X. Orthopedics

A. General

[192] IMPROVED BONE CEMENT FATIGUE RESISTANCE VIA CONTROLLED STRENGTH INTERFACES

William G. Winter, MD; Francis W. Cooke, PhD; Hirotsugu K. Yasuda, PhD; Elizabeth A. Friis, PhD; Charles D. Graber, BS
Department of Orthopaedic Surgery, University of CO School of Medicine, Veterans Affairs Medical Center, Department of Surgery, Orthopaedic Section Denver, CO 80220; Department of Surgery, University of Kansas School of Medicine, Orthopaedic Research Institute, Inc., Wichita, KS 67214; Department of Chemical Engineering, University of Missouri-Columbia, Columbia, MO 65211; Orthopaedic Research Institute, Inc., Wichita, KS 67214; email: fcooke@via-christi.org; yasuda@ecvax2.ecn.missouri.edu; lfriis@via-christi.org

Sponsor: Department of Veterans Affairs, VA Rehabilitation Research and Development Service, Washington, DC 20420
(Project #A796-RA)

PURPOSE--The problem to be addressed in this study is to find a way to modify commercial bone cement to increase its fatigue resistance, especially its resistance to low stress, high-cycle fatigue without impairing its intraoperative handling characteristics.

METHODOLOGY--The basic methodology of the study consists of adding thin fibers to polymethyl methacrylate- (PMMA) based bone cement to improve its resistance to fatigue crack growth and thus increase its fatigue life. The fibers impede crack growth because the interface between the fiber and the cement matrix ruptures or debonds as the propagating crack approaches it. This serves to blunt the crack locally and also absorbs some of the energy of propagation that is the driving force for crack growth. If sufficient interfaces are present, the blunting and energy absorbing interactions can slow the overall rate of crack growth and increase the observed fatigue life of the cement. Optimization of this process depends on creating interfaces with moderate strength. The experimental approach is to modify the surfaces of the fibers before adding them to the bone cement by exposing them to low-pressure, low-temperature ionized gasses, (i.e., plasmas). This renders the surfaces reactive. When the cement-fiber composite cures, some degree of bonding occurs at these reactive surfaces producing an interface with the desired strength. The effectiveness of these interfaces in impeding crack growth is evaluated by measuring the strength of the composite in the presence of cracks (fracture toughness testing) and by measuring the actual fatigue life.

  A secondary, but very important, concern is that the fiber additions can increase the viscosity of bone cement during mixing. This can interfere with proper mixing and hinder intrusion of the cement into the bone during surgery. To circumvent these effects, the size, composition, and loading of the fibers must be manipulated.

PROGRESS--During the second year of this study techniques and procedures were developed. Long (5 m) tows of polyethylene terephthalate (PET) fibers were procured and methods of handling, cleaning, chopping (to 1 mm length), and incorporation into bone cement were developed. These short fibers have supplanted the longer, 6 mm, fibers used during most of the first year of this study. Fibers are used in the "as cleaned condition" (controls) or are surface-treated by exposure to plasmas of oxygen, air, or argon before being added to the cement. Both fracture toughness and fatigue tests are in progress using molded (not machined) specimens. Attempts are also underway to assess the interface strength by the Iosipescu shear test and the short beam bend test. Electron spin resonance is being used to study the chemistry of the free radical (peroxide) accelerator reaction at the surface of the treated fibers. The conventional toluidene accelerator and a novel alternative accelerator, 4 methyamino phenethyl alcohol, are being evaluated.

RESULTS--A great many difficulties were encountered in preparing and testing composite specimens throughout the first 2 yrs of the study. These problems have been largely overcome, and testing is well underway. It has been demonstrated that reducing the length of the PET fibers to at least 0.5 mm improves the mixing and handling of composite bone cement without impairing its fracture toughness. It was also shown that machining (instead of molding) the notch in composite fracture toughness specimens leads to subsurface crazing around the embedded fibers. This, in turn, lowers the stress required for fatigue precracking.

  Early fatigue results indicate that the endurance limit (at 1×106 cycles) of neat bone cement is 11.5 MPa. This is approximately 50 percent higher than the value that can be inferred from most of the prior work reported in the literature. This improvement is attributed to the greater care used in preparing and aligning the specimens in the present study.

FUTURE PLANS--The principal endeavor during the third year of the study will be the determination of the fatigue lives and endurance limits of fiber/bone cement composites following different surface treatments.

RECENT PUBLICATIONS FROM THIS RESEARCH

[193] DISUSE-INDUCED ARTICULAR CARTILAGE ATROPHY, OVER-EXERCISE, AND ARTHRITIS

David S. Howell, MD
VA Medical Center, Miami, FL 33125

Sponsor: Department of Veterans Affairs, VA Rehabilitation Research and Development Service, Washington, DC 20420
(Project #A739-2RA)

PURPOSE--We seek to measure biomechanical parameters in surface, middle, and deep layers of articular cartilage in disuse atrophy using a canine knee model caused by rigid immobilization for 6 wks, as well as disuse followed by 3 wks of conservative versus aggressive remobilization. We seek also to determine in the same animal preparation and articular cartilage layers metalloproteinase activity levels and cellular calcium fluxes with special reference to correlative changes. Critical changes in calcium flux, biomechanical properties, and elevation of endogenous proteases will be related to Dr. Setton's theory on loss of protective prestresses in disuse cartilage.

METHODOLOGY--We will determine the activities of the matrix metalloproteinases and TIMP in the articular cartilage of all joints at three zones through the cartilage layer to establish a baseline and to test for direct production of proteolytic agents following disuse or disuse with aggressive or conservative remobilization. We will then determine the site-specific material properties of articular cartilage at three zones in the articular cartilage layer in compression and swelling experiments. Measurements of the compressive modulus (Ha), Poisson's ratio (v), and Young's modulus (E) will be used to test for effects of periods of joint disuse, and disuse with aggressive or conservative remobilization, on the mechanics of the cartilage extracellular matrix. Fluorescent confocal scanning laser microscopy will be used to measure the transient changes in Ca2+ in chondrocytes in situ at three zones in the cartilage layer during compression of cartilage explants. This protocol will be repeated for all experimental cartilage to establish normal, site-specific mechanochemical interactions and to test whether these interactions are disrupted following periods of disuse or conservative or aggressive remobilization, and to test correlations with levels of metalloproteinases and proteoglycan content. A biphasic finite element model will be used with site-specific material parameters measured for articular cartilage to calculate the relevant mechanical parameters (e.g., fluid-flow, stress, strain, fluid pressure) at the cell-matrix interace for chondrocytes in the extracellular matrix. In this manner, we will use our site-matched data for mechanically induced Ca2+ signaling in situ and predictions for mechanical signals to establish a baseline mechanochemical coupling, and investigate the hypotheses that the mechanochemical interactions change in surface zone cartilage during periods of joint immobilization to turn on proteolysis.

PROGRESS--Intense work on measuring disuse in comparison with normal femoral condylar cartilage, tangential versus radial zone cartilage, shows reduced cell size between zones. Chondrone cell volume compression modulus, relaxation time, trypan blue exclusion, and microbiomechanic investigations are all underway.

RECENT PUBLICATIONS FROM THIS RESEARCH

[194] IMMUNOLOGICAL RESPONSES TO IMPLANT BIOMATERIALS FOLLOWING ARTHROPLASTY

Paul H. Wooley, PhD; Z. Song, MS; S. Sud, MS; Sam Nasser, MD
VA Medical Center, Detroit, MI 48201; Department of Orthopaedic Surgery, Wayne State University School of Medicine, Detroit, MI 48201; email: pwooley@oncgate.roc.wayne.edu

Sponsor: Department of Veterans Affairs, VA Rehabilitation Research and Development Service, Washington, DC 20420
(Project #A797-RA)

PURPOSE--Aseptic loosening, the loss of support by the surrounding bony architecture due to osteolysis, is the single most common complication of total joint replacement. Until recently, aseptic loosening has been theorized to be the result of surgical technique and prosthetic design. However, recent findings have focused attention upon the biological responses to plastics and metals in the pathogenesis of prosthetic loosening. The objective of this study is to identify the key components of the biological responses to implanted biomaterials by examining the hypothesis that implanted biomaterials elicit an inflammatory cellular immune reaction that may lead to osteolysis with eventual failure of the prosthesis.

METHODOLOGY--Immunological and molecular techniques are being used to assess both the serological and cellular responses directed against biomaterials in persons with aseptic loosening. We are examining failed prosthetic implants for the presence of antibodies bound to the implant surface, or reactive with proteins bound to the implant. Electrophoresis (SDS-PAGE) is performed to determine the number and molecular weight of bound proteins extracted from the surface of failed polyethylene implants recovered during surgical revision procedures. Specific proteins in the extracts are analyzed by immunoblot and Western blot techniques. Celluar responses to biomaterials are measured in peripheral blood mononuclear cells (PBMC) from subjects with painful or loosened total joint prostheses, and from those pre-operatively qualified for total joint replacement. Cells are cultured in vitro with polymethylmethacrylate (PMMA) cement, ultra high molecular weight polyethylene (UHMWPE), cobalt-chrome alloy (Co-Cr), and titanium (Ti) particles, and the prolifeative and cytokine responses to each biomaterial are calculated. We are also developing an animal model of the response to wear debris, and using gene therapy as a novel approach to combat the inflammatory reaction in tissue surrounding the prosthesis.

PROGRESS--Extracts from 52 failed prostheses have been evaluated for bound proteins. Cellular responses to biomaterials in particulate form have been evaluated in 185 subjects. Changes in the responses 1 yr after implantation have been assessed in 32 subjects. Image analysis techniques and RT-PCR assays have been established to evaluate the air pouch model. Gene therapy using the IRAP contrakine gene has been preliminarly evaluated in the model.

RESULTS--Type I collagen was detected on the majority of failed prostheses (45/52), while aggrecan proteoglycans were rare (8/52). Bound immunoglobulin was detected frequently (41/52), and antibodies reactive against the UHMWPE bound proteins were detected in autologous sera from 26/38 implant recipients. Most autoantibodies were directed against high molecular weight proteins. There was an association between the presence of antibodies to implant-bound proteins and the occurrence of serum antibodies to type I collagen. Cellular activity in the subjects revealed that the response to PMMA and to CoCr was significantly higher in the revision surgery group compared with the primary surgery group. In vitro IL-1 production in response to stimulation with PMMA reflected the proliferation results closely, and levels of IL-6 were observed in cells stimulated with Ti-6Al-4V particles. High responses to PMMA were observed in subjects undergoing revision surgery due to loosened or painful prostheses, while responses to PMMA in those undergoing revision surgery due to mechanical failure or infection were not elevated. The murine air pouch response to prosthesis wear debris proved to be an accurate model of the inflammation in periprosthetic tissue. Gene therapy techniques resulted in the production of the contrakine IRAP within the pseudosynovial tissue, with a concommitant improvement in several parameters of the inflammatory reaction in the air pouch.

RECENT PUBLICATIONS FROM THIS RESEARCH

[195] BIOCHEMICAL ANALYSIS OF SYNOVIAL ACTIVATION IN JOINT DYSFUNCTION

Christopher H. Evans, PhD, DSc
VA Medical Center, Pittsburgh, PA 15240

Sponsor: Department of Veterans Affairs, VA Rehabilitation Research and Development Service, Washington, DC 20420
(Project #A52-8RA)

PURPOSE--Prior research in this laboratory has generated the hypothesis that interleukin-1 (IL-1) is an important autocrine mediator of the activation of mesenchymal cells by wear particles, and of the osteolysis that follows activation of these cells. These mechanisms are thought to drive the process of aseptic loosening of prosthetic joints. It follows that inhibition of IL-1 should protect against the osseous changes that produce aseptic loosening. The interleukin-1 receptor antagonist (lL-1 Ra) is a naturally occurring inhibitor of IL-1. We suggest that IL-1 Ra should be a useful agent with which to prevent aseptic loosening, and that its delivery by gene transfer is the most expeditious way to realize its potential in this regard.

