Journal of Orthopedic Research a 4 6 5 6 Raven Press, Ltd., New York 0 1990 Orthopaedic Research Society
Transplantation of the Rabbit Medial Collateral Ligament. 11. Biomechanical Evaluation of Frozen/Thawed Allografts “P. Sabiston, *C. Frank, *tT. Lam, and TN. Shrive *The Joint Injury and Diseases Group, Faculty of Medicine, Department of Surgery, Division of Orthopaedics and ?Faculty of Engineering, Department of Civil Engineering, The University of Calgary, Calgary, Alberta, Canada
Summary: This study was designed to test the mechanical behavior of a frozed thawed bone-medial collateral ligament (MCL)-bone allograft in a rabbit model and to compare allograft behavior with contralateral unoperated ligaments as well as a group of normal bone-ligament-bone preparations prior to freezing. Twenty-five mature rabbits received similarly frozen boneMCGbone allografts and were subsequently allowed to heal without immobilization in groups of five for 3, 6, 12, 24, or 48 weeks after transplantation. A series of biomechanical tests was performed on each allograft, contralateral, and external normal control ligament. Results demonstrated that allografts were significantly tighter than controls at 3 weeks and they remained tight over time. All allografts, however, had inferior structural and material behaviors as compared with controls. Allograft insertional bone and substance deteriorated to about 60% of contralateral strength at 12 and 48 weeks. Bony insertions became the most common site of graft weakness. Both stress at failure and the elastic modulus of allograft substance similarly reached a plateau at those intervals, but at about 30%of contralateral controls. Cyclic and load relaxation properties of allografts, on the other hand, did not deteriorate and were, in fact, indistinguishable from contralateral and normal values at 48 weeks. Comparison with a series of fresh autografts suggests that, with the exception of this viscoelastic recovery, allografts were mechanically inferior to autografts in their healing at the intervals tested. Collectively, these results suggest that allograft MCL healing in this composite model is dynamic but slow, featuring changes in bone and soft tissue over the first year after transplantation. Although “viable” and vascular, allografts have not normalized mechanically and, in fact, appear to have reached a new equilibrium whereby properties are neither improving nor deteriorating. Despite this apparent equilibrium, the trend for some viscoelastic recovery provides hope that allografts may have further potential for improvement. Longer-term studies of MCL allograft mechanical behaviors are clearly required. Key Words: Rabbit-Medial collateral ligament-Allografts-Freezekhaw .
plants have not been common. Instead, most studies have concentrated on normal mechanical baselines of potential transplant tissues in humans (4,6,18,19) or in animal models (2,7,27,31), and the mechanical effects of pretransplant sterilization or preservation (16,17,22,29,30). Only a few studies have tested what happens to the beha’ior of these tissues after transplantation and all of
Despite the long history of biological and clinical interest in dense connective tissue transplantation (23,24), biomechanical assessments of such transAddress correspondence and reprint requests to Dr. C. Frank, Department of Surgery, The University of Calgary, Health Sciences Centre, 3330 Hospital Drive NW, Calgary, Alberta, Canada T2N 4N 1.
46
ALLOGRAFT LIGAMENT TRANSPLANTATION (II)
them have evaluated only the anterior cruciate ligament (ACL). Webster and Werner, for example, replaced canine ACLs with freeze-dried flexor tendons and found that they had 29% of control failure strength at 9 months posttransplantation (3 1). Shino and coworkers found similar values for frozen free patellar tendon allografts at 30 and 52 weeks after grafting (27). Curtis and co-workers (7) found dramatic early weakening of freeze-dried fascia lata allografts with a subsequent trend to get stronger but, like tendon replacements, these allografts reached only a fraction of normal ACL strength. Like both Webster and Shino, they felt that, although allografts were inferior to controls, their similarity in strength to commonly used tendon autografts provided sufficient support for their continued investigation and use. Nikolau and co-workers (18),searching for better results from allografting ligaments, examined frozen bone-ACL-bone allografts in dogs. They found steady increases in strength and stiffness for 36 weeks, reaching values within 10% of contralateral controls. It was noted, however, these controls failed at less than 50% of literature values for canine ACL (1,5,13,20). Jackson and co-workers studied a freeze-dried bone-ACL-bone model in goats (16), and found that allografts had only 25% of normal ACL strength at 1 year. Thorson et al. (29) reported even poorer results in a frozen canine ACL model and suggested the possibility of a secondary immune arthritis. Collectively, therefore, these studies suggest that allograft replacements of the ACL are mechanically inferior to control ACLs for at least 1 year after transplantation due to weakening of either insertions (16,18), midsubstance (29,31), or both (27). It is not clear, however, whether these results may be specific to the highly complex ACL (15,33), specific to the canine or goat models used, or may be due to some other immune (29), mechanical, or treatment variables, such as immobilization (7,16, IS). This study was designed to test a relatively simple, homogeneous, and well-characterized allograft ligament model; the rabbit medial collateral ligament (MCL). Our goal was to examine a number of mechanical characteristics in this model over time, as a measure of the mechanical changes in a ligament allograft in an extra-articular environment. Finally, we wanted to compare these changes with contralatcral, normal, and autograft controls (26), and correlate mechanical with biological changes of
47
both these autografts and allografts, reported elsewhere (25), in order to determine the association, if any, between graft “viability” and graft biomechanics . MATERIALS AND METHODS Animal Model The MCL complexes of both legs of 25 mature (12-18 month old; 4.2 2 0.2 kg) female New Zealand white rabbits were studied. The right leg MCLs were designated as the experimentals in every animal and were harvested from donors as bone-ligament-bone complexes. While major histocompatibility testing was not done prior to transplantation, animals were obtained from four different suppliers and siblings did not act as donors or recipients for one another. Distribution of Animals Five allograft ligament transplant animals were studied at each of five time intervals posttransplantation: 3, 6, 12, 24, and 48 weeks (total = 25 animals). Experimental Procedures Under halothane general anesthesia, all animals had their right knee shaved and prepared with iodine. An anteromedial incision was used to expose the MCL. Muscle and fascia were dissected from around both MCL insertion sites and a small circular saw was used to removed rectangles of insertional bone from both tibia and femur without disrupting ligament substance. The distal femoral and proximal tibia1 cuts were 5 mm from the joint line and were used as landmarks to control graft tension during reinsertion. Sharp dissection was used to separate the MCL from the underlying medial meniscus and the bone-MCL-bone complex was removed (Fig. 1). Corners of graft beds were rounded with a drill to minimize the risk of fractures. Donor MCL complexes were then rinsed in phosphatebuffered saline (PBS), placed in sterile 20 ml culture tubes without any cryoprotectants, and the tubes were then wrapped in two sterile towels. Similar to a freezing and storage protocol documented elsewhere (9), towels were placed in a freezer at - 70°C and frozen at approximately - l”C/min. They were
J Orthop Res, Vol. 8, N o . 1, 1990
P . SABISTON ET AL.
