heat and irradiation (9, 68, 91, 97). Tissue penetration is also a challenge for any sterilization process that employs a gas or liquid germicide.
Gamma irradiation. The virucidal and bactericidal effects of gamma irradiation are created via two mechanisms (10). The primary mechanism is direct alteration of nucleic acids leading to genome dysfunction and destruction. A secondary mechanism is the generation of free radicals, primarily from liquid water, contributing to the sterilizing abilities of gamma irradiation. This secondary effect is not realized, however, when an article has been lyophilized or is frozen at the time of irradiation. It is in the frozen and freeze-dried states that the vast majority of tissue is presented for sterilization by irradiation (9).
The differences in efficacy as a function of physical state is well characterized by the plasma component industry (55). In the frozen or freeze-dried state, the virucidal effects of gamma irradiation are directly due to genomic destruction and constitutes a first order process with respect to dose. In the liquid state, however, the formation of free radicals contributed to the virucidal capacity of the treatment. This generation of free radicals is suggested to cause the deleterious effects to the tissue. Taken together, items presented for gamma irradiation sterilization in the frozen or freeze dried state will require significantly higher doses to achieve the same effect as would be realized if the item were in the liquid, hydrated state.
In tissue banking, most of the research has centered on the inactivation of HIV. HIV is a retrovirus that is fairly resistant to destruction by gamma irradiation (18).




21
Although relatively low doses of gamma irradiation ( 1IMrad) are capable of killing most classes of microorganisms, studies have shown that greater than 3.0 Mrad is required to eliminate the chance of transmission of HIV from infected tissue (55, 91). Despite the presence of published data, most tissue banks that use gamma irradiation as a means of secondary sterilization expose the tissue to a dose of only 1.5 2.5 Mrad.
In choosing the irradiation dose, tissue banks must consider the effects the treatment will have on the biomechanical properties of the tissue. These have been demonstrated to be dose dependent. At doses below 2.5 Mrad, the biomechanical effects, defined as a reduction in graft integrity as measured by either axial compression or tensile strength, to most tissue appear to be small. Unfortunately, this is a sub-lethal dose for HIV and other viruses. At doses greater than 3.0 Mrad, tissues can lose a significant amount of strength, ranging from 25% to 75% as compared to untreated controls, depending upon the conditions under which it was irradiated This data suggests that the dose required to assure the complete inactivation of contaminating viruses is above the dose at which gamma irradiation starts to produce detrimental effects to the grafts biomechanical properties.
Ethylene oxide. The use of ethylene oxide (EtO) persists within the tissue
banking industry, despite rapidly accumulating data that suggests EtO not only damages tissue structure and function, but also poses environmental and health risks (110). The primary mechanism by which EtO kills viruses and bacteria is through alkylation of purine and pyrimidine moieties leading to DNA and RNA dysfunction. In addition, secondary mechanisms include enzyme inactivation through alkylation of amino acid




22
residues. These effects make EtO an effective sterilant against bacteria, spores, molds, yeasts, and viruses including HIV (10).
Ethylene oxide and its by-products, ethylene glycol an ethylene chlorohydrin, are mutagens and are considered to be cytotoxic. For this reason, the FDA imposed a limit of 250 ppm as the amount of residual EtO that could remain on a medical device. Recent studies have shown that even in very low concentration (=25 ppm) EtO was toxic to fibroblasts (97). In addition, bone remodeling studies have demonstrated that EtO treatment reduced bone in-growth by 68%, despite no detectable residual EtO or its byproducts (<20 ppm). This inhibition of remodeling is speculated to be due to alkylation of the amino acids in bone morphogenic proteins and other osteoinductive messengers.
In addition to inhibiting osteoinduction (e.g. the ability of allograft bone to induce de novo bone formation at the site of implantation), EtO sterilization is implicated in an even more serious side effect. Ethylene oxide sterilized grafts have been associated with an immunologically induced synovitis following graft implantation (57). In this reaction, patients develop a persistent synovial effusion several months to years following graft implantation. The synovial fluid changes to an orange or brown color and contains collagenous debris, neutrophils, and lymphocytes. Patients tend not to develop elevated peripheral WBC and infection of the graft site is not evident. This condition is refractory to common treatments for inflammation (non-weightbearing, anti-inflammatory agents, aspiration) and is only resolved by removal of the graft. In one study by Jackson and coworkers, detectable levels of ethylene chlorohydrin were found upon removal of a graft fourteen months post-implantation (57).




23
In addition to the detrimental effects of EtO on tissue, its use has become highly regulated by the Environmental Protection Agency (EPA) and the Occupational Safety and Health Administration (OSHA). The use of EtO in hospitals and industry as a method of sterilizing heat labile instruments is being phased out and replaced with other types of non-toxic cold sterilizers, a trend which will most likely extend into the tissue banking industry.
Purging and soaking. Purging grafts of their cellular components and soaking them in antimicrobial solutions comprise the most common pathogen reducing steps employed in allograft production today (15). Most grafts produced within the United States today are processed with at least some form of marrow element purge and antimicrobial soak, except when prohibited by the nature of the graft such as those grafts preserved to maintain cell viability. Typical treatments include warm water lavage, hydrogen peroxide, isopropyl alcohol, iodine, and antibiotic soaks (Table 1-4). Table 1-4. Chemicals used in the processing of human tissue, the concentrations used, and their intended action.
Chemical Treatment Concentration Effect
Warm water lavage NA removes blood and lipids H202 3% removes blood lipids, bactericide, virucide Isopropyl alcohol 70% bactericide, virucide Iodine 10% bactericide, virucide
The first and most basic mechanism by which these steps accomplish their
intended purpose is via simple dilution of the microbes or viruses. The in vivo viremia associated with HIV is of greatest quantity in early infection just prior to antibody seroconversion (7, 15, 111). For tissue routinely recovered and processed for graft




24
production, the bone marrow represents the greatest reservoir for the virus. By removing the marrow elements from the graft, the number of infectious units is substantially reduced. As mentioned, the only grafts known to have transmitted HIV in humans have been those which were not purged of their marrow elements prior to implantation. In fact, there have even been cases where patients have received thoroughly lavaged kidneys from donors who were later found to be infected with HIV and did not contract the disease (125).
In addition to reducing the potential viral load in a graft, purging also reduces the amount of HLA expression a graft carries (29, 47). By removing the majority of the marrow from bone grafts, the antigenicity of the graft and therefore the potential for sensitization is reduced.
Alcohols (ethyl and isopropyl) are among the most common solutions to be used in the production of allografts today. Alcohols are effective at reducing the viability of a broad spectrum of bacteria and are very effective against enveloped viruses such as HIV
(10). Among the alcohols that are completely miscible with water in all proportions, propyl alcohols are the strongest disinfectants. Although these alcohols are directly cytotoxic, their solubility in water and volatility allows for their removal to such an extent that final concentrations are well below their ability to produce toxic effects.
Hydrogen peroxide (H202) is frequently used as an antimicrobial and defatting agent. In vivo it is an active germicide, found in the saliva, milk, and phagocytes, and is found in other tissues as a result of metabolism. There are at least two mechanisms involved with H202 antimicrobial activity. The first is the oxidation of chloride in bacteria to form hypochlorite, a well-characterized germicide, and water. The second is




25
the generation of free hydroxyl radicals, the strongest oxidant known, which then attacks membrane lipids, nucleic acids, and proteins leading to pathogen inactivation (13). In addition to its germicidal effects, the vigorous release of oxygen gas from H202 lends to its ability to clean and debrided fatty and bloody grafts (7, 15, 95).
Iodine containing compounds are used less frequently in tissue processing. This is primarily due to their tendency to discolor grafts. This unpleasant side effect can be averted with the addition of a decolorizing step with ascorbic acid, wherein molecular iodine is reduced into the colorless iodide ion. Only the 12 species is believed to be active as a germicide, where it reacts with several types of functional groups on proteins, nucleic acids, and unsaturated fatty acids. Although reactions to iodine have been reported since the 1800's, they are typically associated with either massive amounts of iodine exposure or intake over an extended period (70).
The major deficiency with these types of chemical treatments is their lack of thorough penetration into the tissue. Because they are in liquid form, they can only function as surface inactivators. The matrix of bone and other soft tissues is highly complex and therefore does not lend itself to complete penetration by solutions. Several tissue banks have attempted to overcome these complexities with solvents, detergents, and mechanical mechanisms aimed at increasing penetration. These augmentations have led to improved success on a tissue specific basis, however large grafts continue to present a challenge for the industry.
The Optimal Sterilizing Process
The ideal model for allograft sterilization would incorporate several key
properties in achieving a safe and effective tissue for transplantation. Below are listed




26
these characteristics. Table 1-5 compares these characteristics to the aforementioned sterilization processes.
1. The process would need to be effective at removing and/or inactivating a wide range
of bacterial and viral pathogens contained on and in the tissue.
2. This process would not result in the graft being toxic to the recipient due to residual
chemicals.
3. The process would not significantly reduce the biomechanical strength of the graft,
which would lead to graft collapse.
4. The process would not adversely alter bone morphogenic protein, a protein
endogenous to bone which is responsible for induction of new bone formation.
5. The process would be robust, meaning small changes to the system or graft would not
alter the ability of the process to yield a sterile product.
6. The latter portion of the process would be executed in the graft's final container,
eliminating the possibility of adventitious contamination of the graft.
Accomplishing Tissue Sterilization
Because of the risk of disease transmission, an essential first step in developing an allograft based drug delivery system is to effectively sterilize the starting material without altering the beneficial properties of tissue. Therefore, prior to the specifically investigating allografts as drug delivery systems, the following basic research was conducted to establish an effective way to sterilize human bone and therefore remove this specific barrier to further development of the project.




27
Table 1-5. A summary of the sterilization processes discussed in this chapter compared against the criteria presented for the optimal sterilization process.
MethoologyFinal Methodology Sterilant Toxicity Biomechanical BMP Robustness Final Container
Gamma Bacteria Good
Gamma
Viruses No Poor Poor Moderate Yes Irradiation
Moderate
Ethylene
Ethylene Good Yes Moderate Poor Moderate Yes Oxide
Purging and Weak to Moderate No Good Good Poor Possible Soaking
Enhancing the potency of hydrogen peroxide. As most sterilization processes are detrimental to the structural and/or biological properties of allograft bone it was therefore necessary to develop a sterilization process that minimizes these. Hydrogen peroxide has been used in tissue banking since its formalization in the 1950's and almost all bone allografts are treated with this solution at various concentrations ranging from 1
- 35 % (15, 87, 95). Through the extensive use of hydrogen peroxide, this compound has demonstrated its compatibility with bone allografts, yet its effectiveness as a sterilant had not been realized. It has been suggested that ultrasonic energy enhances the bactericidal and sporicidal effects of hydrogen peroxide (10). In this study, a reduction in the D-value for the Bacillus sterothermophilus (106) spore was calculated for samples treated with 6% H202 in the presence and absence of ultrasonic energy. This spore was chosen due to its well characterized and accepted resistance to peroxide sterilization (62). Samples were treated with 2 ml of 6% H202 at 450C over a given range of time and the reaction was stopped with the addition of sterile water and transferred to trypticase soy broth for culture at 560C. All samples were performed in triplicate.




28
The D-value obtained for the samples run in the presence of ultrasonic energy was 0.83 1.66 minutes (Table 1-6). This compared favorably to the D-value obtained for the samples run in the absence of ultrasonic energy (>10 min). This reduction in the time required for the inactivation of spores may allow for a practical method of sterilizing allografts without adversely effecting their desired attributes. Table 1-6. Comparison of spore inactivation with 6% hydrogen peroxide in the presence and absence of ultrasonic energy. Positive (+) results were identified by turbidity of the media and confirmed by subculturing the broth to solid media. Negative (-) results were determined by the media retaining clarity over the seven days of incubation. Assay sensitivity was determined to be < 5 organisms. Treatment Treatment Time (min) D0 5 10 15 20 30 40 50 60 value Sonication + + + + - - -- - - - - <1.6
No
+ + + + + + + + + + + + + + + + + + + + + + + + + + + >10 sonication
Effects of residual lipids on the activity of hydrogen peroxide. This study
examined the potential for residual lipids to reduce the effectiveness of hydrogen peroxide at inactivating Bacillus sterothermophillus spores. Whole femora and tibiae were surgically removed from human cadaveric bone donors and debrided of extraneous soft tissue. The bone tissue was then ground yielding a bone slurry with the consistency of an oily paste. A section of this bone paste was removed and thoroughly cleaned of residual fat content using warm (450C) acetone. The cleaned bone slurry was mixed with the untreated bone paste in various weights to yield samples containing 0, 10, 30, and 60% residual fat. All samples were verified using an exhaustive volatile extraction with gravimetric analysis. One gram of each sample was added to test tubes containing a 106 inoculum of spores and was treated with 2 ml of 6% hydrogen peroxide (40oC) in the




29
presence of ultrasonic energy (45Khz) for multiple time points. Each time point was run
in triplicate. The reaction was stopped for a given time point by the addition of 20 ml of
sterile water and the inoculum was transferred to trypticase soy broth for a seven day
incubation at 560C for seven days. Controls included sterile water (negative control),
inoculated water (positive control), and H202 without bone tissue.
The results of the study indicate that lipids prolong the contact time required for
the complete inactivation of B. sterothermophilius spores (Table 1-7). The data
generated from this study supports the hypothesis that removing endogenous lipids from
cortical bone will increase sterilization efficiency. By lowering the contact time required
for sterilization, the potential adverse effects of the sterilant (reduction in tissue strength)
may be minimized. Although this assay confirmed the ability of hydrogen peroxide
toinactivate spores in a suspension of bone material, the ability to accomplish this in a
more relevant context, using an intact bone model was needed.
Table 1-7. Approximated D-values for B. sterothermophilus as a function of residual fat content remaining in a homogenized bone slurry, when sterilized in a 6% hydrogen peroxide solution at 420C in the presence of ultrasonic energy. Positive (+) results were identified by turbidity of the media and confirmed by subculturing the broth to solid media. Negative (-) results were determined by the media retaining clarity over the seven days of incubation.
Treatment Time (min) D Treatment
0 5 10 15 20 30 40 50 60 value Neg to...------ ----- --------------------------- NA
Control
PosControl + + + + + + + + + + + + + + + + + + + + + + + + + ++ >10 Nobone + + + + + - ,- ,- - - - 1.66
0%fat + + + + + + + + ------------------- 2.5
10%fat + + + + + + + + + + - -- --- - ---- ------ 3.33
30%fat + + + + + + + + + + + - - -- --------------- 3.33
60% fat + + + + + + + + + + + + + + + +-- -- - - 6.66




30
A model for sterilization efficacy in cortical bone. Definitively demonstrating the efficacy of a liquid sterilization process for human cortical bone has historically been difficult. In this experiment the use of a machined segment of human cortical bone carrying a B. sterothermophilus (106) biological indicator was evaluated for its potential uses to support claims of allograft sterility (Figure 1-2). The device was prepared by cutting a cortical segment from the anterior ridge of the tibia in a cadaveric bone donor. This segment represents the thickest portion of cortical bone encountered in the body and is thus the most difficult to penetrate and sterilize. A cylindrical hole was machined into the end of the bone, longitudinal to the axis. A second segment of cortical bone was machined into a cylindrical pin with a diameter slightly larger than that of the hole. A partial slit was cut into the pin allowing a biological indicator to be placed within. The pin was then forced under compression into the machined hole and exposed to the sterilization process. A control was also run using only sterile water to evaluate if the spores were appreciably being washed off the strip. In addition, a tracing dye was used to evaluate the path of the liquid through the device.
The results from the controls indicate that the extent of washout that occurred was minimal and did not significantly effect the introduced bioburden. Recovery studies showed that on average 8 x 105 spores were recovered using the saline control. The samples exposed to the sterilization process did not demonstrate growth after incubation for seven days in TSB indicating process efficacy. The sensitivity of the assay was < 5 organisms.




31
Cortical bone pin Slit for BI Cortical bone block Figure 1-4. Model for testing the efficacy of a liquid sterilization process for cortical bone.




32
Effects of sterilization on allograft biomechanics. The purpose of this study was to identify the effects of preservation and sterilization processes on the strength of cortical bone. This work is essential in determining what treatments are acceptable for the graft to be exposed to during the processing and drug loading steps of production. Treatments that significantly reduce strength must be avoided in the graft preparation/drug loading process.
Femora and tibiae were isolated from 18 different human cadaveric donors and machined with a lathe into 203 pins that were 4.0mm in diameter and 10mm in length. The pins were then exposed to treatment that may be used in the graft preparation or drug loading process. Following treatment the ultimate failure load under axial compression was determined. Axial compression testing was adapted from ASTM D695-91 and performed on an MTS 858 (Eden Prairie, MN) servohydraulic mechanical test apparatus.
The results demonstrate that pressure assisted hydrogen peroxide perfusion did not reduce the compressive strength of the cortical bone pins (Figure 1-3). Gamma irradiation did significantly reduce the strength of the tissue and therefore an alternative method should be sought for terminal sterilization of the graft. Lyophilization, in contrary to expectation, significantly increased the axial strength of the tissue. This is a promising result as lyophilization is hypothesized to be a critical component to maximize the amount of drug that can be loaded into bone. Summary of Tissue Sterilization
These sets of experiments demonstrate that the effect of mild germicides can be
enhanced to the level of sterilization through the addition of ultrasonic energy. The effect of residual lipid content was also characterized. These two studies demonstrate that




33
400
HControl 350 15Lyophilization D Irradiation 300
IPAHP S250
S200
150 -....
.............,...., .......... ,
5o:.......
100
50
0
Treatment
Figure 1-5. Treatment groups and mean ultimate strength during axial compression testing. Control a group consisting of no preservation or sterilization treatments was included (n=51). Lyophilization freeze drying to reduced the residual moisture content of the graft to below 2% (n=51). Gamma irradiation a sterilizing dose of 3.5 Mrad (n=50). PAHP Pressure assisted hydrogen peroxide treatment employed exposing the tissue to a 6% solution of hydrogen peroxide at 400C for 30 minute under oscillating pressure and ultrasonic energy (n=51). Error bars indicate + 2 X standard error.




34
sterilization in an organic environment is possible with hydrogen peroxide when ultrasonic energy is employed and that there is a direct relationship between the contact time required for sterilization and the lipid content of the graft. Strength evaluation confirmed that this treatment did not significantly alter the biomechanical properties of the tissue. Lastly, a model was developed that confirmed the effectiveness of the process to kill spores deep within matrix of the tissue. This is significant because it answers two questions definitively. First, the sterilant is able to penetrate the tissue; secondly, the peroxide reaches the inner portions of the tissue in concentrations sufficiently high to result in sterilization within a reasonable time.
Summary
The aims of this project are (1) to identify factors that may predispose allograft bone implants to post-operative infection; (2) to identify the optimal antimicrobial for further investigation within an allograft based system (3) to evaluate a potential drug loading procedure, and (4) to characterize the release profile of an antibiotic from cortical bone. The rationale for this approach is that allograft features such as surface texture and fat content provide a mechanism for protecting microorganisms from chemoprophylaxis and that by addressing these issues, a more infection resistant graft can be developed. In addition, by changing the route of administration from systemic to local (graft delivered), effective concentrations can be attained without realizing systemic side effects. Through a drug loading procedure whereby a solid form of the drug is contained throughout the inner matrix of the graft, sustained release may be attainable. It is hoped that this work could lead to new treatment options for orthopaedic surgeons that would reduce infection rates and mitigate the associated financial, emotional, and physical burdens.




CHAPTER 2
EVALUATION OF ALLOGRAFT COMPOSITION FOR FACTORS THAT
MITIGATE ANTIMICROBIAL CHEMOPROPHYLAXIS Introduction
Allograft implants have a higher rate of post-operative infection than do metallic implants. Several studies have demonstrated that this difference is not attributable to non-sterile allografts being supplied to surgeons for implantation (111, 112, 116). One area that may explain this difference in incident rate is composition of the allograft. Material and device attributes can influence post-operative infection rates (21, 30, 103). To date, allograft composition has not been evaluated for the presence of identifiable and controllable factors that influence the incidence of post-operative infections.
One difference between allografts and metallic implants that can affect infection rates is lipid content. The amount of fat that remains on a graft is a function of the tissue type and extent of cleaning to which the graft was exposed during processing. Cancellous bone has a fat content of 70-90% w/w, whereas the fat content of cortical bone is lower, 6-9% w/w. Fat content is not required for proper graft function and is removed to the extent possible during graft preparation (7, 15, 95). Despite the effort to remove the fat, most bone grafts still carry a significant amount after processing (15). The amount of residual fat that remains in a graft is a function of both graft type and processing methodology employes, and is therefore highly variable. Bacteria can be
35




36
predicted to partition into fat reservoirs carried on the allograft and remain protected from the single dose of antibiotics commonly given to surgical patients. Furthermore, the relative hydrophilicity of cefazolin (Figure 2-1), the most commonly used antibiotics in allograft prophylaxis, cannot partition into these fatty tissues (20, 27, 71). Even if bacteria do not preferentially partition into fat, any organism incidentally surrounded by lipids on the graft would be shielded from these hydrophilic antibiotics.
Another factor that can influence the incidence of allograft infection is the surface architecture. The surface features of allografts may have the ability to protect microorganisms. Cancellous bone is sponge-like in texture with many irregular features, while cortical bone has a dense, regular structure (Figure 2-2). In contrast, metallic implants are usually polished for a smooth surface, decreasing surface area and providing better presentation of a potential contaminant. In addition, allografts are now often machined to contain thread profiles or grooves to allow for insertion or prevent slippage after implantation, greatly increasing their surface area (Figure 2-3). It is possible that bacteria could become lodged inside these crevices and be protected from optimal antibiotic concentrations as well as phagocytosis by neutrophils and macrophages.
The increased incidence of infection can be hypothesized to be due to routine chemoprophylactic procedures being less effective on allograft tissue versus metallic implants. Irregularities on the surface of the bone and residual lipids may be responsible for this decrease in antibiotic effectiveness. Because of advances in tissue processing, finished graft surface architecture and lipid content can be controlled or altered to minimize these disadvantages. For this reason, there is now merit in evaluating these factors for their potential influence on chemoprophylaxis used in allograft surgery.




37
Nq 0 I N -CH2-C-NH S N N
SN CH2S S CH3
C -ONa
II
0
Figure 2-1. Molecular structure of cefazolin.




38
Figure 2-2. Coronal section through a femur showing spongy cancellous bone surrounded by dense cortical outer layer.




39
Figure 2-3. Examples of threaded allografts (interference screws).




40
Materials and Methods
General Experimental Overview
Implants were produced with differing lipid contents and architectural features and were tested in vitro. These implants were then contaminated with a known amount of bacterial bioburden and incubated in baths dynamically controlled to maintain relevant antibiotic concentrations. After completion of the exposure to the antibiotic treatment, the samples were removed and bioburden analysis was performed to determine the logio reduction for each treatment group.
Preparation of Treatment Groups
Three architecturally distinct surfaces were evaluated during this study; (1) the smooth, regular surface of the periosteal side of cortical bone, (2) the irregular and spongy surface of the endosteal side of cortical bone, and (3) cortical bone that had been machined to carry threads. These were selected because they represent the bone surface features that predominate in orthopaedic allograft surgery. To test the effects of lipid residue on antibiotic activity, one half of the specimens in each of these architecturally distinct groups were exposed to a cleaning process that removed the tissues' endogenous lipids. This combination of architectural and compositional variation resulted in six randomized treatment groups (Table 2-1).
To prepare the specimens, diaphyses of human femora, tibiae and fibulae were initially prepared in standard fashion by removing any extraneous muscle, ligamentous attachments and the periosteum. For the periosteal and endosteal specimens, bone segments were then cut from the diaphyseal flare of the femora and tibia yielding specimens that were approximately 15 mm long, 7 mm in wide, and >3 mm in thickness




41
(Figure 2-4a). The periosteal surface of these specimens did not contain any mechanical alteration aside from the debridement procedure. To prepare the threaded specimens, fibulae were cross-sectioned at 15 mm intervals. However, they remained circumferentially intact to allow for the application of threads to the inner lumen. These segments were machined to carry a thread profile of 600 x 0.5mm in depth along their entire lumenal surface (Figure 2-4b). This thread profile is typical of those found on threaded allograft bone and increases the surface area by approximately a factor of two. These segments were then sectioned longitudinally yielding two crescent shaped specimens of threaded cortical bone (Figure 2-4c).
Table 2-1. Allograft reatment group specification prepared in this study.
Architectural Composition
Treatment Physiologic Lipid Free
Cortical bone void of any soft-tissue
Periosteal Cortical bone void of any soft-tissue attachments or periosteum and
Periosteal
attachments or periosteum. cleaned free of endogenous lipids (<5% of physiologic).
Bone from the medullary canal Bone from the medullary canal consisting of the cortical-cancellous
Endosteal consisting of the cortical-cancellous consisting of the cortical-cancellous
interface, interface, cleaned free of endogenous lipids (<5% of physiologic).
Cortical bone with machined threads,
Threaded Cortical bone with machined threads. cleaned free of endogenous lipids (<5% of physiologic).
To achieve different levels of endogenous composition, half of the bone segments were treated in a 3% hydrogen peroxide bath at 400C for 30 minutes followed by a volatile extraction with acetone at 400C for 30 minutes. The other half were left untreated and served as physiologic tissue controls. These samples contained all of the endogenous lipids that remain following the standard debridement procedure. For




42
7mm
Endosteal
15mm
Periosteal
3mm
Figure 2-4a. Preparation of bone samples. Preparation of periosteal and endosteal segments from diaphyseal sections of femora and tibiae.




43
0.5 mm
Figure 2-4b. Representation of a thread profile typically found on threaded cortical allografts.




44
15mm
Figure 2-4c. Preparation of the threaded cortical specimens from the diaphyses of fibulae. Threading is applied to the intact tissue, which is then sectioned longitudinally to create two halves.




45
cortical bone this is approximately 6-9% and for cancellous bone it is generally greater than 70%. Following treatment, the samples were placed into individually labeled cryovials and frozen at 200C. Aseptic technique was used throughout specimen preparation to minimize the introduction of adventitious bioburden. Sample Inoculation
Inoculum preparation. Staphylococcus aureus was utilized in this study because of its relevance as the most frequently isolated pathogen from orthopaedic wound infections (26, 102, 103, 108). A single colony of S. aureus was transferred to a blood agar plate and streaked for isolation. This culture was allowed to incubate for 1824 hours at 370C. Colonies were then selected and transferred into 10 mL of normal saline. This suspension was vortexed for 1 minute to reduce bacterial aggregation. The concentration of this suspension was then adjusted by either the addition of normal saline or bacteria to yield a 1.0 McFarland suspension. This suspension was then diluted 1:10 to give a final concentration of 5-10 x 106 CFU/ml. The suspensions were prepared fresh daily and remained refrigerated.
Sample inoculation. The samples were placed in a biological safety cabinet, where 25 pL of inoculum was transferred onto each via a sterile pipette. The samples were then allowed to air dry in the hood for 60 minutes prior to further testing. Inoculum controls were also performed by adding 25pL of inoculum to 50 mL of sterile saline, vortexing, and plating 100ptL of the resulting suspension directly onto solid media for quantification.




46
Antibiotic Treatment
A dynamic model was used to recreate the concentrations of antibiotic found in bone after the intravenous administration of a prophylactic dose of cefazolin (Ancef@). Cefazolin was chosen because of its prophylactic use in allograft procedures (71, 102, 103). In this model (Figure 2-5), the specimens were placed into a vacuum flask containing 145 mL of phosphate buffered saline (pH 7.4) and an initial concentration of 18 pg/mL of cefazolin. This is the reported maximum concentration at the wound site following typical prophylaxis (27, 71, 92, 105, 121). A water bath surrounding the flask maintains the temperature at 370C throughout the experiment. Antibiotic-free PBS was then introduced into the vessel through an intravenous infusion set connected to a glass rod that delivered the solution into the vessel while a second port allowed the excess mixture to leave the vessel, thus maintaining a constant volume. A stir bar placed at the bottom of the vessel ensured uniform antibiotic distribution throughout the experiment. The infusion rate was set at 0.9 mL/min based on the half-life of cefazolin being 1.8 hours, to produce a concentration profile in the vessel that simulates the pharmacokinetic profile (71). The following equation was used to determine the rate:
C, = Co e-1
Where: Co is the initial concentration of antibiotic (p.g/mL)
C, is the concentration at any given time (yg/mL)
R is the rate of influx (mL/min)
V is the volume of the vessel (min-)
t is time (min)
The term R/V is equal to the elimination rate constant (-0.0064 min' for
cefazolin). Because the volume is fixed, the equation can be solved to determine the rate of fresh saline needed to approximate the half-life of cefazolin.




47
4
Infusion of fresh saline
Outflow of Specimen excess
saline/antibiotic
37 oC
Water Bath LOW Stir Bar
I I
4 Heat/Stir Plate
Figure 2-5 In-vitro model for the approximation of antibiotic concentrations following the administration of a single dose of a drug with first-order elimination kinetics.