METHODOLOGY--In the in vitro studies, a retrovirus (MFG-IRAP) is used to transfer a gene encoding the human IL-1 Ra protein to synovial fibroblasts. These cells then secrete constitutively large amounts of human IL-1 Ra into their culture medium. The cells are then challenged with a number of activating stimuli including phorbol myristate acetate (PMA), 1L-1 and particles, generated from prosthetic joints. The degree to which the IL-1 Ra gene prevents the synthesis of IL-1, matrix metalloproteinases (MMPs) and prostaglandin E2 (PGE2), is then noted.

  The recombinant retrovirus is produced by cloning a hTL-1 Ra cDNA into the backbone of the MFG retrovirus. This is used to transduce the cell cultures. Expression of IL-1 Ra is assessed by ELISA. PMA, IL-1, or particles are then added to the cultures, and the expression of IL-1, MMPs, and PGE2 measured by standard.

PROGRESS--The MFO-IRAP vector efficiently transduces the IL-1 Ra gene into primary cultures of rabbit and human synovial fibroblasts, as well as chondrocytes and bone marrow hematopoietic stem cells. Transduction of fibroblasts was found to be greatly enhanced by centrifuging for 2 hr during the period of viral infection. Transduction of the rabbit synovlal cell line HlG-82 generated a stable line (HIG-IRAP) that constitutively produces large amounts of IL-1Ra (>100 ng IL-1Ra/106 cells/day) and provides a useful experimental tool. Expression of IL-1 Ra not only inhibited induction of mediators by cells exposed to wear particles but, surprisingly, also protected them from the cytotoxic effects of high doses of these particles. This suggests that the latter process is subject to biological control and may be apoptotic in nature.

IMPLICATIONS--Aseptic loosening is the major reason that joint replacements fail. This project investigates one way to prevent it.

[196] AN IN-VIVO MODEL FOR CARTILAGE REGENERATION

Hunter H. McGuire, Jr., MD; Charles L. McDowell, MD; Jennifer S. Wayne, PhD; Paul G. Kiritsis, MD
Hunter Holmes McGuire VA Medical Center, Richmond, VA 23249

Sponsor: Department of Veterans Affairs, VA Rehabilitation Research and Development Service, Washington, DC 20420
(Project #A799-3RA)

PURPOSE--The reason for this study is to evaluate the potential for development and maturation of cartilage or cartilage-like tissue on an articular surface in-vivo, with the study joint allowed full mobility.

METHODOLOGY--The model for this project is the articulating surface of the dog patella, selected for the early phase of our investigations because the mechanical stresses upon the dog patellae are very large, much like the human counterpart. Also, this model provides sufficient material for the extensive histologic, biomechanical, and chemical studies required to evaluate the tissue which grows on the subchondral bone surface after all normal cartilage has been removed. The uncovered subchondral bone surface on the experimental patella is prevented from coming into contact with the femur by inserting high density, polyethylene spacers. Thus, a shielded space is created in the moving joint that allows new tissue to grow in an otherwise normal joint where mechanical stress will not damage the new tissue.

PROGRESS--The project is entering phase three of a five-step study. Phase one answered the question, will new tissue grow on a subchondral bone surface in a shielded environment. That question has been answered affirmatively with histologic, biomechanical, and biochemical evidence of a neo-cartilage tissue growing to completely fill the shielded area on the subchondral bone surface. Phase two utilized the refined model to produce useful study material. The object of this phase was to learn at what time point the new tissue would reach maximum maturation in this stress-shielded environment.

RESULTS--We have determined that 12 wks after the experimental operation, the new tissue reaches a peak of maturation during stress shielding and begins to fibrillate at its surface, and there is increasing evidence of injury to the joint caused by the shielding devices, the polyethylene spacers. With this observation completed we have begun to remove the shielding devices, so that the new tissue can be exposed to the contralateral joint surface covered by normal hyaline cartilage. This third phase will include the same histologic, biomechanical, and biochemical assessment of the new tissue's ability to survive and function after having been subjected to compression and motion stress.

FUTURE PLANS--The next phase of this study will be devoted to an evaluation of the response of the new tissue at three time points (4, 8, and 12 wks) after the shielding devices have been removed. Of course, we hope the new tissue will continue to change into a mature form of fibro-cartilage and will survive to act like normal hyaline cartilage on the joint surface. If it does not, we plan to try to modify the new tissue to act more like normal hyaline cartilage by chemical means with growth factors, and the like.

[197] EFFECT OF RESTORING POSTERIOR TIBIAL TENDON FUNCTION IN AN ACQUIRED FLATFOOT MODEL

H. Niki, MD; R.P. Ching, PhD; B.J. Sangeorzan, MD
Department of Orthopaedics, Biomechanics Lab, Seattle, WA 98104; Puget Sound VA Medical Center, Seattle, WA 98108-1597

Sponsor: None listed

PURPOSE--This study examines the effect of restoring posterior tibial tendon (PTT) function in an acquired flatfoot model. Rupture of the PTT is associated with progressive or acquired flatfoot deformity. Treatments include various fusions, lateral column lengthening, ligament reconstruction, and tendon transfer. This study uses an acquired flatfoot model to test the hypothesis that restoring PTT function (simulating repair) will significantly affect the kinematic orientation of the hindfoot complex in heel strike, stance, and heel off.

METHODOLOGY--Eight human cadaver lower leg and foot specimens were obtained. Gross inspection and radiographic screening were performed to rule out pre-existing pathology. The soft tissue was stripped from around the tibial shaft of each specimen exposing the remaining tendons to the hindfoot level. Carbon fiber pins were placed into the calcaneus, talus, navicular, and cuboid and positioned so as not to interfere with the joint motions. An acrylic rod was inserted into the intramedullary canal and crosslocked to permit axial compressive loading of the foot.

  To create an acquired flatfoot model, the spring ligament was divided and the specimen subjected to cyclical loading on an MTS hydraulic load frame. Peak loads of 2.5 times body weight were applied for 5000 cycles at 1 Hz. A radiolucent load frame was used to apply a static 75N compressive load on the foot while taking lateral and A-P radiographs of each specimen both before and after flattening. These radiographs were used to document the degree of deformity for each foot.

  Simulated axial compressive and tendon loads were applied to each specimen using a custom acrylic foot loading frame that allowed for three different loading positions for each foot: heel strike, stance, and heel off. These positions were achieved by applying different combinations of axial compressive and tendon loads in appropriate ratios while placing each foot on an inclined surface. The degree of incline selected for each position included 6° of plantar flexion for heel strike, 7° of dorsiflexion for stance, and 10° of dorsiflexion for heel off. Tendon loads were calculated from physiological cross-sectional areas, reported peak forces, and normalized EMG values.

  Each specimen was mounted within the foot loading frame, and nylon cables with tendon clamps were used to connect the tendons of the foot to regulated pneumatic cylinders. A central pneumatic cylinder was used to compressively load the foot via the tibial intramedullary rod. 3-D motion sensors were attached to each of the four carbon fiber pins. These electromagnetic sensors were used to track the 3-D motions of the hindfoot bones. The transmitter was mounted onto the loading frame so that its Z-axis aligned with the long axis of the tibial shaft, the X-axis was directed along the medullary canal of the second metatarsal, and the Y-axis was perpendicular to these two axes (approximately medial-lateral). This unloaded position of the foot was the initial position. 3-D motion (bone orientation) data was collected for each of three conditions: initial position, all tendons loaded except the PTT (All-PTT), and all tendons loaded including the PTT (All+PTT). This protocol was repeated for each of three different foot positions (heel strike, stance, and heel off).

  For each test, all of the motion data was transformed such that both the All-PTT and All+PTT data were reported relative to the initial position. A non-parametric, Wilcoxon Signed Rank paired analysis was then used to statistically compare the angular orientation of the four bones for both conditions. This analysis was performed for all three loading positions including heel strike, stance, and heel off.

RESULTS--Small but statistically significant changes in the angular orientation of the four-bone complex were observed when the PTT tendon was loaded (All+PTT) as compared to unloaded (All-PTT) for all three foot positions. These differences were typically less than 1 degree, particularly during heel strike and stance simulation. The greatest changes in angular orientation resulting from loading the PTT were seen during heel off. This was not surprising since the PTT is most actively recruited during this phase of the gait cycle. Talar (3.6°±1.2, p=0.011) and calcaneal (2.3°±0.5, p=0.011) internal rotation together with talar plantar flexion (1.4°±0.9, p=0.011) was observed when the PTT was loaded. Eversion of the navicular (3.9°±2.0, p=0.011), cuboid (2.4°±0.9, p=0.011), and calcaneus (2.0°±0.8, p=0.011) was also observed at heel off.

CONCLUSIONS--Restoring function to the PTT in an acquired flatfoot model had little effect on the kinematic orientation of the hindfoot complex. Although there were statistically significant differences between the All-PTT and All+PTT conditions, these differences were typically less than a few degrees. These results suggest that restoring PTT tendon function alone may not fully correct a radiographically documented flatfoot deformity.

[198] PRECONDITIONING AS A TECHNIQUE TO MINIMIZE TOURNIQUET-INDUCED MUSCLE INJURY

Carlos A. Guanche, MD; Richard V. Baratta, PhD
Bioengineering Laboratory, Department of Orthopaedic Surgery, Louisiana State University Medical Center, New Orleans, LA 70112

Sponsor: Foundation for Sports Medicine Education and Research

PURPOSE--Tourniquets are frequently applied in extremity surgery to minimize blood loss and to improve surgical field visualization. Unfortunately, a degree of risk is associated with the ischemic period experienced by the limb during tourniquet application. The pupose of this research is to develop a methodology to minimize the risks associated with the use of tourniquets during extremity surgery.

METHODOLOGY--Preconditioning is a technique which consists of a brief period of ischemia (5-10 min), followed by re-perfusion for an equal amount of time before the prolonged ischemic episode. Our research regarding this technique includes incremental steps that address immediate effects in an animal model through clinical application during arthroscopic surgery.

PROGRESS--The intraoperative effects of muscle preconditioning have been investigated in a cat model, where the medial gastrocnemius muscle strength is tested during and immediately after tourniquet application. Our results show that preconditioned muscles maintain a significantly higher (p<0.005) ability to contract 1 hr after tourniquet application.

  A prospective, randomized study is being performed to assess the effect of preconditioning on the leg muscles of arthroscopy patients.

RECENT PUBLICATIONS FROM THIS RESEARCH

[199] A SURVEY OF PARENTS OF CHILDREN WHO USE ORTHOSES AND PROSTHESES

Claire B. Kopp, PhD; Toya Wyatt, PhD; Dani Hodge, MA; Donald McNeal, PhD
Rehabilitation Engineering Program, Rancho Los Amigos Medical Center, Downey, CA 90242

Sponsor: National Institute of Disability and Rehabilitation Research, U.S. Department of Education, Washington, DC 22202

PURPOSE--This study focuses on three specific issues: 1) to identify parents' perceptions of minority community views of health, illness, the importance of motor skills, and the implications of orthopedic disabilities for children; 2) to identify parents' personal views of health, illness, motor skills, and orthopedic disabilities with reference to children; 3) to identify the specific parent actions taken to secure child services with respect to orthotic and prosthetic devices.

METHODOLOGY--The study involves parents of children who have orthopedic impairments and who range in age from the preschool years through late adolescence. Upon signing a consent form, parents are interviewed in their homes and at their convenience. Data analyses will involve reliability checks of the instrument, descriptive statistics for subgroups (e.g., ethnic group membership, age of child), relevant dimensions, and if numbers allow, multiple regression analyses to identify the variables that contribute to parents accessing the system for assistive devices for their children.

PROGRESS--Information packets have been designed to acquaint parents and others with the goals of this study. In addition to parents, target recipients include community leaders, directors of regional centers, and others who might directly or indirectly be involved with families appropriate for this study. Each packet contains a letter that briefly describes the project and requests parent participation, a fact sheet that provides details about the project, and a postage-paid return postcard. Packets are available in English and Spanish.

  Resource guides have been designed to provide key access information to parents about local service organizations, support groups, and legal resources that cater to the needs of individuals with handicapping conditions. In addition, the guides supply practical information about the prevention of child abuse, consumer resources, disaster preparedness, housing assistance and tenants' rights, community resources geared to combat harmful gang activities, and neighborhood programs focused on literacy and English as a second language. The guides will be provided to parents as a benefit for participating in the study.