48
FIG. 1. Schematic drawing of allograft medial collateral ligament transplant procedure showing how bone-MCL-bone graft was removed from donor and subsequently reinserted into recipient using cortical screws.
then maintained at - 70°C for 1 week until implantation. Prior to implantation, donor MCL allograft complexes were thawed rapidly in sterile 37°C PBS. Grafts were fixed with 2.0 mm cortical screws, supplemented with a small washer over the femoral insertion, and tension was then adjusted by altering graft position if the joint was found to be abnormally lax or tight on manual valgus testing. In all cases, wounds were closed in layers using interrupted 4-0 nylon suture. A soft dressing was applied and animals were allowed to mobilize with unrestricted cage (65 x 45 x 30 cm) activity. Wounds were inspected on the third postoperative day and all animals were weighed regularly as a general screening index of health. Anteroposterior and lateral radiographs of operated knees were obtained 3 to 6 weeks postoperatively to monitor internal fixation and graft incorporation. Any animal with signs of infection, displaced grafts, or long bone fractures was excluded from further study and replaced by another animal. Controls
Left contralateral MCL complexes of each animal served as an internal control. We cannot refer to these contralaterals as “normal ligaments” de-
J Orthop Res, Vol. 8, N o . I , 1990
spite their being unoperated, due to the possibility of secondary effects from the treatments described. We will therefore refer to them as contralateral controls throughout. A separate series of five unoperated mature animals, sacrificed soon after supply, provided baseline external “time zero” controls of both fresh and frozen normal ligament biomechanical controls. This was done to establish baselines with which to compare the effect of contralateral ligament surgery and to test the effect of freezing on normal ligament properties. Five fresh control MCL complexes (right legs) were tested immediately while five frozen controls (left legs) were stored for 1 week at - 70°C and thawed in 37°C PBS prior to mechanical testing. No difference was found between these fresh and frozen controls for any parameter studied and there was no statistical difference between either of these normal controls and contralaterals. Contralaterals, however, were not static over time and showed some changes in mechanical properties between various healing intervals (see the Results section, Figs. 5-7). Specimen Preparation
All experimental and control MCL specimens were tested fresh. On the day of testing, animals were sacrificed with an overdose of intravenous barbiturate (275 mg/kg). Hindlimbs were disarticulated, the left limb (contralateral control) placed in a sealed container to prevent drying, while the right limb (transplant) was prepared for immediate testing. All soft tissues, including muscles and fascia, were removed from the femur and tibia, excluding the collateral ligaments, the cruciate ligaments, and the menisci. Bones were transected 4-5 cm from the joint line. To facilitate surface strain measurement, parallel dye lines were applied at approximately 5 mm intervals onto the surface of the medial collateral ligament, transverse to the longitudinal axis, with Verhoeff elastin stain. The tibia was then potted in a specially designed clamp using polymethylmethacrylate (PMMA) and this clamp was attached to the crosshead of a materials testing system (Model 1122, Instron Corp., Canton, MA, U.S.A.). The knee flexion angle was adjusted to approximately 60” and the cross-head lowered to place the femoral end of the specimen into a second clamp, where it was fixed with PMMA. The longitudinal axis of the MCL was aligned with the load axis of
ALLOGRAFT LIGAMENT TRANSPLANTATION (II) the testing system using a system of slide plates (Fig. 2). With the knee joint fixed in this position, the MCL was isolated by sequentially cutting and removing both menisci, lateral collateral, and both cruciate ligaments. The knee joint was taken through two cycles of loading and unloading at an extension rate of 1 mmlmin, first compressing the joint by 4 N , and then reversing to a tensile force of 2 N taken up by the ligament. This procedure was designated a compression-tension cycle. The crosshead position where the isolated MCL first registered a tensile load (tO.05 N) was defined as “ligament zero” and was used as the starting position for all subsequent mechanical tests on the ligament. Ligament laxity was determined by measuring the amount of cross-head displacement between this point at which the MCL took up tension and the point where articular surfaces were measurably compressed using maximal load scale sensitivity (compression of 0.05 N). A small portion of medial
49