48
The specimens were held in the vessel for 7.5 hours allowing the antibiotic concentration to drop below the MIC90 (71). After treatment the specimens were removed using aseptic technique and quantitative bioburden analysis was performed. To control for the wash-off of organisms from the samples during the treatment, controls were added that were not exposed to antibiotic. These controls served as the starting point in determining the reduction in bacterial population due to the antibiotic.
Bioburden Quantification
Samples were placed into a sterile conical tube and filled with 15 mL of sterile
saline. The tube was then vortexed on the highest setting for 3 minutes. The conical tube was then filled with an additional 35 mL of sterile saline and vortexed. Using aseptic technique, 10 piL and 100 pL were removed and plated onto 5% sheep's blood agar in duplicate. From the remaining saline 1 mL was passed through a 0.45 Pm filter under vacuum and rinsed with 50 mL of sterile saline. The filter was then transferred to an absorbent pad containing 1 mL of tryptic soy broth. Both sets of cultures were held for 48 hours at 37oC and quantified for CFU.
MIC and Quantification of Cefazolin
The microdilution broth technique was used to determine the minimum inhibitory concentration (MIC) of cefazolin for the organism used in the study. This technique was also used to determine unknown antibiotic concentrations. A stock trypticase soy broth was prepared containing cefazolin at a concentration of 100pg/mL. Dilutions were then made with antibiotic free broth that resulted in the following antibiotic concentrations including 10pg/ml, 7.5 pg/ml, 5 ptg/mL, 4pg/ml, 3pg/ml, 2.5 pg/ml, 2pg/ml, 1.5tg/ml,




49
1.0 pg/ml, 0.5pg/ml, 0.25pg/ml, 0.125ptg/ml. Inoculum was added to each of these tubes resulting in a bacterial concentration of 1-5 X 105 cfu/ml. All samples were run in duplicate.
The 96-well plates were incubated at 370C for 18 hours to allow for growth.
Positive (no antibiotic) and negative (no inoculum) controls were included. Following incubation, the wells were examined for growth as indicated by turbidity. The MIC was determined by the lowest concentration of antibiotic that prevented growth. Requirements for accepting the test results were: (1) all of the tubes with antibiotic concentrations greater than the MIC must be without growth, (2) all of the tubes with antibiotic concentrations less than the MIC must have growth, (3) the purity and identity of the organism in the first tube demonstrating growth from the MIC must be verified, (4) the negative and positive controls must demonstrate no growth and growth respectively,
(5) the concentration of the original suspension must be 1-10 X 105 CFU/ml as determined through by inoculum controls, and (6) replicates from the same sample must be in agreement meaning both samples must have the same inhibitory endpoint.
For samples with an unknown amount of cefazolin, the concentration of the most dilute well showing no growth was assumed to be the MIC. The concentration of the original sample was then determined by multiplying the MIC by the dilution factor of that well. Additional sets of dilutions were performed with these samples to more precisely define the endpoint.
Statistical Analysis
Statistical analysis was performed to determine if there was difference in bacterial reduction between material treatment groups. Analysis was performed using a




50
commercially available software package (Statistica '99, Statsoft Inc.). A one-way ANOVA was used to determine if a significant difference existed in the observed log reductions between the treatment groups. A Newman-Keuls test was used to determine specifically which treatment groups had a significant difference. For all tests significance was defined as a p value less than 0.05. This test assumes that all errors are independently and normally distributed. During the analysis of the results there were no major departures from these assumptions.
Results
The MIC of cefazolin to the particular strain of S. aureus used in this study was determined to be 1.0 pg/mL. This was similar to previously reported MIC for this organism (71, 93, 121). Based on this data, the validity of the model was established by sampling the solution from the flask and determining cefazolin concentrations over the course of the study. Figure 2-6 demonstrates that the model exposed the tissue to antibiotic concentrations that were similar to those found in-vivo. Due to the inherent limitations of sensitivity with the bioassay used in the quantification of cefazolin, verification of antibiotic concentration at the later time-points could not be accomplished. Concentrations determined by earlier time-points, however, provide sufficient data to demonstrate the model was simulating first-order release kinetics with an antibiotic halflife of 1.8 hours.
Although the amount of inoculum added to each specimen was controlled for,
there was substantial variation in the amount recovered from the saline controls between




51
treatment groups (Table 2-2). This difference in recoverable bioburden can be attributed to differences between the starting materials. There are two mechanisms for the loss of organism with this model. First, the organism can be washed-off during the treatment. Secondly, certain materials have the potential to retain bacteria during the extraction phase, prohibiting their detection upon culture. Table 2-2. Mean logo bioburden for the control samples and antibiotic treated samples, and loglo reduction for each treatment group. The number of replicates (n) for each treatment group is also listed. LFE = lipid-free endosteal, LE = lipid containing endosteal, LFP = lipid free periosteal, LP = lipid containing periosteal, LFT = lipid-free threaded, and LT = lipid containing threaded. The standard error of the mean is indicated in parentheses.
LFE LE LFP LP LFT LT Control (CFU) 5.24 5.53 6.19 5.72 5.10 5.37 n=3 (.13) (.13) (.16) (.13) (.12) (.13)
Antibiotic (CFU) 4.47 3.70 3.27 4.33 4.10 4.19
n=6 (.09) (.10) (.09) (.10) (.09) (.09)
Log Reduction 0.7 2.0 2.8 1.4 1.0 1.2 Log Reduction
(.16) (.19) (.19) (.16) (.16) (.19)
The logo reduction for each treatment type was calculated from the difference between the control group with no antibiotic and the treatment group with antibiotic (Figure 2-7). From this analysis, the lipid-free periosteal group had the largest reduction at 2.8 logs. This was significantly more than the periosteal group where the lipid was not removed, which had a logo reduction of 1.4. Both of the threaded groups, lipid and lipidfree, had similar loglo reductions at 1.2 and 1.0 respectively. The lipid-free endosteal specimen had a significantly lower logo reduction, 0.7, than its lipid containing counterpart, 2.0.




52
3.5
3
2.5
0T
2
~1.5
0.5
0
LFE LE LFP LP LFT LT Treatment
Figure 2-7. Summary of the average log0 reduction values attained for each treatment group (see table 2-1 for key to abbreviations). Error bars indicate + 2 standard error.




53
In addition to the 7.5 hour study, the lipid-free periosteal samples were also evaluated at intermediate timepoints to elucidate the reduction kinetics. Figure 2-8 describes the change in bacterial population as a function of time using the dynamic model. Controls were also included in this experiment to determine changing bioburden on samples that were not exposed to antibiotic.
Discussion
The hypothesis for this set of experiments is that complex architectural features and residual lipids serve to undermine the effectiveness of cefazolin by shielding bacteria from optimal concentrations. Therefore, grafts with the most regular features (smooth periosteal grafts) and those treated to remove endogenous lipids should have the highest reduction in bioburden. This hypothesis was not supported by the data from this study. The lipid-free periosteal specimens did have the greatest reduction in bioburden, however, the second largest reduction in bioburden was observed with the lipidcontaining endosteal group. This is unusual in that the surface of these specimens are the most convoluted and carry the highest amount of endogenous lipid among the specimens evaluated.
This apparent paradox may be partially explained by an alternate hypothesis. The ability of the graft to absorb the starting inoculum may be responsible for the high reduction in bioburden observed with the physiologic endosteal group. This surface has the highest lipid content of the three studied (between 70-90%). The surface of this tissue is therefore less receptive to absorbing externally applied aqueous solutions such as the inoculum. It is possible that the bacterial suspension coated the surface of the tissue and did not penetrate appreciably into the matrix. This idea is further supported by the lipid-




54
+ Antibiotics a Control
1.0E+07 .lOE+06
S1.0E+405
1.OE+04
1.OE+03
1.OE+02
0 100 200 300 400 500 Time (min)
Figure 2-8. Change in bioburden over time for the lipid-free periosteal group. For each time-point two replicates were tested. Error bars indicate range.




55
free endosteal group, which had lowest reduction in bioburden. Although the endosteal group has the most complex architecture, unlike its lipid-containing counterpart, the lipid-free specimens were not guarded against the absorption of the inoculum. In fact, quite the opposite is true. When the fat is removed from cancellous bone like that found on the endosteal surface, the remaining matrix resembles a sponge. This allows for substantial absorption of an aqueous solution. During sample inoculation it was observed that inoculum was absorbed faster and to a greater extent on samples that had been cleaned of residual lipids as compared to samples that were not cleaned. Because the organisms were not taken deeply into the tissues containing lipid, particularly the endosteal specimens, they were not afforded the same level of protection from the surrounding tissue.
Most bone allografts are preserved by lyophilization (9, 15, 95). Because there is a significant decrease in the strength of bone that is dry, most allograft package inserts call for the rehydration of the tissue prior to implantation (9). Because this process can be lengthy, up to 24 hours, it is often started prior to surgery. If contamination of the rehydrating solution occurs, the clean, dry, cancellous graft will absorb more bioburden than other types of grafts that do not have the same capacity to retain fluid. Perhaps even more important than the amount of bioburden that is absorbed is the extent to which it is absorbed. Clean cancellous grafts will allow the rehydration solution to fully penetrate the matrix due to its inherent porosity. This is in contrast to cortical grafts that, due to the density of the bone, will not absorb solutions to the same extent.
This pattern would not be observed to the same extent with the periosteal
specimens for several reasons. First, periosteal bone is composed of dense cortical bone




56
that is regular and relatively non-porous. Therefore, the amount and extent of absorption is minimized. Secondly, the amount of endogenous lipids carried on this type of bone in its natural state is approximately 10-fold less than natural cancellous bone. These two factors would mitigate the shielding effect that was observed with the cancellous bone.
The two types of threaded specimens both performed equally. The antibiotic was significantly less effective on these two groups when compared to the lipid-free periosteal samples even though the material was similar in composition to this group. There are two possible explanations for this. First, the thread profile used increased the surface area by a factor of 2. Secondly, the cuts made by the threads cross through the Haversian canals that run in a parallel fashion down the length of cortical bone (Figure 2-9a). These canals serve as the conduit for blood vessels in dense bone. By cutting into the bone, these canals are exposed increasing the porosity and providing a pathway for the entry of microorganisms. In periosteal bone that has not been machined to carry threads, these Haversian canals remain below the cortex of the bone which is only penetrated sporadically by nutrient foramina and Volkmann's canals (Figure 2-9b).
Lastly, the kinetic data provide insight into how the bacterial bioburden changed over the course of the study. As significant as the reduction in bioburden for samples treated with cefazolin was the proliferation that was associated with untreated tissue. This suggests that in the absence of antibiotics, bone provides a suitable environment for the proliferation of bacterial contamination. The substantial bioburden remaining after treatment may also suggest that a single does of cefazolin may not provide adequate




57
a.
Volkmann's Canal Haversian Canals b.
Figure 2-9. Longitudinal diagram of cortical bone. a) Cortical bone that has been machined to carry threads. Note how the thread profile exposes Haversian canals to the exterior. b) Cortical bone with its periosteal surface intact. Note only the sporadic communication of the Haversian canals with the periosteal surface through nutrient foramina.




58
assurance that contamination introduced onto the graft during surgery will be removed to a benign level, however, the level of kill was lower than exptected in all groups. This is likely attributable to the organisms not being in a favorable state for replication and consequently antimicrobial incorporation.
Conclusions
The data acquired in this study support the hypothesis that antibiotics are significantly more effective on cortical bone tissue with regular surface features and minimal porosity than on those tissues with greater architectural complexity. However, the data did not conclusively support the hypothesis that residual lipids carried on the tissue decreased antibiotic effectiveness. In fact, the data suggests that lipids may prevent bacterial absorption into the deep matrices of tissue and thus increase their susceptibility to an antimicrobial agent. For this reason, surgeons and other health-care providers responsible for allograft preparation should take care to guard against contamination during the reconstitution of clean cancellous bone grafts as introduced bacteria may migrate further into these tissues than with other less porous tissues.
Future work in this area should focus on more deliberate and proactive methods of preventing post-operative infection. These data suggest that a single dose of cefazolin may not be adequate to assuredly eliminate microbial contamination on bone allografts. The development of a method to uniformly load an allograft with antibiotic may have significant effects on the incidence of post-operative infections involving allografts and may fill the void left by systemic antimicrobial therapy.




CHAPTER 3
EVALUATION OF CHEMOTHERAPEUTIC AGENTS FOR USE IN A CORTICAL BONE ALLOGRAFT DRUG DELIVERY SYSTEM Introduction
Orthopaedic implants containing an antibiotic for either therapeutic or
prophylactic delivery have been used since the 1970's (61). In this time there have been improvements to both the devices and their application. Today, the majority of local antimicrobial therapy used in orthopaedic surgery is accomplished with polymethylmethacrylate (PMMA) gentamicin beads (11, 72). This system provides many advantages over the more conventional systemic antimicrobial therapy. With local delivery, systemic toxicity is avoided because serum drug concentrations are 10-100 times lower than when administered through the conventional route (33, 34, 53). Because the drug reservoir is located at the site requiring therapy, high tissue concentrations are achieved. In fact, organisms reported to be resistant to a drug at systemically attainable plasma concentrations may indeed be sensitive to the drug at concentrations found at the wound site via local delivery (61). This is in contrast to the scenario that accompanies intravenous therapy where drug concentrations in poorly perfused wound tissue are often much lower than plasma levels (53).
However, this delivery system is not without its limitations. During preparation of the standard beads, dry powders of both PMMA and an antibiotic are mixed with water to form a cement. This reaction is very exothermic resulting in beads reaching temperatures in excess of 1000C during the drying process (32). For reasons pertaining 59




60
to the stability of the drug, exposure to such high temperatures places significant constraints on the antibiotic that may be used. Additionally, PMMA beads serve no function other than being a non-resorbable carrier for the antibiotic. Because of this, additional surgery is generally required to remove the device after the drug has been delivered, adding the expense and risk of a second procedure (61).
Allograft based delivery systems have the major advantage over PMMA in that allografts are bioincorporable, meaning they will eventually incorporate into the recipients' own vital tissues at the site of implantation (15, 17, 22, 50). This eliminates the need for a second surgical procedure to remove the device. In addition, allografts are indicated in orthopaedic procedures for reasons other than drug delivery. Specifically, allografts are frequently used to lend structural support or promote new bone formation. These added advantages are in addition to the previously mentioned benefits associated with local drug delivery systems. Table 3-1 summarizes and compares the advantages of an allograft based delivery system and the PMMA system over systemic drug delivery.
The use of allograft bone as a carrier for antimicrobial agents has previously been proposed, however limitations with tissue preparation and drug loading have prevented the development of a tissue based delivery system (73). Recent improvements in processing technology have allowed for allograft bone to be effectively cleaned and potentially loaded with a wide range of drugs useful in these procedures. These critical advances now permit the investigation of the allograft as a potential drug delivery system.
Fundamental to the further development of a tissue based drug delivery system for the treatment or prevention of orthopaedic infections is the optimal selection of an antimicrobial agent. This theoretical chapter proposes ideal characteristics of an




61
antimicrobial agent that are necessary for success with an allograft based delivery system
and evaluates potential drug candidates against these characteristics.
Table 3-1. A comparison of the advantages and disadvantages associated with the PMMA and allograft based drug delivery s stems.
PMMA Based System Allograft Based System
Advantages Disadvantages Advantages Disadvantages
Locally high drug High temperatures Locally high drug Potential for disease concentrations (>1000C) are required concentrations transmission for formulation
Decreased risk of Second surgery is Decreased risk of Restrictions on the systemic side effects required for removal systemic side effects availability of donated following drug tissue expenditure
Allows for primary No functional purposes Allows for primary Difficulty in drug wound closure other than drug delivery wound closure loading and material homogeneity
Avoids compliance Avoids compliance issues issues Incorporates into the
recipients tissue
eliminating the need for
a second surgical
procedure for removing
the device
Provides structural
support at the graft site
Promotes new bone
formation
Indications for Use
When selecting an antibiotic for use in any situation it is important to first define
the desired effect. This is particularly true when developing a tissue based delivery
system. Antibiotics have several potential indications for use with allografts and these
different indications correspond to different antimicrobial requirements, such as duration
of drug treatment, antimicrobial spectrum, and stability. These differences are
summarized in Table 3-2.




62
Table 3-2. A summary of proposed characteristics for antimicrobial candidates for different surgical indications.
Antimicrobial
Indication Spectrum Duration of Activity Stability Requirements SpectrumI
Must be stable in
1-4 weeks
Graft Preservation Very broad preoperatively solution at 2-80C for several weeks.
Must remain stable
Simple Prophylaxis Primarily Staphylococci < 24 hours through the loading process.
Staph and Gram Must remain stable Directed Prophylaxis negative organisms 3-5 days through the loading process.
Must remain stable
through the loading
Narrow, only the process. Must also Therapeutics infecting organism 2-3 weeks remain stable at body temperature after
implantation.
Graft preservation. Allografts, being aseptically harvested and processed are
frequently treated with a solution containing one or more antibiotics to prevent incidental
contamination during recovery or processing leading to an infection in the recipient.
Tissue banks often mistakenly refer to this type of treatment as "cold sterilization".
However, due to the inherent limitations with regard to bacterial resistance, sterility
assurance levels (SAL) comparable with pharmaceutical or medical device standards
(typically less than 1 contaminated product in 1,000,000) are not attainable. Nonetheless,
for certain tissues this practice remains as an important step in preventing disease
transmission. Graft preservation requires that the antibiotic(s) cover as broad a spectrum
of organisms as is possible and that the drug remains stable in solution at refrigeration
temperatures for up to several weeks.
Simple prophylaxis. Here, an assumably sterile allograft is loaded with
sufficient antibiotic to prevent infection due to contamination introduced during surgery.
In this case, the allograft is independently indicated in the surgery and the antibiotic
loaded into the tissue simply replaces or improves upon systemic perioperative




63
chemoprophylaxis. This application requires an antibiotic with activity against the staphylococci, due to its prevalence with this type of infection, and a short release pattern. This method is employed sporadically by surgeons who reconstitute their allografts with an antibiotic solution prior to implantation. Currently, there are no allografts pre-loaded with antibiotics that are available commercially.
Directed prophylaxis. In this proposed application, an allograft containing an antibiotic would be used specifically for the prevention of infection in a wound where one would likely occur if no prophylaxis were administered. This setting would be analogous to the use of the PMMA- gentamicin beads in preventing infection of an open fracture. Here again, the tissue based delivery system would be advantageous because it would not require a second surgery for removal and lending structural and healing support to the fracture site. An appropriate antibiotic would need to provide effective coverage for 72 hours against the staphylococci and Gram negative organisms, as up to 50% of these types of infections are polymicrobial.
Therapeutics. Lastly, tissue based delivery systems could be used to treat an
existing, long term condition such as chronic osteomyelitis. Current treatment for certain types of osteomyelitis includes a primary operation for wound debridement and placement of PMMA-gentamicin beads followed by a second procedure for bead removal and placement of a bone allograft. Tissue based delivery systems could eliminate the need for the second operation by delivering a therapeutic dose of antibiotic carried within the prescribed allograft, which would then be followed by normal graft incorporation. Although the spectrum of activity would not need to be broad because the offending




64
organism would be known, the duration of therapy would need to be sustained for several weeks.
Characteristics of an Optimal Antimicrobial Agent
Spectrum of activity. The spectrum of the antibiotic should ideally be
appropriate for the target organism. When the indication is prophylaxis, the antibiotic should provide coverage over a broad spectrum of organisms and be definitively effective against the most commonly encountered organisms. Staphylococci, specifically S. aureus and S. epidermidis, are consistently isolated from post-operative orthopaedic wound infections (4). S. epidermidis, which is typically not associated with pathogenesis, is frequently found in iatrogenic infections involving orthopaedic implants. In addition, between 30 to 50% of infections involve polymicrobial contamination that includes a Gram negative organism (108). This significantly reduces the field of potential candidates, as many antimicrobials are effective against organisms of only one Gram's classification.
If the indication for the device is to treat an existing infection, the antibiotic need only be effective against the isolated organism. Here, a pre-operative culture could be obtained and the sensitivity of the organism confirmed prior to the implantation of the device. It is worth reiterating that sensitivity testing performed in the context of attainable antibiotic plasma concentrations may not accurately reflect the susceptibility of the organism to the antibiotic at the more relevant tissue concentrations. This is because most antibiotics are effective against an increasingly broad spectrum of organisms as the concentration of the agent is increased. However, toxicity prevents the systemic administration of the drug at these high levels and therefore organisms not effected by the antibiotic at attainable serum levels are reported to be resistant.




65
Adverse reactions. Antibiotic hypersensitivity is of particular concern when the drug is administered via an implantable device. Due to the release and distribution kinetics of the drug from an implantable delivery system, acute hypersensitivity may not be detectable during the surgical procedure. In addition, confirming the source of a hypersensitivity reaction to be the implanted antibiotic may prove difficult when confounded by other potential sources of hypersensitivity such as the infusion of blood products. Depending upon the severity, a hypersensitivity reaction could necessitate an emergent second procedure to remove the device. For these reasons, it is important to select an antibiotic that has both a low incidence and severity of hypersensitivity reactions. In addition, the implantation of such a device into patients with a known or suspected allergy to the incorporated drug is contraindicated.
Systemic toxicity is of little concern in the selection of an antibiotic for use with a tissue based delivery system, provided the antibiotic has been approved for uses systemically. This is due to the comparatively low plasma levels (1-10% of plasma levels attained with systemic administration) that are attained during the local delivery of an antibiotic (61, 82).
The potential for the antibiotic to cause local toxicity must be addressed in drug selection. When allograft tissue is employed, it is important that the antibiotic does not inhibit the incorporation of the graft. Specifically, graft incorporation could be hampered via two mechanisms. First, the antibiotic could alter the biological properties of the allograft resulting in sub-optimal incorporation (5, 64). Additionally the antibiotic could produce local cytotoxicity resulting in an inflammatory response that could adversely affect wound healing (64, 96, 104). For these reasons, antibiotics that are known to




66
adversely effect orthopaedic tissue should be excluded from use with a tissue based delivery system.
Stability. The stability of the drug needs to be considered from several
perspectives. First the drug must be stable through the loading and preservation phases of graft production. These phases involve dissolving the drug in a warm solution (35450C), freeze-thaw cycles and lyophilization. Harsh conditions are avoided during tissue preparation due to the delicate nature of the graft and its biological activity (7, 9, 15, 91). After the drug is loaded into the tissue, it must remain stable until use. Lyophilization or freezing are the preferred methods of preservation. Lastly, the stability of the drug in vivo must be considered if a prolonged release rate is desired as elevated temperatures or enzymatic breakdown of the drug are possible (48, 78, 79).
Physical and chemical considerations. The potency of the drug is important because the loading capacity of the graft is limited. The optimal drug would be bactericidal at very low concentrations ( minimum bactericidal concentration or MBC below 2 pg/ml). Drugs having minimum inhibitory or bactericidal concentrations greater than 50 ptg/ml are less useful because the graft can simply not hold an effective dose for release. As a rule, the longer drug release needs to be sustained, the more potent a drug is needed.
The hydrophilicity of the drug will also greatly effect its release characteristics
from the tissue. Using a porous matrix as a model for diffusional release into an aqueous environment, the release rate will increase as a function of solubility (84). Therefore, modulation of release kinetics from tissue based systems could be accomplished by selecting compounds with the appropriate aqueous solubility. More lipophilic




67
compounds would generate a sustained release pattern while a more hydrophilic compound would result in faster diffusion and release. For this same reason, the ionization state of the drug at physiologic pH should also be considered. Antimicrobial Candidates
Based on the above criteria, several antibiotics previously proposed for possible use in the local management of orthopaedic infection were evaluated for their compatibility in an allograft based drug delivery system.
Cephalosporins. Cefazolin (Ancef@), a first generation cephalosporin, is the most commonly used antibiotic for the prevention of post-operative orthopaedic infections (14, 16, 27, 103). It exhibits an adequate activity against Staphylococci and Streptococci. In addition, the MIC90 (minimum inhibitory concentration for 90% of the organisms typically encountered) is between 1-4 pg/ml making it sufficiently potent to be used in a tissue based delivery system (71). Cefazolin has limited activity against Gram negative organisms. Second and third generation cephalosporins have increasingly more activity against Gram negative organisms, however, their potency towards Gram positive organisms is diminished. For this reason any single cephalosporin would be a less than optimal choice. Although they posses little systemic toxicity and have been shown to not impair allograft incorporation, cephalosporins are associated with hypersensitivity reactions(14).
Penicillins. The penicillins are grouped into three classes. The naturally
occurring penicillins (penicillins G and V) have a narrow spectrum of activity against Gram positive organisms and Gram negative cocci. Like the cephalosporins, penicillins are dependent upon the integrity of their 1-lactam ring for antimicrobial activity. If this




68
ring is damaged either by acid or bacterial enzymes, penicilloic acid is produced, which lacks bactericidal activity. Because the majority of clinical isolates of S. aureus are found to produce 3-lactamase (penicillinase), the use of the naturally occurring penicillins has been limited in the surgical setting. Therefore, the penicillinase-resistant penicillins are more useful against P3-lactamase producing Staphylococcal infections. The broad-spectrum penicillins have been developed to be effective against both Gram positive and negative organisms, but this group, like the natural penicillins, generally lacks resistance to penicillinase.
Apart from spectrum, a more significant limitation to the use of the penicillins is their ability to induce hypersensitivity reactions that can range in severity from trivial to fatal. The overall incidence of penicillin induced hypersensitivity is as high as 8%, resulting in over 300 deaths per year.
Ciprofloxacin. Ciprofloxacin is synthetic fluoroquinolone that exhibits its
antibacterial effects by inhibiting DNA gyrase. This antibiotic has several characteristics that warrant its consideration for use in a tissue based system. First, it is a broad spectrum antibiotic that is effective against both staphylococci as well as many Gram negative rods. Additionally, ciprofloxacin is a potent antibiotic, effective at low concentrations (between 1-2 jtg/ml for most pathogens) (32, 40, 65). The drawback to ciprofloxacin is its association with permanent cartilage degeneration. Because of the potential proximity of the implant to a weight bearing chondral surface, high concentrations of ciprofloxacin may be attained in the joint space which could result in damage to the cartilage.




69
Tetracycline. The tetracyclines are effective against a very broad spectrum of organisms. This makes their use in prophylaxis, where a wide variety of organisms may be encountered, a tempting choice. However, tetracyclines avidly chelate to calcium which would prevent complete release of the drug from a bone delivery system. This binding could result in a subsequent lack of graft incorporation as tetracycline inhibits bone growth.
Vancomycin. Vancomycin is an ideal candidate for the irradication of
orthopaedic Staphylococcal infections as it is effective against almost all Gram positive organisms and resistance is rare. In addition, vancomycin is potent at low concentrations (usually less than 1 pg/mL) making it unlikely to cause adverse reactions (81, 85, 117). Unfortunately, vancomycin is completely ineffective against Gram negative organisms, thus prohibiting its use as the sole antibiotic in an orthopaedic delivery system unless the identity of the offending organism is known. Vancomycin does act synergistically with the aminoglycosides making its use in a polyantimicrobial system attractive (49, 76, 109, 118).
Gentamicin. The aminoglycosides possess several favorable characteristics for use in a tissue based delivery system. Gentamicin and the other aminoglycosides are effective against a wide variety of bacteria including Gram negative organisms and Staphylococci. This efficacy is obtained at very low concentrations (typically < 2 pg/ml) providing the opportunity for a sufficient reservoir of drug to be carried within the allograft for an extended release (60, 101, 114, 119).
Although toxicity is associated with systemic administration, gentamicin is safe when used in a local delivery scenario (34). This is due to the very low and often non-




70
detectable plasma levels that follow implantation. This has been demonstrated clinically by the wide use of gentamicin and tobramycin with PMMA(33, 34, 61, 82).
The physicochemical characteristics of the aminoglycosides are similarly
appealing, because gentamicin has been shown to be stable at elevated temperatures and through lyophilization (58, 74, 123). In addition, it is a hydrophilic compound that allows for its complete dissolution from the matrix in a timely manner (11). Perhaps the only drawback to gentamicin is that it may elute too quickly from the allograft, preventing its use in the treatment of existing osteomyelitis, which would require therapeutic levels to be maintained for 2-3 weeks.
Discussion
Local drug delivery for the treatment and prevention of orthopaedic infections has many advantages over systemic antibiotic therapy. These advantages have been realized clinically with the gaining popularity of the PMMA-gentamicin bead, which is now available commercially in Europe under the trade name Septopal (Merck, Darmstadt, Germany). However, the limitations associated with PMMA allow opportunities for improvement of these systems. Most obviously, the need for a second surgical procedure to remove the device increases health care costs and places the patient at further risk from surgical complications. The proposed allograft based delivery system would eliminate this need for a second procedure. Additionally, allograft bone can be used to lend structural support and promote new bone formation at the wound site. For these reasons, development of a tissue based delivery system could provide surgeons and their patients with a superior alternative to the PMMA system.




71
The use of allograft bone as a carrier for antimicrobial agents has previously been proposed, however limitations with tissue preparation and drug loading have prevented the development of a tissue based delivery system. Recent advances in allograft processing technology make it possible to effectively clean cortical bone and subsequently load it with a drug. Selection of the optimal antimicrobial agent is the next obvious step in the development of this system. Although there is considerable literature available treating the topic of antibiotic selection for use in the PMMA system, to date, no work has been found that evaluates potential antimicrobials for use in a tissue based system. Because significant differences exist in drug loading methodologies, and because there is the potential for the agent to effect either the allograft or the wound site in a way that would prevent graft incorporation, drug evaluation specifically pertaining to the tissue based system is necessary.
This review was conducted using the criteria generated by the four specific
indications previously described. Each of these indications had specific requirements pertaining to antimicrobial spectrum, duration of treatment, and drug stability. Due to regulatory constraints, the commercial development of an allograft based delivery system that used different drugs for each of the indications is likely not feasible. Therefore a list of requirements that satisfied the needs of all of the indications was developed (Table 33). These "master requirements" are a product of the most stringent criteria from each category. This allowed the antimicrobial candidates to be compared against one set of criteria versus four. In developing this master list, the requirements for graft preservation were excluded from consideration. This was done because the use of antibiotics in the




72
preservation of allografts is not considered to be a delivery system and does not carry the
regulatory burden associated with a drug delivery system (38).
The master set of requirements contains four categories; antimicrobial spectrum,
drug stability, potential for adverse reactions, and the potency of the drug. Less critical is
the duration of therapy required. Duration was omitted because there is the potential that
a sufficiently potent drug could have its release profile modified to meet the needs of
three separate indications. This could be accomplished by adding excipients that
restricted the release of drug from the bone. In this case, were the release rate could be
slowed, having a drug that has a high aqueous solubility and therefore has an inherently
fast release profile would be beneficial.
Table 3-3. Requirements of an antimicrobial drug for use in an allograft based drug delivery system. Included are the requirements for the individual indications and one master set of requirements that satisfies all of these needs.
Antimicrobial Adverse
Indication Spectrum Stability Reactions Potency Spectrum I Reactions
Must not result in Potency is not
Simple Primarily Must remain stable through hypersensitivity a significant Prophylaxis Staphylococci the loading process. or site toxicity, factor or site toxicity. factor
Staph and Gram Must not result in
Directed Staph and Gram Must remain stable through heresti Moderately negative hypersensitivity Prophylaxis gai the loading process. rsitiitypotent organisms or site toxicity.
Must remain stable through
Narrow, only the Must remain stable through Must not result in Potent
the loading process. Must also hypersensitivity Potent Therpeutcs nfecinghypersensitivity
Therapeutics infecting remain stable at body MIC<3 pg/ml
organism temperature after implantation. or site toxicity.
Must remain stable through
Staph and Gram Must remain stable through Must not result in
Master negative the loading process. Must also hypersensitivity Potent negativehypersensitivity Requirements gai remain stable at body or sitMIC<3 pg/ml
Reuirement organisms temperature after implantation. osietxcy
Each of the antibiotics was evaluated against this master set of criteria. The
antibiotics selected for evaluation have previously been suggested for use in local
delivery systems employing PMMA and represent the most obvious candidates for




73
review. Gentamicin meets the requirements of an antibiotic for use in an allograft based drug delivery system. Studies have demonstrated that gentamicin is sufficiently stable and highly potent (72). Although it is associated with toxicity when delivered systemically, plasma concentrations encountered with local delivery of this drug are well below toxic levels and are often not detectable. Additionally, gentamicin has not affected wound healing when used with the PMMA system clinically. The antimicrobial spectrum of gentamicin is adequate to cover both the relevant Gram positive and negative organisms typically encountered, however, the emergence of resistant strains to the aminoglycosides may warrant reevaluation of this drug in the future.
Cefazolin has the major disadvantage of being too narrow in antimicrobial
spectrum. Although the later generation of cephalosporins are effective against Gram negative organisms, they have diminished activity against Gram positive organisms. Hypersensitivity reactions associated with the penicillins exclude their consideration for used in an implantable drug delivery system. In addition, the growing number of penicillinase producing organisms would necessitate the inclusion of a (3-lactamase inhibitor such as clavulanic acid. This, coupled with the high MIC's that are required for some penicillins, makes this an overall poor candidate.
Tetracycline and ciprofloxacin are both associated with toxicity reactions specific for orthopaedic tissue. Ciprofloxacin is associated with permanent cartilage degeneration when given systemically to immature subjects. For this reason it is not indicated for use on children under the age of 17 or by women who are pregnant. Because local drug concentrations delivered via an implantable system can reach levels 100 times greater than when administered systemically, this degenerative effect may be more pronounced.




74
Likewise, tetracycline is known to inhibit bone growth. In addition, it binds avidly to calcium, which would in turn prevent its complete release from the device. This would likely result in poor graft incorporation. Therefore, tetracycline should be excluded from further consideration.
Vancomycin is an optimal choice for the treatment of infections caused by Gram positive organisms. Although there has been the emergence of vancomycin resistance with enterococci, in general most Gram positive organisms are sensitive. Unfortunately, vancomycin has no effect on Gram negative organisms. Synergism with the aminoglycosides makes it an attractive option if multiple antibiotics were used. However, the use of more than one antibiotic raises concerns about the synchronization of their release and the limitation on the absolute amount of drug that is possible to load into an allograft.
Conclusions
The results of this review indicate that gentamicin is currently the optimal
antibiotic for inclusion in an allograft based delivery system. This selection is based on its favorable characteristics with respect to antimicrobial activity, stability, hypersensitivity and local toxicity, potency, and aqueous solubility. In addition, gentamicin has a proven record of being safe and efficacious in the prevention and treatment of orthopaedic infections when used with the PMMA system.
The selection of an antimicrobial, based on a review of the literature, by no means guarantees its success in practice and therefore a significant amount of research remains to confirm this selection. Bacterial resistance as well as the possibility of the development of a better antimicrobial mandates that this selection be continually




75
reexamined. The next step in the development of an allograft based delivery system is to evaluate the characteristics of gentamicin with respect to drug loading and release patterns.