  Many hundreds of information packets have been distributed to interested community agencies, who in turn, forwarded the packets to parents who might participate in the study. To date, more than 70 parents have indicated an interest. Telephone contacts are made after receipt of postcards. A preliminary screening identifies children who meet subject criteria (e.g., an orthopedic problem, normal intellectual functioning or mild developmental/cognitive delay, chronological ages between 2 and 17 years). Approximately 50 percent of responding parents have children who do not meet subject criteria. Most frequently, the children have major neuromuscular disorders and/or severe cognitive impairments.

  Thirty interviews have been completed with parents who represent diverse cultural and ethnic backgrounds; their children have various kinds of orthopedic conditions. Although the interviews are lengthy, parental commitment to the interview is impressive. Apparently parents welcome the opportunity to talk about their child and their parenting experiences. Recruitment continues in order to meet the study target goal of a minimum of 100 participants.

  The interview data base is being established. Code books and variable lists are being developed, and to date, information on 40 variables has been entered.

B. Hip Implants

[200] FATIGUE STRENGTH OF COMPOSITE FEMORAL COMPONENTS FOR HIP ARTHROPLASTY

Harold D. Schutte Jr., MD; Robert A. Latour Jr., PhD; J.M. Kennedy, PhD; S.B. Biggers Jr., PhD
Ralph H. Johnson VA Medical Center, Charleston, SC 29401; Departments of Bioengineering amd Mechanical Engineering, Clemson University, Clemson, SC 29634, email: robert.latour@ces.clemson.edu

Sponsor: Department of Veterans Affairs, VA Rehabilitation Research and Development Service, Washington, DC 20420
(Project #A670-3RA).

PURPOSE--The purpose of this research is to investigate how the design of composite femoral components influence fatigue behavior, and how components should be mechanically tested in vitro to closely simulate in vivo stress states. The goal of this research program is to develop the necessary technology base to enable low stiffness/high strength composite femoral components to be confidently designed and tested for clinical use with the potential for reducing stress-shielding-induced proximal bone resorption following hip arthroplasty. The reduction of bone resorption following arthroplasty should promote longer femoral component life and significantly reduce complications in revision surgery.

METHODOLOGY--In Phase 1 of our research, a detailed anatomic 3-D computer model of the proximal femur with an in situ composite femoral component is being developed to investigate relationships between composite femoral component design and fatigue performance under simulated in vivo loading conditions. In Phase 2, similar numerical modeling and analysis techniques will be applied to determine the proper experimental parameters to be used in Phase 3 for actual fatigue testing of composite stems for model verification. Test parameters and fixture design will be studied to provide component stress states during fatigue testing that closely match those expected in vivo. Once the proper experimental parameters have been established, composite femoral components will be fabricated and fatigue tested in Phase 3. Results are expected to provide a detailed understanding of how composite structural design will influence fatigue behavior under simulated in vivo loading conditions. This knowledge is essential to enable composite femoral component technology to be developed for successful clinical application toward the goal of improving hip joint replacements.

PROGRESS--Phases 1 and 2 have been completed, yielding composite femoral components to be closely simulated in Phase 3. Under Phase 3, a new thick composite plate fabrication hot-press has been prepared for operation and utilized for the fabrication of trial composite laminates. Thick composite plate molds have also been designed and are currently being produced. The molds will be used for the fabrication of the thick composite plates that will be machined into the composite femoral components for fatigue testing.

PRELIMINARY RESULTS--Experimental verification testing demonstrates that the developed computer models accurately predict the mechanical behavior of the femur. Preliminary results of the computer models have been obtained to study the femoral model response under the three load-bearing phases of gait (heel-strike, mid-stance, and toe-off) and stair-climbing. Additionally, the preliminary results from fabricated AS4/APC2 composite plates have shown the new lamination press to produce quality laminates as is necessary for the fabrication of the composite femoral components that will be tested.

FUTURE PLANS--Now that the femoral computer model is completed and verified, it will be utilized during the upcoming year for fatigue modeling studies to evaluate relationships between composite femoral component design and fatigue performance. Composite femoral components will then be fabricated according to the computer model design results. The design and fabrication of the experimental test apparatus will also be completed within the upcoming year and be utilized to experimentally study and document fatigue behavior of the new femoral component designs.

[201] THE EXAMINATION OF EXPLANTED, UNCEMENTED ORTHOPAEDIC PROSTHESES

John P. Collier, DE
Dartmouth Biomedical Engineering Center, Thayer School of Engineering, Dartmouth College, Hanover, NH 03755-8000

Sponsor: Department of Veterans Affairs, VA Rehabilitation Research and Development Service, Washington, DC 20420
(Project #A473-3DA)

PURPOSE--The overall objective of this study is to assess the long-term feasibility of porous coating as a mechanism for fixing orthopaedic prostheses to bone. Additionally, the role of design and material in the performance of these prostheses is being determined through this analysis.

METHODOLOGY--This examination of clinically retrieved, porous-coated hip and knee prostheses will assess the importance of such variables as material composition, design, location of porous coating, pore size, and surface roughness on the resulting interface between the prosthesis and bone. The study will address the issues of stress shielding, ion release, and wear debris formation, and will clarify, where possible, causal relationships with prosthesis parameters.

  Retrieved prostheses are fixed in formalin and examined macroscopically and graded for wear, corrosion, fretting, and tissue adherence. The metal components are mapped for both soft and hard tissue apposition prior to being embedded, cut into standard histological sections, and stained. Serial sections permit the assessment of bone resorption or of stress shielding for prostheses that are retrieved intact as post-mortem specimens. The polyethylene bearing surfaces are graded for damage and defects and then sectioned on the microtome to determine the degree of polyethylene consolidation.

PROGRESS--We presented several recent studies at the American Academy of Orthopaedic Surgeons meeting in San Francisco in February 1997. These studies highlight the results of our current investigations and will be published.

The Impact of Sterilization Method on Clinical Damage of UHMW Polyethylene Knee Components
  Seventy-six retrieved, unicompartmental polyethylene bearings were rated macroscopically for clinical wear. They represent 7 different manufacturers and 13 different designs. A Jung microtome was used to shave 200 micron thin films from the cross section of a variety of the polyethylene bearings. The thin sections were examined under reflected light at 10-60× for the presence of a subsurface white band, shown to be an indicator of gamma ray-induced oxidation. Sterilization method was determined from manufacturer information and by oxidation spectra from Fourier Transform Infrared Spectroscopy (FTIR).

RESULTS--Fifty-three of the knees were gamma irradiated. Their average in vivo duration was 65 mo. Cracking and/or delamination were present in 57 percent (30/53). Complete wear-through of the polyethylene occurred in 21 percent (11/53), and the white subsurface band was apparent in 84 percent (16/19).

  The remaining 23 knees consisted of 3 designs from the same manufacturer; all were implanted within the time period that this company is known to have been using EtO for sterilization. Their average duration was 155 mo. Cracking and/or delamination was absent. Both complete wear-through of the polyethylene and the white subsurface band were absent in all cases.

CONCLUSION--Retrieved polyethylene knee components gamma irradiated in air are observed to have a high incidence of cracking and delamination (>50 percent), leading at times to complete wear-through of the bearing. Knee components sterilized with EtO have little evidence of fatigue wear even after in vivo durations of >15 yrs.

Shelf Versus In Vivo Oxidation of Polyethylene Orthopedic Inserts
  Retrieved ultra high molecular weight polyethylene (UHMWPE) inserts were sectioned by microtome and analyzed microscopically for the presence of a subsurface white band, which signifies oxidative degradation of the material. FTIR spectroscopy was used to determine the oxidation products as a function of depth into the inserts. Uniaxial tensile testing of thin sections measured the UTS and elongation of the material. Hexane extraction of retrieved insert thin sections was used to differentiate oxidation products from in vivo absorbed species. Never-implanted UHMWPE inserts were analyzed in a manner similar to that for the retrieved inserts. Insert sterilization dates were obtained from the manufacturers.

RESULTS--More than 25 retrieved and 20 never-implanted UHMWPE inserts gamma sterilized in air have been evaluated. Oxidation in retrieved inserts is more complex than in shelf-aged inserts. Retrievals can exhibit surface oxidation, either alone or in addition to a subsurface oxidation maximum. Surface oxidation, a result of in vivo duration, follows a diffusion gradient from the surface of the component. It is a result of absorbed species (esters) that are extractable with hexane. All components show a trend of increasing oxidation and decreasing mechanical properties with increasing time on the shelf prior to implantation. Large differences in clinical wear (particularly delamination and cracking) also appear to relate to pre-implant shelf time.

CONCLUSION--Retrieved inserts exhibit a large variation in oxidation and mechanical integrity. Inserts of the same design and material show extremes of performance in vivo. While gamma sterilization in air is detrimental to polyethylene in general, the process does not always lead to degradation in vivo. This study differentiates the very near surface oxidation (absorbed species) from in vivo degradative oxidation reactions with the polyethylene. Variations in polyethylene insert history (pre-implant shelf time vs. in vivo duration) are found to be largely responsible for mechanical property changes that lead to variations of performance in vivo.

PROJECT SUMMARY--We have evaluated more than 500 retrieved components this year bringing our total above 5,000. We continue to gain insight into the factors which influence implant performance including: design, wear in service as well as the polyethylene material variations. Our recent publications have focused primarily on the effect of sterilization on the polymer bearings of the joint replacements.

RECENT PUBLICATIONS FROM THIS RESEARCH

[202] IMPLANT RETRIEVAL RESEARCH, RESULTS AND CONCLUSIONS, 1988-98: A SPECIAL REPORT

John P. Collier, DE
Dartmouth Biomedical Engineering Center, Thayer School of Engineering, Dartmouth College, Hanover, NH 03755-8000

Sponsor: Department of Veterans Affairs, VA Rehabilitation Research and Development Service, Washington, DC 20420
(Projects #A473-1DA; #A473-2DA; #A473-3DA; #A473-4DA: 1988-98)

PURPOSE--Arthritis of the hip and knee is a crippling disease for which joint replacement represents a compelling option to relieve pain and increase mobility. In fiscal year 1986, replacements for 3,068 hips and 362 total knees were performed in the VA system alone. Of these, a significant number would fail, due to a breakdown of the interface between the prosthesis and the host bone of the patient. According to representatives of the three largest manufacturers, nearly half of their orthopaedic implant sales in 1986 were porous-coated designs. While this represented a significant use of these prostheses by orthopaedic surgeons, there was no central resource available for evaluation of their performance, and virtually no clinical information in the literature. This need prompted the VA to establish a center to conduct follow-up studies of the failed devices, to provide direct feedback to the implanting surgeons, and to improve the overall medical care of those undergoing joint replacement.

  For the past decade, The Center for the Examination of Retrieved Orthopaedic Prostheses has been the largest program of its type in the country, receiving approximately 600 components per year from more than 700 participating surgeons: in the first 8 years of this VA-funded program, we evaluated more than 5,000 orthopaedic prostheses. This research has led to a number of unanticipated and important findings about the performance of artificial hips and knees.

  At the Center's inception in 1988, the predominant joint replacement technologies were evenly divided between cemented and uncemented/porous coated implants. Fixation of the implants was the greatest problem in joint replacement, and our original objective was to determine whether porous coating resulted in improved long-term fixation of orthopaedic prostheses to bone. Our analysis provided a way of identifying problems with implants and allowed evaluation of the performance of competing technologies, including porous coated versus non-porous coated implants; bead/pore size of the porous coating; and titanium versus cobalt.

  Since then, concerns regarding the integrity of porous coatings, the longevity of hydroxyapatite surfaces, the potential for modular connections to fret and corrode, and most recently, the impact of gamma radiation sterilization on the performance of polyethylene (PE) bearing surfaces, have all arisen from the examination of retrieved prostheses, while joint replacements have increased to over 350,000 each year in the US. We have built a large database to track the influence of design, material, and fabrication techniques on clinical outcomes and expanded the scope of our research to include the overall design and performance of the artificial joints and the impact of PE quality and sterilization upon them.