femoral condyle was then removed, care being taken to avoid damage to the femoral ligament insertion, to allow in situ measurement of midsubstance MCL cross-sectional area with a specially designed cross-sectional area measuring device (28). An environment tank surrounding the clamping system was then flooded with 0.9% phosphatebuffered saline (PBS) at 35°C and pH 7.2. A television camera connected to a video dimension analyzer (VDA Model 303, Instrumentation f o r Physiology and Medicine Inc., San Diego, CA, U.S.A.) (34) was then positioned perpendicular to the longitudinal axis of the ligament to quantify separation of surface dye lines (32). The video output was recorded for later analysis. The analog load signal from the load cell was digitized at 50 Hz with a multiprogrammer (Model HP6944, HewlettPackard, Rockville, MD, U.S.A.) and stored in a hard disk for subsequent analysis (Fig. 3). Mechanical Test Protocol
The test sequence for each MCL complex began with cycling 30 times between “ligament zero” and a fixed deformation of 0.68 mm (approximately 3% average strain) at an extension rate of 10 mmlmin. This led to a duty cycle of about 4 s of loading time and 4 s of unloading time for each cycle. The mechanical state of the complex at the 30th cycle was taken to represent a relative steady state. This steady state was defined mathematically by identical hysteresis areas under two successive loaddeformation cycles (usually reached after about 15 cycles) as determined by the automatic Instron integrator. Immediately following this cyclic test, specimens were distracted at 10 mm/min to 0.68 mm from “ligament zero,” where they were held and allowed to load-relax for 1,200 s. Ligaments were then returned to “ligament zero” for about 10 min and loaded to failure at an extension rate of 20 mm/ min. Load, deformation, and midsubstance strain data were recorded during all three tests, and the modes of failure recorded from video replays. Data Analysis
FIG. 2. Clamping system showing a mounted test specimen with surface dye lines on ligament that were used for strain analysis.
The cyclic peak load in the 30th cycle was normalized to that in the 1st cycle and was used as one index of cyclic behavior. The difference between the final (at 1,200 s) and the initial load, divided by the initial load during the load-relaxation test, were used to determine the percentage of load relaxation.
J Orthop Res, Vol. 8, No. 1, 1990
P. SABISTON ET AL.
50
FIG. 3. Schematic representation of the mechanical testing measuring systems.
V WINDOWS
*
DISPLACEMENT
TV MONITOR
I
STRAIN
Area under the failure curve up to the point of maximum load was defined as failure energy. Engineering stress was obtained by dividing the load by the measured ligament cross-sectional area (measured at midsubstance) and engineering strain calculated from the Video Dimension Analyzer (VDA). A chord modulus was computed from the slope of the stress-strain curve between 6 and 8% strain. Methods of Statistical Analysis Differences between time intervals for allograft and contralateral ligaments were tested using oneway analysis of variance with significant differences being further tested using a least square difference multiple comparison procedure with an overall 01 level of 0.05. Comparisons between operated and contralateral ligaments were made at each time interval using paired f tests. For the purposes of discussion, comparisons between allografts and the autografts published elsewhere (26) were carried out using similar procedures. RESULTS Allograft Ligament Laxity Allograft ligaments show a tendency to be tighter at all intervals of healing in comparison to their contralateral ligaments. When results for laxity were pooled for all intervals, it was shown that allografts had a mean laxity of 0.07 0.02 mm compared to 0.15 ? 0.03 mm for pooled contralaterals (Table 1). With the exception of the 3 week interval, no statistical differences were found when laxity was examined at each healing interval. At 3 weeks, the allografts were significantly tighter (p < 0.05).
*
J Orthop Res, Vol. 8, N o . I , 1990
There was no subsequent trend statistically for the MCL allografts either to tighten or loosen even though the mean values remained lower than the contralateral mean values. Allograft Structural Properties
Prefailure The normalized cyclic peak loads of allograft ligaments were not statistically different from the contralateral ligaments at all intervals studied. While these normalized cyclic peak loads tended to increase between 12 and 48 weeks of healing (Fig. 4), this change was not statistically significant. Normalized load relaxation results, on the other hand, showed that allografts relaxed significantly (p < 0.05) more than contralaterals at most intervals of healing and showed some recovery toward contralateral values at the longest interval studied (Fig. 5). No statistical difference was observed at 3 and 48 weeks due to the variability in the measured values at those intervals.
Failure Failure testing demonstrated significantly lower failure loads and energy absorbed to failure of all TABLE 1. Ligament laxity (expressed in mrn 2 SEM) Weeks of healing
Allograft (n = 5)
3 6 12
0.05 i 0.02 0.08 t 0.04 0.07 t 0.02
24 48
0.06 i 0.03 0.06 f 0.05
Contralateral (n = 5) 0.27 0.15 0.13 0.20 0.27
t 0.09" 2 0.09 f 0.03 f
0.06
2 0.11
Normal control MCL laxity is 0.34 2 0.19 mm ( n Denotes p < 0.05.
=
8).
ALLOGRAFT LIGAMENT TRANSPLANTATION (II)
51
500
Y Q
w n 0
65
., 0.75
W
0.50
LT
-
3
*
cl"ATERAL (' pc0.05)
v
20.251
0
-
I
10
-
I
2 0
.