CHAPTER 4
OPTIMIZATION OF GENTAMICIN LOADING INTO CORTICAL BONE Introduction
The development of an allograft drug delivery system is contingent upon the successful loading of the drug into the allograft. Historically, this task has been a difficult one to accomplish (73). Conventional cortical bone allografts are far from ideal drug release devices. One major obstacle associated with bone as a drug delivery system is the amount of lipids and cellular elements that remain within the matrix even after processing. If fat remains trapped to any appreciable extent, the effective volume that is available for drug loading is reduced. This was demonstrated in previous work wherein the rehydration of cortical bone samples that had been cleaned of residual lipids and lyophilized were compared to bone specimens that were lyophilized without cleaning. The reconstitution of the two groups was then evaluated over time. The group that had been cleaned free of endogenous lipids had a significantly faster reconstitution time than did those samples that carries endogenous levels of lipids (Figure 4-1).
Lipids may also block or impede the uptake of a water-soluble drug into the matrix and conversely impede the release of a non-polar compound. Therefore, the removal of lipids and other cellular debris from the matrix of cortical bone prior to loading is a prerequisite to successful use as a drug carrier. Earlier studies demonstrated that effective cleaning procedures were possible without altering the beneficial properties of the tissue (15, 75, 95).
76




77
60
50 Lipid Free
+ Endogenous Lipids
40
0
.E 30
0 20
10
0
0 10 20 30 40 50 60 Time (min)
Figure 4-1. The effects of endogenous lipid content in cortical bone on rehydration. Error bars indicate standard deviation and n=14.




78
A major challenge that remains is the ability to completely load the matrix with the compound in question. Simply soaking bone the in a solution of the drug does not provide effective penetration of the entire matrix in a reasonable time. This can be visualized histologically by allowing a dye to penetrate into the internal matrices of bone. As is evident by examination of figure 4-2, complete penetration of cortical bone through passive diffusion is not complete at 8 hours. Figure 4-3 demonstrates a cortical bone pin that has been completely perfused with a tracing dye. Long drug loading times raise concerns over the stability of both the drug and tissue while in a hydrated state at temperatures above freezing. Because the volume within cortical bone is limited, incomplete loading would result in grafts with less than the optimal dose of a drugs. In addition, the variability in release profiles that would be associated with grafts that were incompletely loaded with drug would likely be too high to permit the device acceptance from regulatory authorities. Because the density of cortical bone as a material is rather uniform from donor to donor and because this density is directly related to the porosity of the matrix, it could be hypothesized that at the amount of drug within the bone approaches the maximum amount possible, this variation will decrease. Therefore, a process is needed to effectively load drug into intact cortical bone in a consistent and timely manner.




79
Figure 4-2. Micrograph of cortical bone taken from the cross-section of a human femur. The bone was allowed to soak in safranin for 8 hours. Note the partial staining (dark areas depicted with arrows) of the matrix with approximately half of the haversian canals remaining unstained.




80
Figure 4-3. Micrograph of cortical bone taken from the cross-section of a human femur. This bone was subjected to a pressure/vaccum perfusion process. Note the complete staining (dark areas) of the matrix with all of the haversian canals retaining dye.




81
The optimal drug loading process should contain several key features:
1. The process allows for the maximum amount of drug possible to be loaded into
the matrix.
2. The process results in tissue containing a consistent and known amount of drug.
3. The process results in a tissue that contains low residual moisture, as moisture is
associated with decreased tissue and drug stability.
4. The loading process does not alter the effectiveness of drug.
5. The process does not adversely effect the biomechanical properties of the tissue.
In previous work, it has been demonstrated that by thoroughly cleaning the tissue of cellular debris, removing the moisture in bone prior to drug loading, and through the use of negative pressure, the rate at which a solution is perfused, as well as the rate of perfusion through the internal matrix of cortical bone, can be increased relative to that seen with bone containing all of its endogenous debris (Figure 4-4). However, because of the limitations on space available for drug loading, it may sometimes be desirable to have a greater amount of drug inside the matrix than the solubility would permit if the drug were simply perfused into the matrix. Therefore a system is needed that can place a drug inside the tissue at a higher concentration than would be permitted based solely on solubility. To accomplish this, the following loading mechanism is proposed:
* Clean, dry bone is perfused with a saturated solution of the drug. During this first
step, the amount of drug loaded approaches the maximal amount as predicted by bone
volume and solubility.




82
* The tissue is then removed from the solution and lyophilized. This causes the drug to
crystallize on the walls of the internal matrix (Haversian and Volkmann's canals).
* The tissue would then be exposed to additional cycles of drug loading and
lyophilization to maximize the amount of crystallized drug added.
This model assumes that the perfusion of the matrix with drug will be complete to the extent allowable by the internal volume. It also assumes that drug will not be lost in the lyophilization process. This process is detailed in Figures 4-5a-e.
The stability of both the tissue and the drug need to be considered this process. Due to the use of both multiple freeze/thaw and drying cycles, it is possible to damage the drug and/or the tissue (9). This is particularly true for the protein components of bone that are susceptible to denaturation caused by phase separation (51, 52, 90, 98). Conversely, the deposition of drug crystals within the internal matrix of the bone may enhance the initial biomechanical properties of the tissue. A primary consideration in the selection of gentamicin is its well established stability through heating and drying cycles (11, 72). It is therefore hypothesized that the drug loading process as described above, will deleteriously affect neither the tissue nor the antibiotic.




83
80 70 60 S50 40
30
20 -- Vac
1No Vac
10
0
0 10 20 30 40 50 60 Time (min)
Figure 4-4. The effects of vacuum on cortical bone on rehydration. Error bars indicate standard deviation and n=14.




84
Empty matrix of cortical bone
a.
0 00 0 0 0 0 0 0 0 0
S0 0 0 0 0 0 0 0 0 0 0 0
* 0 0 0 0 0 0 0 0 0 0 0 0
Saturated drug solution
b.
Figure 4-5. A schematic of the proposed drug loading process. a) represents the internal matrix of clean cortical bone, b) represents that matrix filled with a saturated solution of drug, c) depicts crystal formation on the walls of the matrix following lyophilization, d) shows the same matrix being filled with a saturated drug solution for a second time, e) depicts the increase in crystal mass following a second lyophilization cycle.




85
Drug crystal formation i Exclusion of water by lyophilization C.
d.
Additional drug crystal Exclusion of water by
formation I ophilization e.




86
The specific aims of this work are (1) to determine whether multiple loading and lyophilization cycles will maximize the amount and minimize the variation of drug loaded into cortical bone, (2) to determine if the loading process is detrimental to the biomechanical properties of cortical bone and (3) to determine if the antimicrobial activity of gentamicin is altered during the drug loading process.
Materials and Methods
General Experimental Overview
Bone samples were machined into cylindrical segments of identical volume and surface area. The tissues were then loaded with antibiotic solutions using the aforementioned process for a total of three loading cycles. After each drying cycle, the mass gained as a result of drug addition was determined gravimetrically. In addition, residual moisture levels were determined for each lyophilization cycle to account for mass changes due to incomplete tissue dehydration. Deionized sterile water in place of the antibiotic solution was used as a negative control to ensure the mass did not change due to an artifact of the treatment. Following drug loading, the minimum inhibitory concentration of gentamicin was determined and compared to antibiotic that was not subjected to the loading process to demonstrate if there had been a change in potency. The bone was also subjected to mechanical testing to determine if the drug loading process had altered its ultimate compressive strength. Sample Preparation
To prepare the specimens, diaphyses of human femora and tibiae were initially prepared in standard fashion by removing any extraneous muscle, ligamentous




87
attachments, and the periosteum. Cortical bone segments were then cut with an oscillating bone saw from the diaphysis yielding specimens that were approximately 50 mm in length, 7 mm in cortical width, and 7 mm in depth. These were then machined on a lathe into cylinders measuring 5 mm in length and 5mm in diameter.
These pins were treated to remove blood elements and residual lipids. This step consisted of exposing bone segments to a 6% hydrogen peroxide bath at 400C for 30 minutes followed by an exhaustive lipid extraction with 400C acetone for 15 minutes per extraction. Following treatment, the samples were placed into individually labeled bottles, lyophilized, and held at room temperature until they were used for testing. Antimicrobial Preparation
Lyophilized, USP grade gentamicin sulfate powder was purchased from Sigma
Chemical and stored at 2-80C until use. The structure of gentamicin sulfate is depicted in Figure 4-6. On the day of the first drug loading cycle, a 200 mg/ml solution of gentamicin in sterile deionized water was prepared. The solution was mixed immediately prior to use to ensure all of the drug was in solution. The concentration of gentamicin in the solution was verified by fluorescence polarization immunoassay (FPIA) commercially available from Abbott Diagnostics.
Because gentamicin is completely soluble in water and because a saturated solution is needed for the subsequent loading cycles, acetone was used to limit the solubility of gentamicin. For the second and third drug loading cycles, a 25% v/v mixture of acetone to water was prepared to which gentamicin was added to saturation. The resulting solution from this mixture contained 75 mg/ml of gentamicin at 250C.




88
HO, CH3 CH3NHO ,
HOO Geniamicin_ R R'
NH HO 0 H2S C = CH CH C2 = H CH3
0 "" NH2 C2A = H H
S0
RNHCH H2N
I
R/
Figure 4-6. Molecular structure of gentamicin sulfate.




89
Drug Loading
The mass of the lyophilized cylinders was recorded and each was placed into
individual test tubes. To these test tubes, 1 ml of the gentamicin containing solution was added. Deionized water controls were also included to account for changes in mass or tissue strength that were not attributable to the presence of the drug. The test tubes were then placed into a vacuum chamber and a vacuum was applied. The cylinders were allowed to stabilize at a vacuum of <200 torr for 5 minutes ensuring complete outgassing of the matrix. After stabilization, the pressure in the chamber was returned to atmospheric level. The tissues were held at atmospheric pressure for 5 minutes and then a vacuum was reapplied. This vacuum/atmospheric pressure cycle was repeated a total of three times (Table 4-1). Following the loading process, the tissue samples were removed from the test tubes, wiped free of residual surface solution and placed into 30 cc glass containers for lyophilization. After lyophilization the mass of the samples was recorded. This process was repeated for a total of 2 loading cycles. After each cycle, samples were removed for destructive biomechanical and drug stability analysis. Table 4-1. Summary of the drug loading process used in this experiment. The loading parameters describe the materials and solutions used for each cycle. The loading process is also listed. This process was used on each of the drug loading cycles.
Loading Cycle Loading Parameters Loading Process Used for Each Starting Material Solution Cycle
Cycle 1 Lyophilized 200 mg/ml 1. Submerge in solution.
cortical bone gentamicin in H2O 2. Vacuum (200 torr) x 5 min. Randomized 75 mg/ml 3. Atmospheric pressure x 5 min. Cycle 2 specimens from gentamicin in 25% 4. Vacuum (200 torr) x 5 min.
Cycle 1 acetone 5. Atmospheric pressure x 5 min. Randomized 75 mg/ml 6. Vacuum (200 torr) x 5 min. Cycle 3 specimens from gentamicin in 25% 7. Atmospheric pressure x 5 min.
S Cycle 2 acetone 8. Lyophilize




90
Minimum Inhibitory Concentration Determination
A microdilution broth technique was used to determine the minimum inhibitory concentration (MIC) of gentamicin for the Staphylococcus aureus used in the study. Samples collected from the drug loading experiment were eluted in a flask containing sterile 0.9% saline. The elution step was expedited by placing the samples on a rotator set at 30 RPM. The eluent was then analyzed for antibiotic concentration and MIC determination.
A stock trypticase soy broth was prepared containing gentamicin at a
concentration of 100pg/ml. Dilutions were then made with antibiotic free broth that resulted in the following antibiotic concentrations; 20tg/ml, 14 tg/ml, 10 pg/mL, 8 pg/ml, 6 pg/ml, 5 pg/ml, 4 pg/ml, 3 pg/ml, 2.0 pg/ml, 1.0 pg/ml, 0.5 pg/ml, 0.25 Pg/ml. Inoculum was added to each of these tubes resulting in a bacterial concentration of 1-5 x 105 cfu/ml. The addition of this inoculum resulted in a two-fold reduction in antibiotic concentration. All samples were run in duplicate.
Using 96-well plates, the samples were incubated at 370C for 18 hours to allow for growth. Positive (no antibiotic) and negative (no inoculum) controls were included. Following incubation, the wells were examined for growth as indicated by turbidity. The MIC was determined by the lowest concentration of antibiotic that prevented growth. Requirements for accepting the findings were (1) all of the tubes with antibiotic concentrations greater than the MIC must be without growth, (2) all of the tubes with antibiotic concentrations less than the MIC must have growth, (3) the purity and identity of the organism in the first tube demonstrating growth from the MIC must be verified, (4)




91
the negative and positive controls must demonstrate no growth and growth respectively,
(5) the concentration of the original suspension must be 1-10 X 10s cfu/ml as determined by inoculum controls, and (6) replicates from the same sample must be in agreement meaning both samples must have the same inhibitory endpoint.
Biomechanical Analysis
Following treatment the ultimate failure load under axial compression was
determined. The method employed was adapted from the American Society for Testing and Materials (ASTM) test number D695-91 for determining the compressive strength of a material and was performed on an MTS 858 servohydraulic mechanical test apparatus. Samples were not rehydrated prior to testing. Load was applied under displacement control and applied at 25mm/minute in line with the axis of the bone (Figure 4-7). Ultimate strength was determined by the following equation: Pmax
Compression .
Where.: Pmax is the maximum load attained in Newtons
d is the diameter of the specimen
Statistical Analysis
Statistical analysis was performed to determine if there was a difference in the
amount of drug loaded into the samples per loading cycle and to determine if there was a change in strength per cycle. Analysis was performed using a commercially available software package (Statistica '99, Statsoft Inc.). A two-way ANOVA with repeated measureswas used to determine if a significant difference existed between the




92
AXIAL COMPRESSION
F
LEVELING MOUNT
SPECIMEN STEEL PLATE LOAD CELL
Figure 4-7. Loading scheme for the axial compression testing of cortical bone specimens.




93
treatment groups for the two outcome parameters (drug load and bone strength as a function of loading cycle). For all tests significance was defined as a p value less than
0.05.
Results
The first drug loading cycle resulted in a positive change in mass of the specimens of 1.2 mg/sample with a standard deviation of 0.5 mg (Figure 4-8). This was significantly greater than the untreated samples which on average lost 0.5mg/specimen (SD =0.5). This mass gain for the samples loaded with gentamicin continued for the second loading cycle where the cumulative mass gain increased to 1.7 mg/specimen (SD = 1.7) for the gentamicin samples while the water control specimens had a cumulative loss of 1.3 mg/specimen (SD =1.2). On the third and final loading cycle the gentamicin treated samples had a total mass gain of 3.2 mg/specimen (SD = 1.0) while the water control samples ended with a net loss of 1.8 mg/sample (SD = 0.8).
While the multiple loading process successfully loaded on average a significantly greater amount of gentamicin into the bone than a single loading cycle (3.2 mg/sample vs
1.2 mg/sample), the variation in the amount loaded also increased. The change in mass after the first loading cycle had a range from 0 mg/specimen to 3 mg/specimen. By the third loading cycle this range had increased to -3mg for the minimum to 6 mg for the maximum.
The change in mass provides an estimate of the amount of drug that is being loaded into the bone, but it does carry some inherent sources of error. First the




Full Text

PAGE 1

INTEGRATING BIOTECHNOLOGY AND PHARMACEUTICS: DEVELOPMENT OF THE BIOCOMPATIBLE ALLOGRAFT AS AN ORTHOPEDIC DRUG DELIVERY SYSTEM By CHARLES RANDAL MILLS 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 1999

PAGE 2

To organ and tissue donors and their families who in their time of sorrow give compassionately to relieve the suffering of others.

PAGE 3

ACKNOWLEDGMENTS I would like to extend my sincere gratitude to Dr. Gayle Brazeau for her continuous support throughout my graduate education. I would also like to thank Dr. Hochhaus Dr. Derendorf and Dr. Duggan for their time and effort they put forth on my committee. I would like to thank my family for their support Most importantly I offer my sincerest appreciation to Anna Radloff for always believing in me. I would like to thank Regeneration Technologies Inc. for its financial support without which this work would not have been possible. I would like to thank Michael Roberts for his efforts and insight on this and many other projects. Lastly I would like to thank Jamie Grooms for serving as my mentor and for encouraging me to follow my VJSIOn 111

PAGE 4

TABLE OF CONTENTS ACKNOWLEDGMENTS ... ........................... .... ........ . ........ ....................... ........ .... iii ABSTRACT .... .... ..... ......... .................. .... . ........ ...... ... ....... .... . ... ......... ...... v i l AN INTRODUCTION TO THE ALLOGRAFT AS A DRUG D ELIVERY SYSTEM ............... ... ........... . ........ .... . ........... .................. . ....... 1 Proposed Solution ..... ....... . .... ..... .... .... ....... ...... ....... .... .... ........ .... ....... .... ..... 8 Barriers to Development ...... ............. . .................................. .... ..... ... ... . ..... 15 Concentration .... ....... . ......... .................. .... ..... . .......... ............... .... . ....... .... 23 Accomplishing Tissue Sterilization ..... ........... ....... . ...... .... ...... . . ......... .... .... 26 Summary ... .......................... . ..... ..... ....... .... ............ ........ ...... .... ......... . ..... .... 34 2 EVALUATION OF ALLOGRAFT COMPOSITION FOR FACTORS THAT MITIGATE ANTIMICROBIAL CHEM OPROPH YLAXIS ............ ..... .............. ..... 35 Introduction ... .... ....... . ..... ...... . . ............ ... .......................... .... . ............. 35 Materials and Methods ...... ... .................... ............... ........ ..... .... . ..... ....... ...... 40 Results ......................... .... ....... ......... ...... ...... .... ........... ..... .... ............ .... . ......... ... 50 Di sc ussion . ...... ....... ............ ................ . ..................... ....... ...... . . .................... 53 Conclusions ... ...................... .......... .......... .... . .... .... ........ .... ......... ..... ............... 58 3 EVALUATION OF CHEMOTHERAPEUTIC AGENTS FOR USE IN A CORTICAL BONE ALLOGRAFT DRUG DELIVERY SYSTEM .... . ....... 59 Introduction ........ ....... ... . ................ ........... . ...... ...... .... .... ......... . . ......... 59 Discussion . .... ......... ......... ....... ..... ..... .... .... ... . ........ .... .... ..... .................. 70 Conclusions . ...... ....... ..... . .... . .......... .......... . .......... ... .... ............ ....... . ....... ... 74 4 OPTIMIZATION OF GENTAMICIN LOADING INTO CORTICAL BONE . ..... 76 Introduction ....... ....... . ..... ................................... ..... . .... .... .... . ..... .................. 76 Materials and Methods . ..... . ................... .... ............ ...... .... .... ..... . ..... ........ 86 Results ............ .... .......................... ..... . ..... . ..................... .... ........................ ..... 93 Discussion ..... ........ ... ......... ............... .................... ..... ...... ... ...... ........ ...... .... .... .... 96 Conclusions ................. . . . ............. . .... ............ ........ .... .... ............... ..... 100 IV

PAGE 5

5 IN VITRO PHARMACOKINETIC MODELING OF GENTAMICIN RELEASE FROM CORTICAL BONE ALLOGRAFTS ......... .... .. .......... ..... .......... 101 Introduction .... .. ........ .. .. .. ......................................... .. ...... .... .... ..... .. .. .. .......... ...... 101 Materials and Methods ...... ..... .. ..... .. .. .............. .... .. .... .... ...... ..... ................. .. ............ 106 Results ............. ... ............................ ..... .. ......... .............. ......... ....................... ...... 109 Discussion ............. ....... .. .. .. .. .. ... .. .......... .. .... .. ........... ..... ........... ........ .. .. ........ .. ..... 117 Conclusions ............... ...... .. .. .. .......... .......... ..... .. .. .. .. ....................... ........... ........ ..... 122 6 CONCLUSIONS AND A DISCUSSION ON AREAS OF POTENTIAL APPLICATION AND FUTURE WORK ..... .... ................................... .. ........ ...... .... 123 Conclusions ................. ...... .... ..... ...... .. .. .. ....... ... .. ..... .. ..... ......... ...... ..... ........ .......... 123 Future Work ..... .. .. ..... .... ..... ... .. ......... ..................... .. .. .. .. ......... ........ .... .... .. ... ......... 129 REFERENCES .... .. ......... ........................ .. .. ..... ........ ... .. .... ...... ..... .. ......... ............. ..... 131 BIOGRAPHICAL SKETCH .. .. .. ................... .............. .. .... ......... .. ....... ....... ..... .......... 141 V

PAGE 6

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy INTEGRATING BIOTECHNOLOGY AND PHARMACEUTICS : DEVELOPMENT OF THE BIOCOMPA TIBLE ALLOGRAFT AS AN ORTHOPEDIC DRUG DELIVERY SYSTEM By Charles Randal Mills December 1999 Chairman: Gayle Brazeau Major Department: Pharmaceutics The motivation behind the development of an allograft based drug delivery system (ABDDS) is to treat or prevent orthopaedic infections using a biocompatible local delivery system. Currently, the majority of local antimicrobial therapy is accomplished with the polymethylmethacrylate (PMMA)-gentamicin bead. However an additional surgical procedure is generally required to remove the device after the drug has been delivered. An ABDDS would have the major advantage over PMMA in that allografts are bioincorporable eliminating the need for a second surgical procedure to remove the device. The influence of allograft architecture and composition on the effectiveness of cefazolin is evaluated. The results indicate that cefazolin effectiveness may be hampered by architectural features, however residual lipids may positively influence cefazolin effectiveness. It is hypothesized that this is the result of endogenous lipids blocking the Vl

PAGE 7

uptake of contaminations into the allograft, making the bacteria accessible to higher concentration of the antibiotic. By changing the route of delivery from systemic to local, it is hypothesized that the effectiveness of antimicrobial therapy can be enhanced Gentamicin was selected for evaluation with an allograft based drug delivery system because of its appropriate antimicrobial spectrum low incidence of provoking hypersensitivity reactions, and potency. A novel drug loading procedure was evaluated for its ability to impregnate cortical bone segments with this drug. This set of experiments demonstrated that gentamicin could be loaded into cortical bone without altering its activity or lowering the strength of the bone. Lastly, the release profile of gentamicin from cortical bone was evaluated. An in vitro model was used to establish elution kinetics. The data suggest that the release profile was consistent with that predicted by a bi-exponential, diffusional based model. This data however, also demonstrats that the release profile is rapid and that modification of the device to attain a more sustained release is likely needed. Together this work establishes the foundation for the further development of an allograft based drug delivery system. It is hoped that this research will ultimately provide surgeons and their patients with a superior alternative to synthetic antibiotic impregnated cements for the treatment or prevention of orthopaedic infections Vll

PAGE 8

CHAPTER 1 AN INTRODUCTION TO THE ALLOGRAFT AS A DRUG DELIVERY SYSTEM Osteomyelitis and surrounding soft tissue infections are a significant concern for orthopaedic surgeons. Infection can arise from an acute traumatic injury such as a puncture wound, following invasive surgery or be secondary to a predisposing condition such as sickle cell anemia or diabetes. Of particular concern is the infection rate with some types of open fractures. Current treatment for this type of wound typically involves the use of radical debridement of the wound site, prophylactic intravenous antibiotic, and open-wound irrigation and suction (16, 34, 53, 103, 122) Even with this meticulous care, the risk of subsequent infection for certain types of fractures is unacceptability high at nearly 50% (82). The use of orthopaedic implants derived from human donor bone (allografts) also carries a significant risk of infection. Of the 500,000 surgical procedures using allograft that are performed each year, approximately 10% result in the development of an iatrogenic infection (31, 68, 80, 100). The severity of the infection can range from subclinical only being identified by culture to severe requiring graft removal and life saving antimicrobial chemotherapy. The majority of such infections are mild and can be resolved with prolonged antibiotic therapy, usually lasting approximately 6 weeks. It is estimated that the additional charges for medication laboratory testing, extended hospital

PAGE 9

2 stay and follow-up examinations increase the cost of treatment by 20 % (25 42 56 88). Increased recovery time also leads to the potential for lost income for the patient. In addition to the costs associated with iatrogenic allograft infection, there is increased morbidity and mortality (107). The use of extended antibiotic therapy in sufficient concentrations to eradicate the infection can lead to toxicity or hypersensitivity reactions (34) For example, high levels of gentamicin can lead to both renal toxicity and ototoxicity (28 43 66, 106, 113). Also, there is a substantial risk to the patient associated with any additional surgical procedure especiall y those requiring general anesthesia (44 69 89 115) Physician concern about infection often leads to the use of sub-optimal alternatives to allografts. These consist of materials that lack the ability to incorporate such as metallic implants (31 ). Long-term this may lead to a poorer surgical outcome than would have been realized if a tissue graft was utilized Other more radical procedures such as limb salvage for patients with osteosarcoma are deferred for limb amputation because of this increased risk ( 41 ). In the case of neoplasm, the concern o v er infection is sometimes heightened further by adju v ant chemotherapy that may ha v e immunosuppressive side effects Current Antimicrobial Treatments The rationale of chemoprophylaxis is to attain an effective concentration of antibiotic at the time of wound contamination. During this time the antibiotic reduces the quantity of introduced organisms to a number that can be cleared by the body's immune system without developing purulence. Four parameters need to be considered to achieve a

PAGE 10

3 desired therapeutic end result including timing, route of administration, duration of treatment, and selection of antibiotic Early studies by Burke demonstrated the importance of timing in the effectiveness of chemoprophylaxis (24). He administered a single dose of penicillin at various times before and after the inoculation of penicillin-sensitive Staphylococcus aureus in simulated surgical wounds of guinea pigs. Preor perioperative administration of antibiotic resulted in lesions histologically identical to lesions induced by attenuated controls. A three hour postponement in the administration of antibiotic resulted in lesions similar to those in animals not receiving antibiotics, thus study establishing the critical dependence of prophylactic efficacy on the timing of drug administration For effective antimicrobial prophylaxis, adequate drug concentrations must be present in the tissues at the onset and throughout the operative procedure (23). The majority of surgical protocols call for the initial dose to be administered parenterally immediately prior to the operation (71 ). The optimal situation is to have peak concentrations (Cmax) well above the minimum inhibitory concentrations for the relevant organisms at the time of wound contamination and maintain these therapeutic concentrations throughout the procedure. For this reason, the half-life and time to Cmax for the chosen drug must be factored into the decisions of when and how much antibiotic to administer during these procedures. A single dose of antibiotics before or during the procedure is recommended for prophylaxis in most surgical procedures (71, 102, 103, 108, 122). The logic behind a single dose regime is that contamination ends after the wound is closed. Studies comparing single-dose with multi-dose regimens have demonstrated no difference in

PAGE 11

4 rates of postoperative infection (93). However, these studies did not specifically investigate allograft surgery and it was recognized by those authors that for some surgical procedures, the number of doses of antimicrobials required for optimal prophylaxis has not been precisely defined. There are several considerations when selecting the appropriate antibiotic for prophylaxis. The optimal prophylactic antibiotic should ( 1) be effective against relevant microorganisms; (2) attain sufficient local tissue concentrations; (3) result in minimal side effects; ( 4) be cost effective, and ( 5) not be likely to select for virulent organisms. Since Staphylococcus aureus, a Gram-positive cocci, is isolated from 35-55% of orthopaedic wound infections, it would clearly need to fall within the spectrum of an appropriately selected antibiotic (26). In addition, polymicrobial infection is also common with a reported incidence of 50% in allograft wound infections ( 108). Cefazolin (Ancet) is the most commonly used antibiotic for prophylaxis in orthopaedic surgery (71). It is a first generation cephalosporin effective against gram positive organisms including most species of staphylococci. It also has a wide range of effectiveness against gram negative organisms. Of the first generation cephalosporins, it has the longest half-life at approximately 1.8 hours. Typically, 1-2 grams is given within 30 minutes prior to the incision and a second dose is administered if the lasts longer than 3hours(71, 122) The two main side effects associated with cefazolin are allergic reactions and antibiotic-associated colitis (AAC) (19). AAC is rarely a problem when the drug is given as a single intravenous prophylactic dose. High serum concentrations of cefazolin can

PAGE 12

5 cause seizures and renal function should be considered before a multiple dose regime is initiated (16) Limitations of Current Practice The obvious limitation of the current strategies in preventing post-operative complications is the persistence of an unacceptably high rate of infection. Although it is recognized that there is a chance of infection any time natural barriers to infection are compromised a goal for all surgical procedures should be to achieve infection rates no higher than those found with the cleanest procedures. In with some types of wounds this is currently not being approached Specifically, there is nearly a 50% rate of infection following the reduction of some open fractures (82). Allograft use is associated with a higher infection rate compared with similar synthetic prosthetics. One reason for the ineffectiveness of chemoprophylaxis on allograft tissue is the physical difference between synthetic and tissue implants Current recommendations for prophylactic antibiotic administration in orthopedic procedures do not differentiate between those procedures that utilize allograft tissue from those that involve synthetic implants The differences between the two types of can be hypothesized to cause prophylactic chemotherapy to be less effective with allografts thereby resulting in a higher incidence of infection There are several important features of allograft cortical bone that are dissimilar to metallic implants (Table 1-1 ). An examination of these specific features reveals that it may not be appropriate to consider allografts and synthetics equivalent for the purpose of chemoprophylaxis. An appreciation of the complexity of cortical bone is requisite to understand how these differences influence the effectiveness of systemic therapy.