METHODOLOGY--Upon arrival at the Center, retrieved implants are soaked in formalin for 48 hours, then air dried and examined macroscopically. The prostheses are evaluated for both soft and hard tissue ingrowth to the implant. The metal is rated for burnishing, scratching, abrasion, and fretting; the PE is rated for burnishing, scratching, abrasion, pitting, delamination, creep, and cracking. Both metal and PE are graded on a scale of 0 (absence of the phenomenon) to 3 (presence of severe amounts of the phenomenon).

  Depending on the initial evaluation results and specific implant characteristics, further analysis may be required. To determine the extent of bone ingrowth, and amount of metal and PE debris, implants are embedded, sectioned and stained in preparation for histological evaluation. Modular components are evaluated for evidence and severity of corrosion. PE is measured for wear and tested for chemical and mechanical properties. All data are entered into the database mentioned above.

Overall Design and Performance of Hips
  Prior to 1977, the only alternative to direct contact between the metal of the prosthesis and the bone of the patient was the bone cement polymethylmethacrylate (PMMA), inserted as a luting agent between the two and allowed to polymerize in situ. While this material increased comfort for the patient and provided the surgeon with better assurance of a good fit, it proved to have poor fatigue properties; a significant number of cemented prostheses failed at the cement/bone interface.

  In 1977, an alternative was made available in the form of a porous metal coating applied first to femoral hip prostheses. Unlike cemented prostheses that require no biological response for fixation, porous coated devices require ingrowth by bone and fibrous tissue (osseointegration) to develop an interface maintained by the body and therefore not subject to this fatigue failure. Osseointegration is considered successful if bone tissue grows directly into the pores of the coating. Well-organized fibrous tissue-ingrowth can also provide satisfactory implant fixation in some cases. Our research has shown that the primary requirement for satisfactory ingrowth is initial rigid fixation with a tight press fit of porous coating against bleeding bone. In a recent study of 1,050 retrieved porous-coated, uncemented hip components with in vivo durations greater than 2 months (521 acetabular cups, 536 femoral stems), 218 of the cups (42 percent) and 275 of the 536 stems (51 percent) displayed some extent of bone ingrowth.

  In revision hip prostheses, it can be difficult to obtain the required tight press fit of porous coating to bone because of the large, irregular proximal femoral canal and an irregular diaphyseal geometry. In these cases, fully porous-coated prostheses can afford the necessary contact of coating against living host bone. Also, proximally coated femoral hip prostheses, when used with femoral allografts, may place the porous coating against dead bone and, therefore, bone ingrowth cannot reasonably be expected. However, substantial fibrous ingrowth from an endosteal response can be expected, and may be sufficient to fix the prosthesis to the graft. In addition, the firm fixation of acetabular components in revision surgery can be more difficult than the fixation of femoral components. Acetabular components fixed with screws are not immune to micromotion, which can generate metal debris, as evidenced by blackening of the tissue in the region of the screw holes.

  The early porous-coated prostheses had an average pore size of less than 50 µm. Early retrievals displayed fibrous tissue ingrowth but no ingrowth by bone. Animal experiments and evaluation of human physiology indicated that pore sizes of at least 100 µm would be required to permit bone and its attendant blood supply to grow into the prostheses. Subsequently, pore sizes were increased, and later retrievals had evidence of bone ingrowth. Today's prostheses have pore sizes ranging from 100 to >1,000 µm. Histologic evidence indicates that for most hip and knee components, pore size variation within this range is not critical to the performance of the prosthesis, and examples of dense bone ingrowth for most every type of component has been observed.

  The surface roughness of the porous coating is, however, a function of the powder size and pore size of the coating. It is apparent that larger pore size prostheses have a smoother surface than small pore size prostheses. Therefore, prostheses which rely only on the frictional fit of the coating against reamed bone must take this into consideration. Prostheses with large pore size and relatively smooth surfaces may need screws, threads, or pegs to provide initial stabilization until ingrowth fixation is established.

  Current interest in the use of coatings, particularly plasma-sprayed hydroxyapatite (HA) and tricalcium phosphate (TCP), on femoral and acetabular hip prostheses is based upon a large series of studies showing calcium phosphate ceramics to be osteoconductive in animals. There is a general consensus that calcium coatings on smooth-surfaced prostheses produce improved mechanical fixation for up to 12 weeks postoperatively. At longer periods, the results are mixed: some studies show increased bond strength, others show none.

  The results of the use of HA coatings on porous-surfaced prostheses are somewhat more difficult to interpret. In our early animal studies, there was more bone ingrowth with HA coating, but the mechanical fixation strengths were similar at all time periods. Also, it was generally observed that the HA/bone bond interface was stronger than either the HA-implant interface or the HA coating itself. HA has been shown to be resorbable in animal experiments, and it appears that this is also the case in the clinical setting. How long the resorption takes is subject to a large number of variables, including the type of bioceramic applied, the degree of crystallization, the thickness of the coating, and its level of density and coherence.

  It appeared to us that the weak link in some systems was the HA-implant bond. Examination of six retrieved HA-coated components demonstrated direct bone adherence to the HA. Areas not in contact with bone appear to absorb first. Rates of resorption were quite variable, with examples from some manufacturers demonstrating complete resorption after 6-9 months in situ and components from another manufacturer demonstrating considerable HA at 48 months. Histologically, the bone/HA bond appears to be quite as secure as bone ingrowth; removal of both systems requires removal of surrounding bone due to the strength of that bond. Although HA appears to be osteoconductive, it does not appear that HA can be considered the sole long-term mechanism for fixation of prostheses, due to its limited initial bond strength to the substrate and the ability of the host to resorb the material over time. Therefore a stable implant geometry and a textured titanium surface suitable for bone adherence may be critical for long-term fixation.

Material Composition
  In the early 1980s, the desire for reduced femoral hip-stem stiffness lead to the increasing use of titanium alloys (Ti 6A1 4V, ASTM F-136-79) in place of the stiffer cobalt-chrome materials (HS 21, ASTM F-75). Although concerns were expressed for its use in cemented applications because of its flexibility and associated increased risk of failure of the cement mantel and the potential for metal ion release if the stem became loose, there has been little concern about its application for use in press-fit or biologic fixation applications. The fatigue characteristics of titanium alloy are excellent under cyclic loading, and it is a very workable metal, easily milled to tight tolerances.

  Titanium has achieved acceptance over cobalt alloy for dental implants, due to the belief that the likelihood of osseous integration is greater; laboratory testing in animals and a study of porous plugs of both materials in humans has tended to support the merits of titanium. The trend toward the use of the more easily machinable titanium alloys presented the opportunity to compare the ingrowth frequencies of the two alloys in components of similar design, manufacturer, porous coating type, and pore size.

  We examined 109 acetabular components for ingrowth, 46 composed of cobalt alloy, 63 of titanium alloy. All had durations in vivo >2 mo, and no severe wear of the metal or PE was evident. The average implant duration of the devices was 34 mo for the cobalt and 37 mo for the titanium. Eight of the 46 cobalt alloy cups (17 percent) and 36 of the 63 titanium devices (57 percent) presented bone ingrowth. With design variables minimized, the frequency of bone ingrowth into titanium alloy components was more than three times greater than those of cobalt, supporting the contention that porous coated acetabula composed of titanium are more likely to have successful biologic fixation than those of cobalt alloy. Femoral hip components do not present as close an opportunity for direct comparison as acetabular cups, and both titanium and cobalt alloy components are often observed to have become bone ingrown. Ingrowth of knee components appears to be directly related to quality of fixation, especially in the tibial component. Well-fixed titanium components were observed to demonstrate a greater frequency of ingrowth than cobalt alloy devices but direct comparisons have not been possible.

  First-generation titanium-alloy femoral hip components with integral heads were seen to suffer scratching, burnishing, and loss of head sphericity to a greater degree than their cobalt-alloy counterparts. From the mid-1980s to date, it appeared that the solution of a cobalt-alloy head on a titanium stem was satisfactory for both press-fit and biologic ingrowth applications. The titanium alloy stem provided reduced stiffness, and the tougher cobalt-alloy femoral head provided improved wear resistance against the PE. This trend toward modular prostheses that permit variable neck lengths through separate heads and stems mated at the time of surgery, provided manufacturers the opportunity to combine dissimilar metals. Since then, modular hip prostheses have become the standard cementless design and account for a large percentage of the cemented hip market as well. This reflects a compelling interest within the surgical community in the technical advantages accorded by modular design. The 2-piece acetabular components with separate shells and bearings, and femoral stems with modular heads, have led more recently to designs with increasing numbers of components. Potential benefits of this increased modularity include intraoperative flexibility in implant geometry resulting in improved fit, optimization of materials, simplified revision surgery, and, with improved patient outcome, a long-term reduction in overall costs.

  Although modular hip prostheses have become dominant over one-piece configurations, the use of these devices has not been without complications. Several researchers have reported the failure of modular and factory-assembled acetabular components through disassociation of the liner and shell. Reports of corrosion and fretting of modular femoral stems, and the contributory roles of these two phenomena in the generation of debris, also are common in the literature. Modularity requires a mechanical coupling of parts by the surgeon at the time of surgery; as a result, the attention to surgical cleanliness and the technique used to assemble the parts may influence the ultimate longevity of the component in service. In addition, the quality of the materials and the accuracy of the machining of the mating surfaces are likely to play a role in performance. Modular components inherently have more potential modes of failure than one-piece devices, and each modular connection is a potential site for fretting, corrosion, and disassociation.

  One major concern arising during development of the cobalt-alloy head/titanium-alloy stem femoral hip components was their potential for multi-alloy crevice corrosion, deterioration occurring within crevices and other shielded areas of metal surfaces exposed to corrosives such as body fluids. Crevice corrosion may be differentiated from fretting corrosion in that the former requires no relative motion between surfaces, whereas the latter occurs from a combination of motion and corrosion. However, the amount of motion required to produce fretting is so small (on the order of microns) that it may be difficult if not impossible for a visual analysis to eliminate the potential for fretting as a contributor to corrosion of retrieved components. A crevice corrosion site must be wide enough to allow liquid to enter but narrow enough to create a stagnant zone: openings a few thousandths of an inch or less in width are sufficient.

  The mechanism for crevice corrosion of both similar and dissimilar metals, involves the dissolution of the metal alloy into its respective ions and the reduction of oxygen to hydroxide ions (Oxidation M-->M+e; Reduction O2+2H20+4e-->4OH-). The oxygen within the crevice is depleted because of restricted convection. This tends to produce an excess of positive charge in the solution, balanced by the migration of negative chloride ions into the crevice. The fluid in the crevice can contain 3 to 10 times as much chloride as the bulk solution and possess a pH approaching 1. The corrosion that occurs at the interface between a titanium-alloy stem and a cobalt-alloy head is galvanic in nature and occurs when the passive oxide layer is breached. Metals or alloys that depend on oxide films or passive layers for corrosion resistance, such as CoCr and Ti alloys, are particularly susceptible to crevice corrosion.

  The initiation of crevice corrosion requires, in addition to a breakdown of the passive film, the presence of an aqueous environment of body fluids. The continuation of corrosion requires that the fluids be constantly, though very slowly, replaced. One ramification of the corrosion is that fluid from inside the taper with a high metal ion content of the constituents of both the cobalt alloy and titanium alloys will be released into the host.

  A continued examination of more than 650 modular femoral components tends to indicate that some designs demonstrate a higher frequency of corrosion than others, independent of the material combinations. It appears that taper geometry and quality of the fit may be dominant factors controlling the extent of the fretting that can lead to corrosion through breakdown of the passive oxide layer as previously described. However, there is still an interesting difference in the frequency of corrosion of mixed metal components versus single-alloy devices of identical taper geometry from a single manufacturer. In this recent study, 651 femoral hip prostheses, 411 of which were modular and 240 of which were one piece, were analyzed for fretting, corrosion, wear, and disassociation. Mixed-alloy components had a higher incidence of corrosion than any other combination, nearly 10 times higher than the all-cobalt-alloy systems, and 5 times higher than the all-titanium-alloy prostheses. When the total number (n=15) of retrieved all-cobalt alloy devices was matched with an equivalent number of mixed-alloy devices with similar durations from the same manufacturer, it was found that even with taper geometry and durations closely matched, the frequency of corrosion was significantly higher in the mixed-alloy group. Whether the difference in corrosion frequency is related to any galvanic effect or to the difference in modulus of elasticity of the two materials cannot be determined from the components.