I
t
3 0
40
- I
0
50
1 0
WEEKS POST-TRANSPLANTATION FIG. 4. Normalized cyclic peak loads (30th vs. 1st cycles) of allograft vs. contralateral MCLs at various healing intervals (mean t SEM).
allograft ligaments, compared with contralaterals (Figs. 6 and 7). Allograft complex strength and energy absorbing capacity was found to be highest at the earliest (3 week) healing interval. Between 12 and 48 weeks posttransplantation, the failure load in the allograft ligament complex was between 5040% of contralateral values. In the same time period, the energy absorbed to failure was about 32% of contralaterals. Examination of the mechanism of failure revealed that the majority of contralateral ligaments (80%) failed in substance. Early (3 to 6 weeks) allograft ligament complexes most commonly failed through the screw fixation at either femoral or tibia1 host graft sites. From 12 weeks posttransplantation onward, all failures occurred within the transplanted complex either in the substance of the ligament or through one of the ligament insertion sites. By 24 weeks, the most common mode of failure was avul-
=P
Allograft Material Properties Allograft ligaments had larger mean crosssectional areas than contralaterals at all healing intervals (Table 2). These differences contributed to a marked decrease in the failure stress of all allograft ligaments, reaching a plateau of about 30% 10% of contralaterals at 12, 24, and 48 weeks. These failures can be subdivided by failure modes for better comparison of mechanical properties of different parts of MCL complexes (Table 3). Figure 10 depicts the composite stress-strain curves for allograft and contralateral ligaments up to 8% strain. Comparison of the chord moduli between 6 and 8% strain showed a significant difference between contralateral and allograft ligaments
*
1000 h
*z (3
50
a I
5 0
40
sion of the femoral insertion and this site was even more definite at 48 weeks (Fig. 8). The composite load-deformation curves (Fig. 9) show similar strengths and stiffnesses of allografts between 12 and 48 weeks of healing.
100
a
a 9
30
FIG. 6. Failure load of allograft vs. contralateral MCLs at various healing intervals (mean t SEM).
E E
0
2 0
WEEKS POST-TRANSPLANTATION
I
W z W W
.pF
500
a
0
10
20
30
40
50
WEEKS POST-TRANSPLANTATION FIG. 5. Normalized load-relaxation (percent load relaxed at 1,200 s vs. initial peak load) of allograft vs. contralateral MCLs at various healing intervals (mean i SEM).
2a LL
(" pc0.05)
0 0
10
20
30
5 0
4 0
WEEKS POST-TRANSPLANTATION FIG. 7. Energy absorbed to failure of allograft vs. contralatera1 MCLs at various healing intervals (mean i SEM).
J Orthop Res, Vol. 8, No. 1 1990 ~
52
P . SABISTON ET A L . TABLE 2. Midsubstance ligament cross-sectional area (expressed in mm2 SEM)
*
Weeks of healing 3 6 12 24 48
Allograft (n = 5 ) 7.0 t 10.0 2 8.3 2 12.8 2 8.2 2
1.8 0.5 1.2 2.0 0.5
Contralateral (n = 5) 3.7 4.3 3.3 2.8 3.3
0.2 0.8" s 0.2" +- 0.2" 2 0.5" 2 2
..
CONT
3
6
24
12
Normal control MCL cross-sectional area is 3.2
4 8
WEEKS POST-TRANSPLANTATION FIG. 8. Bar graph representing different modes of failure of allografts at each healing interval vs. pooled contralateral MCLs. Note the increasing trend for insertional (mainly femoral) failures at 12 weeks and beyond. Also note, however, that some allograft complexes failed in midsubstance at all intervals.
beyond 6 weeks posttransplantation, with allografts having about 3040% of contralateral values at 12, 24, and 48 weeks (Fig. 11). DISCUSSION
Allograft Behavior Like previous biomechanical studies on replacement of ligaments with various allograft tissues (7,16,18,27,29), present results confirm that allografted ligament complexes do not return to normal within the first year after transplantation. While some properties are recovering slowly during that period, others are not, suggesting either an extremely slow recovery process, or the possibility of a permanent mechanical deficiency. A detailed look at individual parameters will shed some light on these possibilities. I
400 I
0
1
2
3
4
5
DEFORMATION (mm) FIG. 9. Composite, nearly identical, load-deformation curves of allografts at 12 and 48 week healing intervals vs. composite load-deformation curve of contralaterals (mean 2 SEM). Results demonstrate an apparent plateau in allograft recovery.
J Orthop Res, Vol. 8 , No. 1, 1990
-t
0.2 mm2
( n = 8). a
Denotes p < 0.05.