PAGE 13

6 The porosity of allograft material makes it possible for contaminants that are absorbed into the matrix during graft reconstitution to not be immediately available to the antibiotic. Although studies have demonstrated that antibiotic concentrations reach prophylactic levels within bone tissue using the standard prophylactic protocol, this is perhaps an incorrect assumption in the graft itself (14, 16, 27, 99, 121). In fact it seems highly unlikely that this is possible considering the half-life of cefazolin and that only a single dose is typically administered. Although the recipient's bone is usually well vascularized, no vascular connections are made to the implanted bone by the surgeon. This leaves diffusion as the only mechanism for delivering antibiotic into the newly grafted material. Metallic implants, in contrast, are solid and non-porous. Therefore, metallic implants would not have the ability to harbor organisms in a protective manner from systemically administered antibiotics. Table 1-1. Compositional and architectural features that are unique to allografts and the h h"d t "hb ri 1ypot es1ze assoc1a 10n wit actena m ection. Feature Hypothesized Association with Bacterial Infection Porosity Bacteria sequestered deep within the internal matrix of the tissue will be protected from optimal concentrations of antibiotic and as well as the recipients immune system. Bacteria sequestered within complex surface features of the graft, such as threads will Architecture be protected from optimal concentrations of antibiotic and as well as the recipients immune system. Endogenous Lipids Protection of microorganisms from hydrophilic compounds such as cefazolin and gentamicin Similar to porosity, the surface features of allografts appear to have the ability to shelter microorganisms. When examined at close proximity it is evident that the surface of bone is irregular, containing peaks and valleys that increase surface area. It is possible that bacteria can reside in these valleys, protected from optimal antibiotic concentrations and phagocytosis by neutrophils and macrophages. Metallic implants are usually

PAGE 14

7 polished for a smooth surface, decreasing surface area and providing better presentation of a potential contaminant to the antibiotic ( 1, 67). A third unique feature of allografts is the presence of residual lipids. The amount of fat that remains on a graft is a function of the tissue type and of the extent of cleaning to which the graft was exposed. Cancellous or trabecular bone has a fat content of 7090% w / w, whereas the fat content of cortical bone is lower, 6-9% w/w. Fat content is not necessary for proper graft function and is removed to the extent possible during graft preparation ( 15). The amount of residual fat that remains in allografts is highly variable and depends upon the processing facility's methods and graft type. Despite efforts to remove fat, most bone grafts still carry a significant amount after processing. Bacteria can potentially partition into fat reservoirs carried on the allograft and remain protected from the antibiotic, particularly if only single dose is given. This would be augmented by the relative insolubility of water-soluble cephalosporins in lipids (27). Even if bacteria do not preferentially partition into fat, organisms that become surrounded by lipids carried on the graft would be protected for this reason. The Concept of Local Therapy Orthopaedic implants containing an antibiotic for either therapeutic or prophylactic delivery have been used since the 1970s ( 61 ). During that time there have been significant improvements to both the devices and their application. Currently, the majority of local antimicrobial therapy used in orthopaedic surgery is accomplished with the polymethylmethacrylate (PMMA)-gentamicin bead (11, 72). This system provides many advantages over the more conventional systemic antimicrobial therapy. With local delivery, systemic toxicity is avoided because serum drug concentrations are 10-100 times lower (33, 34, 53). Because the drug reservoir is located at the site requiring

PAGE 15

8 therapy, high tissue concentrations are achieved only in the location they are needed. Organisms reported to be resistant to a drug at systemically attainable plasma concentrations may be sensitive to the drug at concentrations found at the wound site via local delivery (61). For example if a bacterial strain was resistant to an antibiotic because it was able to product an enzyme that inactivated the drug, the addition of more drug would eventually saturate the enzyme allowing the accumulation of the antibiotic. This is in contrast to the scenario that accompanies intravenous therapy where drug concentrations in poorly perfused wound tissue are often much lower than plasma levels (53). However, targeted delivery systems are not without limitations. During preparation of the standard bead, dry powders of both PMMA and an antibiotic are mixed with water to form a cement. This reaction is very exothermic resulting in beads reaching temperatures in excess of 100 C during the drying process (32). For reasons pertaining to the stability of the drug, exposure to such high temperatures places significant constraints on the antibiotic that may be used. Additionally, PMMA beads serve no function other than being a non-resorbable carrier for the antibiotic. Because of this, additional surgery is generally required to remove the device after the drug has been delivered, adding the expense and risk of a second procedure (61). Proposed Solution As shown in Figures 1-1 a-c, an allograft based drug delivery system would have the major advantage over PMMA in that allografts are bioincorporable, meaning they will eventually integrate into and become part of the recipient's own tissues (15, 17, 22, 50). This eliminates the need for a second surgical procedure to remove the device, as

PAGE 16

9 Figure 1-1 a. Immediate postoperative radiograph of an extensive defect of the tibial diaphysis. The oval-shaped, dark area at the center is a void in the structure of the bone itself. This void is filled with a combination of demineralized bone matrix and cancellous chips (non-radioevident).

PAGE 17

10 Figure 1-1 b. Three week postoperative radiograph of the tibial diaphysis shown in 1-la. The size and extent of the v oid has decreased noticeably as new bone formation has occurred.

PAGE 18

11 Figure 1-1 c. Six week postoperative radio graph of the tibial diaphysis shown in 1-1 a-b. Remodeling has completely filled the void with new bone restoring the tibia to a structural unit.

PAGE 19

12 Figure 1-2. Postoperative radiograph of a threaded cortical dowel fusing L4 and L5 (top); An example of a threaded cortical dowel allograft (bottom).

PAGE 20

13 with the PMMA bead. In addition, allografts are indicated in orthopaedic procedures for reasons other than drug delivery. Allografts are frequently used to lend structural support or promote new bone formation (Figure 1-2). These added advantages do not come at the expense of previously mentioned benefits associated with local drug delivery systems. Table 1-2 summarizes and compares the advantages of an allograft based delivery system and the PMMA system over systemic drug delivery. Table 1-2. A comparison of the advantages and disadvantages associated with the PMMA and allograft based drug delivery systems. PMMA Based System Allograft Based System Advantae:es Disadvantae:es Advantae:es Disadvantae:es Locally high drug High temperatures Locally high drug Potential for disease concentrations ( > I 00 C) are required concentrations transmission for formulation Decreased risk of Second surgery is Decreased risk of Restrictions on the systemic side effects required for removal systemic side effects availability of donated following drug tissue expenditure Allows for primary No functional purposes Allows for primary Difficulty in drug wound closure other than drug delivery wound closure loading and material homogeneitv A voids compliance A voids compliance issues issues Incorporates into the recipients tissue eliminating the need for a second surgical procedure for removing the device Provides structural suooort at the graft site Promotes new bone formation Indications for Use There are several possible indications for an allograft based drug delivery system, raging from simple graft preservation to therapeutics. The following list briefly describes four major uses of antibiotics with allografts.

PAGE 21

14 1. Graft preservation. Allografts, being aseptically harvested and processed are frequently treated with a solution containing one or more antibiotics to prevent incidental contamination during recovery or processing leading to an infection in the recipient. 2. Simple prophylaxis An assumably sterile allograft could be loaded with sufficient antibiotic to prevent infection due to contamination introduced during surgery. The allograft is independently indicated in the surgery and the antibiotic loaded into the tissue simply replaces or improves upon systemic perioperative chemoprophylaxis. 3. Directed prophylaxis. An allograft containing an antibiotic would be used specifically for the prevention of infection in a wound where one would likely occur if no prophylaxis were administered. This setting would be analogous to the use of the PMMAgentamicin beads in preventing infection of an open fracture. The tissue based delivery system would have the two fold advantage of not requiring a second surgery for removal and lending structural and healing support to the fracture site. 4. Therapeutics. Tissue based delivery systems could be used to treat an existing condition such as chronic osteomyelitis. Current treatment for certain types of osteomyelitis includes a primary operation for wound debridement and placement of PMMA-gentamicin beads followed by a second procedure for bead removal and placement of a bone allograft. Tissue based delivery systems could eliminate the need for the second operation by delivering a therapeutic dose of antibiotic carried within the prescribed allograft, which would then be followed by normal graft incorporation.

PAGE 22

15 Barriers to Development The use of allograft bone as a carrier for antimicrobial agents has previously been proposed, however limitations with tissue preparation and drug loading have prevented the development of a tissue based delivery system(73). The major barrier in developing an allograft based drug delivery system is the potential for disease transmission that is associated with any biological material. Potential for Disease Transmission Annually, over 500,000 allografts are surgically implanted and the vast majority are done without serious complication (8, 15, 17, 31, 59, 111, 116). Viral transmission, predominantly HIV and HCV, is a rare event (1 in 1 million for HIV) (15, 59 111) However, because these diseases are incurable as of now, their statistically small risk of transmission remains a significant concern in the operations of tissue banks. There are two cases known to date where grafts from HIV infected individuals have resulted in disease transmission. The first incident was in 1984 when a femoral head removed from one patient during hip arthroplasty was subsequently used for spinal fusion in a second patient (7). This bone segment was neither tested for HIV (no test was licensed for HIV in 1984) nor processed by a tissue bank. In addition, the donor was not screened for symptoms or high risk behaviors associated with HIV, both of which were present (7, 36, 37). The second incident occurred in 1985 In this case, a young man involved in a robbery was fatally shot. Four solid organs were procured and all four recipients seroconverted prior to their death. In addition, 46 tissue grafts were produced, however only three transmissions have been reported. Each of the three grafts implicated with

PAGE 23

16 transmission were large, non-purged grafts with largely intact marrow reservoirs. These grafts were preserved via freezing and had no secondary sterilization. Interestingly, four grafts of this type were implanted, indicating one was a non-transmission (a non-purged femoral head). The other grafts, all of which were purged to some extent and freeze dried, did not result in disease transmission (7). In this second transmission, the only available test for HIV was the HTLV-111 antibody test that is far less sensitive than HIV tests used today (36, 39, 54, 63). At the time of implantation, the donor tested negative with this assay. However, subsequent retesting of banked white blood cell preparations by polymerase chain reaction (PCR), were positive for HIV proviral DNA. Based on this evidence, it is estimated that the donor had contracted HIV approximately 3 weeks prior to donation and was at that time, in the time period from when a person is infected to when the infection is detectable. When taken together, these two incidents indicate that HIV can be transmitted through implantation of infected allografts. Also evident is the potential reduction of this infectivity through tissue processing. Limiting Risk by Controlling the Supply To reduce the potential for disease transmission and increase the safety of the supply of tissues made available to implanting surgeons, the Food and Drug Administration (FDA) in 1993 mandated that blood from each donor be tested for the presence of antibodies to the human immunodeficiency virus type 1 and type 2 (HIV 1 and 2), hepatitis B virus (HBV) surface antigen, and antibodies to the hepatitis C virus (HCV) (35-38) Many of the recovery agencies and processing have elected to test for disease markers beyond those required by the FDA (Table 1-3).

PAGE 24

17 The use of this complex battery of tests greatly reduces the window period of an infection (Figure 1-1 ). A donor in the window period has contracted the disease and is infectious although will not test positive to markers for the disease. This type of result is often referred to as a false negative. As the disease progresses the donor will begin to test positive for the disease, a process termed seroconversion. Consequently these tests narrow the window period and serve as a more reliable indicator of donor thus reducing the chance of yielding a false negative. Table 1-3. An example serological profile run on potential tissue donors Donors fh k ldd t ft 1 testmg positive to an I one o t ese mar ers are exc u e pnor o irra re ease. Test Name Marker Detected HIV-DNA by PCR Human Immunodeficiency Virus Proviral DNA HIV Y2 Ab Human Immunodeficiency Virus 1 and 2 Antibody HBsAg Hepatitis B Surface Antigen HBcAb Hepatitis B Virus Core Antibody HCV Hepatitis C Virus Antibody HTLV I & II Antibodies Human T-Cell Lymphotropic Virus Type I & II RPR Antibodies to Treponemal palladium In addition to serological testing, each potential donor is given a physical examination and a comprehensive medical and social history is obtained to exclude donors with "high risk" factors for infectious diseases (15, 59, 111 ). Lastly, bacterial and fungal testing is done on processed grafts prior to their release for implantation The potential for transmission of HIV through allograft implantation was more likely during in the earlier stages of the epidemic, before reliable screening assays were available and the importance of excluding donors with high-risk symptoms and behaviors

PAGE 25

10 -C: 8 :::I 0 E 6 ci, > ;:; 4 RI i ci, a::: 2 0 0 Window I I ..... '~ ... 2 ,' I / \ \ \:' .. \ 18 4 6 Weeks After Exposure --HIV-DNA ----HIV Ag HIV Ab --------8 10 Figure 1-3. Idealized serological profile following HIV infection. The solid horizontal line indicates the theoretical level of detection (sensitivity). The bracket denotes the window period for HIV when tested with the PCR assay. Note that the window period is longer with the HIV antibody te st.

PAGE 26

19 was realized. The two cases discussed earlier would likely not happen today. Both of the donors would have been excluded if PCR testing for HIV was available and performed. In addition the first donor would have been excluded based on his medical and social history, which revealed past IV drug abuse and generalized lymphadenopathy. It is also speculated that the second donor who tested negative for antibodies to HTLV-III would have tested positive for antibodies to HIVI with the current more sensitive testing assays. Controlling the Risk through Processing Although donor screening has led to a dramatic increase in tissue safety, it has inherent limitations. For one, it is not practical nor possible to test for every potential pathogen In addition, the emergence of new, yet unrecognized pathogens is a certainty due to increased resistance to antibiotics. For these reasons, it is prudent to incorporate steps in the production of a tissue product that address the possibility of a contaminated starting material or the introduction of an adventitious contaminant. Limitations to Bone Sterilization Bone does not lend itself to sterilization due to a number of factors First, the number of potential contaminants is high. Most materials that are sterilized such as plastics or metals, do not have a substantial reservoir for viral contamination. Human tissue obviously does carry the potential for significant viral contamination. This is an important factor as viruses are highly resistant to some sterilization processes that are effective against bacteria such as irradiation (6, 45, 68, 91, 111) Another consideration is the delicate nature of the proteins carried within the bone. Bone morphogenic proteins or BMP's that enhance new bone formation are necessary for proper graft. It is recognized that several common sterilization processes alter or eliminate the beneficial

PAGE 27

20 properties of these proteins ( 64, 110). Decreasing strength and sufficient penetration are the other considerations for bone sterilization. The biomechanical strength of tissue can be effected by heat and irradiation (9, 68, 91, 97). Tissue penetration is also a challenge for any sterilization process that employs a gas or liquid germicide. Gamma irradiation. The virucidal and bactericidal effects of gamma irradiation are created via two mechanisms (10). The primary mechanism is direct alteration of nucleic acids leading to genome dysfunction and destruction. A secondary mechanism is the generation of free radicals, primarily from liquid water, contributing to the sterilizing abilities of gamma irradiation. This secondary effect is not realized, however, when an article has been lyophilized or is frozen at the time of irradiation. It is in the frozen and freeze-dried states that the vast majority of tissue is presented for sterilization by irradiation (9). The differences in efficacy as a function of physical state is well characterized by the plasma component industry (55). In the frozen or freeze-dried state, the virucidal effects of gamma irradiation are directly due to genomic destruction and constitutes a first order process with respect to dose. In the liquid state, however, the formation of free radicals contributed to the virucidal capacity of the treatment. This generation of free radicals is suggested to cause the deleterious effects to the tissue. Taken together, items presented for gamma irradiation sterilization in the frozen or freeze dried state will require significantly higher doses to achieve the same effect as would be realized if the item were in the liquid, hydrated state. In tissue banking, most of the research has centered on the inactivation of HIV. HIV is a retrovirus that is fairly resistant to destruction by gamma irradiation ( 18).

PAGE 28

21 Although relatively low doses of gamma irradiation (:;:j lMrad) are capable of killing most classes of microorganisms, studies have shown that greater than 3.0 Mrad is required to eliminate the chance of transmission of HIV from infected tissue (55, 91). Despite the presence of published data, most tissue banks that use gamma irradiation as a means of secondary sterilization expose the tissue to a dose of only 1.5 2.5 Mrad. In choosing the irradiation dose, tissue banks must consider the effects the treatment will have on the biomechanical properties of the tissue. These have been demonstrated to be dose dependent. At doses below 2.5 Mrad, the biomechanical effects, defined as a reduction in graft integrity as measured by either axial compression or tensile strength, to most tissue appear to be small. Unfortunately, this is a sub-lethal dose for HIV and other viruses At doses greater than 3.0 Mrad, tissues can lose a significant amount of strength, ranging from 25% to 75% as compared to untreated controls, depending upon the conditions under which it was irradiated This data suggests that the dose required to assure the complete inactivation of contaminating viruses is above the dose at which gamma irradiation starts to produce detrimental effects to the grafts biomechanical properties. Ethylene oxide. The use of ethylene oxide (EtO) persists within the tissue banking industry, despite rapidly accumulating data that suggests EtO not only damages tissue structure and function, but also poses environmental and health risks (110). The primary mechanism by which EtO kills viruses and bacteria is through alkylation of purine and pyrimidine moieties leading to DNA and RNA dysfunction. In addition, secondary mechanisms include enzyme inactivation through alkylation of amino acid

PAGE 29

22 residues. These effects make EtO an effective sterilant against bacteria, spores molds, yeasts and viruses including HIV ( 10). Ethylene oxide and its by-products, ethylene glycol an ethylene chlorohydrin are mutagens and are considered to be cytotoxic. For this reason the FDA imposed a limit of 250 ppm as the amount of residual EtO that could remain on a medical device. Recent studies have shown that even in very low concentration (~25 ppm) EtO was toxic to fibroblasts (97) In addition, bone remodeling studies have demonstrated that EtO treatment reduced bone in-growth by 68%, despite no detectable residual EtO or its by products ( < 20 ppm). This inhibition of remodeling is speculated to be due to alkylation of the amino acids in bone morphogenic proteins and other osteoinductive messengers. In addition to inhibiting osteoinduction ( e .g. the ability of allograft bone to induce de novo bone formation at the site of implantation) EtO sterilization is implicated in an even more serious side effect. Ethylene oxide sterilized grafts have been associated with an immunologically induced synovitis following graft implantation (57). In this reaction patients develop a persistent synovial effusion se v eral months to years following graft implantation. The syno v ial fluid changes to an orange or brown color and contains collagenous debris neutrophils and lymphocytes. Patients tend not to develop elevated peripheral WBC and infection of the graft site is not evident. This condition is refractory to common treatments for inflammation (non-weightbearing anti-inflammatory agents aspiration) and is only resolved by removal of the graft. In one study by Jackson and coworkers detectable levels of eth y lene chlorohydrin were found upon removal of a graft fourteen months post-implantation (57).

PAGE 30

23 In addition to the detrimental effects of EtO on tissue its use has become highly regulated by the Environmental Protection Agency (EPA) and the Occupational Safety and Health Administration (OSHA). The use of EtO in hospitals and industry as a method of sterilizing heat labile instruments is being phased out and replaced with other types of non-toxic cold sterilizers a trend which will most likely extend into the tissue banking industry Purging and soaking. Purging grafts of their cellular components and soaking them in antimicrobial solutions comprise the most common pathogen reducing steps emplo y ed in allograft production today (15). Most grafts produced within the United States today are processed with at least some form of marrow element purge and antimicrobial soak except when prohibited by the nature of the graft such as those grafts preserved to maintain cell viability. Typical treatments include warm water lavage hydrogen peroxide, isopropyl alcohol iodine and antibiotic soaks (Table 1-4). Table 1-4 Chemicals used in the processing of human tissue the concentrations used, and their intended action. Chemical Treatment Concentration Effect Warm water lavage NA removes blood and lipids H202 3 % removes blood lipids, bactericide virucide Isopropyl alcohol 70 % bactericide virucide Iodine 10% bactericide, virucide The first and most basic mechanism by which these steps accomplish their intended purpose is via simple dilution of the microbes or viruses The in vivo viremia associated with HIV is of greatest quantity in early infection just prior to antibody seroconversion (7 15, 111). For tissue routinely recovered and processed for graft

PAGE 31

24 production the bone marrow represents the greatest reservoir for the virus. By removing the marrow elements from the graft, the number of infectious units is substantially reduced. As mentioned the only grafts known to have transmitted HIV in humans have been those which were not purged of their marrow elements prior to implantation In fact, there have even been cases where patients have received thoroughly lavaged kidneys from donors who were later found to be infected with HIV and did not contract the disease (125) In addition to reducing the potential viral load in a graft purging also reduces the amount of HLA expression a graft carries (29 47). By removing the majority of the marrow from bone grafts the antigenicity of the graft and therefore the potential for sensitization is reduced. Alcohols ( ethyl and isopropyl) are among the most common solutions to be used in the production of allografts today. Alcohols are effective at reducing the viability of a broad spectrum of bacteria and are very effective against enveloped v iruses such as HIV ( 10). Among the alcohols that are completely miscible with water in all proportions propyl alcohols are the strongest disinfectants. Although these alcohols are directly cytotoxic their solubility in water and volatility allows for their removal to such an extent that final concentrations are well below their ability to produce toxic effects Hydrogen peroxide (H202 ) is frequently used as an antimicrobial and defatting agent. In vivo it is an active germicide found in the saliva milk, and phagocytes, and is found in other tissues as a result of metabolism. There are at least two mechanisms invol v ed with H202 antimicrobial activity The first is the oxidation of chloride in bacteria to form hypochlorite a well-characterized germicide, and water. The second is

PAGE 32

25 the generation of free hydroxyl radicals the strongest oxidant known, which then attacks membrane lipids nucleic acids, and proteins leading to pathogen inactivation (13). In addition to its germicidal effects, the vigorous release of oxygen gas from H202 lends to its ability to clean and debrided fatty and bloody grafts (7, 15, 95) Iodine containing compounds are used less frequently in tissue processing. This is primarily due to their tendency to discolor grafts. This unpleasant side effect can be averted with the addition of a decolorizing step with ascorbic acid, wherein molecular iodine is reduced into the colorless iodide ion. Only the I2 species is believed to be active as a germicide where it reacts with several types of functional groups on proteins, nucleic acids, and unsaturated fatty acids. Although reactions to iodine have been reported since the 1800's they are typically associated with either massive amounts of iodine exposure or intake over an extended period (70). The major deficiency with these types of chemical treatments is their lack of thorough penetration into the tissue. Because they are in liquid form, they can only function as surface inactivators. The matrix of bone and other soft tissues is highly complex and therefore does not lend itself to complete penetration by solutions. Several tissue banks have attempted to overcome these complexities with solvents, detergents, and mechanical mechanisms aimed at increasing penetration. These augmentations have led to improved success on a tissue specific basis however large grafts continue to present a challenge for the industry. The Optimal Sterilizing Process The ideal model for allograft sterilization would incorporate several key properties in achieving a safe and effective tissue for transplantation. Below are listed

PAGE 33

26 these characteristics. Table 1-5 compares these characteristics to the aforementioned sterilization processes. 1. The process would need to be effective at removing and/or inactivating a wide range of bacterial and viral pathogens contained on and in the tissue. 2. This process would not result in the graft being toxic to the recipient due to residual chemicals 3. The process would not significantly reduce the biomechanical strength of the graft, which would lead to graft collapse. 4 The process would not adversely alter bone morphogenic protein, a protein endogenous to bone which is responsible for induction of new bone formation. 5. The process would be robust, meaning small changes to the system or graft would not alter the ability of the process to yield a sterile product. 6. The latter portion of the process would be executed in the graft's final container, eliminating the possibility of adventitious contamination of the graft. Accomplishing Tissue Sterilization Because of the risk of disease transmission, an essential first step in developing an allograft based drug delivery system is to effectively sterilize the starting material without altering the beneficial properties of tissue. Therefore, prior to the specifically investigating allografts as drug delivery systems, the following basic research was conducted to establish an effective way to sterilize human bone and therefore remove this specific barrier to further development of the project.

PAGE 34

27 Table 1-5. A summary of the sterilization processes discussed in this chapter compared t th t t d fi th f l t T f agams e en ena presen e or e op 1ma s en 1za 10n 1 Jrocess. Methodology Sterilant Toxicity Biomechanical BMP Robustness Final Container Gamma Bacteria Good Irradiation Viruses No Poor Poor Moderate Yes Moderate Ethylene Good Yes Moderate Poor Moderate Yes Oxide Purging and Weak to Moderate No Good Good Poor Possible Soaking Enhancing the potency of hydrogen peroxide. As most sterilization processes are detrimental to the structural and/or biological properties of allograft bone it was therefore necessary to develop a sterilization process that minimizes these. Hydrogen peroxide has been used in tissue banking since its formalization in the 1950's and almost all bone allografts are treated with this solution at various concentrations ranging from 1 -35 % (15, 87, 95). Through the extensive use of hydrogen peroxide, this compound has demonstrated its compatibility with bone allografts, yet its effectiveness as a sterilant had not been realized It has been suggested that ultrasonic energy enhances the bactericidal and sporicidal effects of hydrogen peroxide (10). In this study, a reduction in the D-value for the Bacillus sterothermophilus ( 106 ) spore was calculated for samples treated with 6% H202 in the presence and absence of ultrasonic energy. This spore was chosen due to its well characterized and accepted resistance to peroxide sterilization (62). Samples were treated with 2 ml of 6% H202 at 45C over a given range of time and the reaction was stopped with the addition of sterile water and transferred to trypticase soy broth for culture at 56C. All samples were performed in triplicate.

PAGE 35

28 The D-value obtained for the samples run in the presence of ultrasonic energy was 0.83 1.66 minutes (Table 1-6) This compared favorably to the D-value obtained for the samples run in the absence of ultrasonic energy ( > 10 min). This reduction in the time required for the inactivation of spores may allow for a practical method of sterilizing allografts without adversely effecting their desired attributes Table 1-6. Comparison of spore inactivation with 6% hydrogen peroxide in the presence and absence of ultrasonic energy. Positive(+) results were identified by turbidity of the media and confirmed by subculturing the broth to solid media. Negative(-) results were determined by the media retaining clarity over the seven days of incubation. Assay sensitivity was determined to be < 5 organisms. Treatment Treatment Time (min) D-0 5 10 15 20 30 40 50 60 value Sonication + + + + ------------<1.6 No + + + + + + + + + + + + + + + + + + + + + + + + + + + >10 sonication Effects of residual lipids on the activity of hydrogen peroxide. This study examined the potential for residual lipids to reduce the effectiveness of hydrogen peroxide at inactivating Bacillus sterothermophillus spores. Whole femora and tibiae were surgically removed from human cadaveric bone donors and debrided of extraneous soft tissue The bone tissue was then ground yielding a bone slurry with the consistency of an oily paste A section of this bone paste was removed and thoroughly cleaned of residual fat content using warm ( 45C) acetone. The cleaned bone slurry was mixed with the untreated bone paste in various weights to yield samples containing 0, 10, 30, and 60% residual fat. All samples were verified using an exhaustive volatile extraction with gravimetric analysis. One gram of each sample was added to test tubes containing a 106 inoculum of spores and was treated with 2 ml of 6% hydrogen peroxide ( 40C) in the

PAGE 36

29 presence of ultrasonic energy (45Khz) for multiple time points Each time point was run in triplicate. The reaction was stopped for a given time point by the addition of 20 ml of sterile water and the inoculum was transferred to trypticase soy broth for a seven day incubation at 56C for seven days. Controls included sterile water (negative control), inoculated water (positive control), and H202 without bone tissue. The results of the study indicate that lipids prolong the contact time required for the complete inactivation of B. sterothermophilius spores (Table 1-7). The data generated from this study supports the hypothesis that removing endogenous lipids from cortical bone will increase sterilization efficiency By lowering the contact time required for sterilization, the potential adverse effects of the sterilant (reduction in tissue strength) may be minimized Although this assay confirmed the ability of hydrogen peroxide toinactivate spores in a suspension of bone material, the ability to accomplish this in a more relevant context, using an intact bone model was needed. Table 1-7. Approximated D-values for B sterothermophilus as a function ofresidual fat content remaining in a homogenized bone slurry, when sterilized in a 6% hydrogen peroxide solution at 42C in the presence of ultrasonic energy. Positive(+) results were identified by turbidity of the media and confirmed by subculturing the broth to solid media Negative (-) results were determined by the media retaining clarity over the seven d f b ays o mcu at10n. Treatment Time (min) D Treatment value 0 5 1 0 15 20 3 0 4 0 5 0 60 N eg --------------------------N A Co ntr o l Po s Co ntrol + + + + + + + + + + + + + + + + + + + + + + + + + + + >10 No bone + + + + + -----------------1.66 0 % fat + + + + + + + + -----------------2.5 10% fat + + + + + + + + + + + + ------------3.33 3 0% fat + + + + + + + + + + + ----------3.3 3 6 0 % fat + + + + + + + + + + + + + + + + + ----------6 66

PAGE 37

30 A model for sterilization efficacy in cortical bone. Definitively demonstrating the efficacy of a liquid sterilization process for human cortical bone has historically been difficult. In this experiment the use of a machined segment of human cortical bone carrying a B. st e roth e rmophilus (106 ) biological indicator was evaluated for its potential uses to support claims of allograft sterility (Figure 1-2). The device was prepared by cutting a cortical segment from the anterior ridge of the tibia in a cadaveric bone donor. This segment represents the thickest portion of cortical bone encountered in the body and is thus the most difficult to penetrate and sterilize. A cylindrical hole was machined into the end of the bone, longitudinal to the axis A second segment of cortical bone was machined into a cylindrical pin with a diameter slightly larger than that of the hole. A partial slit was cut into the pin allowing a biological indicator to be placed within. The pin was then forced under compression into the machined hole and exposed to the sterilization process. A control was also run using only sterile water to evaluate if the spores were appreciably being washed off the strip. In addition a tracing dye was used to evaluate the path of the liquid through the device. The results from the controls indicate that the extent of washout that occurred was minimal and did not significantly effect the introduced bioburden. Recovery studies showed that on average 8 x 105 spores were recovered using the saline control. The samples exposed to the sterilization process did not demonstrate growth after incubation for seven days in TSB indicating process efficacy The sensitivity of the assay was < 5 orgamsms.

PAGE 38

Co rtic a l b o n e pin 31 S lit for Bl Co rti ca l bo n e bl ock Figure 1-4. Model for testing the efficacy of a liquid sterilization process for cortical bone.

PAGE 39

32 Effects of sterilization on allograft biomechanics. The purpose of this study was to identify the effects of preservation and sterilization processes on the strength of cortical bone. This work is essential in determining what treatments are acceptable for the graft to be exposed to during the processing and drug loading steps of production. Treatments that significantly reduce strength must be avoided in the graft preparation/drug loading process. Femora and tibiae were isolated from 18 different human cadaveric donors and machined with a lathe into 203 pins that were 4.0mm in diameter and 10mm in length. The pins were then exposed to treatment that may be used in the graft preparation or drug loading process. Following treatment the ultimate failure load under axial compression was determined. Axial compression testing was adapted from ASTM D695-91 and performed on an MTS 858 (Eden Prairie, MN) servohydraulic mechanical test apparatus. The results demonstrate that pressure assisted hydrogen peroxide perfusion did not reduce the compressive strength of the cortical bone pins (Figure 1-3) Gamma irradiation did significantly reduce the strength of the tissue and therefore an alternative method should be sought for terminal sterilization of the graft. Lyophilization, in contrary to expectation, significantly increased the axial strength of the tissue. This is a promising result as lyophilization is hypothesized to be a critical component to maximize the amount of drug that can be loaded into bone. Summary of Tissue Sterilization These sets of experiments demonstrate that the effect of mild germicides can be enhanced to the level of sterilization through the addition of ultrasonic energy. The effect of residual lipid content was also characterized. These two studies demonstrate that

PAGE 40

350 300 250 a. '1s 200 u. 1ij ] 150 100 50 0+---33 control ISi Lyoph ilization CJ Irrad iation l:IPAHP ,,.,,..,.,.,.,.,,..,.,.,, .,.,.,.,,,..,,,..,.,.,.,.,. ,.,.,.,.,.,.,.,,.,,.,.,,. ,..,,,._,,.,.,.,,.,,.,,./,. ... ,.,.,,,,.,,., .,,,, ,., .,.,.,.,.,.,.,.,.,., /,u.,.,,u.U//' ,,,,._,_,,,..,_,_,_,,,_,, ,,.,..,..,.,.,.,.,.,.,,,,.,.,,,.,,,,;,.,-;,,.,.,.,,,. "l----+----f.,,.,.,, .,.,,.,.,,,..,_,,,_;_,., .......... .,, .,.,, T r e atme n t .,.,..,.,.,,..,.,.,,,,,,._,/.,.,.,.,.,.,.,.,..,/_,.,_ .,,._,,,,.,.,,.,,.,,.,,.,,.,,.,,..,,.,.,.,.,,.,,.,.,.,.,.,,./' .,.,.,,.,.,..,,.,.,.,.,., .,,._,,.,.,..,.,,,,.,,,,, ,'///////////,1///,',,.,.,.,,,,,,-.,_ '//"//////////,1".1",1"1'.I',',.,,,,,.,. ,..,,,..,,..,,..,,.,,..,,.,,..,,.,.,.,..,,.,.,.,..,, ///////h'.'l'.',',1',/,1'////,l'l,',I'., ;;;/l'h',l'l'h"1";'/',',','/,'1','I//'/ ,..,.,,..,,.,.,..,,..,,., .,, .,,,,,.,,..,.,,,,., .. ,, .,.,_,.,,..,.,.,.,/.,.,.,..,..,.,,,.,,..,.,,.,,.,,,,,,,, .,,.,,.,,.,.,,.,.,.,.,,.,,.,.,.,.,,.,,.,,.,,.,,.,,,.,.,.,,.,.,. /./////.//////////,N//,11',".",I', .,_,,.,.,,,,,,,,,,,,.,,.,,,.,,.,,;.,;;;;-;;,.-, ,.,.,.,.,.,.,..,,.,.,.,.,.,.,,.,..,,.,.,.,.,.,.,.,._ ,.,.,..,,.,.,..,.,.,.,.,.,.,.,.,.,.,.,.,.,.,.,.,., .,.,.,.,.,,.,., .,.,.,.,.,.,.,.,.,.,.,.,.,,..,.,., ,.,..,.,,.,.,.,,,',','/////,'////,,,,, Figure 1-5. Treatment groups and mean ultimate strength during axial compression testing. Control -a group consisting of no preservation or sterilization treatments was i ncluded (n = 51) Lyophilization freeze drying to reduced the residual moisture content of the graft to below 2 % (n=51). Gamma irradiation -a sterilizing dose of 3.5 Mrad (n=50). PAHP Pressure assisted hydrogen peroxide treatment employed expo s ing the tissue to a 6 % solution of hydrogen peroxide at 40 C for 30 minute under oscillating pressure and ultrasonic energy (n=51 ). Error bars indicate 2 X standard error.