  Our retrieval analysis results show that the crevice provided by the tapers of modular prostheses is a harsh environment that can lead to corrosion if the metallurgy or design is not optimal. Several factors appear to be important in accelerating the rate of corrosion, including:

  Relative motion between head and stem of tapers will concurrently provide a constant breakdown of the corrosion resistant. The rapid intergranular attack of the very heterogeneous sinter-heat-treated materials as well as the relatively rapid attack of the cast-alloy heads on the titanium-alloy stems indicate that heterogeneous alloy structures will be more susceptible to rapid corrosion attack than those that are more homogeneous.

  A second problem of corrosion is the potential for the long-term loss of mechanical fixation of the head to the stem. No prosthesis is known to have lost the head/stem bond by corrosion, although several retrievals have heads that were loose on their tapers. There is no reason to believe that the corrosion rate should decrease with time, and our data show that the longer the implant duration, the greater the corrosion observed. Once firm fixation is lost, fretting will add to the rate of material loss and may be expected to obliterate the pitting characteristic of this type of corrosion. Elimination of fluid from the crevice environment will eliminate the potential for corrosion, as would the return to fixed-head prostheses. However, the fact that many modular components have been in service for years demonstrating little or no evidence of corrosion indicates that proper attention to detail can result in satisfactory longevity of taper-type modular connections.

  The metal backings of acetabular components can reduce the available PE thickness, often to an alarming extent. The pitting and cracking of thin PE surfaces have some similarities to tibial and patellar bearings, and creep-related deformation occurs more frequently in thin PE components.

  Bartel et al. described the effect of metal backing on the stresses in PE, finding that head size was an important variable: smaller head sizes have smaller contact areas and consequently higher stresses. On the other hand, larger metal heads demand the use of thinner PE articulating surfaces. Stresses in the PE were increased enormously as thickness was reduced below 6 mm. They also examined variables of congruency between the femoral head and the PE insert as well as stiffness of the PE. They found that increasing the stiffness of the PE and reducing congruency between the femoral head and the PE insert each increased the shear stresses. They acknowledged that perfect congruency between ball and cup would be difficult to replicate, because of inherent variations in the manufacturing processing, and that fortunately, congruency increased as wear took place, resulting in lower stress and lower wear conditions.

Overall Design and Performance of Knees
  Immediately following the early successes of the porous-coated hip prostheses, some failed total knee prostheses were suspected to have also been the result of problems with the cement-metal interface, and application of porous coatings to total knee components was initiated. The first such available in the U.S. were the porous-coated anatomic prostheses (PCA), a cobalt alloy prosthesis provided with a sintered, large-pored, beaded porous coating. Shortly thereafter, other manufacturers developed a variety of pore sizes of cobalt alloy bead and wire mesh titanium coatings, leading to a series of investigations to determine the optimal pore size, whether or not cobalt alloy was superior or inferior to titanium alloy, and whether a fiber mesh surface was more or less conducive to bone ingrowth than a beaded surface.

  Analysis of 505 retrieved knee prostheses indicates that bone ingrowth occurs independent of pore size in the range currently available and independent of whether the coating is composed of cobalt or titanium alloy. Both beaded and wire mesh systems demonstrated the ability to permit bony fixation. Bone ingrowth of multiple screw-fixed, titanium tibial prostheses was seen with greatest frequency (75 percent), but no determination could be made as to whether this was more a function of the material composition or the type of fixation. Experience with other orthopedic prostheses suggests that mechanical fixation is the dominant variable in these systems.

  The geometry of the femoral knee prostheses provides sufficient fixation and resistance to micromotion so that ingrowth of the large variety of porous surfaces appears reliable. Well-fixed patellar prostheses also demonstrate bone ingrowth, but their ingrowth interfaces appear at early risk to breakdown, due to the presence of PE wear debris generally associated with the metal-backed PE design.

  The tibial prostheses evidently present a much more difficult fixation problem. The quality of bone in the proximal tibia often varies widely between medial and lateral compartments, and loads rock the prostheses both in the mediolateral as well as the anteroposterior planes. The most common appearance of tissue ingrowth of these prostheses is that of a fibrous interdigitation of varying quality. Those plateaus poorly anchored to the bone generally present the appearance of a fibrous membrane and osteoclastic activity in the adjacent bone. Prostheses that were better anchored, either by screws or central pegs, characteristically present the appearance of well-interdigitated fibrous tissue, which may be absent of any osteoclastic activity. Only those devices with very firm mechanical fixation and contact with good quality bone present bony ingrowth.

  Retrieval analysis frequently revealed severe wear of the PE of the tibial insert, with the resulting wear debris often found at the interface between the metal backing and tibia. The long-term fixation of these prostheses by either bone or well-interdigitated fibrous tissue appears to be at risk to the factors released by wear debris. Early work by Willert and Semlitsch indicates that in the presence of wear debris, the host can be expected to release prostaglandins and collagenase that may well break down the interface of even well-fixed prostheses.

Material Composition
  With the initiation of porous-coated total knee components, prostheses that had previously been composed entirely of PE required the application of a metal backing to which the porous coating could be applied. For cemented implants, a metal backing became desirable to permit bearing replacement without disruption of the implant fixation. Of necessity, this backing took the place of some of the previous PE thickness. The problems of patellar resurfacing are more severe than tibial or acetabular replacement, as there is insufficient room for both a metal backing and a thick PE insert without either removing sufficient bone stock that fracture of the patella becomes a very real possibility, or over-stuffing of the joint, resulting in pain and limited flexion. Therefore, many of the current designs provide a very thin PE bearing at risk for failure. Examination of components retrieved for PE failure or separation as well as those components retrieved for other reasons, often reveal the presence of particulate PE wear debris at the margins of the bone ingrowth surface, leading to concerns for long-term breakdown of this interface.

  The metal backing typically reduces the space available for PE by 3-4 mm. As a result, the PE thickness is always less than the 6 mm cited as necessary for nonconforming surfaces by Wright and Bartel. At the fixating lugs and margins of the component, the PE thickness may be less than 3 mm. Additionally, creep reduces the PE thickness, thereby further increasing the stresses causing early fatigue failure.

  Testing of total knee components indicates that patellofemoral articulation produces the greatest contact stresses at higher flexion angles as a function of increased resultant loads. There is also a concomitant reduction in contact area as the patella moves into the intercondylar notch, thus losing support at the apex and shifting contact to two points at the periphery. The increase in contact stress in flexion can be significant. Most dome-shaped components are designed to have their greatest conformity in extension. Additionally, designs with convex condylar and patellar geometries can produce point contact that results in very high contact stresses. To avoid high contact stresses in the patella, it is important that components be designed to match femoral geometries in both extension and flexion. Drawing the metal of the patellar groove further into flexion serves to increase the contact area and decrease the contact stresses in this region. In some instances, however, this may not leave enough space to preserve the posterior cruciate ligament. Another way to increase contact area in flexion is to modify the dome of the patella component so that it more closely matches the curve of the femoral condyles in the frontal plane.

  Reports of the high incidence of PE failure of metal-backed patellar components rekindled interest in cemented, all-PE designs. The trend was based on the false assumption that the metal backing was the source of all problems and that all-PE components, relatively independent of design, would provide improved performance simply because of increased PE thickness. Our analysis of 104 retrieved patellar components, combined with high-resolution, contact stress testing, indicates that surface geometry is as critical to performance as PE thickness. Contact stress analysis indicated that under loads of 4× body weight, all but the most congruent designs produced stresses in the PE considerably in excess of its yield strength. The least congruent designs produced more than twice the contact stress of the most congruent designs.

  In the early 1980s, the possibility that aseptic loosening, the major cause of total knee failure, was influenced by the flexibility of the all-PE tibial component was examined. Researchers concluded that metal-backed components would reduce the stress in the supporting bone and bone cement compared to all-PE designs. In 1985, the importance of PE thickness and geometry was analyzed in several separate studies, and researchers agreed on a recommended minimum PE thickness of 8 mm.

  This PE thickness, when combined with a metal backing, results in a relatively thick component requiring significant bone resection. Surgeons have been taught to resect as little bone as possible and manufacturers realize they must sell what the market demands. Therefore, many of the current designs provide surgeons with less than the recommended minimum PE thickness.

  Our analysis indicated that the vast majority of failed, metal-backed tibial components were of a modular type and had been fitted with the thinnest, or close to the thinnest, available PE inserts. This is most likely the result of the surgeon's desire to resect as little tibial bone stock as possible, combined with a lack of retrieval data documenting the potential for failure of thin PE inserts caused by high-contact stress. While a study by Goldstein has shown that the quality of tibial bone decreases with increasing depth of resection, a careful analysis of the data indicates that these changes are small within the first 2 cm of the tibial surface. Therefore, it is not unreasonable to suggest that surgeons sacrifice several millimeters of bone to enable the insertion of a thicker PE bearing.

  High stresses in thin PE bearings in noncongruent knee designs are predictable. Further, our analysis of retrieved tibial components revealed that the less congruent, higher contact stress-producing devices presented greater wear. Finite element analysis demonstrated that the stress intensity in the PE decreases dramatically with increasing thickness. An increase of as little as 4 mm of PE (from 3 to 7 mm) reduced the maximum Von Mises stress by as much as 27 percent, and the maximum contact stress by 14 percent. Retrieval analysis also revealed that the fixation of the PE inserts to the metal backing does not eliminate motion between the two, and that an additional, initial articulation is provided in many of these modular prostheses. Over time, the PE often creeps significantly: more than 1 mm of creep into screw holes was seen in some components. While this creep may eventually help the PE to lock in place, its extent, and the potential for screw/PE contact and fretting wear, raises additional concerns.

  Thin PE tibial inserts and designs with low conformity contribute to high stresses within the PE, increasing the risk for fatigue failure. Additionally, surface heat treatment appears to be a factor in the extent of delamination in high-stress designs. Interestingly, however, even the most congruent designs with the thickest PE bearings sometimes fail by fatigue of the PE after years in service. The PE of these components appeared yellowed and brittle as well as cracked and delaminated, leading us to the conclusion that the white, highly ductile PE from which the components were fabricated changed with time through some mechanism.

  Laboratory testing of total knee components indicates that many designs produce contact stresses that exceed the yield strength of ultra-high molecular weight PEs (UHMWPEs). Stress on the bearing material in the knee is more complex than in the hip, because the kinematics of the knee articulating surfaces include both rolling and sliding, while the hip only slides. The rolling/sliding motion of the knee results in cyclical tension and compression at the articulating PE surface and shear below the surface. Typically, the intensity of these stresses is directly influenced by the contact stress at the surface of the bearing, which is in large part a function of implant design and surface conformity, the outcome of which is directly related to the quality of the PE.

  Most knee designs are of the fixed-bearing configuration in which the PE-articulating surface is firmly affixed to bone. Unfortunately, in a fixed-bearing knee, low contact stress resulting from increased conformity, limits flexibility in terms of alignment (specifically rotation), mating of components of different sizes, and implant positioning. This typically leads to a compromise in which the conformity, and therefore the contact area, is reduced to avoid overconstraining the joint, which might lead to loosening. Just how much contact area to sacrifice and how best to achieve that compromise remains a matter of contention, as evidenced by the myriad of different designs available to the surgeon.

  Although highly conforming designs produce lower contact stresses through increased contact area, they also may produce excessive stresses at the bone-prosthesis interface, running the risk of long-term loosening. Fully conforming devices match articulating geometries in both the frontal and sagittal planes, producing area contact, while moderately conforming geometries typically match profiles only in one plane, producing line or point contact. The frontal plane profiles in moderately conforming tibiofemoral designs are either curve-on-curve (CoC) or flat-on-flat (FoF). Moderately conforming patellofemoral designs are typically dome-shaped, producing line contact in extension and point contact in flexion. In previous works, testing of virgin components demonstrated that fully conforming tibiofemoral and patellofemoral components produced dramatically lower contact stresses than their moderately conforming counterparts. In the tibiofemoral designs, no difference was seen between the FoF or CoC designs in neutral alignment, but the latter were more forgiving of varus-valgus and rotational malalignment than the FoF geometries. In the patellofemoral designs, anatomic fixed bearings proved to be much more sensitive to rotational malalignment than the dome-shaped configurations.