Similar to all other models (7,16,27,29), allograft complexes were structurally weaker than contralaterals at all intervals studied. Like allografted fascia lata (7), these results show that complexes got weaker during the first 12 weeks after transplantation. After that time, similar to Shino (27) but unlike the improvement suggested by Nikolau (18) and Webster (31), a plateau in complex strength at about 60% of contralateral loads was observed between 12 and 48 weeks posttransplantation. This is below the 88% plateau reported by Nikolau (18) but is similar to the 24 week results of Curtis (7). All other replacements of ACLs have been found to have less than 35% of control strength at various intervals after implantation (16,27,29). Comparison of such failure percentages alone, however, is misleading. For example, the failure loads alone do not reflect changes in failure modes that occur during the healing processes and instead simply indicate the strength of the "weakest link" in the bone-ligament-bone complex. The current model showed a shift in failure modes (Fig. 8), suggesting that tissues are definitely changing over time. Most allografts fractured through grafts or through screw holes during the first 6 weeks of healing at over 80% of control values. After 6 weeks, there was a dramatic increase in insertional failures (ligament pulling out with small pieces of insertional bone) usually at the femoral end. In essence, this demonstrates that screw holes are no longer the weakest link in the rabbit MCL complex after 6 weeks of healing, since the femoral insertion then becomes even weaker (at 60% of contralateral complex strength). Just as importantly, however, it was found that some samples failed in ligament substance at all intervals, but at quite different loads. When these substance failures are compared (Table 3), it is clear that allograft ligament substance weakened during the first few months of healing and reached a plateau that ranged between 60-75% of
ALLOGRAFT LIGAMENT TRANSPLANTATION (II)
53
TABLE 3. Ligament complex failures allowing comparison by mode of failure Allograft Weeks of healing
n
Substance Fracture Insertion Substance Fracture Insertion Substance Fracture Insertion Substance Fracture Insertion Substance Fracture Insertion
3
6 12 24 48
2 3 0 1 4 0 1 0 4 1 0 4 2 0 3
Contralateral
Peak load -+ SEM (N)
Peak stress i SEM (MPa)
295 2 7 239 2 35
75 f 4 31 2 8 30 24 i 3 23 25 i 8 40 13 2 2 26 20? 1
-
326 234 2 20 -
172 192 i 36 25 1 179 2 6 193 i 1 173 2 16
contralateral substance failure values. Collectively, this suggests early structural weakness of perifixational bone in this model, with subsequent deterioration of both insertional bone and ligament substance. Femoral insertional bone was obviously the biggest weak link in longer-term samples (Fig. 8). This is consistent with other investigations, which showed various combinations of substance (7,27, 29,31), insertional (16,18,27,31), and bony failures (16) but it is the first suggestion that there may be a sequence of structural changes in such composite allografts. While degradation or removal of allograft bone and ligament could explain this deterioration, it seems more likely that some “creeping substitution” by host tissue is taking place. This possibility 80
cn
n
Peak load f SEM (N)
Peak stress 2 SEM (MPa)
368 i 12 314 ? 13 369 i 23 368 f 7 263 2 13 333 f 8 313 f 14
1 1 1 t 18 82 6 93 i 5 90 2 38 81 ? 5
3 2 0 3 2 0
5 0 0 5 0 0
5 0
*
-
122 i 3 105 7 -
*
-
-
0
is supported by evidence that the mass and the cross-sectional area of the allograft complex is increased at four intervals (Table 2) and extensive information, provided elsewhere (25), that both matrix and insertions have shown both morphological and some biochemical evidence of cellular ingrowth, revascularization, and return toward normality. If allograft bone is being replaced by host bone (3,12,14,21), it is not surprising that the much thinner cortical bone of the femoral insertion in this rabbit model would be affected first. Deterioration of midsubstance stress-strain properties of MCL allograft complexes further supports the observation that ligament substance, as a material, is also being altered. The failure stress at midsubstance of the allografts was 74% of contralateral
I
I
3 J 15 3
n
40-
0
[r
0
2
4
6
8
1 0
STRAIN (%) FIG. 10. Composite stress-strain curves of 12 and 48 week healing allograft vs. contralateral MCLs (mean 5 SEM), shown up to 8% strain. Note the inferiority of allografts. Also note no statistical difference between these two allograft intervals.
I
lo
g
5
4w
o GOUT
3
6
12
24
48
WEEKS POST-TRANSPLANTATION FIG. 11. Bar graph representing midsubstance chord modulus (mean i SEM) between 68% strain of allografts at various healing intervals vs. pooled contralaterals. Inferiority of allograft midsubstance is statistically significant at 6 and 48 week intervals.
J Orthop Res, Vol. 8, No. I , 1990
54
P . SABISTON ET A L .
by 3 weeks after transplantation (Table 3) and decreased to about 30 & 10% at all subsequent intervals. The moduli of allograft midsubstances chord (slopes of stress-strain curves taken between 6-8% of tensile strain) were similarly inferior (Fig. 11) and demonstrated an early deterioration to a similar plateau. These normalized values give strong evidence that allograft substance is being altered during the first few weeks after transplantation but then reaches a relative equilibrium, which lasts for at least 1 year in the rabbit. Another unique observation in the present study is the quantification of MCL complex laxity (Table 1). Results suggest that, unlike virtually all other studies of allograft replacements, which show some degree of graft laxity (2,16,29), MCL allograft complexes are, if anything, tighter than contralateral ligaments. This is most likely due to graft surgical placement but it could also be due to graft contraction during the early healing process (8). A third possibility is that there may have been an undetectable difference in joint surfaces, such as undetectable early swelling of articular cartilage, that could have secondarily decreased our measurement of ligament laxity. As with the speculation regarding rapid ligament contraction, this seems unlikely. The possibility of a relaxation difference of allografts during the measurement of laxity can be ruled out, since both the cyclic and the load relaxation behavior (Figs. 4 and 5 ) show that, if anything, allografts should relax more quickly than controls. While prefailure data support failure results in many ways, with allograft values consistently below normal, there was one important difference. Unlike failure data, which reached a plateau below normal levels, both cyclic and load relaxation behavior recovered. In the latter case, relaxation was significantly greater in allografts from 6 to 24 weeks, but became indistinct statistically from contralaterals at 48 weeks of healing. The cyclic peak load was even less affected, being indistinguishable from contralaterals at any interval (Fig. 4). Assuming that these tests have some functional significance in vivo, these results suggest that allograft complexes are either maintained or are remodeling for the cyclic loads of everyday life, rather than the highly unusual loads of failure. The apparent trend for late recovery of these properties also suggests, importantly, that complexes may be changing for many months after transplantation in ways that may be too subtle to be seen during failure tests.