PAGE 41

34 sterilization in an organic environment is possible with hydrogen peroxide when ultrasonic energy is employed and that there is a direct relationship between the contact time required for sterilization and the lipid content of the graft. Strength evaluation confirmed that this treatment did not significantly alter the biomechanical properties of the tissue Lastly, a model was developed that confirmed the effectiveness of the process to kill spores deep within matrix of the tissue. This is significant because it answers two questions definitively. First, the sterilant is able to penetrate the tissue; secondly, the peroxide reaches the inner portions of the tissue in concentrations sufficiently high to result in sterilization within a reasonable time Summary The aims of this project are (1) to identify factors that may predispose allograft bone implants to post-operative infection; (2) to identify the optimal antimicrobial for further investigation within an allograft based system (3) to evaluate a potential drug loading procedure, and ( 4) to characterize the release profile of an antibiotic from cortical bone. The rationale for this approach is that allograft features such as surface texture and fat content provide a mechanism for protecting microorganisms from chemoprophylaxis and that by addressing these issues, a more infection resistant graft can be developed. In addition, by changing the route of administration from systemic to local (graft delivered), effective concentrations can be attained without realizing systemic side effects Through a drug loading procedure whereby a solid form of the drug is contained throughout the inner matrix of the graft, sustained release may be attainable. It is hoped that this work could lead to new treatment options for orthopaedic surgeons that would reduce infection rates and mitigate the associated financial, emotional, and physical burdens.

PAGE 42

CHAPTER2 EVALUATION OF ALLOGRAFT COMPOSITION FOR FACTORS THAT MITIGATE ANTIMICROBIAL CHEMOPROPHYLAXIS Introduction Allograft implants have a higher rate of post-operative infection than do metallic implants. Several studies have demonstrated that this difference is not attributable to non-sterile allografts being supplied to surgeons for implantation (111 112, 116). One area that may explain this difference in incident rate is composition of the allograft. Material and device attributes can influence post-operative infection rates (21, 30 103) To date allograft composition has not been evaluated for the presence of identifiable and controllable factors that influence the incidence of post-operative infections One difference between allografts and metallic implants that can affect infection rates is lipid content. The amount of fat that remains on a graft is a function of the tissue type and extent of cleaning to which the graft was exposed during processing. Cancellous bone has a fat content of 70-90% w /w, whereas the fat content of cortical bone is lower, 6-9% w /w. Fat content is not required for proper graft function and is removed to the extent possible during graft preparation (7, 15, 95) Despite the effort to remove the fat, most bone grafts still carry a significant amount after processing ( 15). The amount of residual fat that remains in a graft is a function of both graft type and processing methodology employes, and is therefore highly variable. Bacteria can be 35

PAGE 43

36 predicted to partition into fat reservoirs carried on the allograft and remain protected from the single dose of antibiotics commonly given to surgical patients. Furthermore the relative hydrophilicity of cefazolin (Figure 2-1 ) the most commonly used antibiotics in allograft prophylaxis cannot partition into these fatty tissues (20 27 71 ). Even if bacteria do not preferentially partition into fat any organism incidentally surrounded by lipids on the graft would be shielded from these hydrophilic antibiotics Another factor that can influence the incidence of allograft infection is the surface architecture The surface features of allografts may have the ability to protect microorganisms Cancellous bone is sponge-like in texture with many irregular features while cortical bone has a dense regular structure (Figure 2-2). In contrast metallic implants are usually polished for a smooth surface decreasing surface area and providing better presentation of a potential contaminant. In addition, allografts are now often machined to contain thread profiles or grooves to allow for insertion or prevent slippage after implantation greatly increasing their surface area (Figure 2-3) It is possible that bacteria could become lodged inside these crevices and be protected from optimal antibiotic concentrations as well as phagocytosis by neutrophils and macrophages The increased incidence of infection can be hypothesized to be due to routine chemoprophylactic procedures being less effective on allograft tissue versus metallic implants. Irregularities on the surface of the bone and residual lipids may be responsible for this decrease in antibiotic effectiveness. Because of advances in tissue processing finished graft surface architecture and lipid content can be controlled or altered to minimize these disadvantages. For this reason there is now merit in evaluating these factors for their potential influence on chemoprophylaxis used in allograft surgery.

PAGE 44

37 f cefazolin t cture o 2 1 Molecular s ru Figure

PAGE 45

38 Figure 2-2. Coronal section through a femur showing spongy cancellous bone surrounded by dense cortical outer layer.

PAGE 46

39 Figure 2-3. Examples of threaded allografts (interference screws).

PAGE 47

40 Materials and Methods General Experimental Overview Implants were produced with differing lipid contents and architectural features and were tested in vitro. These implants were then contaminated with a known amount of bacterial bioburden and incubated in baths dynamically controlled to maintain relevant antibiotic concentrations. After completion of the exposure to the antibiotic treatment, the samples were removed and bioburden analysis was performed to determine the log10 reduction for each treatment group. Preparation of Treatment Groups Three architecturally distinct surfaces were evaluated during this study; (1) the smooth, regular surface of the periosteal side of cortical bone, (2) the irregular and spongy surface of the endosteal side of cortical bone and (3) cortical bone that had been machined to carry threads. These were selected because they represent the bone surface features that predominate in orthopaedic allograft surgery. To test the effects of lipid residue on antibiotic activity one half of the specimens in each of these architecturally distinct groups were exposed to a cleaning process that removed the tissues' endogenous lipids. This combination of architectural and compositional variation resulted in six randomized treatment groups (Table 2-1 ). To prepare the specimens, diaphyses of human femora, tibiae and fibulae were initially prepared in standard fashion by removing any extraneous muscle, ligamentous attachments and the periosteum For the periosteal and endosteal specimens, bone segments were then cut from the diaphyseal flare of the femora and tibia yielding specimens that were approximately 15 mm long, 7 mm in wide, and > 3 mm in thickness

PAGE 48

41 (Figure 2-4a) The periosteal surface of these specimens did not contain any mechanical alteration aside from the debridement procedure. To prepare the threaded specimens, fibulae were cross-sectioned at 15 mm intervals. However, they remained circumferentially intact to allow for the application of threads to the inner lumen. These segments were machined to carry a thread profile of 60 x 0.5mm in depth along their entire lumenal surface (Figure 2-4b ). This thread profile is typical of those found on threaded allograft bone and increases the surface area by approximately a factor of two. These segments were then sectioned longitudinally yielding two crescent shaped specimens of threaded cortical bone (Figure 2-4c ). T bl 2 1 All ft t tm t 'fi t d. th tud a e -. ogra rea en group spec1 1ca 10n prepare m IS S ly. Architectural Composition Treatment Physiologic Lipid Free Cortical bone void of any soft-tissue Periosteal Cortical bone void of any soft-tissue attachments or periosteum and attachments or periosteum. cleaned free of endogenous lipids ( < 5 % of physiologic) Bone from the medullary canal Bone from the medullary canal Endosteal consisting of the cortical-cancellous consisting of the cortical-cancellous interface, cleaned free of endogenous interface. lipids ( < 5 % of physiologic) Cortical bone with machined threads Threaded Cortical bone with machined threads. cleaned free of endogenous lipids ( < 5% of physiologic) To achieve different levels of endogenous composition half of the bone segments were treated in a 3% hydrogen peroxide bath at 40C for 30 minutes followed by a volatile extraction with acetone at 40 C for 30 minutes. The other half were left untreated and served as physiologic tissue controls These samples contained all of the endogenous lipids that remain following the standard debridement procedure. For

PAGE 49

I I I I I I I :11111 7mm I I I I I I I : -42 ---------En dos teal ------------P e r i o stea l --Figure 2-4a. Preparation of bone samples --Preparation of periosteal and endosteal segments from diaphyseal sections of femora and tibiae. 15 mm 3mm

PAGE 50

43 0 5 mm Figure 2-4b Representation of a thread profile typically found on threaded cortical allografts

PAGE 51

44 15mm Figure 2-4c. Preparation of the threaded cortical specimens from the diaphyses of fibulae Threading is applied to the intact tissue which is then sectioned longitudinally to create two halves

PAGE 52

45 cortical bone this is approximately 6-9% and for cancellous bone it is generally greater than 70%. Following treatment the samples were placed into individually labeled cryo vials and frozen at 20 C. Aseptic technique was used throughout specimen preparation to minimize the introduction of adventitious bioburden. Sample Inoculation lnoculum preparation. Staphylococcus aureus was utilized in this study because of its relevance as the most frequently isolated pathogen from orthopaedic wound infections (26 102 103, 108) A single colony of S. aureus was transferred to a blood agar plate and streaked for isolation. This culture was allowed to incubate for 1824 hours at 3 7 C Colonies were then selected and transferred into 10 mL of normal saline. This suspension was vortexed for 1 minute to reduce bacterial aggregation. The concentration of this suspension was then adjusted by either the addition of normal saline or bacteria to yield a 1 0 McFarland suspension. This suspension was then diluted 1: 10 to give a final concentration of 5-10 x 106 CFU / ml. The suspensions were prepared fresh daily and remained refrigerated. Sample inoculation. The samples were placed in a biological safety cabinet where 25 L of inoculum was transferred onto each via a sterile pipette The samples were then allowed to air dry in the hood for 60 minutes prior to further testing Inoculum controls were also performed by adding 25 L of inoculum to 50 mL of sterile saline vortexing, and plating 1 OOL of the resulting suspension directly onto solid media for quantification.

PAGE 53

46 Antibiotic Treatment A dynamic model was used to recreate the concentrations of antibiotic found in bone after the intravenous administration of a prophylactic dose of cefazolin (Ancef). Cefazolin was chosen because of its prophylactic use in allograft procedures (71, 102, 103). In this model (Figure 2-5), the specimens were placed into a vacuum flask containing 145 mL of phosphate buffered saline (pH 7.4) and an initial concentration of 18 g/mL of cefazolin This is the reported maximum concentration at the wound site following typical prophylaxis (27, 71, 92, 105, 121). A water bath surrounding the flask maintains the temperature at 37C throughout the experiment. Antibiotic-free PBS was then introduced into the vessel through an intravenous infusion set connected to a glass rod that delivered the solution into the vessel while a second port allowed the excess mixture to leave the vessel, thus maintaining a constant volume. A stir bar placed at the bottom of the vessel ensured uniform antibiotic distribution throughout the experiment. The infusion rate was set at 0.9 mL/min based on the half-life of cefazolin being 1.8 hours, to produce a concentration profile in the vessel that simulates the pharmacokinetic profile (71). The following equation was used to determine the rate: Where: Co is the initial concentration of antibiotic (g / mL) C1 is the concentration at any given time (g/mL) R is the rate of influx (mL/min) Vis the volume of the vessel (min-') t is time (min) The term RIV is equal to the elimination rate constant (-0.0064 min-1 for cefazolin). Because the volume is fixed, the equation can be solved to determine the rate of fresh saline needed to approximate the half-life of cefazolin.

PAGE 54

Specimen 47 Infusion of fresh saline OQ Outflow of excess saline / antibiotic 31 c Water Bath Heat/Stir Plate Figure 2-5 In-vitro model for the approximation of antibiotic concentrations following the administration of a single dose of a drug with first-order elimination kinetics.

PAGE 55

48 The specimens were held in the vessel for 7.5 hours allowing the antibiotic concentration to drop below the MIC90 (71). After treatment the specimens were removed using aseptic technique and quantitative bioburden analysis was performed. To control for the wash-off of organisms from the samples during the treatment, controls were added that were not exposed to antibiotic. These controls served as the starting point in determining the reduction in bacterial population due to the antibiotic. Bioburden Quantification Samples were placed into a sterile conical tube and filled with 15 mL of sterile saline. The tube was then vortexed on the highest setting for 3 minutes. The conical tube was then filled with an additional 35 mL of sterile saline and vortexed. Using aseptic technique, 10 Land 100 L were removed and plated onto 5% sheep's blood agar in duplicate. From the remaining saline 1 mL was passed through a 0.45 m filter under vacuum and rinsed with 50 mL of sterile saline. The filter was then transferred to an absorbent pad containing 1 mL of tryptic soy broth Both sets of cultures were held for 48 hours at 37C and quantified for CFU. MIC and Quantification of Cefazolin The microdilution broth technique was used to determine the minimum inhibitory concentration (MIC) of cefazolin for the organism used in the study. This technique was also used to determine unknown antibiotic concentrations. A stock trypticase soy broth was prepared containing cefazolin at a concentration of 1 OOg/ mL. Dilutions were then made with antibiotic free broth that resulted in the following antibiotic concentrations including lOg/ml, 7 5 g/ml, 5 g / mL, 4g / ml, 3g / ml, 2.5 g / ml, 2g / ml, l.5g/ml,

PAGE 56

49 1.0g/ml 0 5 g/ml, 0.25 g/ml, 0.125 g / ml. Inoculum was added to each of these tubes resulting in a bacterial concentration of 1-5 X 105 cfu/ml. All samples were run in duplicate. The 96-well plates were incubated at 3 7 C for 18 hours to allow for growth. Positive (no antibiotic) and negative (no inoculum) controls were included. Following incubation, the wells were examined for growth as indicated by turbidity The MIC was determined by the lowest concentration of antibiotic that prevented growth. Requirements for accepting the test results were: (1) all of the tubes with antibiotic concentrations greater than the MIC must be without growth, (2) all of the tubes with antibiotic concentrations less than the MIC must have growth, (3) the purity and identity of the organism in the first tube demonstrating growth from the MIC must be verified, (4) the negative and positive controls must demonstrate no growth and growth respectively, (5) the concentration of the original suspension must be 1-10 X 105 CFU/ml as determined through by inoculum controls, and (6) replicates from the same sample must be in agreement meaning both samples must have the same inhibitory endpoint. For samples with an unknown amount of cefazolin, the concentration of the most dilute well showing no growth was assumed to be the MIC The concentration of the original sample was then determined by multiplying the MIC by the dilution factor of that well. Additional sets of dilutions were performed with these samples to more precisely define the endpoint. Statistical Analysis Statistical analysis was performed to determine if there was difference in bacterial reduction between material treatment groups. Analysis was performed using a

PAGE 57

50 commercially available software package (Statistica '99, Statsoft Inc.). A one-way ANOV A was used to determine if a significant difference existed in the observed log reductions between the treatment groups A Newman-Keuls test was used to determine specifically which treatment groups had a significant difference. For all tests significance was defined asap value less than 0 05. This test assumes that all errors are independently and normally distributed. During the analysis of the results there were no major departures from these assumptions. Results The MIC of cefazolin to the particular strain of S. aureus used in this study was determined to be 1.0 g/mL. This was similar to previously reported MIC for this organism (71, 93, 121). Based on this data, the validity of the model was established by sampling the solution from the flask and determining cefazolin concentrations over the course of the study. Figure 2-6 demonstrates that the model exposed the tissue to antibiotic concentrations that were similar to those found in-vivo. Due to the inherent limitations of sensitivity with the bioassay used in the quantification of cefazolin, verification of antibiotic concentration at the later time-points could not be accomplished. Concentrations determined by earlier time-points, however, provide sufficient data to demonstrate the model was simulating first-order release kinetics with an antibiotic half life of 1 8 hours. Although the amount of inoculum added to each specimen was controlled for, there was substantial variation in the amount recovered from the saline controls between

PAGE 58

51 treatment groups (Table 2-2). This difference in recoverable bioburden can be attributed to differences between the starting materials There are two mechanisms for the loss of organism with this model. First, the organism can be washed-off during the treatment. Secondly certain materials have the potential to retain bacteria during the extraction phase, prohibiting their detection upon culture. Table 2-2. Mean log1o bioburden for the control samples and antibiotic treated samples and log10 reduction for each treatment group. The number ofreplicates (n) for each treatment group is also listed. LFE = lipid-free endosteal, LE= lipid containing endosteal, LFP = lipid free periosteal LP= lipid containing periosteal, LFT = lipid-free threaded, and LT = lipid containing threaded. The standard error of the mean is indicated h m parent eses. LFE LE LFP LP LFT LT Control (CFU) 5 24 5 .53 6 .19 5 .72 5.10 5.37 n=3 (.13) ( .13) ( .16) ( 13) ( 12) ( .13) Antibiotic (CFU) 4.47 3 70 3.27 4 .33 4 .10 4 .19 n=6 (.09) (.10) ( 09) (.10) (.09) (.09) Log Reduction 0.7 2.0 2.8 1.4 1.0 1.2 ( 16) ( 19) ( 19) ( 16) ( 16) (.19) The log10 reduction for each treatment type was calculated from the difference between the control group with no antibiotic and the treatment group with antibiotic (Figure 2-7). From this analysis, the lipid-free periosteal group had the largest reduction at 2.8 logs This was significantly more than the periosteal group where the lipid was not removed, which had a log1o reduction of 1.4. Both of the threaded groups, lipid and lipid free, had similar log10 reductions at 1.2 and 1.0 respectively The lipid-free endosteal specimen had a significantly lower log1 0 reduction, 0.7, than its lipid containing counterpart 2.0.

PAGE 59

52 3 5 3 -2.5 = Q '.C 2 u ,e 1.5 .s -->---0.5 0 LFE LE LFP LP LFT LT Treatment Figure 2-7 Summary of the average log10 reduction values attained for each treatment group (see table 2-1 for key to abbreviations). Error bars indicate 2 standard error.

PAGE 60

53 In addition to the 7.5 hour study, the lipid-free periosteal samples were also evaluated at intermediate timepoints to elucidate the reduction kinetics. Figure 2-8 describes the change in bacterial population as a function of time using the dynamic model. Controls were also included in this experiment to determine changing bioburden on samples that were not exposed to antibiotic. Discussion The hypothesis for this set of experiments is that complex architectural features and residual lipids serve to undermine the effectiveness of cefazolin by shielding bacteria from optimal concentrations. Therefore, grafts with the most regular features (smooth periosteal grafts) and those treated to remove endogenous lipids should have the highest reduction in bioburden. This hypothesis was not supported by the data from this study. The lipid-free periosteal specimens did have the greatest reduction in bioburden, however the second largest reduction in bioburden was observed with the lipid containing endosteal group This is unusual in that the surface of these specimens are the most convoluted and carry the highest amount of endogenous lipid among the specimens evaluated. This apparent paradox may be partially explained by an alternate hypothesis. The ability of the graft to absorb the starting inoculum may be responsible for the high reduction in bioburden observed with the physiologic endosteal group. This surface has the highest lipid content of the three studied (between 70-90% ) The surface of this tissue is therefore less receptive to absorbing externally applied aqueous solutions such as the inoculum. It is possible that the bacterial suspension coated the surface of the tissue and did not penetrate appreciably into the matrix. This idea is further supported by the lipid-

PAGE 61

l .OE+-07 l .OE+-06 l.OE+-05 = .. "C ; l .OE+-04 .c Q = l .OE+-03 54 I Antibiotics o Control I l .OE+-02 +------.-----.--------.------.------, 0 100 200 300 400 500 11me (min) Figure 2-8. Change in bioburden over time for the lipid-free periosteal group. For each time-point two replicates were tested. Error bars indicate range.

PAGE 62

55 free endosteal group, which had lowest reduction in bioburden. Although the endosteal group has the most complex architecture, unlike its lipid-containing counterpart, the lipid-free specimens were not guarded against the absorption of the inoculum. In fact, quite the opposite is true. When the fat is removed from cancellous bone like that found on the endosteal surface, the remaining matrix resembles a sponge. This allows for substantial absorption of an aqueous solution During sample inoculation it was observed that inoculum was absorbed faster and to a greater extent on samples that had been cleaned of residual lipids as compared to samples that were not cleaned. Because the organisms were not taken deeply into the tissues containing lipid, particularly the endosteal specimens, they were not afforded the same level of protection from the surrounding tissue. Most bone allografts are preserved by lyophilization (9, 15, 95). Because there is a significant decrease in the strength of bone that is dry, most allograft package inserts call for the rehydration of the tissue prior to implantation (9). Because this process can be lengthy, up to 24 hours, it is often started prior to surgery. If contamination of the rehydrating solution occurs, the clean, dry, cancellous graft will absorb more bioburden than other types of grafts that do not have the same capacity to retain fluid. Perhaps even more important than the amount of bioburden that is absorbed is the extent to which it is absorbed. Clean cancellous grafts will allow the rehydration solution to fully penetrate the matrix due to its inherent porosity. This is in contrast to cortical grafts that, due to the density of the bone, will not absorb solutions to the same extent. This pattern would not be observed to the same extent with the periosteal specimens for several reasons. First, periosteal bone is composed of dense cortical bone

PAGE 63

56 that is regular and relatively non-porous. Therefore, the amount and extent of absorption is minimized Secondly, the amount of endogenous lipids carried on this type of bone in its natural state is approximately IO-fold less than natural cancellous bone. These two factors would mitigate the shielding effect that was observed with the cancellous bone. The two types of threaded specimens both performed equally The antibiotic was significantly less effective on these two groups when compared to the lipid-free periosteal samples even though the material was similar in composition to this group. There are two possible explanations for this. First the thread profile used increased the surface area by a factor of 2. Secondly, the cuts made by the threads cross through the Haversian canals that run in a parallel fashion down the length of cortical bone (Figure 2-9a). These canals serve as the conduit for blood vessels in dense bone By cutting into the bone, these canals are exposed increasing the porosity and providing a pathway for the entry of microorganisms. In periosteal bone that has not been machined to carry threads these Haversian canals remain below the cortex of the bone which is only penetrated sporadically by nutrient foramina and Volkmann's canals (Figure 2-9b). Lastly, the kinetic data provide insight into how the bacterial bioburden changed over the course of the study As significant as the reduction in bioburden for samples treated with cefazolin was the proliferation that was associated with untreated tissue. This suggests that in the absence of antibiotics bone provides a suitable environment for the proliferation of bacterial contamination. The substantial bioburden remaining after treatment may also suggest that a single does of cefazolin may not provide adequate

PAGE 64

57 Volkmann's Canal Haversian Canals b Figure 2-9. Longitudinal diagram of cortical bone a) Cortical bone that has been machined to carry threads. Note how the thread profile exposes Haversian canals to the exterior. b) Cortical bone with its periosteal surface intact. Note only the sporadic communication of the Haversian canals with the periosteal surface through nutrient foramina.

PAGE 65

58 assurance that contamination introduced onto the graft during surgery will be removed to a benign level however the level of kill was lower than exptected in all groups This is likely attributable to the organisms not being in a favorable state for replication and consequently antimicrobial incorporation. Conclusions The data acquired in this study support the hypothesis that antibiotics are significantly more effective on cortical bone tissue with regular surface features and minimal porosity than on those tissues with greater architectural complexity. However, the data did not conclusively support the hypothesis that residual lipids carried on the tissue decreased antibiotic effectiveness. In fact, the data suggests that lipids may prevent bacterial absorption into the deep matrices of tissue and thus increase their susceptibility to an antimicrobial agent. For this reason surgeons and other health-care providers responsible for allograft preparation should take care to guard against contamination during the reconstitution of clean cancellous bone grafts as introduced bacteria may migrate further into these tissues than with other less porous tissues. Future work in this area should focus on more deliberate and proactive methods of preventing post-operative infection. These data suggest that a single dose of cefazolin may not be adequate to assuredly eliminate microbial contamination on bone allografts. The development of a method to uniformly load an allograft with antibiotic may have significant effects on the incidence of post-operative infections involving allografts and may fill the void left by systemic antimicrobial therapy

PAGE 66

CHAPTER3 EVALUATION OF CHEMOTHERAPEUTIC AGENTS FOR USE IN A CORTICAL BONE ALLOGRAFT DRUG DELIVERY SYSTEM Introduction Orthopaedic implants containing an antibiotic for either therapeutic or prophylactic delivery have been used since the 1970's (61). In this time there have been improvements to both the devices and their application. Today, the majority of local antimicrobial therapy used in orthopaedic surgery is accomplished with polymethylmethacrylate (PMMA) gentamicin beads ( 11, 72). This system provides many advantages over the more conventional systemic antimicrobial therapy. With local delivery, systemic toxicity is avoided because serum drug concentrations are 10-100 times lower than when administered through the conventional route (33, 34, 53). Because the drug reservoir is located at the site requiring therapy, high tissue concentrations are achieved. In fact, organisms reported to be resistant to a drug at systemically attainable plasma concentrations may indeed be sensitive to the drug at concentrations found at the wound site via local delivery (61). This is in contrast to the scenario that accompanies intravenous therapy where drug concentrations in poorly perfused wound tissue are often much lower than plasma levels (53). However, this delivery system is not without its limitations. During preparation of the standard beads, dry powders of both PMMA and an antibiotic are mixed with water to form a cement. This reaction is very exothermic resulting in beads reaching temperatures in excess of 100C during the drying process (32). For reasons pertaining 59

PAGE 67

60 to the stability of the drug, exposure to such high temperatures places significant constraints on the antibiotic that may be used. Additionally, PMMA beads serve no function other than being a non-resorbable carrier for the antibiotic. Because of this, additional surgery is generally required to remove the device after the drug has been delivered, adding the expense and risk of a second procedure (61). Allograft based delivery systems have the major advantage over PMMA in that allografts are bioincorporable, meaning they will eventually incorporate into the recipients' own vital tissues at the site of implantation (15, 17, 22, 50). This eliminates the need for a second surgical procedure to remove the device. In addition, allografts are indicated in orthopaedic procedures for reasons other than drug delivery. Specifically, allografts are frequently used to lend structural support or promote new bone formation. These added advantages are in addition to the previously mentioned benefits associated with local drug delivery systems. Table 3-1 summarizes and compares the advantages of an allograft based delivery system and the PMMA system over systemic drug delivery. The use of allograft bone as a carrier for antimicrobial agents has previously been proposed, however limitations with tissue preparation and drug loading have prevented the development of a tissue based delivery system (73). Recent improvements in processing technology have allowed for allograft bone to be effectively cleaned and potentially loaded with a wide range of drugs useful in these procedures. These critical advances now permit the investigation of the allograft as a potential drug delivery system. Fundamental to the further development of a tissue based drug delivery system for the treatment or prevention of orthopaedic infections is the optimal selection of an antimicrobial agent. This theoretical chapter proposes ideal characteristics of an

PAGE 68

61 antimicrobial agent that are necessary for success with an allograft based delivery system and evaluates potential drug candidates against these characteristics. Table 3-1. A comparison of the advantages and disadvantages associated with the PMMA d ll ft b d d d r t an a ogra ase rug e 1very s, s ems. PMMA Based System Allograft Based System Advantaees Disadvantaees Advantages Disadvantages Locally high drug High temperatures Locally high drug Potential for disease concentrations ( > 100 C) are required concentrations transmission for formulation Decreased risk of Second surgery is Decreased risk of Restrictions on the systemic side effects required for removal systemic side effects availability of donated following drug tissue expenditure Allows for primary No functional purposes Allows for primary Difficulty in drug wound closure other than drug delivery wound closure loading and material homogeneity A v oids compliance A voids compliance i s sues issues Incorporates into the recipients tissu e eliminating the need for a second surgical procedure for removing the device Provides structural suooort at the graft site Promotes new bone formation Indications for Use When selecting an antibiotic for use in any situation it is important to first define the desired effect. This is particularly true when developing a tissue based delivery system Antibiotics have several potential indications for use with allografts and these different indications correspond to different antimicrobial requirements such as duration of drug treatment, antimicrobial spectrum, and stability. These differences are summarized in Table 3-2.

PAGE 69

62 Table 3-2. A summary of proposed characteristics for antimicrobial candidates for d ffi t l d. f 1 eren surg1ca m 1ca 10ns. Indication Antimicrobial I Duration of Activity I Stability Requirements Spectrum 1-4 weeks Must be stable in Graft Preservation Very broad preoperatively solution at 2-8C for several weeks Must remain stable Simple Prophylaxis Primarily Staphylococci < 24 hours through the loading process. Staph and Gram Must remain stable Directed Prophylaxis 3-5 days through the loading negative organisms process Must remain stable through the loading Therapeutics Narrow only the 2-3 weeks process Must also infecting organism remain stable at body temperature after implantation. Graft preservation. Allografts, being aseptically harvested and processed are frequently treated with a solution containing one or more antibiotics to prevent incidental contamination during recovery or processing leading to an infection in the recipient. Tissue banks often mistakenly refer to this type of treatment as "cold sterilization". However, due to the inherent limitations with regard to bacterial resistance, sterility assurance levels (SAL) comparable with pharmaceutical or medical device standards (typically less than 1 contaminated product in 1,000,000) are not attainable. Nonetheless, for certain tissues this practice remains as an important step in preventing disease transmission. Graft preservation requires that the antibiotic(s) cover as broad a spectrum of organisms as is possible and that the drug remains stable in solution at refrigeration temperatures for up to several weeks Simple prophylaxis Here, an assumably sterile allograft is loaded with sufficient antibiotic to prevent infection due to contamination introduced during surgery In this case, the allograft is independently indicated in the surgery and the antibiotic loaded into the tissue simply replaces or improves upon systemic perioperative

PAGE 70

63 chemoprophylaxis. This application requires an antibiotic with activity against the staphylococci due to its prevalence with this type of infection and a short release pattern. This method is employed sporadically by surgeons who reconstitute their allografts with an antibiotic solution prior to implantation. Currently, there are no allografts pre-loaded with antibiotics that are available commercially. Directed prophylaxis. In this proposed application an allograft containing an antibiotic would be used specifically for the prevention of infection in a wound where one would likely occur if no prophylaxis were administered This setting would be analogous to the use of the PMMAgentamicin beads in preventing infection of an open fracture. Here again the tissue based delivery system would be advantageous because it would not require a second surgery for removal and lending structural and healing support to the fracture site. An appropriate antibiotic would need to provide effective coverage for 72 hours against the staphylococci and Gram negative organisms, as up to 50 % of these types of infections are polymicrobial. Therapeutics. Lastly, tissue based delivery systems could be used to treat an existing long term condition such as chronic osteomyelitis. Current treatment for certain types of osteomyelitis includes a primary operation for wound debridement and placement of PMMA-gentamicin beads followed by a second procedure for bead removal and placement of a bone allograft. Tissue based delivery systems could eliminate the need for the second operation by delivering a therapeutic dose of antibiotic carried within the prescribed allograft which would then be followed by normal graft incorporation. Although the spectrum of activity would not need to be broad because the offending

PAGE 71

64 organism would be known, the duration of therapy would need to be sustained for several weeks. Characteristics of an Optimal Antimicrobial Agent Spectrum of activity. The spectrum of the antibiotic should ideally be appropriate for the target organism. When the indication is prophylaxis, the antibiotic should provide coverage over a broad spectrum of organisms and be definitively effective against the most commonly encountered organisms. Staphylococci, specifically S. aureus and S. epidermidis, are consistently isolated from post-operative orthopaedic wound infections (4). S. epidermidis, which is typically not associated with pathogenesis is frequently found in iatrogenic infections involving orthopaedic implants. In addition, between 30 to 50% of infections involve polymicrobial contamination that includes a Gram negative organism (108). This significantly reduces the field of potential candidates, as many antimicrobials are effective against organisms of only one Gram's classification. If the indication for the device is to treat an existing infection the antibiotic need only be effective against the isolated organism. Here, a pre-operative culture could be obtained and the sensitivity of the organism confirmed prior to the implantation of the device. It is worth reiterating that sensitivity testing performed in the context of attainable antibiotic plasma concentrations may not accurately reflect the susceptibility of the organism to the antibiotic at the more relevant tissue concentrations. This is because most antibiotics are effective against an increasingly broad spectrum of organisms as the concentration of the agent is increased. However, toxicity prevents the systemic administration of the drug at these high levels and therefore organisms not effected by the antibiotic at attainable serum levels are reported to be resistant.