  Clinical correlations have shown that fully conforming tibiofemoral and patellofemoral designs result in a lower incidence of pitting, delamination, and cracking than less conforming devices. Examination of retrieved, fully conforming tibiofemoral designs with mobile bearings showed burnishing of the metal articulating surface, indicating that the PE bearings are, in fact, moving with respect to the tibial baseplate. It is this motion that permits the fully conforming tibial component to avoid transmitting high shear forces to the bone-prosthesis interface. Additionally, these designs are very forgiving of rotational malalignment. The absence of significant scratching or burnishing on the metal articular surfaces of most of the mobile patellar bearings, however, suggests that these devices may not actually be rotating in service. The principal advantage of the mobile bearing in the patella may instead be that it is extremely forgiving of variations in surgical alignment and, once the components are engaged, these devices may simply rotate to a preferential alignment and stay there.

  In tests of 33 total knee components of varying manufacturers and designs, we found that several designs produced contact stresses above the uniaxial yield strength of UHMWPE (21 MPa). Given these high contact stresses and the data relating high contact stresses to component failure, it is worth asking why more components do not fail in vivo. There are probably a number of reasons for this. Contact stresses that exceed the uniaxial yield strength of PE do not necessarily imply that the shear stresses, which are responsible for material failure, are above the yield strength in shear. But even if the shear stresses are above yield, yield in PE is characterized by plastic deformation and not by brittle failure. Components tested in the laboratory are often loaded to points well above the yield strength without any readily apparent catastrophic effects: obviously, they can exceed the yield stress some finite number of times before noticeable damage begins to appear. Because of individual variations in activity and gait alteration, it is possible that many components do not experience large, cyclic loads frequently enough to accumulate damage. Total knee recipients may place less stringent demands on their prosthetic knees by favoring them or by using aids such as canes or walkers. They also may limit or aid the deep flexion activities, such as rising from a chair or walking up stairs, that produce higher stresses. In reality, most prostheses probably experience loads well below three times the body weight for most of their service. It also has been shown that at higher flexion angles the patellar tendon engages the femur and carries almost half of the resultant patellar load, thus reducing the severity of the load on the articular patellar surface.

  Variations in the material properties of the PE as a result of the extent of consolidation also may dramatically influence the susceptibility of the bearing to fatigue failure. It may be that the variations in failure rates for designs with similar contact stresses are due to variations in PE quality. A combination of a low contact stress design and high-quality material may be the ultimate prescription for long-term survival of PE bearing surfaces in the knee.

Polyethylene Quality
  Improved PE in the form of UHMWPE has been the material of choice for the past two decades. Its high abrasion resistance and impact strength, along with its toughness, ease of fabrication, and nontoxicity seemed to ensure that bearing surfaces fabricated from this material would serve well in hip and knee prostheses. Unfortunately, there is little documentation in the orthopaedic literature regarding the uniformity, material properties, or quality of the PE used in these applications. Recently, the increased frequency of fatigue-related wear damage, the failure of PE bearing components, and the problems associated with the resulting wear debris, have shifted the focus of current research to improving the wear characteristics of the PE components.

  As part of our analysis of retrieved acetabular components, a combination of reflected and transmitted light was used to examine the inner surface of the PE. In doing so, it became obvious that many specimens contained internal flaws in the shape of spherical voids that can be seen at high magnification as isolated particles that did not fully fuse to the surrounding material. Because PE is more vulnerable to oxidation when processed at temperatures above 150 °C, manufacturers keep the processing temperature below this level. It seems, however, that the temperature being used is often too low to completely melt all the particles. In one study, 79 percent of the PE inserts examined showed voids, 45 percent in high concentrations. These occurred in almost all types of acetabular components and were often associated with the presence of cracking. With further inspection, the voids were found to be relatively uniform and were not a surface phenomenon.

  Examination of more than 300 components for wear and the extent of the fusion defects, however, revealed little correlation between the two, possibly because a second phenomenon, a white band that appeared to follow the contours of the most highly damaged components, was a more dominant variable. Laboratory testing revealed differences in the static properties of samples with and without defects with large samples indicating a reduction of strength and ductility of up to 20 percent in some samples with defects. It is likely that the defects play a greater role in fatigue situations that have not been tested in the laboratory to any significant extent.

  The white band, a zone of material more opaque than the surrounding PE, became visible due to the light scattering from the microcracking generated during the sectioning process: it follows the contour of the component at a depth of approximately 1 mm. The location of the band varied in retrievals, appearing most frequently in the flange region and on the articulating surface, and least frequently on the surface directly apposed to a metal backing.

  Two never-implanted, 14-year-old components, still sealed in orignal packaging, were examined. They were from the same manufacturing lot and differed only in that one was gamma sterilized in air (GSA) and the other was never sterilized. Only the GSA component had the band. This pair of components confirmed the effect of GSA seen in retrievals.

  To further investigate this phenomenon, we selected 150 PE acetabular components, sterilized by various methods including GSA and ethylene oxide (EO), from a retrieval base of >600 to represent a range of manufacturers and in vivo durations. The heavily damaged retrievals with rim-cracking and delamination were GSA; none of the EO-sterilized components showed either rim-cracking or delamination. Correspondingly, none of the EO-sterilized components showed the subsurface white band, while 68 percent of the GSA components did. Upon further investigation and discussion with the manufacturers, components exhibiting the band were confirmed to be GSA and most were over 3 years post-sterilization. Of the retrieved components showing delamination, 95 percent had the band, and it was present in 100 percent of the components showing cracking-type damage. Statistical contingency tests confirmed that both cracking and delamination were highly dependent on the presence of the subsurface white band. Other modes of wear did not significantly correlate with its presence.

  Measurement by infrared spectroscopy of oxidation (CO area/mil thickness) versus depth into components revealed that the GSA components with the subsurface white band have substantially elevated oxidation levels. Previous research supports the idea of GSA-induced oxidation: free radicals formed by irradiation persist for years, reacting with oxygen over time. Oxidation has a great effect on the performance of the material. Mechanical testing of white-banded components reveals a marked decrease in elongation and ultimate tensile strength (UTS). Stresses applied by normal service loading and the existence of the interface between ductile and brittle material may eventually be a source of delamination of hips. Although delamination is currently not a predominant mode of failure in hips, this subsurface loss of elongation would be critical in knee components where maximum stresses occur subsurface.

  As with hips, analysis of 182 retrieved, and 40 never-implanted tibial components revealed that 63 percent of the retrievals and 90 percent of the never-implanted components exhibited the subsurface white band. Of these, all had been GSA and were at least 3 years post-sterilization. Statistical contingency tests again confirmed that component cracking and delamination were highly dependent on the presence of this subsurface opaque band. The band was present in 89 percent of the components showing signs of cracking and in 91 percent of the components with delamination.

  Mechanical testing revealed the biggest changes for GSA components. UTS and percent elongation were significantly less than the ASTM specifications (27 MPa and 200 percent, respectively) in the white band region of these components. Percent elongation was nearly zero in the opaque zone, indicating severe PE embrittlement. Tensile testing of the microtomed vertical cross sections confirm a difference in properties as a function of depth. Failure of the subsurface white band region before failure of either the clear outer layer or the center zone indicates reduced ductility of the subsurface zone relative to the others. Sections taken without including white-band material stretched much further before failure.

  The band was found in sections cut with a microtome but not in those cut with a diamond-impregnated abrasive disk, indicating the importance of sample preparation in such testing. Vertical sections prepared with the diamond-disk did not have any cracks in the subsurface region, but since they also failed first in the this region, they confirm a difference in ductility as a function of depth without the confounding effects of microtome-induced microcracks.

  Earlier studies have reported high subsurface shear stresses at the same location as the white band. Our cyclic testing confirms what clinical studies indicate: fatigue occurs preferentially in the subsurface region of PE tibial components. White-banded, GSA components failed by fatigue after only a few hundred cycles, compared to 10,000 cycles for band-free components.

  From these investigations, we determined that GSA acts as both an initiator and an accelerator of PE oxidation and that there is a need for improvement in the GSA process used by the orthopaedic implant industry. Sterilizing in an oxygen-free atmosphere has potential for reducing the immediate effects of this oxidation. Also, low-dose radiation appears to cause less oxidation than higher dosages, probably by breaking fewer PE chains, thereby reducing the number of oxidation sites. Furthermore, non-gamma radiation sterilization techniques such as EO and Gas Plasma appear to have little effect on oxidation.

  In addition to providing increased longevity of all prostheses using PE as a bearing surface, implementation of these modifications to sterilization procedures may actually eliminate the fatigue failure of PE tibial components, now the most common mechanism for the breakdown of total knee prostheses.

CONCLUSION--Understanding of the complexities of successful joint replacement has improved substantially since 1988. While fixation issues are still a major concern, many more potential problems have been identified. Debris generation from PE wear is considered the biggest one today; other problems result from attempts to customize implants for a more individualized fit. Modular implants, while allowing an immediate custom fit, introduce the possibility of component corrosion, fretting, screws loosening, and so forth, and screw holes can provide pathways for the migration of PE and other types of debris. Identification of such potential problems with various designs has been an important function of retrieval analysis.

  Our retrieval rate has increased by well over a factor of two during the decade of VA funding. We have published more than 10 papers and several book chapters, and have presented results of the Center's research in more than 50 papers and presentations at the American Academy of Orthopaedic Surgeons and other pertinent meetings. In addition, our implant retrieval lab is visited on a regular basis by orthopedic surgeons and the heads of research and development of orthopaedic implant manufacturers interested in the performance of these devices.

[203] THE EFFECTS OF TRANSVERSE, ANTERIOR COLUMN, AND POSTERIOR COLUMN ACETABULAR FRACTURES ON THE STABILITY OF THE HIP JOINT

Kevin A. Thomas, PhD; Mark S. Vrahas, MD; Kirstin K. Widding, MS
Bioengineering Laboratory, Department of Orthopaedic Surgery, Louisiana State University Medical Center, New Orleans, LA 70112; email: kthoma@lsumc.edu

Sponsor: Orthopaedic Trauma Association, Rosemont IL, 60018

PURPOSE--Post-traumatic arthrosis may result from unreduced acetabular fractures. This is especially true if the fracture crosses a portion of the acetabulum necessary for weightbearing. However, the exact portion of the acetabulum necessary for normal weightbearing is not known. At the very least, normal weightbearing requires a stable hip joint. Thus, an acetabular fracture that affects hip joint stability can be assumed to involve an important weightbearing area. Such a fracture, if left unreduced, most likely would lead to arthrosis. The purpose of this study was to assess how various transverse, anterior column, and posterior column acetabular fractures affect hip joint stability.

METHODOLOGY--A total of 24 fresh-frozen cadaver hip joints were used in this investigation. Each was mounted in a mechanical testing machine so that compressive forces of various magnitudes could be applied. The femur was mounted vertically on a freely moving instrumented x-y displacement table. Stability of the hip joint depended upon only the intrinsic stability of the articulation between the femur and the acetabulum. The acetabulum was mounted in a fixture that allowed simulation of the various fractures and allowed positioning of the joint so that the magnitude and direction of the peak forces seen during normal gait could be applied. For each of the fracture types, the specimens were loaded in compression four times at each of three loads (800, 1,200, and 1,600 N). Each specimen was first tested intact, and again following each of the various simulated fractures. Simultaneous recordings were made of the load, actuator displacement, and translation of the femur. A hip was considered stable if no dislocations occurred over the four loading cycles.

  Transverse acetabular fractures (OTA classification 62-B1) were evaluated in 12 specimens. Each specimen was tested with successive transverse fractures having roof-arc angles of 60, 50, 40, 30, and 20°. The roof-arc angles were determined radiographically. Anterior column fractures (OTA 62-A3) were evaluated in six specimens. For each specimen, very low, low, intermediate, and high fractures were tested successively. Posterior column fractures (OTA 62-A2) also were evaluated in six specimens; only very low posterior column fractures were tested.