J Orthop Res, Vol. 8, No. 1. 1990
Comparison of Allograft with Autograft Failure Behavior It is beyond the scope of this paper to present all autograft properties. Those properties are presented elsewhere (26) for review. For the purpose of this paper, it is most important to compare allografts with autografts in terms of behavior, where important similarities and differences can be seen. This comparison allows a better understanding of the effects of the model (bone-MCL-bone free graft) on MCL properties as distinct from the variables involved with an allograft complex. Whereas allografts showed an early deterioration in structural strength and a subsequent plateau at about 60% of contralateral failure loads between 12 and 48 weeks, autografts were, on average, better. At 3 weeks, like allografts, autografts failed at approximately 75% of the strength of contralaterals. By 48 weeks, however, in contrast to the allografts, the autografts had reached over 90% of contralaterals in terms of strength. Failure modes of allografts were also different from that of the autografts in longer-term healing groups. Like allografts, autografts failed either in substance or through the femoral screw hole area for the first 6 weeks. Unlike allografts, however, autografts then showed some recovery of both the substance and the insertions between 24 and 48 weeks of healing, with the majority of failures occurring through bone away from insertions at 48 weeks. The major weakening of the insertions seen in allografts (73% were insertional failures between 12 and 48 weeks) was not seen in autografts (27% were insertional failures over the same period). Reasons for these differences between allografts and autografts remain to be defined. We speculate that explanations could be either “biological” (e.g., immune-mediated changes, or changes in graft cellularity/graft cell behavior) or ‘‘mechanical.” The former is the most common explanation for differences in any type of allograft performance and has been studied extensively in other tissues such as bone (3,14,21,23,24).The likelihood that cell behavior does influence graft maintenance or recovery, combined with the fundamental difference between allografts and autograft cell biology (immunologically), and the fact that autografts were fresh and thereby very cellular, while allografts were frozed thawed and thereby less cellular (9), makes this an
ALLOGRAFT LIGAMENT TRANSPLANTATION (II) appealing explanation for slight allograft “inferiority. The latter possibility, that differences may also relate to differences in mechanical loading of grafts, is somewhat unique. This speculation relates to the qualitative observation that allografts were not the “ideal fit” seen in autografts. Despite being from age- and sex-matched donors, allograft bone plugs varied slightly in size and shape and were likely loaded slightly differently than original host tissues. These possibilities require further investigation.
55
REFERENCES
”
The Extra-Articular Model Finally, there are both negative and positive aspects to this extraarticular ligament model, which should be discussed. One negative aspect of this model is that the MCL complex is small, making it technically difficult to handle. In addition, this small tissue mass and thin bone may be much more easily influenced by degradative and reparative processes than larger complexes. Another important consideration is the relatively sedentary nature of the rabbit, possibly inhibiting graft incorporation during healing. Conversely, unlike studies utilizing immobilization (18,27,3I), allowance of early cage activity could have damaged graft fixation. Positive aspects of this model are its previously documented normal and healing mechanical baseline ( l O , l l ) , its relative anatomic simplicity, and its presumably “advantaged extra-articular, vascular environment. As such, it therefore represents a quantifiable model that may well represent the “best case” for examining the processes and possibly “the fastest” rates of healing of a boneligament-bone allograft. As with ACL models, however, each of these factors will have to be examined in more detail in order to understand their relevance to human ligament allografting. Acknowledgment: The authors gratefully acknowledge the financial support of the Medical Research Council of Canada (#MA 9858), the CORE/ACORE Award from the Canadian Orthopaedic Foundation, the Alberta Heritage Foundation for Medical Research, and the Foundation of the Alberta Children’s Hospital. We would also like to acknowledge the assistance of Mr. E. Damson, Dr. M. Andersen, Ms. P. Edwards, Mr. J. Matyas, Mrs. D. McDonald, Mrs. J. Swainger, Judy Crawford, and Mrs. B. Walley in the completion of this work.