PAGE 72

65 Adverse reactions. Antibiotic hypersensitivity is of particular concern when the drug is administered via an implantable device. Due to the release and distribution kinetics of the drug from an implantable delivery system, acute hypersensitivity may not be detectable during the surgical procedure. In addition, confirming the source of a hypersensitivity reaction to be the implanted antibiotic may prove difficult when confounded by other potential sources of hypersensitivity such as the infusion of blood products Depending upon the severity a hypersensitivity reaction could necessitate an emergent second procedure to remove the device. For these reasons, it is important to select an antibiotic that has both a low incidence and severity of hypersensitivity reactions. In addition, the implantation of such a device into patients with a known or suspected allergy to the incorporated drug is contraindicated. Systemic toxicity is of little concern in the selection of an antibiotic for use with a tissue based delivery system, provided the antibiotic has been approved for uses systemically. This is due to the comparatively low plasma levels (1-10% of plasma levels attained with systemic administration) that are attained during the local delivery of an antibiotic (61, 82) The potential for the antibiotic to cause local toxicity must be addressed in drug selection. When allograft tissue is employed, it is important that the antibiotic does not inhibit the incorporation of the graft. Specifically graft incorporation could be hampered via two mechanisms First, the antibiotic could alter the biological properties of the allograft resulting in sub-optimal incorporation (5, 64). Additionally the antibiotic could produce local cytotoxicity resulting in an inflammatory response that could adversely affect wound healing (64, 96, 104). For these reasons, antibiotics that are known to

PAGE 73

66 adversely effect orthopaedic tissue should be excluded from use with a tissue based delivery system Stability. The stability of the drug needs to be considered from several perspectives. First the drug must be stable through the loading and preservation phases of graft production These phases involve dissolving the drug in a warm solution (3545 0 C), freeze-thaw cycles and lyophilization. Harsh conditions are avoided during tissue preparation due to the delicate nature of the graft and its biological activity (7 9, 15, 91). After the drug is loaded into the tissue it must remain stable until use. Lyophilization or freezing are the preferred methods of preservation. Lastly, the stability of the drug in vivo must be considered if a prolonged release rate is desired as elevated temperatures or enzymatic breakdown of the drug are possible ( 48, 78, 79). Physical and chemical considerations. The potency of the drug is important because the loading capacity of the graft is limited. The optimal drug would be bactericidal at very low concentrations ( minimum bactericidal concentration or MBC below 2 g/ml) Drugs having minimum inhibitory or bactericidal concentrations greater than 50 g/ml are less useful because the graft can simply not hold an effective dose for release. As a rule, the longer drug release needs to be sustained, the more potent a drug is needed. The hydrophilicity of the drug will also greatly effect its release characteristics from the tissue. Using a porous matrix as a model for diffusional release into an aqueous environment the release rate will increase as a function of solubility (84). Therefore, modulation of release kinetics from tissue based systems could be accomplished by selecting compounds with the appropriate aqueous solubility. More lipophilic

PAGE 74

67 compounds would generate a sustained release pattern while a more hydrophilic compound would result in faster diffusion and release For this same reason, the ionization state of the drug at physiologic pH should also be considered. Antimicrobial Candidates Based on the above criteria, several antibiotics previously proposed for possible use in the local management of orthopaedic infection were evaluated for their compatibility in an allograft based drug delivery system. Cephalosporins. Cefazolin (Ancef), a first generation cephalosporin, is the most commonly used antibiotic for the prevention of post-operative orthopaedic infections (14, 16, 27, 103). It exhibits an adequate activity against Staphylococci and Streptococci. In addition, the MIC9o (minimum inhibitory concentration for 90% of the organisms typically encountered) is between 1-4 g/ml making it sufficiently potent to be used in a tissue based delivery system (71 ). Cefazolin has limited activity against Gram negative organisms. Second and third generation cephalosporins have increasingly more activity against Gram negative organisms, however, their potency towards Gram positive organisms is diminished. For this reason any single cephalosporin would be a less than optimal choice. Although they posses little systemic toxicity and have been shown to not impair allograft incorporation, cephalosporins are associated with hypersensitivity reactions(l4) Penicillins. The penicillins are grouped into three classes. The naturally occurring penicillins (penicillins G and V) have a narrow spectrum of activity against Gram positive organisms and Gram negative cocci. Like the cephalosporins, penicillins are dependent upon the integrity of their P-lactam ring for antimicrobial activity. If this

PAGE 75

68 ring is damaged either by acid or bacterial enzymes, penicilloic acid is produced, which lacks bactericidal activity. Because the majority of clinical isolates of S. aureus are found to produce P-lactamase (penicillinase), the use of the naturally occurring penicillins has been limited in the surgical setting. Therefore, the penicillinase-resistant penicillins are more useful against P-lactamase producing Staphylococcal infections. The broad-spectrum penicillins have been developed to be effective against both Gram positive and negative organisms, but this group, like the natural penicillins, generally lacks resistance to penicillinase. Apart from spectrum, a more significant limitation to the use of the penicillins is their ability to induce hypersensitivity reactions that can range in severity from trivial to fatal. The overall incidence of penicillin induced hypersensitivity is as high as 8%, resulting in over 300 deaths per year. Ciprofloxacin. Ciprofloxacin is synthetic fluoroquinolone that exhibits its antibacterial effects by inhibiting DNA gyrase. This antibiotic has several characteristics that warrant its consideration for use in a tissue based system. First, it is a broad spectrum antibiotic that is effective against both staphylococci as well as many Gram negative rods. Additionally, ciprofloxacin is a potent antibiotic effective at low concentrations (between 1-2 g/ml for most pathogens) (32 40, 65). The drawback to ciprofloxacin is its association with permanent cartilage degeneration. Because of the potential proximity of the implant to a weight bearing chondral surface high concentrations of ciprofloxacin may be attained in the joint space which could result in damage to the cartilage.

PAGE 76

69 Tetracycline. The tetracyclines are effective against a v ery broad spectrum of organisms. This makes their use in prophylaxis where a wide variety of organisms may be encountered a tempting choice. However, tetracyclines avidly chelate to calcium which would prevent complete release of the drug from a bone delivery system. This binding could result in a subsequent lack of graft incorporation as tetracycline inhibits bone growth. Vancomycin. Vancomycin is an ideal candidate for the irradication of orthopaedic Staphylococcal infections as it is effective against almost all Gram positive organisms and resistance is rare. In addition, vancomycin is potent at low concentrations (usually less than 1 g/mL) making it unlikely to cause adverse reactions (81, 85, 117). Unfortunately vancomycin is completely ineffective against Gram negative organisms, thus prohibiting its use as the sole antibiotic in an orthopaedic delivery system unless the identity of the offending organism is known Vancomycin does act synergistically with the aminoglycosides making its use in a polyantimicrobial system attractive (49 76, 109 118). Gentamicin The aminoglycosides possess several favorable characteristics for use in a tissue based delivery system. Gentamicin and the other aminoglycosides are effective against a wide variety of bacteria including Gram negative organisms and Staphylococci. This efficacy is obtained at very low concentration s (typically < 2 g / ml) providing the opportunity for a sufficient reservoir of drug to be carried within the allograft for an extended release (60 101, 114 119) Although toxicity is associated with systemic administration, gentamicin is safe when used in a local delivery scenario (34). This is due to the very low and often non-

PAGE 77

70 detectable plasma levels that follow implantation. This has been demonstrated clinically by the wide use of gentamicin and tobramycin with PMMA(33, 34, 61, 82). The physicochemical characteristics of the aminoglycosides are similarly appealing, because gentamicin has been shown to be stable at elevated temperatures and through lyophilization (58, 74, 123). In addition, it is a hydrophilic compound that allows for its complete dissolution from the matrix in a timely manner (11 ). Perhaps the only drawback to gentamicin is that it may elute too quickly from the allograft, preventing its use in the treatment of existing osteomyelitis, which would require therapeutic levels to be maintained for 2-3 weeks. Discussion Local drug delivery for the treatment and prevention of orthopaedic infections has many advantages over systemic antibiotic therapy. These advantages have been realized clinically with the gaining popularity of the PMMA-gentamicin bead, which is now available commercially in Europe under the trade name Septopal (Merck, Darmstadt, Germany). However, the limitations associated with PMMA allow opportunities for improvement of these systems. Most obviously the need for a second surgical procedure to remove the device increases health care costs and places the patient at further risk from surgical complications. The proposed allograft based delivery system would eliminate this need for a second procedure. Additionally, allograft bone can be used to lend structural support and promote new bone formation at the wound site. For these reasons, development of a tissue based delivery system could provide surgeons and their patients with a superior alternative to the PMMA system.

PAGE 78

71 The use of allograft bone as a carrier for antimicrobial agents has previously been proposed however limitations with tissue preparation and drug loading have prevented the development of a tissue based delivery system. Recent advances in allograft processing technology make it possible to effectively clean cortical bone and subsequently load it with a drug. Selection of the optimal antimicrobial agent is the next obvious step in the development of this system Although there is considerable literature available treating the topic of antibiotic selection for use in the PMMA system, to date no work has been found that evaluates potential antimicrobials for use in a tissue based system Because significant differences exist in drug loading methodologies, and because there is the potential for the agent to effect either the allograft or the wound site in a way that would prevent graft incorporation drug evaluation specifically pertaining to the tissue based system is necessary. This review was conducted using the criteria generated by the four specific indications previously described. Each of these indications had specific requirements pertaining to antimicrobial spectrum duration of treatment and drug stability Due to regulatory constraints the commercial development of an allograft based delivery system that used different drugs for each of the indications is likely not feasible. Therefore a list of requirements that satisfied the needs of all of the indications was d veloped (Table 33). These "master requirements" are a product of the most stringent criteria from each category. This allowed the antimicrobial candidates to be compared against one set of criteria versus four. In developing this master list the requirements for graft preservation were excluded from consideration. This was done because the use of antibiotics in the

PAGE 79

72 preservation of allografts is not considered to be a delivery system and does not carry the regulatory burden associated with a drug delivery system (38). The master set of requirements contains four categories; antimicrobial spectrum, drug stability, potential for adverse reactions, and the potency of the drug. Less critical is the duration of therapy required. Duration was omitted becau se there is the potential that a sufficiently potent drug could have its release profile modified to meet the needs of three separate indications. This could be accomplished by adding excipients that restricted the release of drug from the bone. In this case, were the release rate could be slowed, having a drug that has a high aqueous solubility and therefore has an inherently fast release profile would be beneficial. Table 3-3. Requirements of an antimicrobial drug for use in an allograft based drug delivery system. Included are the requirements for the individual indications and one t f h f fi 11 fh d mas er set o reqmrements t at sa 1s 1es a 0 t ese nee s. Indication Antimicrobial I Stability Adverse I Potency Spectrum Reactions Simple Primarily Must remain stable through Must not result in Potency is not Prophylaxis Staphylococci the loading process hypersensitivity a significant or site toxicity factor Directed Staph and Gram Must remain stable through Must not result in Moderately negative hypersensitivity Prophylaxis organisms the loading process or site toxicity potent Narrow, only the Must remain stable through Must not result in Therapeutics infecting the loading process. Must also hypersensitivity Potent organism remain stable at body or site toxicity. MIC < 3 g/ml temperature after implantation. Staph and Gram Must remain stable through Must not result in Master negative the loading process. Must a lso hypersensitivity Potent Requirements organisms remain stable at body or site toxicity. MIC < 3 g/ml temperature after implantation Each of the antibiotics was evaluated against this master set of criteria. The antibiotics selected for evaluation have previously been suggested for use in local delivery systems employing PMMA and represent the most obvious candidates for

PAGE 80

73 review. Gentamicin meets the requirements of an antibiotic for use in an allograft based drug delivery system. Studies have demonstrated that gentamicin is sufficiently stable and highly potent (72) Although it is associated with toxicity when delivered system i cally, plasma concentrations encountered with local delivery of this drug are well below toxic levels and are often not detectable. Additionally, gentamicin has not affected wound healing when used with the PMMA system clinically. The antimicrobial spectrum of gentamicin is adequate to cover both the relevant Gram positive and negative organisms typically encountered, however, the emergence of resistant strains to the aminoglycosides may warrant reevaluation of this drug in the future. Cefazolin has the major disadvantage of being too narrow in antimicrobial spectrum. Although the later generation of cephalosporins are effective against Gram negative organisms, they have diminished activity against Gram positive organisms. Hypersensitivity reactions associated with the penicillins exclude their consideration for used in an implantable drug delivery system. In addition, the growing number of penicillinase producing organisms would necessitate the inclusion of a ~-lactamase inhibitor such as clavulanic acid. This, coupled with the high MIC's that are required for some penicillins, makes this an overall poor candidate. Tetracycline and ciprofloxacin are both associated with toxicity reactions specific for orthopaedic tissue. Ciprofloxacin is associated with permanent cartilage degeneration when given systemically to immature subjects. For this reason it is not indicated for use on children under the age of 17 or by women who are pregnant. Because local drug concentrations delivered via an implantable system can reach levels 100 times greater than when administered systemically, this degenerative effect may be more pronounced.

PAGE 81

74 Likewise tetracycline is known to inhibit bone growth. In addition it binds avidly to calcium, which would in tum prevent its complete release from the device. This would likely result in poor graft incorporation. Therefore tetracycline should be excluded from further consideration Vancomycin is an optimal choice for the treatment of infections caused by Gram positive organisms. Although there has been the emergence of vancomycin resistance with enterococci in general most Gram positive organisms are sensitive. Unfortunately, vancomycin has no effect on Gram negative organisms. Synergism with the aminoglycosides makes it an attractive option if multiple antibiotics were used. However, the use of more than one antibiotic raises concerns about the synchronization of their release and the limitation on the absolute amount of drug that is possible to load into an allograft. Conclusions The results of this review indicate that gentamicin is currently the optimal antibiotic for inclusion in an allograft based delivery system. This selection is based on its favorable characteristics with respect to antimicrobial activity stability hypersensitivity and local to x icity potency and aqueous solubility. In addition gentamicin has a proven record of being safe and efficacious in the prevention and treatment of orthopaedic infections when used with the PMMA system The selection of an antimicrobial, based on a review of the literature, by no means guarantees its success in practice and therefore a significant amount of research remains to confirm this selection. Bacterial resistance as well as the possibility of the development of a better antimicrobial mandates that this selection be continually

PAGE 82

75 reexamined. The next step in the development of an allograft based delivery system is to evaluate the characteristics of gentamicin with respect to drug loading and release patterns.

PAGE 83

CHAPTER4 OPTIMIZATION OF GENTAMICIN LOADING INTO CORTICAL BONE Introduction The development of an allograft drug delivery system is contingent upon the successful loading of the drug into the allograft. Historically, this task has been a difficult one to accomplish (73). Conventional cortical bone allografts are far from ideal drug release devices. One major obstacle associated with bone as a drug delivery system is the amount of lipids and cellular elements that remain within the matrix even after processing If fat remains trapped to any appreciable extent, the effective volume that is available for drug loading is reduced. This was demonstrated in previous work wherein the rehydration of cortical bone samples that had been cleaned of residual lipids and lyophilized were compared to bone specimens that were lyophilized without cleaning. The reconstitution of the two groups was then evaluated over time. The group that had been cleaned free of endogenous lipids had a significantly faster reconstitution time than did those samples that carries endogenous levels of lipids (Figure 4-1 ). Lipids may also block or impede the uptake of a water-soluble drug into the matrix and conversely impede the release of a non-polar compound. Therefore, the removal of lipids and other cellular debris from the matrix of cortical bone prior to loading is a prerequisite to successful use as a drug carrier. Earlier studies demonstrated that effective cleaning procedures were possible without altering the beneficial properties of the t issue (15, 75, 95). 76

PAGE 84

60 50 ,,....._ Cl) :) 40 <.) s '-' i:: 30 "
PAGE 85

78 A major challenge that remains is the ability to completely load the matrix with the compound in question. Simply soaking bone the in a solution of the drug does not provide effective penetration of the entire matrix in a reasonable time This can be visualized histologically by allowing a dye to penetrate into the internal matrices of bone. As is evident by examination of figure 4-2 complete penetration of cortical bone through passive diffusion is not complete at 8 hours. Figure 4-3 demonstrates a cortical bone pin that has been completely perfused with a tracing dye. Long drug loading times raise concerns over the stability of both the drug and tissue while in a hydrated state at temperatures above freezing Because the volume within cortical bone is limited, incomplete loading would result in grafts with less than the optimal dose of a drugs. In addition the variability in release profiles that would be associated with grafts that were incompletely loaded with drug would likely be too high to permit the device acceptance from regulatory authorities. Because the density of cortical bone as a material is rather uniform from donor to donor and because this density is directly related to the porosity of the matrix it could be hypothesized that at the amount of drug within the bone approaches the maximum amount possible, this variation will decrease Therefore, a process is needed to effectively load drug into intact cortical bone in a consistent and timely manner.

PAGE 86

79 Figure 4-2. Micrograph of cortical bone taken from the cross-section of a human femur. The bone was allowed to soak in safranin for 8 hours. Note the partial staining ( dark areas depicted with arrows) of the matrix with approximately half of the haversian canals remaining unstained.

PAGE 87

80 Figure 4-3. Micrograph of cortical bone taken from the cross-section of a human femur. This bone was subjected to a pressure / vaccum perfusion process. Note the complete staining (dark areas) of the matrix with all of the haversian canals retaining dye.

PAGE 88

81 The optimal drug loading process should contain several key features: 1. The process allows for the maximum amount of drug possible to be loaded into the matrix. 2. The process results in tissue containing a consistent and known amount of drug. 3. The process results in a tissue that contains low residual moisture, as moisture is associated with decreased tissue and drug stability. 4. The loading process does not alter the effectiveness of drug. 5. The process does not adversely effect the biomechanical properties of the tissue. In previous work, it has been demonstrated that by thoroughly cleaning the tissue of cellular debris, removing the moisture in bone prior to drug loading, and through the use of negative pressure the rate at which a solution is perfused, as well as the rate of perfusion through the internal matrix of cortical bone, can be increased relative to that seen with bone containing all of its endogenous debris (Figure 4-4 ). However, because of the limitations on space available for drug loading, it may sometimes be desirable to have a greater amount of drug inside the matrix than the solubility would permit if the drug were s i mply perfused into the matrix Therefore a system is needed that can place a drug inside the tissue at a higher concentration than would be permitted based solely on solubility. To accomplish this, the following loading mechanism is proposed: Clean, dry bone is perfused with a saturated solution of the drug. During this first step, the amount of drug loaded approaches the maximal amount as predicted by bone vo l ume and solubility.

PAGE 89

82 The tissue is then removed from the solution and lyophilized This causes the drug to crystallize on the walls of the internal matrix (Haversian and Volkmann's canals). The tissue would then be expo s ed to additional cycles of drug loading and lyophilization to maximize the amount of crystallized drug added. This model assumes that the perfusion of the matrix with drug will be complete to the extent allowable by the internal volume. It also assumes that drug will not be lost in the lyophilization process This process is detailed in Figures 4-5a-e. The stability of both the tis s ue and the drug need to be considered this process. Due to the use of both multiple freeze / thaw and drying cycles, it is possible to damage the drug and/or the tissue (9) This is particularly true for the protein components of bone that are susceptible to denaturation caused by phase separation ( 51, 52 90 98) Conve r sely the d eposition of drug crystals within the internal matri x of the bone may enhance the initial biomechanical properties of the tissue. A primary consideration in the selection of gentamicin is its well established stability through heating and drying cycles ( 11, 72). It is therefore hypothesized that the drug loading process as described above, will deleteriously affect neither the tissue nor the antibiotic.

PAGE 90

83 80 70 1 60 ,.-._ Oil ---..J ~ 50 5 C: 40 @ c., .... 30 2 "' 20 -+-Vac ---oNo Vac 10 0 0 10 20 30 40 50 60 Time (min) Figure 4 4. The effects of vacuum on c o rtical bone on rehydration. Error bars indicate standard deviation and n=14

PAGE 91

84 \ Empty matrix of cortical bone ) a. .......... ,,.,,. ... Saturated drug solution / b Figure 4-5 A schematic of the proposed drug loading process. a) represents the internal matrix of clean cortical bone b) represents that matrix filled with a saturated solution of drug c) depicts crystal formation on the walls of the matrix following lyophilization, d) shows the same matrix being filled with a saturated drug solution for a second time, e) depicts the increase in crystal mass following a second lyophilization cycle.

PAGE 92

C. d. e. Drug crystal formation/ 85 Exclusion of water by lyophilization

PAGE 93

86 The specific aims of this work are (1) to determine whether multiple loading and lyophilization cycles will maximize the amount and minimize the variation of drug loaded into cortical bone (2) to determine if the loading process is detrimental to the biomechanical properties of cortical bone and (3) to determine if the antimicrobial activity of gentamicin is altered during the drug loading process. Materials and Methods General Experimental Overview Bone samples were machined into cylindrical segments of identical volume and surface area. The tissues were then loaded with antibiotic solutions using the aforementioned process for a total of three loading cycles. After each drying cycle the mass gained as a result of drug addition was determined gravimetrically In addition residual moisture levels were determined for each lyophilization cycle to account for mass changes due to incomplete tissue dehydration. Deionized sterile water in place of the antibiotic solution was used as a negative control to ensure the mass did not change due to an artifact of the treatment. Following drug loading the minimum inhibitory concentration of gentamicin was determined and compared to antibiotic that was not subjected to the loading process to demonstrate if there had been a change in potency. The bone was also subjected to mechanical testing to determine if the drug loading process had altered its ultimate compressive strength Sample Preparation To prepare the specimens, diaphyses of human femora and tibiae were initially prepared in standard fashion by removing any extraneous muscle ligamentous

PAGE 94

87 attachments, and the periosteum. Cortical bone segments were then cut with an oscillating bone saw from the diaphysis yielding specimens that were approximately 50 mm in length 7 mm in cortical width, and 7 mm in depth. These were then machined on a lathe into cylinders measuring 5 mm in length and 5mm in diameter. These pin s were treated to remove blood elements and residual lipids. This step consisted of exposing bone segments to a 6% hydrogen peroxide bath at 40 C for 30 minutes followed by an exhaustive lipid extraction with 40 C acetone for 15 minutes per extraction. Following treatment the samples were placed into individually labeled bottles lyophilized and held at room temperature until they were used for testing. Antimicrobial Preparation Lyophilized USP grade gentamicin sulfate powder was purchased from Sigma Chemical and stored at 2-8 C until use. The structure of gentamicin sulfate is depicted in Figure 4-6. On the day of the first drug loading cycle, a 200 mg/ml solution of gentamicin in sterile deionized water was prepared. The solution was mixed immediately prior to use to ensure all of the drug was in solution. The concentration of gentamicin in the solution was verified by fluorescence polarization immunoassay (FPIA) commercially available from Abbott Diagnostics. Because gentamicin is completely soluble in water and because a saturated solution is needed for the subsequent loading cycles acetone was used to limit the solubility of gentamicin. For the second and third drug loading cycles, a 25% v / v mixture of acetone to water was prepared to which gentamicin was added to saturation. The resulting solution from this mixture contained 75 mg/ml of gentamicin at 25C.

PAGE 95

HO NH 2 HO. 0 60-Q .... NH, 0 ;" RNHCH I R' .... 88 Gen1amicin C1 = H 2so, C2 = C2A = Figure 4-6. Molecular structure of gentamicin sulfate. R RI CH 3 CH3 H CH3 H H

PAGE 96

89 Drug Loading The mass of the lyophilized cylinders was recorded and each was placed into individual test tubes. To these test tubes, 1 ml of the gentamicin containing solution was added. Deionized water controls were also included to account for changes in mass or tissue strength that were not attributable to the presence of the drug. The test tubes were then placed into a vacuum chamber and a vacuum was applied. The cylinders were allowed to stabilize at a vacuum of <200 torr for 5 minutes ensuring complete outgassing of the matrix. After stabilization, the pressure in the chamber was returned to atmospheric level. The tissues were held at atmospheric pressure for 5 minutes and then a vacuum was reapplied. This vacuum/atmospheric pressure cycle was repeated a total of three times (Table 4-1 ). Following the loading process, the tissue samples were removed from the test tubes, wiped free of residual surface solution and placed into 30 cc glass containers for lyophilization. After lyophilization the mass of the samples was recorded. This process was repeated for a total of 2 loading cycles After each cycle, samples were removed for destructive biomechanical and drug stability analysis. Table 4-1. Summary of the drug loading process used in this experiment. The loading parameters describe the materials and solutions used for each cycle. The loading process l I' d Th. d h f h d I d' l IS a SO Iste IS process was use on eac 0 t e rug oa mg eye es. Loading Cycle Loadin2 Parameters Loading Process Used for Each Startin!!: Material Solution Cycle Cycle I Lyophilized 200 mg/ml I. Submerge in solution. cortical bone gentamicin in H O 2. Vacuum (200 torr) x 5 min Randomized 75 mg/ml 3 Atmospheric pressure x 5 min Cycle 2 specimens from gentamicin in 25% 4 Vacuum (200 torr) x 5 min Cyc l e I acetone 5 Atmospheric pressure x 5 min. Randomized 75 mg/ml 6 Vacuum (200 torr) x 5 min Cycle 3 specimens from gentamicin in 25% 7. Atmospheric pressure x 5 min. Cycle 2 acetone 8. Lyophilize

PAGE 97

90 Minimum Inhibitory Concentration Determination A microdilution broth technique was used to determine the minimum inhibitory concentration (MIC) of gentamicin for the Staphylococcus aureus used in the study. Samples collected from the drug loading experiment were eluted in a flask containing sterile 0 9% saline. The elution step was expedited by placing the samples on a rotator set at 30 RPM. The eluent was then analyzed for antibiotic concentration and MIC determination. A stock trypticase soy broth was prepared containing gentamicin at a concentration of 1 OOg/ml. Dilutions were then made with antibiotic free broth that resulted in the following antibiotic concentrations; 20g / ml, 14 g / ml, 10 g / mL, 8 g / ml, 6 g/ml, 5 g / ml, 4 g/ml, 3 g / ml 2.0 g/ml, 1.0 g / ml, 0.5 g/ml, 0.25 g/ml. Inoculum was added to each of these tubes resulting in a bacterial concentration of 1-5 x 105 cfu / ml. The addition of this inoculum resulted in a two-fold reduction in antibiotic concentration. All samples were run in duplicate. Using 96-well plates, the samples were incubated at 37 C for 18 hours to allow for growth. Positive (no antibiotic) and negative (no inoculum) controls were included. Following incubation, the wells were examined for growth as indicated by turbidity The MIC was determined by the lowest concentration of antibiotic that prevented growth. Requirements for accepting the findings were ( 1) all of the tubes with antibiotic concentrations greater than the MIC must be without growth, (2) all of the tubes with antibiotic concentrations less than the MIC must have growth, (3) the purity and identity of the organism in the first tube demonstrating growth from the MIC must be verified (4)

PAGE 98

91 the negative and positive controls must demonstrate no growth and growth respectively, (5) the concentration of the original suspension must be 1-10 X 105 cfu / ml as determined by inoculum controls, and (6) replicates from the same sample must be in agreement meaning both samples must have the same inhibitory endpoint. Biomechanical Analysis Following treatment the ultimate failure load under axial compression was determined. The method employed was adapted from the American Society for Testing and Materials (ASTM) test number D695-9 l for determining the compressive strength of a material and was performed on an MTS 858 servohydraulic mechanical test apparatus. Samples were not rehydrated prior to testing Load was applied under displacement control and applied at 25mm/minute in line with the axis of the bone (Figure 4-7). Ultimate strength was determined by the following equation: pmax (Jco mpr ess i o n = Jr ( fi r Where: Pmax is the maximum load attained in Newtons d is the diameter of the specimen Statistical Analysis Statistical analysis was performed to determine if there was a difference in the amount of drug loaded into the samples per loading cycle and to determine if there was a change in strength per cycle. Analysis was performed using a commercially available software package (Statistica '99, Statsoft Inc ) A two-way ANOVA with repeated measureswas used to determine if a significant difference existed between the

PAGE 99

92 AXIAL COMPRESSION LEVELING MOUNT SPECIMEN I I STEEL PLATE ~ ~----'--~ LOAD CELL Figure 4-7. Loading scheme for the axial compression testing of cortical bone specimens.

PAGE 100

93 treatment groups for the two outcome parameters ( drug load and bone strength as a function of loading cycle) For all tests significance was defined as a p value less than 0.05 Results The first drug loading cycle resulted in a positive change in mass of the specimens of 1.2 mg/sample with a standard deviation of 0.5 mg (Figure 4-8). This was significantly greater than the untreated samples which on average lost 0.5mg / specimen (SD =0.5). This mass gain for the samples loaded with gentamicin continued for the second loading cycle where the cumulative mass gain increased to 1. 7 mg / specimen (SD = 1. 7) for the gentamicin samples while the water control specimens had a cumulative loss of 1.3 mg/specimen (SD = 1.2). On the third and final loading cycle the gentamicin treated samples had a total mass gain of 3.2 mg/specimen (SD= 1.0) while the water control samples ended with a net loss of 1.8 mg / sample (SD = 0.8). While the multiple loading process successfully loaded on average a significantly greater amount of gentamicin into the bone than a single loading cycle (3.2 mg/sample vs 1.2 mg/sample) the variation in the amount loaded also increased. The change in mass after the first loading cycle had a range from O mg/specimen to 3 mg/specimen. By the third loading cycle this range had increased to 3mg for the minimum to 6 mg for the maximum. The change in mass provides an estimate of the amount of drug that is being loaded into the bone, but it does carry some inherent sources of error. First the

PAGE 101

94 6 5 -.-Antibiotic 4 ----0-Control = 3 ._, .. C 2 .c u "' "' .. 0 = = = U -I -2 -3 0 2 3 Loading Cycle Figure 4-8 Mass change of bone samples as a function of loading cycle (n = 15 for each timepoint and error bars indicate standard deviation).