RESULTS--All 24 hips were stable when loaded intact to 800, 1,200, and 1,600 N. For the transverse fractures, for each of the applied loads, the number of stable specimens decreased with each successive fracture. For these fractures, stability was significantly affected by both the location of the fracture (roof-arc angle) and the magnitude of the applied load. For the anterior column fractures, all specimens with very low, low, and intermediate fractures remained stable throughout all tests. With high anterior column fractures, 33 percent were unstable when loaded to 800 N, 67 percent were unstable when loaded to 1,200 N, and all were unstable when loaded to 1,600 N. Overall, the posterior column fractures were the least stable. All six specimens were unstable with the simulated very low fracture (exiting the posterior column just superior to the ischial spine).

  Anterior column fractures were more stable than expected. No specimens were unstable unless the fracture was of the very high type. Fractures at the base of the pubic rami did not affect stability. Only fractures that exited through the iliac crest caused instability. This suggests that fractures at the base of the pubic ramus may be less important then previously thought. Posterior column fractures were much less stable than expected. Even low fractures, that is, fractures at the level of the ischial spine, rendered the hip unstable in all six specimens, even at the lowest applied load.

  The results of this study have direct implications in the treatment of anterior column and T-type fractures. Based on these results, currently only anterior column fractures that exit through the iliac crest are treated surgically. For T-type fractures, the posterior column is the greater concern, and requires reduction and fixation even if the fracture is very low. The anterior column portion of the fracture is of concern only if it very high.

RECENT PUBLICATIONS FROM THIS RESEARCH

C. Knee Implants

[204] MODIFICATION OF POLYETHYLENE TO IMPROVE THE WEAR PERFORMANCE OF JOINT REPLACEMENT PROSTHESES

Myron Spector, PhD; Robert E. Cohen, PhD; Anuj Bellare, PhD; Yot Boontongkong, MS; Natasha Chang, BS
Brockton/West Roxbury VA Medical Center, West Roxbury MA 02132; Brigham & Women's Hospital, Boston MA 02115; Massachusetts Institute of Technology, Cambridge MA 02139; email: spector@ortho.bwh.harvard.edu

Sponsor: Department of Veterans Affairs, VA Rehabilitation Research and Development Service, Washington, DC 20420
(Project #A877-RA)

PURPOSE--This study is an investigation of the effects of molecular orientation on the wear behavior of ultra-high molecular polyethylene (UHMWPE) used in total joint replacement prostheses. The long-term goals are to determine processing conditions for UHMWPE that optimize its performance in total joint replacement prostheses.

METHODOLOGY--Ram-extruded rod stock of medical grade UHMWPE (GUR 4150) was obtained from commercial sources. Slabs were machined from the stock, heated to 170 °C in a vacuum oven and slow cooled to room temperature. Each specimen was characterized using wide angle X-ray diffraction (WAXD), small angle X-ray scattering (SAXS) and differential scanning calorimetric (DSC) analyses.

  The slabs were subjected to plane strain compression, using a channel-die compression apparatus. Selected samples were chosen for wear tests on the three orthogonal surfaces of the rectangular bars produced by channel-die compression. We define these surfaces as loading or LD surface (surface normal along loading or LD direction), constrained or CD surface (surface normal perpendicular to the walls of the channel-die) and flow or FD surface (surface normal along the flow direction).

PROGRESS--We have essentially completed all the experiments proposed in this study. Approximately 35-40 unoriented slabs were prepared for the morphological and wear studies. Each slab was snugly fitted into a channel-die preheated to 90 °C and compressed using an Instron tester to varying compression ratios (CR). We define CR as the ratio of initial height of the slab to the final height of the compressed (or oriented) slab. Each bar was prepared with a different initial height so that the dimensions after compression were identical. Samples with CR of 1.4, 1.95, 3.2, 4.0, and 5.6 were subjected to SAXS, WAXD, and DSC analyses to study evolution of crystallographic orientation.

  Rectangular bars with selected CR of 2.7 and 1.7 were wear tested using a 4 station reciprocating (sliding) wear tester. Wear tests were performed on the LD, CD, and FD surfaces. Each test was performed by sliding cylindrical cobalt-chromium alloy pins on flat UHMWPE bars with an applied load of 90.7 kg. Bovine serum was used as the lubricant.

RESULTS--There was a 5 percent reduction in the percentage crystallinity (or density) in specimens with a CR above 2.0. SAXS and WAXD characterization showed a substantial degree of molecular orientation above CR of 1.4. The degree of orientation increased monotonically until CR of 6.0, after which brittle fracture occurred.

  Wear tests of samples with a CR of 2.7 and 1.7 were compared to standard, uncompressed UHMWPE (CR=1.0). At a sliding distance of 14,813 m, the wear volume of standard UHMWPE normalized by applied load was (0.97±0.31)E-7 mm3/N. The volume of wear particles on LD, CD, and FD surfaces under identical conditions of wear was in the range of (1.27-0.79)E-7 mm3/N. This reveals that molecular orientation does not substantially affect wear rates in UHMWPE.

FUTURE PLANS/IMPLICATIONS--Our results suggest that wear of UHMWPE is governed by morphology at the micrometer length scale rather than the nanometer level morphological parameters, such as degree of molecular orientation. We plan to use a combination of processing conditions and chemistry to improve consolidation of the micrometer-sized resin particles to improve wear performance of UHMWPE prosthetic components.

RECENT PUBLICATIONS FROM THIS RESEARCH

[205] EFFECT OF COMPONENT PLACEMENT ON THE PATELLOFEMORAL JOINT KINETICS WITH TOTAL KNEE ARTHROPLASTY

Harry E. Rubash, MD; Mark C. Miller, PhD
University of Pittsburgh Medical Center; Pittsburgh PA 15260; University of Pittsburgh, Pittsburgh, PA 15261; email: mcmllr@sprite.me.pitt.edu

Sponsor: Department of Veterans Affairs, VA Rehabilitation Research and Development Service, Washington, DC 20420
(Project #A859-RA)

PURPOSE--This research is quantifying the effects of component alignment in total knee arthroplasty (TKA) on patellofemoral joint kinetics and general knee kinematics. The goal is to increase long-term survival of knee replacements by determining the required accuracy in the placement of these components and by evaluating joint replacement component selection.

METHODOLOGY--Cadaveric knee specimens are mounted in a knee simulator after careful specimen preparation. Knee arthroplasty is performed with components from Zimmer, Inc., using standard sizing techniques and recommended instrumentation.

  A specially designed device anchors the femur in the knee simulator and sets the amount of internal-external rotation of the femoral component. Using the wormgear drive mechanism in the device, internal-external rotation of the component can be set to any desired orientation with an accuracy of 0.25 degrees without re-opening the joint capsule.

  After recording the native thickness, the patella is cut in the recommended way, removing the articular surface. A load cell, mounted on the anterior patellar face, supports a plate on which the patellar prosthesis is mounted. This support plate has holes for fixation of the patellar prosthesis at desired lateral-medial positions and also adjusts to restore the native patellar thickness. The joint capsule is closed and a nylon strap is sutured to the central quadriceps tendon. The knee is placed in the simulator with the nylon strap attached to a hydraulic actuator.

  All patellofemoral kinetics and tibiofemoral kinematics are recorded. The load cell measures all six components of load on the patella and two instrumented spatial linkages track all six degrees of freedom of the movement of the patella and tibia with respect to the femur. The actuator of the knee simulator drives the knee through a flexion-extension cycle of 105°. The femoral and patellar components are repositioned after each test and a lateral retincacular release is performed after all alignments have been studied.

PROGRESS--Ten specimens have been tested with all combinations of femoral external rotation of 0, 2.5, and 5° and patellar button placement at the geometric center and at 3.75 mm medial to the geometric center for a posterior cruciate sparing TKA. To compare the effects of TKA system selection and component placement, an ongoing study has tested four with both a system that retains, and a system that sacrifices, the posterior cruciate ligament.

RESULTS--The combination of 2.5° of femoral component external rotation with patellar button medialization reduced the lateral-medial shear force on the patella. This combination would therefore reduce any tendency of patellar subluxation. All combinations of small changes in femoral and patellar component alignment provided generally very good patellar tracking, indicating that implantation could vary and still maintain good short-term surgical outcome.

  Tibial kinematics were altered by variation in femoral component alignment, but not by patellar button placement. Femoral component rotation affected tibial rotation at small flexion angles and tibial varus-valgus rotation at large flexion angles. At larger flexion angles, the tibio-femoral contact becomes more posterior on the femoral condyles. This increasingly posterior contact causes external femoral component rotation to effect increasing tibial varus orientation. No short-term problems in functionality occurred, but potential long-term implications may exist.

  A lateral retinacular release had essentially no effect on kinetics or kinematics. All specimens tracked well before and after the procedure.

IMPLICATIONS--A combination of small amount of external femoral component rotation and patellar button medialization provides good patellofemoral kinematics and load transfer for this TKA design. Tibiofemoral kinematics are clearly affected by femoral component rotation.

FUTURE WORK--Testing with posterior cruciate retaining and sacrificing components will be continued and will include variation in tibial component alignment.

RECENT PUBLICATIONS FROM THIS RESEARCH

[206] EFFECT OF INTRA-ARTICULAR IMPLANTS ON THE HEALTH OF THE KNEE JOINT

M. Stubbs, MD; Mark S. Vrahas, MD; Richard V. Baratta, PhD; H. Zhang, MD
Bioengineering Laboratory, Department of Orthopaedic Surgery, Louisiana State University Medical Center, New Orleans, LA 70112

Sponsor: Smith & Nephew Richards

PURPOSE--Retrograde femoral nails inserted are gaining increasing popularity. After implantation, the device is exposed to the joint environment. The purpose of this study is to established the long-term effects of these implants on the health of the joint's structures.

METHODOLOGY--Specially modified stainless steel screws were inserted through the intercondylar notch into the femora of 18 New Zealand White rabbits. The contralateral limb underwent a sham insertion, with every procedure duplicated, including reaming and threading, with the exception of the insertion of the device. Knees are harvested in groups 6 wks, 3 mo, 6 mo, and 1 yr postimplant and examined histologically for changes in the cartilage, sibchondral bone, and synovium.

PROGRESS--To date, all implants and harvestings have been performed; gross examination of the 6 wk, 3 mo, and 6 mo knees shows no evidence of joint arthropathy. Currently, we are processing and evaluating histological specimens to document whether there are differences between the implanted and sham-operated knees.

D. Arthritis

[207] IMPACT-INDUCED POST-TRAUMATIC ARTHRITIS MODEL

Mark S. Vrahas, MD; Richard V. Baratta, PhD; H. Zhang, MD
Bioengineering Laboratory, Department of Orthopaedic Surgery, Louisiana State University Medical Center, New Orleans, LA 70112

Sponsor: AO Foundation

PURPOSE--Post-traumatic arthritis is a rapid joint degeneration which occurs as a result of a traumatic event. The purpose of our research is to develop a model to predict the impact stress magnitudes that result in post-traumatic arthritis.

METHODOLOGY--An impact tower is used to guide a falling mass that delivers a blow of specific energy to an impactor instrumented with strain gages to record force. Bone cement is used to create a custom impactor to apply uniform stress over the impacted area. The area of impact is measured by making an imprint onto a sheet of paper with a wax marker. Accurate estimation of impact force and area lead to an estimate of impact stress, believed to be the most important factor in predicting the occurrence of post-traumatic changes.

PROGRESS--To date, an in vitro rabbit model was used to develop the impact tower instrumented to record impact force, along with the technique to fabricate custom impactors which provide uniform stress to a given area of the rabbit's femoral condyle. A pilot in vivo study was performed to show that histochemical changes associated with the early development of arthritis could be found in rabbits 6 wks following an impact. Currently, we are evaluating histological specimens from 36 rabbits whose knees have been impacted and who have been kept up to 1 yr post-impact with aim to document whether there are long-term changes in the impacted femoral cartilage.