1. Alm A, Ekstrom H, Gillquist J, Stromberg B: The anterior cruciate ligament. A clinical and experimental study on tensile strength, morphology and replacement by patellar ligament. Acta Chir Scand 14O(suppl445):4-49, 1974 2. Arnoczky SP, Warren RF, Ashlock MA: Replacement of the anterior cruciate ligament using a patellar tendon allograft. J Bone Joint Surg [Am] 68:37&385, 1986 3. Bos GD, Goldberg VM, Powell AE, Heiple KG, Zika JM: The effect of histocompatibility matching on canine frozen bone allografts. J Bone Joint Surg [Am] 65:89-96, 1983 4 . Butler DL, Noyes FR, Waltz KA, Gibbons MJ: Biomechanics of human allograft treatment. Trans Orthop Res Soc 12:128,1987 5. Cabaud HE, Feagin JA, Rodkey WG: Acute anterior cruciate ligament injury and augmented repair. A m J Sports Med 8:395401, 1980 6. Cho KO:Reconstruction of the anterior cruciate ligament by semitendinosus tenodesis. J Bone Joint Surg [Am] 57:608612, 1975 7. Curtis RJ,Delee JC, Drez DJ: Reconstruction of the anterior cruciate ligament with freeze dried fascia lata allografts in dogs. Am J Sports Med 13:408414,1985 8. Dahners LE, Wilson CJ: An examination of the mechanism of ligament and joint contracture. Trans Orthop Res Soc 12:44, 1987 9. Frank C, Edwards P, McDonald D, Bodies D, Sabiston P: Viability of ligaments after freezing: an experimental study in a rabbit model. J Orthop Res 1:95-102, 1988 10. Frank C, Schachar N, Dittrich D, Shrive N, Edward GE: Normal ligament healing in a rabbit model. A morphological, histological, biochemical and biomechanical study. Trans Orthop Res Soc 6:197, 1981 11. Frank C, Woo SL-Y, Amiel D, Gomez MA, Hanvood FL, Akeson WH: Medial collateral ligament healing: a rnultidisciplinary assessment in rabbits. A m J Sports Med 11:379389, 1983 12. Goldberg VM, Lance EM: Revascularization and accretion in transplantation: quantitative studies of the role of the allograft barrier. J Bone Joint Surg [Am] 54:807-816, 1972 13. Grood ES, Butler DL, Noyes FR: Models of ligament repairs and grafts. In: AAOS Symposium on Sports Medicine: The Knee, ed by G Finerman, St. Louis, CV Mosby Co., 1985,pp 169-181 14. Heiple KG, Chase SW, Herndon CH: A comparative study of the healing process following different types of bone transplantation. J Bone Joint Surg [Am] 45:1593-1616, 1963 15. Hollis JM, Lyon RM, Marcin JP, Horibe S, Lee EB, Woo SL-Y: Effect of age and loading axis on the failure properties of the human ACL. Trans Orthop Res Soc 13:83, 1988 16. Jackson DW, Grood ES, Arnoczky SP, Butler DL, Simon TM: Freeze dried anterior cruciate ligament allografts: preliminary studies using a goat model. A m J Sports Med 4:295303, 1987 17. Jackson DW, Grood ES, Wilcox P, Butler DL, Simon TM, Holden JP: The effects of processing techniques on the mechanical properties of bone-anterior cruciate ligament-bone allografts: an experimental study in goats. A m J Sports Med 2:lOl-105, 1988 18. Nikolaou PK, Seaber AV, Glisson R, Ribbeck BM, Bassett FH: Anterior cruciate ligament allograft transplantation. Long-term function, histology, revascularization, and operative technique. A m J Sports Med 14:341360, 1986 19. Noyes FR, Butler DL, Grood ES, Zernicke, RF, Hefzy MS: Biomechanical analysis of human ligament grafts used in
J Orthop Res, Vol. 8, No. I , 1990
P . SABISTON ET AL.
56
20. 21. 22.
23. 24. 25.
26. 27.
knee ligaments repairs and reconstructions. J Bone Joint Surg [Am] 66:344-351, 1984 O'Donoghue, Rockwood CA, Frank GR, Jack SC, Kenyon R: Repair of the anterior cruciate ligament in dogs. J Bone Joint Surg [Am] 48:503-519, 1966 Pappas AM, Beisaw NE: Bone transplantation: correlation of physical and histologic aspects of graft incorporation. Clin Orthop 61:79-91, 1968 Paulos L, France E, Rosenberg T, et al: Comparative material properties of allograft tissues for ligament replacement: effects of type, age, sterilization and preservation. Trans Orthop Res Soc 12:129, 1987 Peer LA: Transplantation of tendons in animals. In: Transplantation ofTissues, ed by LA Peer, Baltimore, Williams & Wilkins, 1955, pp 227-295 Peer LA: Transplantation of tendons in humans. In: Transplantation ofli'ssues, ed by LA Peer, Baltimore, Williams & Wilkins, 1955, pp 296-334 Sabiston P, Frank C, Lam T, Shrive N: Allograft ligament transplantation: a morphological and biochemical evaluation of a medial collateral ligament complex in a rabbit model. Am J Sports Med (in press, 1989) Sabiston P, Frank C, Lam T, Shrive N: A biomechanical study of an extraarticular ligament autograft-an experimental study in a rabbit model. J Orthop Res (in press) Shino K, Kawasaki T, Hirose H, Botoh I, Inoue M, Ono K: Replacement of the anterior cruciate ligament by an allo-
J Orthop Res, Vol. 8, No. I , 1990
28. 29.
30.
31. 32.
33. 34.