PAGE 102

95 movement of a solution into and out of bone will wash away cellular debris from the matrix. This was seen with the water control, which had a cumulative loss of 1.8 mg/sample over the course of the three loading cycles. Secondly, there is variation associated with the extent of sample drying with the lyophilization cycles. This variation is controlled by equipment qualification, cycle validation, and product monitoring for the presence of residual moisture. During the three cycles used in this study each produced grafts with less than 0.5% residual moisture as determined by gravimetric analysis. However, because these variables could not be completely controlled for, exhaustive extractions were performed on the drug-carrying samples from the third loading cycle. The concentration of gentamicin in the eluent was then determined using the previously mentioned FPIA method. From this the amount of gentamicin originally in the bone was calculated. This analysis showed the average amount of gentamicin in each bone to be 3.7 mg (SD= 1.2). This was not significantly different than the estimate of 3 .2 mg derived from gravimetric analysis. The MIC for the strain of S. aureus (ATCC 6358) was determined to be 0.7 g/ml for stock gentamicin solution prior to drug loading. The MIC determinations for the gentamicin eluted from the bone specimens after cycles 1, 2, and 3 were 0.5 g/ml, 0.4 g/ml and 0.4 g/ml respectively (Table 4-2). Two bone samples were tested for each cycle and the MIC for each sample was tested in duplicate. Based on this data there was no observable loss in potency of the gentamicin as a function ofloading cycle

PAGE 103

96 Table 4-2 Results of MIC determinations for gentamicin against S. aureus Control 1 t b. t d t 1 d. samp es were no su ec e o any oa mg process. Loading Cycle Minimum Inhibitory Concentrations Replicate 1 Replicate 2 Geometric Mean Control 0 .68 0.68 0.68 Cycle 1 0.84 0 .31 0.51 Cycle 2 0.21 0.84 0.42 Cycle3 0 .78 0 .23 0.42 The ultimate compressive strength of the bone was also evaluated as a function of loading cycle and is summarized in Figure 4-9. The mean compressive strength of the starting material was 303 Mpa (SD= 76). This was significantly different from the mean strength after any loading cycle. The water controls were included to determine if there was a decrease in strength of the bone due to the multiple lyophilization cycles or an increase in the strength of the bone due to the presence of the antibiotic within the internal matrix. The resulting strength of the drug loaded bone after the third cycle was 250 Mpa (SD= 63) for the drug loaded bone and 215 Mpa (SD= 7) for the water control. Although the treatment resulted in a lower compressive strength, there were no differences between the drug-loaded group and the control group. Discussion Human cortical bone was selected as a potential carrier for the local delivery of gentamicin because of its ability to incorporate into the recipients' tissue ( 12, 17, 100). This would eliminate the need for a second surgical procedure to remove the device after delivery was complete. It was selected over cancellous bone because it is stronger and

PAGE 104

-C. 400.0 350.0 300 0 250.0 -"Cl 200 0 ..J 150 0 .E! ; r.. 100.0 50 0 0 97 --&Antibiotic -0---Control 2 3 Number of Loading Brents Figure 4-9. Cortical bone strength as a function of drug loading cycle. Four replicates were evaluated for each loading cycle. Error bars depict standard deviation.

PAGE 105

98 more regular in density. Theoretically this would allow for the preparation of grafts containing a consistent dose of antibiotic while remaining structurally sound. However realizing this potential requires navigation through several significant obstacles. First the lumenal space within cortical bone can only be loaded with a drug if it is empty. In its natural state, this space is filled with blood lipids and other cellular debris. The removal of this material is therefore a necessary first step in the drug loading process. If the cleaning process is not complete the result will be an allograft that is incompletely and heterogeneously loaded with drug The data offered by the water control samples suggests that the tissue was not completely cleaned of debris prior to the start of the study. This would account for the loss in mass observed with the first two loading cycles. It is unlikely that this decrease in mass was attributable to the loss of mineral from the bone as the contact time in solution was only 30 minutes per loading cycle. The data also supports the hypothesis that multiple drug loading cycles significantly increase the amount of drug that is able to be loaded into cortical bone over a single loading cycle. Only three loading cycles were evaluated due to the length of the procedure Each lyophilization cycle is 48 and a day is needed for each loading procedure. This resulted in an elapsed time of 10 days to load the bone using the three cycles. The development of methods that decrease this time would be beneficial in the manufacturing setting. Of even greater concern than time is the variation in drug content of the samples The gravimetric analysis of how much gentamicin was in each sample covered a range of 9 mg. Of the 23 drug loaded specimens evaluated after the third cycle, one had a cumulative loss of 3 mg. This data was not supported by the FPIA results that provided a

PAGE 106

99 similar estimate of the mean (3 .6 mg / sample) but much tighter range ( 1.4 -4 8 mg / sample). This suggests that gravimetric analysis may be useful for the large scale monitoring of the process but may not provide accurate enough information about the drug content of an individual sample to be useful. The reasons for the loss of mass balance was likely due to variation in the lyophilization and loss of endogenous material from the matrix. The MIC data demonstrated that the process did not lower the activity of gentamicin against S. aureus. This is not surprising as the stability of gentamicin has been previously demonstrated in studies involving the polymethylmethacrylate (PMMA) bead (72, 94). During the preparation of PMMA beads, the drug is exposed to very high temperature ( > 100 C) without any loss in activity. S. aureus was chosen for this study because of its relevance to orthopaedic infections. The Staphylococci are by far most commonly isolated organisms from orthopaedic wounds (26, 31, 108) Therefore any drug delivery system proposed for the treatment or prevention of osteomyelitis would need to provide adequate coverage of these organisms. Historically, gentamicin has been used with success in the treatment of orthopaedic infections involving S. aureus (61, 82). It is critical that the strength the bone is not compromised during the loading process. In this study the loaded specimens were evaluated for their biomechanical strength under axial compression In this procedure, a force is placed on the tissue in the direction of the axis of the bone in-vivo This mode ofloading was selected because it represents the direction that forces are typically applied to a structural allograft in-vivo (8). Because of the presence of the drug in the bone in a solid form, it was hypothesized that the gentamicin containing bone would experience an increase in strength. However,

PAGE 107

100 the data generated from this study did not reveal a statistically significant change in strength. Conclusions The use of multiple drug loading cycles can increase the amount of drug that is able to be loaded into cortical bone as compared to a single loading scheme. This, however, comes at the expense of an increase in specimen to specimen variation The use of gravimetric analysis for estimating the average amount of drug loaded into a population of bone samples appears to have value however the use of a more specific assay such as FPIA is required for the evaluation of individual specimens. Additionally, this drug loading process did not lower the potency of the antibiotic or the biomechanical strength of the tissue. However there is a need for the development of a faster drug loading process one that will not result in as much intersample variation.

PAGE 108

CHAPTERS IN VITRO PHARMACOKINETIC MODELING OF GENT AMICIN RELEASE FROM CORTICAL BONE ALLOGRAFTS Introduction The implantation of medical devices impregnated with therapeutic agents has been proposed for the treatment of several conditions, ranging from osteomyelitis to neoplasm to augmenting bone fusion and osseous void repair(2, 61, 73, 82, 84, 94). The advantage of this type of delivery system is that drug concentrations are highest at the site of action, thus requiring lower concentrations in the general circulation. For some agents, this means that therapeutic concentrations can be obtained at the intended site, with negligible concentrations of the drug found systemically, thus lowering the potential for adverse reactions. In addition, sustained delivery mechanisms could eliminate the need for the use of multiple doses through their role as a depot site. To date, numerous studies have been conducted on the release kinetics of drugs from synthetic bone-replacement material (3, 11, 72, 73, 77, 83). Porous glass, apatatic calcium phosphate, and other self-setting bioactive bone cements have been investigated. Studies have indicated that drug release in therapeutic concentrations is possible from a wide range of bioactive and inert substances. Human bone offers several potential advantages over synthetic material. The most relevant of these is its ability to incorporate and eventually be replaced by the recipient's existing bone matrix, through remodeling (17, 100, 120, 124). Remodeled bone is vascularized, living tissue. This is in sharp contrast to synthetic implants, which either 101

PAGE 109

102 degrade or remain as permanent fixtures. Implants that are biodegradeable are problematic because the void that is left behind is often filled in with scar tissue (94). Non-biodegradable implants can require a second surgical procedure for removal. This is the case with the polymethylmethacrylate (PMMA) beads that are impregnated with antibiotics (34, 61 ). Permanent fixtures, on the other hand, have a tendency to wear on the communicating surface of the bone to which they are attached, leading to the deterioration of the recipients' bone. This potential complication is seen with prosthetic hip replacements that cause post-operative fracture of the femoral shafts into which they were inserted (124) Conventional bone allografts, however are far from ideal drug release systems. In addition to the presence of residual endogenous substances and challenges associated with the loading of cortical bone, conventional bone allografts are irregular in shape, volume, and surface area. These irregularities create a problem with the ability to reproducibly prepare a standard dosage and to attain predictable release rates. To account for these variables special consideration needs to be given the grafts being prepared for the purpose of drug delivery in these situations. The tissue should be machined into standardized sizes and shapes with known volumes and surface areas. In addition, the tissue must be homogenous in density (i.e. from a single bone type), thus being free of transition zones containing a mixture of both cortical and cancellous bone. The objective of these studies was to investigate the release rates of gentamicin and other model compounds, from standardized cortical bone segments. Additionally, the effectiveness of this type of antibiotic delivery system was studied for the elimination of adventitious contamination.

PAGE 110

103 Background Cortical bone is comprised of a series of interconnecting tunnels that supply the permanent cells of bone, the osteocytes, with nutrients. The primary unit of cortical bone is the osteon (Figure 5-1 ). An osteon is centered on a haversian canal, which runs parallel to the axis of the shaft in long bones. This serves as a conduit for blood vessels and nerves. Haversian canals are connected transversely to each other and to the periosteal and endosteal surfaces of the bone via Volkmann's canals. Each haversian canal is surrounded by 4-20 layers of lamellae. Osteocytes reside in small spaces between adjacent lamellae referred to as lacunae. Connecting the lacunae to each other and ultimately back to a haversian canal, are very small, canaliculi. These passageways are formed when bone mineral is laid down around the filopodial processes of the osteocyte. The small size of the canaliculi can only supply enough nutrients to support about 15 osteocytes. This anatomical arrangement forms an interesting matrix when the cellular component is removed. Comparatively, the haversian and Volkmann's canals are much larger than the small canaliculi that connect them to the lacunae. From this it could be predicted that the release of drug from the haversian canals would be fast and that the drug remaining in the lacunae would diffuse through the smaller canaliculi slowly. It is therefore hypothesized that the release profile would be bi-exponential, with a rapid release of drug from haversian canals followed by a slower release from lacunae (Figure 5-2).

PAGE 111

104 The O s teon Ca n a liculi Y o lkmann's Can a l H avers i a n Cana l Figure 5-1. The microarchitecture of cortical bone

PAGE 112

105 Lacunae and k12 Haver s ian and k10 Drug Depo s ition Canaliculi Yolkmann s C ompartm e nt .... Canals .... k21 ko1 Figure 5-2 Proposed two-compartment model for the release a drug from human cortical bone used as a delivery system. When sink conditions are maintained and kio> > k12, k 2 1 and k o can be ignored.

PAGE 113

106 Materials and Methods Overview In this study, cortical bone segments were machined from diaphyseal sections of human cadaveric femora and tibiae. The bone segments were cleaned and impregnated with gentamicin, cefazolin, FITC complexed to 10 and 20 kD proteins (10 kD FITC and 20 kD FITC), and bovine hemoglobin (Hb ). The samples were then placed in elution baths that maintained sink conditions, and the release of drug over time was measured. To evaluate the effectiveness of the antibiotic loaded specimens on eliminating adventitious contamination, samples were first spiked with Staphylococcus aureus. The resulting bioburden was then monitored over time. Sample Preparation To prepare the specimens, diaphyses of human femora and tibiae were initially prepared in standard fashion by removing any extraneous muscle, ligamentous attachments and the periosteum. Cortical bone segments were then cut with an oscillating bone saw from the diaphysis yielding specimens that were approximately 50 mm in length, 7 mm in cortical width and 7 mm in depth. These were then machined on a lathe into cylinders measuring 5 mm in length and 5mm in diameter. These cylinders were treated to remove blood elements and residual lipids as described in earlier chapters. Following treatment the samples were placed into individually labeled bottles, lyophilized and held at room temperature until used for testing. Lyophilized gentamicin sulfate was purchased from Sigma Chemical and stored at 2-8 C until use. A 500 mg/ml solution of gentamicin in sterile deionoized water was

PAGE 114

107 prepared. The solution was mixed for 15 minutes immediately prior to use to ensure solubility. The mass of each lyophilized cylinder was recorded and each was placed into individual test tubes. To these test tubes, 1 ml of the gentamicin containing solution was added. The test tubes were then placed into a vacuum chamber and a vacuum was applied. The cylinders were allowed to stabilize at a vacuum of < 200 torr for 5 minutes ensuring complete outgassing of the matrix. After stabilization, the pressure in the chamber was returned to atmospheric. The tissues were held at atmospheric pressure for 5 minutes then a vacuum was reapplied. This vacuum/atmospheric pressure cycle was repeated a total of three times for an elapsed contact time of approximately 30 minutes. Following the loading process, the tissue samples were removed from the test tubes, wiped free of r esidual surface solution and placed into 30 cc glass containers for lyophilization .. After lyophilization the mass of the samples was recorded to estimate the change in due to the presence of drug. Four compounds of varying molecular weigh were also studied using the same loading procedure; 1) cefazolin sodium at a concentration of 500 mg/mL 2) FITC dextran complex (MW= 10 kD) at a concentration of 500 mg/mL in Tris buffered saline with a pH of 8-8.5 3) FITC-dextran complex (MW= 20 kD) at a concentration of 1000 mg/mL in Tris buffered saline with a pH of 8-8.5 4) bovine hemoglobin, Hb, (MW= 68 kD) in saline a t a concentration of 1000 mg/mL. Each impregnated segment was placed into a glass vial containing 20 cc of phosphate buffered saline (PBS) at a pH of 7.4. Tris buffered saline (pH 8) was used for the FITC specimens. A stir bar was added to each vial to disrupt the static water layer

PAGE 115

108 and to ensure a homogenous distribution of solution. At various time intervals the samples were transferred into vials containing fresh media. The resulting solution from each vial was analyzed for the concentration of the compound. To elucidate the effect of the bone matrix on the slowing of the release of antibiotic from that of simple dissolution, controls were added that evaluated rate of dissolution alone. Cefazolin crystals ( 4 mg) were added to a vial under the same conditions as listed above except in the absence of a device. Samples of the solution were then removed from the vials through a 0.1 m filter and analyzed to determine the cefazolin concentration over time. Fluorescence Polarization Immunoassay (FPIA) commercially available from Abbott Diagnostics was used for the detection and quatification of gentamicin. Cefazolin concentration was approximated by using the bioassay described previously. The FITC and Hb specimens were analyzed spectrophotometrically The FITC specimens were measured at a wavelength of 492 nm and the Hb was measured at 519 nm. Ranges of known concentrations were prepared and a standard curve was generated. Each extraction sol tion was first blanked with extraction buffer from a bone sample that was not impregnated with a compound. This was done to eliminate interference from any potential for substances leaching from the bone other than the compound. Pharmacodynamic Model The effectiveness of an allograft based drug delivery system was investigated using in-vitro model. Cortical bone pins loaded with gentamicin and cefazolin were spiked with 20 L of a solution containing approximately 107 CFU/mL of Staphylococcus aureus (SA) to produce the resulting bioburden of approximately 5 x 105 CFU. A control group of bone pins that did not contain an antibiotic were simultaneously

PAGE 116

109 spiked with bioburden. Additionally, a treatment group was included wherein all of the antibiotic potentially released from the bone was placed into the first starting solution. This group w a s added to determine if there was a specific targeting advantage gained with allograft based delivery. Samples from each of the groups were immediately tested to determine the starting bioburden (T 0). The remainder of the specimens were then placed into flasks containing 1000 mL of sterile saline maintained at 3 7C. The saline in the flasks was exchanged every 30 minutes to prevent the accumulation of effective concentrations of antibiotic within the solution. At various time points samples were removed and subjected to destructive bioburden analysis. Bioburden determination employed the use of a 0.45 m filter flask that allowed for the separation and washing of the bacteria from the antibiotic solution. By using this technique, residual antibiotic is not carried over onto the culture medium, which could possibly cause interference. Results The loading procedure resulted in grafts with a mean gentamicin content of 3.2 .8 mg/specimen. This estimate was determined by gravimetric analysis and confirmed by a destructive elution procedure whereby several specimens were morsellized, then exposed to an exhaustive extraction procedure. The concentration of the resulting eluent was then dete r mined by FPIA. The use of the gravimetric analysis permitted the non destructive evaluation of each of the specimens used in the study, whereas the destructive assay using F P IA allowed for an accurate estimate of the mean drug concentration of the specimens as a population. The same procedure was used to determine the amount of the other compounds that had been loaded into the bone specimens. Cefazolin was present on average at 3.8

PAGE 117

110 0.5 mg/specimen, 10 kD FITC at 5.2 1.2 mg/specimen, 20 kD FITC at 9 1 0 1.9 mg/specimen and Hb at 11.2 2.1 mg/specimen. Since each specimen contained a different amount of the test compound due to the inherent variability associated with loading cortical bone, a method of evaluating the release data from all of the specimens as a group was required. By knowing the starting drug concentration of the individual specimens, the normalization of the release data to "percent released" was possible. This was accomplished by dividing the cumulative amount released for a given time point by the total amount contained within the sample as determined by exhaustive analysis determined at the completion of the study. The release of gentamicin was rapid and followed a bi-exponenetial profile (Figure 5-3). Approximately 90% of the drug had been released within the first three hours. This was followed by a slow, more sustained release over the next 48 hours. The release of cefazolin followed a similar release pattern (Figure 5-4). The dissolution control was in complete solution by the first time point (1 min.). The release of the larger compounds was more protracted as molecular weight increased (Figures 5-5 and 5-6). From the gentamicin data an equation describing the terminal portion of the release profile was fitted, allowing for the estimation of B (0.127) and~ (-0.0007 min). From this, a n~sidual plot (Figure 57) was constructed that allowed for the estimation of A (0 873) and a (-0 0225min). This data was used to formulate a prediction curve that followed the equation: fR1 = 0.873eat + 0 127eP Where: fR1 = the fraction of gentamicin remaining in the bone at time (t) a = 0.0225 min1 /J= 0 0007 min' t = elapsed time in minutes

PAGE 118

111 100% 90 % 80% 70 % = a 60 % ; 50% e 0::: 40 % 30 % 2 0 % 10% :c---------D:-----'IJ: 0 % 0 200 400 600 800 1000 1200 1400 1600 1800 Time (min) Figure 5-3. The observed release profile for gentamicin from cortical bone cylinders under well-st i rred sink conditions. Error bars indicate standard deviation and n=l5 observations for each time point.

PAGE 119

100 90 80 "i: 70 c,s 60 .!: : 50 = ; 40 e a., 30 i::i::: 20 10 112 0-+-----~-----~----~~----~ 0 50 100 Time (min) 150 200 Figure 5-4. The observed release profile for cefazolin from cortical bone cylinders under well-stirred sink conditions. Error bars indicate standard deviation and n=4 observations for each time point.

PAGE 120

113 100 90 80 -t:r--10 kD FITC 'i: -a20 kD FITC .... 70 in .5 60 50 = ; 40 8 30 ci::: 20 Q 10 0 0 5 10 15 20 25 T i me (hrs) Figure 5-5 The o b served release profile for 10 kD and 20 kD FITC from cortica l bone cylinders under well-stirred sink conditions Error bars indicate standard deviation and n = 5 specimens for each time point.

PAGE 121

114 100 90 80 'i: .... 70 in 60 .5 Oil 50 .5 = ; 40 e Q,j 30 20 10 0 0 20 40 60 8 0 100 Time (hrs) Figure 5-6. The observed release profile for bovine hemoglobin from cortical bone cylinders under well-stirred sink conditions. Error bars indicate standard deviation and n= 16 observations for each time point.

PAGE 122

115 100/o y = 0.873e--0.0225x y = 0.127e--0.0007x 0 200 400 800 1000 1200 1400 l(i)() 1 800 Tirre (min) Figure 5-7. Residual plot for gentamicin in the determination of a and A as depicted by the triangles. The boxes represent the data points used to determine the terminal slope.

PAGE 123

116 l .OE+07 l.OE +06 ---------------f-~--w-----------------, l .OE+OS -.----~[:__-----.~---. ~ C ~ on ~ t = ro ~ l ~ I 1.0E+04 -0Gentamicin -lsCefazolin l .OE+03 Cefazolin Solution l .OE+02 l.OE +Ol l .OE+OO 0 100 200 300 400 500 Figure 5-8 Bioburden levels following adventitious contamination of cortical bone pins loaded with antibiotic and subsequently placed into a saline bath. Control samples did not contain an antimicrobial agent. Samples with all of the antibiotic in solution (Cefazolin Solution) were also evaluated to demonstrate the difference in antimicrobial effect due to targeting by the delivery system Error bars indicate standard deviation

PAGE 124

117 The results of the pharmacodynamic portion of this study demonstrate the rapid elimination of bioburden from the samples loaded with gentamicin (Figure 5-8). By the 180 minute time-point there was no detectable bioburden ( assay sensitivity is < 10 CFU) The cefazolin samples also had significant bioburden reductions although> 100 CFU's of SA was detected at each time-point. There was a slight initial reduction in bioburden in the cefazolin solution group however this reduction was not maintained. The control group maintained bioburden levels greater than the starting inoculum throughout the study and no contaminating organisms were identified during in any of the cultures. Discussion The augmentation of allografts by impregnation with drugs or growth factors has been proposed. Potential uses included antibiotic therapy to fight existing infection or as prophylaxis against post-operative infection, growth factor therapy to expedite allograft incorporation, or antineoplastic agents as an adjunct to tumor resection. Until recently, there have been limitations on the extent to which the internal matrix of cortical bone could be cleaned and subsequently loaded with drug. This has prevented the study and use of bone as a potential drug delivery system. However, as newer technology now permits the effective cleaning and loading of cortical bone, investigations into its potential as a delivery system are possible. In this paper, a model for the release kinetics of drug from human cortical bone is proposed and investigated. The basis for the bi-exponential model is derived from the anatomical compartments that remain when the internal matrix of bone is substantially cleaned of its cellular debris. To control for as many variables as possible, only human

PAGE 125

118 cortical bone was used and all of the specimens tested were machined into cylinders of identical size. The results generated in the study support the proposed bi-exponential model. Several in-vitro models exist for the evaluation of drug release from a matrix device, however there has been no clear benefit demonstrated by one over another. During this study an in-vitro, well-stirred model was employed that maintained sink conditions throughout the course of the experiment. When an insoluble matrix is used to carry a soluble drug, release is the result of the medium entering the matrix leading to drug dissolution and finally diffusion out of the matrix (3). Because gentamicin is very water soluble this model may overestimate the release rate of a drug from a device placed into an orthopaedic wound due to limitations on the amount of extracullular fluid that is available. Regardless of the differences that may exist between the in-vitro model and in vivo conditions, the release of gentamicin from the matrix was obviously too rapid for applications that require sustained release. It can be hypothesized that different indications would have different release requirements based upon their intended application. We can estimate these requirements by examining the existing practices that are used with alternate delivery systems to achieve the same goal. The therapeutic application, wherein the device would be used to treat an existing infection, would likely require sustained release for 1-3 weeks to be effective. If the device was being used to prevent an infection where one would likely occur if no treatment were used, therapy would probably not need to be as protracted, lasting 37 days. In simple chemoprophylaxis optimal concentrations only need to be attained during and immediately following the introduction of bioburden (i.e. during and

PAGE 126

119 immediately following surgery) (24, 93, 122). Therefore, to be effective in this setting, an allograft would only need to release an effective dose of the drug for the first few hours following surgery. The release profile obtained for gentamicin in this study would clearly not produce effective concentration over an extended duration. Therefore, it can be assumed that without modification, this device would likely not be useful in the treatment of an existing infection or the prevention of infection from a grossly contaminated wound. However, the sudden burst of antibiotic from the unmodified graft evaluated in this study may be helpful in the prevention of infection in routine allograft surgery. These surgeries are typically elective in nature and are therefore planned in advance. This planning usually affords the surgeon the luxury of operating within a clean wound, under aseptic conditions on an otherwise healthy patient. This greatly reduces the bioburden introduced into the wound and therefore lowers the requirements for effective prophylaxis. The rapid release of gentamicin into the surrounding media that was observed in this study during the first three hours suggests that if implanted, initial concentrations at the wound site would be above that of the minimum inhibitory concentration for most organisms. In fact it is likely that some organisms found to be resistant to gentamicin at typical systemically delivered therapeutic concentrations ( < 10 g/ml) would be sensitive at the much higher concentrations found immediately surrounding the graft. This hypothesis was supported by the data collected with the pharmacokinetic model wherein gentamicin was able to reduce a starting bioburden of greater than 105 CFU of Staph aureus to below detectable limits in less than three hours. The cefazolin

PAGE 127

120 c ontaining devices were not as effective as those containing gentamicin were, however they did significantly reduce the initial bioburden. Although it is recognized that the two models have slight differences that preclude their direct comparison, the data generated in t he previous chapter using a model simulating systemic administration of cefazolin was less effective at reducing the bioburden in nearly double the time. Additionally, the inclusion of a control group that evaluated the antimicrobial effect observed if all of the drug were present in the first starting solution produced results only slightly lower than t he control. This data suggests that the optimal delivery system for the prevention of p ost-operative allograft infection is one of local delivery originating from within the a llograft. The use of this technology for the treatment of conditions such as chronic osteomyelitis or for the prevention of infection following a severe open fracture requires t he system to be modified to allow for a more sustained release. This is because the starting bioburden is higher than that seen with incidental contamination in routine surgical procedures ( 46). There are several potential approaches to attaining slower r elease kinetics. It has been established that the rate of release from an insoluble matrix i s proportional to the solubility of the drug into the surrounding medium (3, 11, 73, 84). Gentamicin is completely water-soluble and consequently releases rapidly. Therefore, t he use of a less water-soluble antibiotic would result in a slower release rate. A potential candidate with lower water solubility is ciprofloxacin. Ciprofloxacin is a fluoroquinolone with a solubility of only 0.16 mg/ml in an aqueous solution (pH of7.4 at 37C) (3, 32). I n addition, ciprofloxacin, a DNA gyrase inhibitor, has a similar spectrum of antimicrobial activity. The draw back to the use of ciprofloxacin for the prevention of

PAGE 128

121 orthopaedic infection is its tendency to cause cartilage degeneration in patients with still maturing skeletal systems and its tendency to chelate with calcium (86). Another potential mechanism that could be used to slow the release rate would be the inclusion of a bioabsorbable polymer. For example, gentamicin could be mixed with a dilute solution of gelatin, loaded into the bone, and lyophilized. The gelatin being less soluble than gentamicin would form an additional matrix within the infrastructure of the bone thus increasing the tortuosity of the device and slowing the release. The use of gelatin or any large polymeric substance, would need to be balanced against the effects that inclusion of the polymer would have on the loading of the drug. Cortical bone is different from other implantable delivery systems in that it must be loaded as an intact structure. Consequently, there are limitations on the size and viscosity of the mixtures that can be effectively loaded. If the concentration of gelatin is too high the viscosity will prevent complete and/or timely loading. Higher molecular weight compounds were included in this study because of their potential relevance to the delivery of growth factors and other biological compounds Since the model is diffusion based, the change in release rate as a function of molecular weight is predicted. Since both the FITC-protein compounds and the hemoglobin have much higher molecular weights than the two antibiotics evaluated it is not surprising that their release was significantly slower that that of gentamicin and cefazolin. Because the remodeling and repair of bone is a lengthy process, this sustained release pattern is desirable to the application of delivering bone growth factors.

PAGE 129

122 Conclusions The results of this study demonstrated the rapid release kinetics found with the small water-soluble compounds of gentamicin and cefazolin. Although this release pattern was rapid, its effectiveness at clearing adventitiously placed bioburden was demonstrated. This type of delivery system may prove useful in the prevention of post operative infection following the routine implantation of an allograft, where the starting bioburden is low and adventitious in nature. Future work should concentrate on developing methods to load the tissue with a consistent dose of drug and to modify the release of the drug to attain a more protracted therapy. In addition a better understanding of the influence of surface area and volume on the release kinetics of a drug from cortical bone is needed to allow for the application of this technology to existing grafts. The ultimate goal of this work should be to formulate biocompatible allografts of functional shapes and sizes that will release therapeutic agents within useful concentrations and for an appropriate length of time.

PAGE 130

CHAPTER6 CONCLUSIONS AND A DISCUSSION ON AREAS OF POTENTIAL APPLICATION AND FUTURE WORK Conclusions The development of an allograft based drug delivery system (ABDDS) was initially motivated by the high rate of post-operative infection associated with allografts These infections are associated with patient morbidity and sometimes mortality, and carry a significant economic burden, each requiring additional treatment ranging from intravenous antibiotics to surgical revision and graft removal. Currently there exist four methods of preventing disease transmission with allografts: donor screening, graft disinfection aseptic handling, and chemoprophylaxis Donor screening for infectious diseases and extensive tissue processing typically result in the production of a sterile allograft. It has therefore been concluded that the cause of post-operative infection is the introduction of adventitious contaminants during surgery and that these organisms persist despite prophylactic antibiotic administration. The failure of traditional prophylaxis is hypothesized to be linked to compositional or textural properties unique to the allograft. This hypothesis is supported by data that show post-operative infection rates using synthetic implants are lower than when allografts are used in similar procedures. There exist other significant applications for an ABDDS in orthopaedic surgery In particular there is an almost 50% rate of infection associated with some types of open fractures (82) Synthetic implants containing an antibiotic have been used in the treatment of these conditions since the 1970s, and today, the majority oflocal antimicrobial therapy 123

PAGE 131

124 is accomplished with the polymethylmethacrylate (PMMA) gentamicin bead (11, 72). With this type of drug delivery therapeutic concentrations are attained at the infection site while systemic toxicity is avoided because serum drug concentrations are 10-100 times lower than with IV administration (33 34, 53). However, the PMMA beads serve no function other than being a non-resorbable carrier for the antibiotic. Because of this additional surgery is generally required to remove the device after the drug has been delivered, adding the expense and risk of a second procedure ( 61 ). An allograft based drug delivery system would have the major advantage over PMMA in that allografts are bioincorporable, meaning they will eventually integrate into and become part of the recipients own vital tissues (15 17 22 50) This eliminates the need for a second surgical procedure to remove the device, as is necessary with the PMMA bead. In addition allografts are indicated in orthopaedic procedures for reasons other than drug delivery. Specifically, allografts are frequently used to lend structural support or promote new bone formation These added advantages do not come at the expense of previously mentioned benefits associated with local drug delivery systems. The aims of this project were to (1) to use a dynamic in-vitro model to identify factors that may predispose allograft bone implants to post-operative infection, (2) to use a rational set of criteria to identify the optimal antimicrobial agent for further investigation within an allograft based system (3) to evaluate the effectiveness of a potential drug loading procedure, and ( 4) to characterize the release profile of gentamicin from cortical bone. It is hoped that this work will serve as the foundation for the future development of a biocompatible drug delivery system that will reduce the incidence or severity of orthopaedic infection and the additional costs that accompany them.

PAGE 132

125 Allograft Composition The overriding hypothesis for this set of experiments was that complex architectural features and residual lipids undermine the effectiveness of cefazolin The data acquired in this study support the hypothesis that antibiotics are significantly more effective when used with cortical bone tissue with regular surface features and minimal porosity than on tissues with greater architectural complexity. However, the data did not conclusively support the hypothesis that residual lipids carried on the tissue decreased antibiotic effectiveness. This apparent paradox may be partially explained by an alternate hypothesis The surface of tissues with high amounts of lipids will be receptive to absorbing externally applied aqueous solutions such as the inoculum. It is possible that the bacterial suspension coated the surface of the tissue and did not penetrate appreciably into the matrix. The data suggests that materials that can absorb contamination to a greater extent will realize a diminished antibacterial effect with short-term exposure to antibiotics. Therefore surgeons and other health-care providers responsible for allograft preparation should take care to guard against contamination during the reconstitution of clean cancellous bone grafts as introduced bacteria may migrate further into these tissues than with other less porous tissues. Additionally the data demonstrates that the application of threads or other surface modifications that both increase surface area and allow exposure to the haversian canals, decrease the ability of cefazolin to reduce bacterial bioburden in-vitro. As allograft processing technology becomes more sophisticated it can be assumed that the complexity of the grafts will increase. This should further encourage researchers to develop and

PAGE 133

126 incorporate antimicrobial strategies into allografts instead of relying upon perioperative chemoprophylaxis to prevent infection. Antibiotic Evaluation The selection of an optimal antimicrobial agent was made by comparing classes of antibiotics against a set of characteristics determined to be necessary for success. The requirements contained four categories: appropriate antimicrobial spectrum drug stability potential for adverse reactions, and drug potency. The results of this review indicate that gentamicin is currently the optimal antibiotic for inclusion in an allograft based delivery system. For the past 30 years gentamicin has developed a record of being safe and efficacious in the prevention and treatment of orthopaedic infections when used in similar local drug delivery systems. However, it is recognized that the selection of an antimicrobial agent based on a review of the literature by no means guarantees its success in practice leaving a significant amount of research remains to confirm this selection. Bacterial resistance as well as the possibility of the development of a better antimicrobial agent mandates that this selection be continually reexamined. In addition the physiochemical properties of gentamicin may prove to be a significant challenge in further development of the ABDDS. This selection may need to be revised to employ a less water-soluble compound if the release kinetics of gentamicin cannot be slowed using an alternate method.