RECENT PUBLICATIONS FROM THIS RESEARCH

E. Low Back Pain

[208] EFFECT OF SURGICAL PROCEDURES ON THE STABILITY OF THE LUMBAR MOTION SEGMENT

Gunnar B.J. Andersson, MD; Avinash G. Patwardhan, PhD; Raghunatha N. Natarajan PhD; Steven A. Lavender, PhD
Rush Presbyterian-St. Luke's Medical Center, Chicago, IL; Loyola University Chicago, Maywood, IL 60153; Hines VA Hospital, Hines, IL 60141

Sponsor: Department of Veterans Affairs, VA Rehabilitation Research and Development Service, Washington, DC 20420
(Project #A830-RA)

PURPOSE--Many surgical procedures of the lumbar spine, such as laminectomy and discectomy, require removal of tissue from the motion segment. When large amounts of tissue are removed, the stiffness of the motion segment is reduced to a degree that there is an abnormal response to an applied load. To compensate for this decrease in stiffness, fusion surgery is sometimes performed. At the present time, there is a clinical controversy about when a fusion should or should not be performed. The purpose of this research is to quantify the segmental instability resulting from the commonly used surgical techniques of facet and disc resection through a combination of experimental and analytical techniques.

METHODOLOGY--We performed experiments on human cadaveric lumbar spine specimens to determine the effects of surgical procedures on the 3-D load-displacement behavior of lumbar motion segments. The experiments simulated different combinations of surgical procedures, including unilateral and bilateral facet removal and disc denucleation. Next, we validated a finite element (FE) model of a lumbar segment by modeling the experimental simulations of these procedures. Finally, the validated FE model was used to conduct a detailed parametric study of the effects of the surgical procedures on the change in stiffness of the lumbar motion segments.

PROGRESS--We have completed a major portion of the tests on lumbar specimens to determine the effects of different surgical procedures on the load-displacement behavior. The geometric shape of a vertebra-disc-vertebra segment was generated from a serial CT scan. The nonlinear FE analyses have been conducted. The model was first used to analyze the effect of graded unilateral facetectomies (25, 50, 75, and 100 percent) on the stiffness of the spine. In addition, at each level of graded facetectomy, nucleotomy was also performed to understand the combined effect of graded facetectomy and nucleotomy.

RESULTS--A large increase in flexibility was observed for all loading modes when as little as 25 percent of the facet was resected and a nucleotomy performed. If the surgical intervention was restricted to facet joints only, this sudden increase in flexibility was observed after 75 percent of unilateral facetectomy. The breakpoint at which there is a sudden increase in flexibility depended on the loading mode. Torsional and flexion moments produced a sudden increase in flexibility after 75 percent of a facet was removed, while the rest of the physiological loads showed a breakpoint after complete unilateral facetectomy. When nucleotomy was combined with graded facetectomy, all loading modes except torsion increased in flexibility. Clinically, these results show that a unilateral 75 percent facetectomy alone can be done without much effect on the stiffness of the spine. If facetectomy is combined with nucleotomy even removal of 25 percent of facet is enough to cause an increase in motion of the segment.

FUTURE PLANS/IMPLICATIONS--Additional experiments will be run as needed to improve the statistical significance of the experimental study. The information generated in this study will be used as a basis for developing recommendations concerning when to fuse and when not to fuse.

RECENT PUBLICATIONS FROM THIS RESEARCH

[209] A MINIMAL EFFORT TEST FOR PATIENTS IN PAIN USING THE BACK ANALYSIS SYSTEM

Lars I.E. Oddsson, DrMedSc; Carlo J. De Luca, PhD; Jonathan Bean, MD
NeuroMuscular Research Center, Boston University, Boston, MA 02215; Physical Medicine and Rehabilitation Service, Boston VA Medical Center, Boston, MA 02130; email: loddsson@bu.edu

Sponsor: Department of Veterans Affairs, VA Rehabilitation Research and Development Service, Washington, DC 20420
(Project #B872-2RA)

PURPOSE--Previous analysis of surface EMG signals of the lumbar back muscles assessed with the Back Analysis System (BAS) has been based on the concept of fatiguing the muscles by investigating the behavior of the initial median frequency and the rate of decrease of the median frequency during fatiguing contractions. The use of such procedures may not be well suited for subjects in acute phases of injury due to obvious hesitation and fear of re-injury, making the assessment of a reliable endurance effort unreliable. In this project, new force insensitive parameters are being investigated that instead reflect aspects of load sharing by describing spectral interactions and imbalances between pairs of muscles of the lumbar back.

  The long-term objectives of this proposal are to: provide clinicians and researchers with a procedure to objectively quantify muscle impairment associated with low back pain in previously non-testable subjects with acute pain so that specific early interventions can be provided thus minimizing the risk that subjects reach a chronic injury state. Also, we seek to provide clinicians with objective information regarding the transition of an injury from an acute to a subacute phase thus assisting in decisions concerning treatment for the individual patient.

METHODOLOGY--A total of 80 controls and 70 LBP patients in an acute phase of injury will be tested. Surface EMG activity will be recorded from six muscle sites of the lumbar back while the subject performs brief isometric incrementing stair case contractions up to 80 percent of ideal body weight in the BAS. The behavior of ratios between spectral properties in contralateral as well as ipsilateral pairs of muscles of the lower back are being investigated. New parameters with clinical relevance describing segmental imbalances and bilateral compensations in spectral parameters have been introduced. A series of questionnaires will be administered to subjects to assess fear-avoidance strategies in different phases of the injury.

PROGRESS--A clinical version of the BAS system has been completed and installed with new software for data collection and database storage. Software development for data analysis to study the behavior of load-sharing parameters on the BAS is near completion. Recruitment and screening of LBP subjects is in progress; 15 controls subjects have been tested.

PRELIMINARY RESULTS--Results thus far have suggested that imbalances are present in both persons with LBP and controls. However, imbalances appear to be compensated across lumbar levels for the latter but not for the former.

FUTURE PLANS--Scheduling and testing of subjects is currently in progress. Testing of controls continues. Development of software for extraction of EMG-based parameters is progressing and should be finished by the end of 1997.

RECENT PUBLICATIONS FROM THIS RESEARCH

  • Development of new protocols and analysis procedures for the assessment of LBP by surface EMG techniques. Oddsson LIE, Giphart JE, Buijs RJC, Roy SH, Taylor HP, De Luca CN. J Rehabil Res Dev 1997;34(4):415-26.

    [210] BACK EXERCISE PRESCRIPTION AND IMPLEMENTATION BY SURFACE ELECTROMYOGRAPHIC PROCEDURES

    Serge H. Roy, ScD, PT; Carlo J. De Luca, PhD; Jonathan Bean, MD
    NeuroMuscular Research Center, Boston University, Boston, MA 02215; Physical Medicine and Rehabilitation Service, Boston VA Medical Center, Boston, MA 02130; email: sroy@bu.edu

    Sponsor: Department of Veterans Affairs, VA Rehabilitation Research and Development Service, Washington, DC 20420
    (Project #B2029-RA)

    PURPOSE--The purpose of this project is to develop a low back pain (LBP) exercise procedure for restoring muscle impairments classified by the Back Analysis System (BAS). This study will considerably advance the BAS technique to include the prescription and implementation of muscle-specific, individualized exercises to improve strength and endurance.

    METHODOLOGY--Phase I of the study will focus on modifying the BAS to implement isometric exercises for reconditioning impaired back extensor muscles. Isometric and isotonic exercises will be developed outside of the BAS to selectively recondition impaired muscle groups. The use of a portable 2-channel EMG unit will be evaluated to enhance muscle activation during the performance of the exercises outside of the BAS. Phase II of the study will evaluate the effectiveness of the exercise protocol by comparing treatment outcomes in patients with LBP randomly assigned to either an experimental exercise group, a generalized strengthening exercise group, or a control group without supervised exercise.

    PROGRESS--We have completed the modifications of the BAS to implement isometric exercises via EMG and force feedback and are currently evaluating its capabilities among test subjects. We have also defined an exercise protocol of trunk exercise strengthening and control based on paraspinal surface EMG measurements to verify patterns of muscle activation and spinal stabilization. These exercises will be implemented among the LBP clinical population utilizing portable EMG biofeedback units. A BAS system has been set up at our clinical site and we have commenced the randomized longitudinal component of the study to evaluate the efficacy of the experimental exercise program vs control exercise.

    PRELIMINARY RESULTS--We are currently working on all aspects of Phase I. We have made considerable modifications to the BAS software in preparation for real-time visual feedback of EMG signals and trunk extension forces during the performance of isometric exercises. Tests have been conducted upon control subjects to determine the extent to which the technique can isolate functional regions of paraspinal muscles that correspond to the to the EMG electrode locations. Subjects have been tested during the execution of a variety of floor exercises commonly prescribed for progressive lumbar stabilization. Subjects are instrumented with surface EMG electrodes during the performance of these exercises to record the activation patterns among eight muscles of the trunk.

    FUTURE PLANS--We are continuing the collection of data at the clinical test site and will evaluate the results upon completion of EMG and outcome data.

    RECENT PUBLICATIONS FROM THIS RESEARCH

    [211] THE QUANTIFICATION AND INTERPRETATION OF BACK MOTION AS AN EVALUATIVE TOOL IN LOW BACK DISORDERS

    Sheldon R. Simon, MD; William S. Marras, PhD; Mohamad Parnianpour, PhD; S. Bose; S.A. Ferguson; K. Keefer Lewis
    The Division of Orthopaedics, and Industrial, Systems and Welding Engineering, The Ohio State University, Columbus, OH 43210; email: Simon.1@osu.edu; wmarras@magnus.acs.ohio-state.edu; parnianpour.1@osu.edu; ferguson.4@osu.edu

    Sponsor: National Institute for Disability and Rehabilitation Research, Rehabilitation Engineering Research, Washington, DC 22202

    PURPOSE--Low back disorders (LBD) are extremely common and limit the ability of millions of Americans to lead productive lives and engage in meaningful work. Impairment ratings of LBD can vary by as much as 70 percent using current systems. Diagnoses and classification schemes are rarely based upon quantitative indicators, and we are unable to easily assess and diagnose LBDs. It is important to evaluate LBDs quantitatively, so that proper treatment can be administered and to minimize the risk of exacerbating the problem.

    METHODOLOGY--Studies were performed that quantified 3-D trunk motion characteristics as a function of asymmetry for controls and persons with LBD to classify those with LBD compared to those without; to test alternative testing protocols; to develop a sincerity-of-effort test; and to benchmark longitudinal recovery.

      Four studies have been performed or are in the process of being performed. In the Task Asymmetry Protocol, subjects bend forward and back upright while controlling their twisting position at five asymmetries including: 0, 15° right and left, and 30° right and left. A new Cardinal Plane Protocol was developed where subjects performed three separate tasks: bending forward and back, bending side to side, and twisting right to left. They were instructed to move as fast as they felt comfortable in a comfortable range of motion. The Sincerity-of-Effort aspect of this project has been developed in the past year. The theory is that unless the test can distinguish sincere effort from the subject, the classification of control and subject may be erroneous. Subjects are required to perform each task twice, once sincerely and once pretending their symptoms are worse or, in the case of controls, pretending they have back pain. The Preliminary Longitudinal Phase evaluates subjects every 2 wks during the first 3 mo of injury, regardless of treatments and/or visits for a total of six visits.

    PROGRESS--The Task Asymmetry Protocol has a total of 717 subjects (350 controls and 367 LBD). A model was developed using a training data set to distinguish between the two groups. The best model for classifying them included: ability to perform each task, twisting range, sagittal range of motion at zero, sagittal extension velocity at zero, sagittal extension acceleration at zero and lateral range of motion at zero. The results of the discriminant function analysis show a 9 percent test set error.

      The Cardinal Plane Protocol elicits dynamic tasks in the transverse and coronal planes of the body, which were not in the asymmetry protocol. The theory behind this protocol is that biomechanically, a LBD may affect one plane of motion more than another, depending on the nature of the injury. Three separate models have been developed using discriminant function to distinguish the two groups in each task separately.

      The Sincerity-of-Effort thus far has 96 controls with at least 10 males and 10 females in 10-yr age groups from 20-59. All LBD subjects tested have been within these age groups. Two models were developed, one for each gender. The sincere men were correctly classified in 86 percent of the cases and the sincere women were correctly classified in 86 percent.

      In the Preliminary Longitudinal Phase, 21 subjects have begun testing in two diagnostic categories, either herniated discs or acute local low back pain.

    RECENT PUBLICATIONS FROM THIS RESEARCH


    Go to top.

    Previous
    Contents
    Next

    Last revised Fri 04/30/1999