PAGE 134

127 Drug Loading This set of experiments demonstrate that the use of multiple drug loading cycles can inc r ease the amount of drug that is able to be loaded into cortical bone as compared to a single loading scheme. The proposed procedure resulted in an increase in drug load of 1.9 mg / specimen over a single loading cycle. This increase in drug load, however, came a t the expense of an increase in specimen to specimen variation. The variation associated with allograft drug loading therefore, remains a significant challenge to the further development of this project as is would almost assuredly be questioned by regulatory authorities. Because the density of cortical bone as a material is rather uniform from donor to donor and because this density is directly related to the porosity of the matrix, it can be hypothesized that as the amount of drug within the bone approaches the maximum amount possible the variation will decrease. This should be a motivating factor for further development of more efficient and complete drug loading procedures. Fortunately, this drug loading process did not lower the potency of gentamicin. This finding is not particularly surprising as the drug loading procedure is relatively mild in nature so that the biological properties of the bone are not affected. The stability of gentamicin has been confirmed under more drastic conditions such as significantly elevated temperature. Due to the fact that allografts are often used to replace or repair structural elements of the human skeleton it is critical that the strength the bone is not compromised during the loading process. The data generated from this study did not reveal a statistically significant change in strength as evaluated by axial compression The presence of the drug in the bone in a solid form was hypothesized to increase the strength of the material however this was not observed. The reason for this may be attributable to

PAGE 135

128 an insufficient amount of drug being loaded into the bone. Perhaps this effect will be realized when more effective methods for drug loading into bone are developed Drug Release Human cortical bone cylinders machined to identical size were used to evaluate the release kinetics of gentamicin. The results of this study lend support to the hypothesis that drug release from human cortical bone follows a bi-exponential equation relating to the two distinct anatomic compartments that remain after cellular debris is removed. The initial release was very rapid and would probably not provide antibiotic coverage for a sufficient length of time to be effective against conditions that involve a high starting bacterial bioburden such as an open fracture or pre-existing osteomyelitis However, this release pattern might be useful in the prevention of post-operative infection following the routine implantation of an allograft, where the starting bioburden is low and adventitious in nature. During this study an in-vitro, model was employed that maintained sink conditions throughout the course of the experiment. When an insoluble matrix is used to carry a soluble drug, release is the result of the medium entering the matrix leading to drug dissolution and finally diffusion out of the matrix (3) Because gentamicin is very water soluble this model may overestimate the release rate of a drug from a device placed into an orthopaedic wound due to limitations on the amount of extracellular fluid that is availab le. Regardless, the use of this technology for the treatment of conditions such as chronic osteomyelitis will require the system to be modified to allow for a more sustained release. The use of a less water-soluble antibiotic such as ciprofloxacin would likely

PAGE 136

129 result in a slower release rate. Another potential mechanism that could be used to slow the release rate would be the inclusion of a bioabsorbable polymer, increasing the tortuosity of the matrix and slowing the release. Future Work Perhaps the most obvious area requiring further investigation of this work is the resulting biocompatiblity of the graft. A major advantage to the AB DDS is the assump t ion that allograft tissue loaded with drug will incorporate into the recipients' tissue as does a conventional allograft. For this to happen, the drug must not alter the allograft or result in toxicity to the surrounding tissue. Previous work has demonstrated that the indiscriminate chemical modification of allografts, as is done during ethylene oxide sterilization, can result in a product that is void of the attributes that supported its original indication and may even lead to toxicity. Therefore, biocompatibility studies should be initiated in the next phase of development as a necessary precaution prior to large scale testing. Future work should also include developing methods to load the tissue with a more consistent dose of drug This product is unlikely to be accepted by regulatory authorities if the dose of antibiotic cannot be assured within reasonable limits. One potential strategy for offsetting (not alleviating) the significance of consistent dosing would be a consistent indication for multiple devices to be used in one wound site. For example, if 5 pellets were always indicated per cubic centimeter of wound space. By having several devices in close proximity, the criticality of device to device variation would be lessened.

PAGE 137

130 The release of the drug obviously needs to be controllable to a greater extent. Generic methods of controlling the release of drug from allografts will permit their application beyond a single agent. In addition, a better understanding of the influence of surface area and volume on the release kinetics of a drug from cortical bone is needed to allow the application of this technology to existing grafts. The ultimate goal here is the production of allografts in functional shapes and sizes that will release therapeutic agents in useful concentrations and for an appropriate length of time.

PAGE 138

REFERENCES 1. Aichelmann-Reidy ME Yukna RA. Bone replacement grafts. The bone substitutes Dent Clin North Am 42 (3): 491-503, 1998. 2 Allababidi S, Shah JC. Efficacy and Pharmacokinetics of Site-Specific Cefazolin Delivery Using Biodegradable Implants in the Prevention of Post-Operative Wound Infections. Pharm Res 15 (2) : 325-334 1997. 3 Allababidi S Shah JC. Kinetics and Mechanism of release from Glyceral Monostearate Based Implants: Evaluation of Release in a Gel Simulating in Vivo Implantation J Pharm Scis 87 (6): 738-744 1998 4. An YH Friedman RJ. Pre v ention of sepsis in total joint arthroplasty. J Hosp Infection 33: 93-108 1996 5. Ari z ono T, Iwamoto Y Okuyama K, Sugioka Y. Ethylene o x ide sterili z ation of bone grafts. Residual gas concentration and fibroblast toxicity. Acta Orthop Scand 65 (6): 6402 1994 6 Arscott E, Anderson S Broad JJ, Parente D. Validating Radiation Sterilization in a Global Marketplace : 68-82, 1999. 7. Asselmeier MA, Caspari RB, Bottenfield S A review of allograft processing and sterilization techniques and their role in transmission of the human immunodeficiency virus. Am J Sports Med 21 (2) : 170-5, 1993. 8. Berry DJ Chandler HP Reilly DT. The Use of Bone Allografts in Two-Stage Reconstruction after Failure of Hip Replacemnets Due to Infection. The J Bone and Joint Surgery: 1460-1468, 1991. 9 Bianchi JR Ross K James E. Keesling J, Mills CR. The Effect of Preservation/Sterilization Processes on the Shear Strength of Cortical Bone Proceedings of the 1999 Bioengineering Conference 42: 407-408 1999 10 Block SS. Disinfection sterilization and preservation 4th ed Philadelphia: Lea & Febiger p. xvi, 1162 1991. 11. Bowyer GW, Cumberland N. Antibiotic Release from Impregnated Pellets and Beads J Trauma 36 (3): 331-335, 1994. 131

PAGE 139

132 12. Brien EW, Terek RM, Healey JH Lane JM Allograft reconstruction after pro x imal tibial resection for bone tumors An analysis of function and outcome comparing allograft and prosthetic recon s tructions. Clin Orthop (303) : 116-27 1994. 13. Britigan BE Ratcliffe HR, Buettner GR, Rosen GM. Binding of myelopero x idase to bacteria: effect on hydroxyl radical formation and susceptibility to oxidant-mediated killing. B i ochim Biophys Acta 1290 (3): 231-40 1996. 14. Bryan CS Morg a n SL Caton RJ Lunceford EM, Jr. Cefazolin v ersus cefamandole for prophylaxis during total joint arthroplasty. Clin Orthop (228): 117-22 1988. 15. Buck BE Malinin TI. Human bone and tissue allografts. Preparation and safety. Clin Orthop (303): 8-17, 1994. 16. Buckley R Hughes GN, Snodgra s s T, Huchcroft SA. Perioperative cefazolin prophyla x is in hip fracture surgery. Can J Surg 33 (2): 122-7, 1990. 17. Buttermann GR, Glazer PA Bradford DS. The use of bone allografts in the spine. Clin Orthop (324): 75-85 1996. 18. Campbell DG Li P. Sterilization of HIV with irradiation: relevance to infected bone allografts. Aust N Z J Surg 69 (7): 517-21, 1999. 19. Cannon SR Dyson PH Sanderson PJ. Pseudomembranous colitis associated with antibiotic prophyla x is in orthopaedic surgery. J Bone Joint Surg [Br] 70 (4) : 600-2 1988. 20 Carone SM Bornstein M Coleman DL, Thomas PN, Boylan JC. Stability of frozen solutions of cefazolin sodium. Am J Hosp Pharm 33 (7): 639-41, 1976 21. Chang CC Merritt K. Infection at the Site of Implanted Materials with and without Preadhered Bacteria. J Orthopaedic Res 12: 526-31 1993. 22. Choong PFM, Sim FH. Limb-Sparing Surgery for Bone Tumors: New Developments Sem Surg One 13: 64-69 1997. 23 Citak MS Cue JI, Peyton JC, Malangoni MA. The critical rel a tionship of antibiotic dose and bacterial contamination in experimental infection. J Surg Res 52 (2) : 127-30 1992. 24 Classen DC, Evans RS Pestotnik SL, Hom SD, Menlove RL Burke JP. The timing of prophylactic administration of antibiotics and the risk of surgical-wound infection. N Engl J Med 326 (5): 281-6 1992. 25. Coello R Glenister H Fereres J, Bartlett C Leigh D Sedgwick J Cooke EM. The cost of infection in surgical patients : a case-control study J Hosp Infect 25 (4): 239-50, 1993. 26. Cordero J Munuera L, Folgueira MD. Influence of bacterial strains on bone infection. J Orthop Res 14 (4): 663-7 1996.

PAGE 140

133 27. Cunha BA, Gossling HR, Pasternak HS Nightingale CH, Quintiliani R. Penetration of cephalosporins into bone Infect 12 (2): 80-4 1984. 28 Dayal VS, Whitehead GL, Smith EL. Gentamicin Progressive cochlear toxicity Can J Otolaryngol 4 (2): 348-51, 1975. 29. Deijkers RLM, Bouma GJ Meer-Prins EMWvd, Huysmans PE Taminiau AHM F.H.J.Claas. Human bone allografts can induce T cells with high affinity for donor antigens. The J Bone and Joint Surgery (Br) : 538-544, 1999. 30. Delmi M, Vaudaux P, Lew DP, Vasey H. Role of Fibronectin in Staphylococcal Adhesion to Metallic Surfaces Used as Models for Orthopaedic Devices J Orthopaedic Res 12: 432-438, 1992. 31. Dick HM Strauch RJ. Infection of massive bone allografts. Clin Orthop (306) : 46-53, 1994. 32 DiMaio FR, O'Halloran JJ Quale JM. In vitro elution of ciprofloxacin from polymethylmethacrylate cement beads J Orthop Res 12 (1): 79-82 1994 33. Donati D Biscaglia R. The use of antibiotic impregnated cement in infected reconstructions after resection for bone tumours J Bone Joint Surg Br 80 (6): 1045-50 1998. 34 Evans RP Nelson CL. Gentamicin-Impregnated Polymethylmethacrylate Beads Compared With Systemic Antibiotic Therapy in the Treatment of Chronic Osteomyelitis Clin Orthopaedics and Related Res 295 : 37-42, 1993. 35. FDA. Guidance for Industry pp. 12-15 1997. 36. FDA. Gu i dance for Industry-Screening and Testing of Donors of Human Tissue Intended for Transplantation, pp. 11, 1997. 37 FDA. Human Tissue Intended for Transplantation Final Rule Vol. 62 pp. 88 1999 38. FDA. Proposed Approach to Regulation of Cellular and Tissue-Based Products 1997. 39 Fiebig E. Safety of the blood supply Clin Orthop (357): 6-18 1998. 40. Firsov AA Vostrov SN Kononenko OV, Zinner SH Portnoy YA. Prediction of the effects of inoculum size on the antimicrobial action of trovafloxacin and ciproflo x acin against Staphylococcus aureus and Escherichia coli in an in v itro dynamic model. Antimicrob Agents Chemother 43 (3): 498-502 1999. 41. Frassica FJ Chao EY Sim FH Special problems in limb-salvage surgery. Semin Surg Oncol 13 (1): 55-63, 1997.

PAGE 141

134 42. French GL, Cheng AF. Measurement of the costs of hospital infection by prevalence surveys. J Hosp Infect 18 Suppl A: 6572, 1991. 43. Fry TL, Fried FA, Gaven BA. Renal toxicity. Tobramycin and gentamicin. Invest Urol 15 (2): 100-3, 1977. 44. Gerber MA, Gastanaduy AS, Buckley JJ, Kaplan EL. Risk ofbacteremia after endotracheal intubation for general anesthesia. South Med J 73 (11): 1478-80, 1980. 45. Goertzen MJ, Clahsen H, Burrig KF, Schulitz KP. Sterilisation of canine anterior cruciate allografts by gamma irradiation in argon. Mechanical and neurohistological properties retained one year after transplantation [published erratum appears in J Bone Joint Surg Br 1995 Nov;77(6):985] [retracted by Fulford P. In: J Bone Joint Surg Br 1997 Sep;79(5):705-6]. J Bone Joint Surg Br 77 (2): 205-12, 1995 46. Gosden PE, MacGowan AP, Bannister GC. Importance of air quality and related factors in the prevention of infection in orthopaedic implant surgery. J Hosp Infect 39 (3): 17380, 1998. 47. Hamlet W, Liu S, Yang R. Destruction of a Cyropreserved Meniscal Allograft: A Case for Acute Rejection. Arthroscopy: The J Arthroscopic and Related Surgery 13 (4): 517521, 1997. 48 Harris AS, Kirstein-Pedersen A, Stenberg P, Ulmsten U, Wingerup L. Preparation, characterization, and stability of new prostaglandin E2 gel for local administration. J Pharm Sci 69 (11): 1271-3, 1980. 49. Hayden MK, Koenig GI, Trenholme GM. Bactericidal activities of antibiotics against vancomycin-resistant Enterococcus faecium blood isolates and synergistic activities of combinations. Antimicrob Agents Chemother 38 (6): 1225-9, 1994. 50. Hejna MJ, Gitelis S. Allograft prosthetic composite replacement for bone tumors. Semin Surg Oneal 13 (1): 18-24, 1997. 51. Heller MC, Carpenter JF, Randolph TW Manipulation of lyophilization-induced phase separation: implications for pharmaceutical proteins. Biotechnol Prog 13 (5): 590-6, 1997. 52. Heller MC, Carpenter JF, Randolph TW. Protein formulation and lyophilization cycle design: prevention of damage due to freeze-concentration induced phase separation. Biotechnol Bioeng 63 (2): 166-74, 1999. 53. Henry SL, Galloway KP. Local antibacterial therapy for the management of orthopaedic infections. Pharmacokinetic considerations. Clin Pharmacokinet 29 (1): 36-45, 1995. 54. Herrera GA, Lackritz EM, Janssen RS, Raimondi VP, Dodd RY, Aberle-Grasse J, Petersen LR. Serologic test for syphilis as a surrogate marker for human

PAGE 142

135 immunodeficiency virus infection among United States blood donors [see comments]. Transfusion 37 (8): 836-40, 1997. 55. Hiemstra H, Tersmette M, Vos AH, Over J, van Berke! MP, de Bree H Inactivation of human immunodeficiency virus by gamma radiation and its effect on plasma and coagulation factors. Transfusion 31 (1 ): 32-9, 1991. 56. Hunt MNdl, Chan A YC, Karran SJ. Postoperative Complications: How Much Do They Cost? Annals of the Royal Colleges of Surgeons of England 68: 199-202, 1986. 57. Jackson DW, Windler GE, Simon TM. Intraarticular reaction associated with the use of freeze-dried, ethylene oxide-sterilized bone-patella tendon-bone allografts in the reconstruction of the anterior cruciate ligament. Am J Sports Med 18 (1 ): 1-10; discussion 10-1, 1990 58. Jones SM, Blazevic DJ, Balfour HH. Stability of gentamicin in serum Antimicrob Agents Chemother 10 (5): 866-7, 1976. 59. Kealey GP. Disease transmission by means of allograft. J Bum Care Rehabil 18 (1 Pt 2): Sl0-1, 1997. 60. Klein RA, Meyer RD. Synergy studies with Pseudomonas aeruginosa resistant to gentamicin and/or carbenicillin Chemotherapy 24 (5) : 283-9, 1978. 61. Klemm KW. Antibiotic Bead Chains. Clin Orthopaedics and Related Res 295: 63-76, 1993. 62. Kokubo M, Inque T, Akers J. Resistance of Common Environmental Spores of the Genus Bacillus to Vapor Hydrogen Peroxide PDA J Pharm Sci & Technology 52 (5): 228-231, 1998. 63. Kroes AC Role of testing for antibody to hepatitis B core antigen to prevent transmission of hepatitis B virus in organ and tissue transplantation. Transplant Proc 28 (5): 2937-8, 1996. 64. Kudryk VL, Scheidt MJ, McQuade MJ, Sutherland DE, VanDyke TE, Hollinger JO Toxic effect of ethylene-oxide-sterilized freeze-dried bone allograft on human gingival fibroblasts. J Biomed Mater Res 26 (11): 1477-88, 1992. 65. Lacy MK, Lu W, Xu X, Tessier PR, Nicolau DP Quintiliani R, Nightingale CH. Pharmacodynamic Comparisons of Levofloxacin, Ciprofloxacin, and Ampicillin against Streptococcus pneumonia in an In Vitro Model of Infection. Antimicrobial Agents and Chemotherapy 43 (3): 672-77, 1999 66. LeBrun M, Grenier L, Gourde P, Bergeron MG, Labrecque G, Beauchamp D. Effectiveness and toxicity of gentamicin in an experimental model of pyelonephritis: effect of the time of administration. Antimicrob Agents Chemother 43 ( 5): 1020-6, 1999.

PAGE 143

136 67. Long M, Rack HJ. Titanium Alloys in Total Joint Replacement -A Materials Sci Perspecti ve. Biomaterials: 1621-39 1998 68. Loty B Courpied JP Tomeno B Postel M Forest M, Abelanet R Bone allografts sterilised by irradiation. Biological properties procurement and results of 150 massi v e allografts Int Orthop 14 (3) : 237-42 1990. 69. Manolio TA, Beattie C Christopherson R Pearson TA. Regional versus general anesthesia in high-risk surgical patients: the need for a clinical trial. J Clin Anesth 1 (6): 414-21, 1989. 70. May SR, Roberts DP DeClement FA, Still JM, Jr. Reduced bacteria on transplantable allograft skin after preparation with chlorhexidine gluconate povidone-iodine and isopropanol. J Bum Care Rehabil 12 (3): 224-8 1991. 71. McEniry DW, Gorbach SL. Cephalosporins in surgery. Prophylaxis and therapy. Drugs 34 (Suppl 2): 216-39, 1987. 72 Meyer JD Falk RF Kelly RM, Shively JE Withrow SJ Demell WS, Kroll DJ, Randolph TW, Manning MC. Preparation and in vitro characterization of gentamycin-impregnated biodegradable beads suitable for treatment of osteomyelitis. J Pharm Sci 87 (9): 1149-54, 1998. 73. Miclau T Dahners LE Lindsey RW. In Vitro Pharmacokinetics of Antibiotic Release from Locally Implantable Materials J Orthopaedic Res 11: 627-632, 1993 74 Mitrzyk B Jarco PE, Rybak MJ. Stability and antimicrobial acti v ity of gentamicin sulfate tobramycin sulfate and amikacin sulfate in polypropylene syringes for use in oncedaily aminoglycoside therapy. Am J Health Syst Pharm 53 (23): 2855-9 1996. 75 Morse BS. A New surfactan t/ Iodophor-Based Viral Inactivation Process (VIP) for Preparation of Bone Allografts. 16th Annual Meeting of the American Association of Tissue Banks 1992. 76. Mulazimoglu L Drenning SD Muder RR. Vancomycin-gentamicin synergism revi s ited : effect of gentamicin susceptibility of methicillin-resistant Staphylococcus aureus. Antimicrob Agents Chemother 40 (6): 1534-5, 1996 77. Nie L Nicolau DP Nightingale CH Browner BO Quintiliani R In Vitro Elution of Ofloxacin from a Bioabsorbable Polymer. Acta Orthopaedic Scandanavica 66 (4): 36568 1995 78 Nitescu P Hultman E Appelgren L Linder LE Curelaru I. Bacteriolo g y drug stability and exchange of percutaneous delivery systems and antibacterial filters in long-term intrathecal infusion of opioid drugs and bupivacaine in "refractory" pain. Clin J Pain 8 (4): 324-37 1992.

PAGE 144

137 79. Noguchi T, Izumizawa K, Fukuda M, Kitamura S, Suzuki Y Ikura H. New method for local drug delivery using resorbable base material in periodontal therapy. Bull Tokyo Med Dent Univ 31 (3): 145-53, 1984. 80. Norman-Taylor FH, Villar RN. Bone allograft: a cause for concern? [editorial]. J Bone Joint Surg Br 79 (2): 178-80, 1997. 81. Odenholt-Tornqvist I, Lowdin E, Cars 0. Postantibiotic sub-MIC effects ofvancomycin, roxithromycin, sparfloxacin, and amikacin. Antimicrob Agents Chemother 36 (9): 1852-8, 1992. 82. Ostermann PA, Seligson D, Henry SL. Local antibiotic therapy for severe open fractures A review of 1085 consecutive cases. J Bone Joint Surg Br 77 ( 1 ): 937 1995. 83. Park E-S, Maniar M, Shah JC. Biodegradable Polyanhydride Devices of Cefazolin Sodium, Bupivacaine, and Taxol for Local Drug Delivery: Preparation, and Kinetics and Mechanism ofln Vitro Release. J Controlled Release 52: 179-89, 1998 84. Park E-S Maniar M, Shah JC. Biodegradable polyanhydride devices of cefazolin sodium, bupivicaine and taxol for local drug delivery: preparation and kinetics and mechanism of in vitro release. J Controlled Release 52: 179-189, 1998. 85. Pelletier LL, Jr., Baker CB. Oxacillin, cephalothin, and vancomycin tube macrodilution MBC result reproducibility and equivalence to MIC results for methicillinsusceptible and reputedly tolerant Staphylococcus aureus isolates. Antimicrob Agents Chemother 32 (3): 374-7 1988. 86 Phy. Physicians' desk reference: PDR. 50th ed. Oradell, N.J.: Medical Economics Co., 1996. 87. Pinkowski JL Rodrigo JJ Sharkey NA, Vasseur PB. Immune response to nonspecific and altered tissue antigens in soft tissue allografts. Clin Orthop (326): 80-5, 1996. 88. Poulsen KB, Bremmelgaard A, Sorenson AI, Raahave D, Petersen N. Estimated costs of Postoperative Wound Infections. Epidemiol. Infect. 113: 283-295, 1994. 89. Prins MH Hirsh J A comparison of general anesthesia and regional anesthesia as a risk factor for deep vein thrombosis following hip surgery: a critical review. Thromb Haemost 64 (4): 497-500, 1990. 90 Randolph TW. Phase separation of excipients during lyophilization: effects on protein stability. J Pharm Sci 86 (11): 1198-203 1997. 91. Rasmussen TJ, Feder SM, Butler DL, Noyes FR. The effects of 4 Mrad of gamma irradiation on the initial mechanical properties of bone-patellar tendon-bone grafts. Arthroscopy 10 (2): 188-97, 1994.

PAGE 145

138 92. Rhoten RL, Murphy MA, Kalfas IH, Hahn JF, Washington JA. Antibiotic penetration into cervical discs. Neurosurgery 37 (3): 418-21, 1995. 93. Riley LH, 3rd. Prophylactic antibiotics for spine surgery: description of a regimen and its rationale. J South Orthop Assoc 7 (3): 212-7, 1998. 94. Sasmor MT, Morain WD, Balestrero LM, Camp BJ. Investigation of a biodegradable, implantable antibiotic delivery system on rate of wound infection Ann Plast Surg 30 (6): 525-30, 1993. 95. Scarborough NL. Current procedures for banking allograft human bone. Orthopedics 15 (10): 1161-7, 1992. 96 Silcox DH, 3rd, Boden SD, Schimandle JH, Johnson P Whitesides TE, Hutton WC. Reversing the inhibitory effect of nicotine on spinal fusion using an osteoinductive protein extract. Spine 23 (3): 291-6; discussion 297, 1998. 97. Simonian PT, Conrad EU, Chapman JR, Harrington RM, Chansky HA. Effect of sterilization and storage treatments on screw pullout strength in human allograft bone. Clin Orthop (302): 290-6, 1994. 98. Skrabanja AT, de Meere AL, de Ruiter RA, van den Oetelaar PJ. Lyophilization of biotechnology products. PDA J Pharm Sci Technol 48 ( 6): 3117, 1994 99. Sorenson TS, Colding H, Schroeder E, Rosdahl VT. The Penetration of Cefazolin, Erythromycin and Methicilin Into Human Bone Tissue. Acta orthop. scand. 49: 549-553, 1978 100. Springfield DS. Allograft reconstructions. Semin Surg Oncol 13 (I): 11-7, 1997. 101. Steinbakk M, Midtvedt T, Lingaas E, Gardsjord G. In vitro activity of ceftazidime, cefotaxime and gentamicin against 11,521 clinical isolates of bacteria. Acta Pathol Microbiol Immunol Scand [B] 95 (6): 337-46, 1987. 102. Strachan CJ Antibiotic prophylaxis in peripheral vascular and orthopaedic prosthetic surgery. J Antimicrob Chemother 31 Suppl B : 65-78, 1993. 103. Strachan CJ. The prevention of orthopaedic implant and vascular graft infections. J Hosp Infect 30 Suppl: 54-63, 1995. 104 Straw RC, Withrow SJ, Douple EB, Brekke JH, Cooper MF, Schwarz PD, Greco DS, Powers BE. Effects of cis-diamrninedichloroplatinum II released from D,L-polylactic acid implanted adjacent to cortical allografts in dogs. J Orthop Res 12 (6): 871-7, 1994. 105. Swoboda SM, Merz C, Kostuik J, Trentler B, Lipsett PA. Does intraoperative blood loss affect antibiotic serum and tissue concentrations? Arch Surg 131 ( 11 ): 116571; discussion 1171-2, 1996

PAGE 146

139 106 Tablan OC, Reyes MP, Rintelmann WF, Lerner AM Renal and auditory toxicity of high dose, prolonged therapy with gentamicin and tobramycin in pseudomonas endocarditis. J Infect Dis 149 (2): 257-63, 1984. 107. Taylor GD Kirkland TA, McKenzie MM, Sutherland B, Wiens RM. The effect of surgical wound infection on postoperative hospital stay. Can J Surg 38 (2): 149-53, 1995. 108. Taylor GJ Bannister GC, Calder S. Perioperative wound infection in orthopaedic surgery. J Hosp Infection 16 (October): 241-247 1990. 109. Thamlikitkul V Synergy study of vancomycin or teicoplanin plus gentamicin against enterococci Staphylococcus aureus and coagulase-negative staphylococci by time-kill method. J Med Assoc Thai 74 (12) : 669-74, 1991. 110 Thoren K Aspenberg P. Ethylene oxide sterilization impairs allograft incorporation in a conduction chamber. Clin Orthop (318) : 259-64, 1995. 111. Tomford WW. Transmission of Disease through Muscoskeletal Transplantation. Portland Bone Symposium-Conference: 410-420 1997. 112. Tomford WW, Thongphasuk J, Mankin HJ Ferraro MJ. Frozen musculoskeletal allografts A study of the clinical incidence and causes of infection associated with their use. J Bone Joint Surg [Am] 72 (8) : 1137-43, 1990 113. Trollfors B Norrby R Bergmark J Hultberg B Linderholm H Comparati v e toxicity of gentamicin and cefotetan. Scand J Infect Dis 18 (2): 139-46 1986. 114. Tselentis J Legakis NJ, Nicolas KJ, Melissinos K Papavassiliou J. Comparative in vitro activity of netilmicin amikacin tobramycin and sisomicin against gentamicin highly resistant enterobacteriaceae. Chemotherapy 26 (6): 409-17, 1980. 115. Vine DL. The cardiovascular risk of general anesthesia. Kans Med 93 ( 5): 159-60 1992. 116. ViroMed Laboratories I. Infectious Diseases Associated with Allograft Tissues. Seminar 1997. 117. Visalli MA, Jacobs MR, Appelbaum PC MIC and time-kill study of acti v ities ofDU-6859a ciprofloxacin levofloxacin sparfloxacin, cefotaxime, imipenem and vancomycin against nine penicillin-susceptible and -resistant pneumococci. Antimicrob Agents Chemother 40 (2): 362-6, 1996. 118. Watanakunakom C, Tisone JC Synergism between vancomycin and gentamicin or tobramycin for methicillin-susceptible and methicillin-resistant Staphylococcus aureus strains Antimicrob Agents Chemother 22 (5): 903-5 1982. 119 Welch DF, Matsen JM. Bactericidal activity of cefoperazone in comparison with other betalactam antibiotics and gentamicin. Clin Ther 3 (Spec) : 127-9 1980.

PAGE 147

140 120. Whitecloud TS 3rd Modem alternatives and techniques for one-level discectomy and fusion. Clin Orthop (359): 67-76 1999. 121. Williams DN, Gustilo RB Beverly R Kind AC. Bone and serum concentrations of five cephalosporin drugs. Relevance to prophylaxis and treatment in orthopedic surgery Clin Orthop (179): 253-65 1983. 122. Woods RK, Dellinger EP. Current guidelines for antibiotic prophylaxis of surgical wounds. Am Fam Physician 57 (11): 2731-40 1998 123. Zbrozek AS Marble DA, Bosso JA. Compatibility and stability of cefazolin sodium clindamycin phosphate, and gentamicin sulfate in two intravenous solutions. Drug Intell Clin Pharm 22 (11): 873-5 1988. 124. Zehr RJ Enneking WF Scarborough MT. Allograft-prosthesis composite versus megaprosthesis in proximal femoral reconstruction Clin Orthop (322): 207-23 1996 125. Zucker K Cirocco R Roth D, Olson L Burke GW, Nery J Esquenazi V Miller J. Depletion of hepatitis C virus from procured kidneys using pulsatile perfusion preservation. Transplantation 57 (6): 832-40, 1994.

PAGE 148

BIOGRAPHICAL SKETCH Charles Randal Mills received his undergraduate education at the University of Florida in microbiology and physiology. Mr. Mills received post-graduate training in clinical laboratory sc i ence from Shands Hospital at the University of Florida and holds licensure as a Clinical L aboratory Technologist. Additionally, Mr. Mills is a Certified Tissue Bank Specialist. After completing his initial education Mr. Mills established the Biomedical Laboratory a t the University of Florida Tissue Bank a multidisciplinary clinical and GLP compliant laboratory specializing in the testing of human tissue for transplantation. He currently serves as Senior Technical Affairs Manager for Regeneration Technologies, Inc. where he has developed the BioCleanse Tissue Processing System. This system allows for the sterilization of allografts without a reduction in their beneficial properties. This greatly reduces the potential for donor-to-recipient transmission of diseases such as HIV and hep a titis. His research interests include pursuing novel non-enveloped viral inactivation strategies for human tissue and developing drug and gene delivery systems that will facilitate the transplantation of solid organs and intact limbs. 141

PAGE 149

I certify that I have read this study and that in my opinion it conforms to acceptab l e standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. GayleB zeau, Chair / Associate Professor of Pharmaceutics I certify that I have read this study and that in my opinion it conforms to acceptab l e standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor ofPhilota 22 <-\.S Guenther Hochhaus Associate Professor of Pharmaceutics I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy it. Q& e.A do fHartmut Derendorf Professor of Pharmaceutics I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully dequat in scope and quality, as a dissertation for the degree of Doctor of Philosophy. December 1999 Denni uggan Associate Professor of Cell Science Dean, Graduate School /