genic tendon graft: an experimental study in the dog. J Bone Joint Surg [Br] 66:672481, 1984 Shrive NG, Lam TC, Damson E, Frank CB: A new method of measuring the cross-sectional area of connective tissue structures. J Biomech Eng 2:104-109, 1988 Thorson EP, Rodrigo JJ, Vasseur HA, Heitter DO: Comparison of frozen allograft versus fresh autogenous anterior cruciate ligament replacement in the dog. Trans Orthop Res Soc 12:65, 1987 Turner WD, Vasseur P, Gorek JE, Rodrigo JJ, Wedell JR: An in vitro study of the structural properties of deep-frozen versus freeze-dried, ethylene oxide-sterilized canine anterior cruciate ligament bone-ligament-bone preparations. Clin Orthop 230:251-256, 1988 Webster DA, Werner FW: Freeze-dried flexor tendons in anterior cruciate ligament reconstruction. Clin Orthop 181:238-243, 1983 Woo SL-Y, Gomez MA, Seguchi Y, Endo CM, Akeson WH: Measurement of mechanical properties of ligament substance from a bone-ligament-bone preparation. J Orthop Res 1:22-29, 1983 Woo SL-Y, Hollis JM, Roux RD, et al: Effects of knee flexion on the structural properties of the rabbit femurACLtibia complex (FATC). J Biomech 637-563, 1987 Yin FCP, Tompkins WR, Peterson KL, Intaglietta M: A video dimension analyzer. IEEE Trans Biomed Eng 19:37& 381, 1972
Transplantation of the Rabbit Medial Collateral Ligament. 11. Biomechanical Evaluation of Frozen/Thawed Allografts “P. Sabiston, *C. Frank, *tT. Lam, and TN. Shrive *The Joint Injury and Diseases Group, Faculty of Medicine, Department of Surgery, Division of Orthopaedics and ?Faculty of Engineering, Department of Civil Engineering, The University of Calgary, Calgary, Alberta, Canada
Summary: This study was designed to test the mechanical behavior of a frozed thawed bone-medial collateral ligament (MCL)-bone allograft in a rabbit model and to compare allograft behavior with contralateral unoperated ligaments as well as a group of normal bone-ligament-bone preparations prior to freezing. Twenty-five mature rabbits received similarly frozen boneMCGbone allografts and were subsequently allowed to heal without immobilization in groups of five for 3, 6, 12, 24, or 48 weeks after transplantation. A series of biomechanical tests was performed on each allograft, contralateral, and external normal control ligament. Results demonstrated that allografts were significantly tighter than controls at 3 weeks and they remained tight over time. All allografts, however, had inferior structural and material behaviors as compared with controls. Allograft insertional bone and substance deteriorated to about 60% of contralateral strength at 12 and 48 weeks. Bony insertions became the most common site of graft weakness. Both stress at failure and the elastic modulus of allograft substance similarly reached a plateau at those intervals, but at about 30%of contralateral controls. Cyclic and load relaxation properties of allografts, on the other hand, did not deteriorate and were, in fact, indistinguishable from contralateral and normal values at 48 weeks. Comparison with a series of fresh autografts suggests that, with the exception of this viscoelastic recovery, allografts were mechanically inferior to autografts in their healing at the intervals tested. Collectively, these results suggest that allograft MCL healing in this composite model is dynamic but slow, featuring changes in bone and soft tissue over the first year after transplantation. Although “viable” and vascular, allografts have not normalized mechanically and, in fact, appear to have reached a new equilibrium whereby properties are neither improving nor deteriorating. Despite this apparent equilibrium, the trend for some viscoelastic recovery provides hope that allografts may have further potential for improvement. Longer-term studies of MCL allograft mechanical behaviors are clearly required. Key Words: Rabbit-Medial collateral ligament-Allografts-Freezekhaw .
plants have not been common. Instead, most studies have concentrated on normal mechanical baselines of potential transplant tissues in humans (4,6,18,19) or in animal models (2,7,27,31), and the mechanical effects of pretransplant sterilization or preservation (16,17,22,29,30). Only a few studies have tested what happens to the beha’ior of these tissues after transplantation and all of
Despite the long history of biological and clinical interest in dense connective tissue transplantation (23,24), biomechanical assessments of such transAddress correspondence and reprint requests to Dr. C. Frank, Department of Surgery, The University of Calgary, Health Sciences Centre, 3330 Hospital Drive NW, Calgary, Alberta, Canada T2N 4N 1.
46
ALLOGRAFT LIGAMENT TRANSPLANTATION (II)
them have evaluated only the anterior cruciate ligament (ACL). Webster and Werner, for example, replaced canine ACLs with freeze-dried flexor tendons and found that they had 29% of control failure strength at 9 months posttransplantation (3 1). Shino and coworkers found similar values for frozen free patellar tendon allografts at 30 and 52 weeks after grafting (27). Curtis and co-workers (7) found dramatic early weakening of freeze-dried fascia lata allografts with a subsequent trend to get stronger but, like tendon replacements, these allografts reached only a fraction of normal ACL strength. Like both Webster and Shino, they felt that, although allografts were inferior to controls, their similarity in strength to commonly used tendon autografts provided sufficient support for their continued investigation and use. Nikolau and co-workers (18),searching for better results from allografting ligaments, examined frozen bone-ACL-bone allografts in dogs. They found steady increases in strength and stiffness for 36 weeks, reaching values within 10% of contralateral controls. It was noted, however, these controls failed at less than 50% of literature values for canine ACL (1,5,13,20). Jackson and co-workers studied a freeze-dried bone-ACL-bone model in goats (16), and found that allografts had only 25% of normal ACL strength at 1 year. Thorson et al. (29) reported even poorer results in a frozen canine ACL model and suggested the possibility of a secondary immune arthritis. Collectively, therefore, these studies suggest that allograft replacements of the ACL are mechanically inferior to control ACLs for at least 1 year after transplantation due to weakening of either insertions (16,18), midsubstance (29,31), or both (27). It is not clear, however, whether these results may be specific to the highly complex ACL (15,33), specific to the canine or goat models used, or may be due to some other immune (29), mechanical, or treatment variables, such as immobilization (7,16, IS). This study was designed to test a relatively simple, homogeneous, and well-characterized allograft ligament model; the rabbit medial collateral ligament (MCL). Our goal was to examine a number of mechanical characteristics in this model over time, as a measure of the mechanical changes in a ligament allograft in an extra-articular environment. Finally, we wanted to compare these changes with contralatcral, normal, and autograft controls (26), and correlate mechanical with biological changes of
47
both these autografts and allografts, reported elsewhere (25), in order to determine the association, if any, between graft “viability” and graft biomechanics . MATERIALS AND METHODS Animal Model The MCL complexes of both legs of 25 mature (12-18 month old; 4.2 2 0.2 kg) female New Zealand white rabbits were studied. The right leg MCLs were designated as the experimentals in every animal and were harvested from donors as bone-ligament-bone complexes. While major histocompatibility testing was not done prior to transplantation, animals were obtained from four different suppliers and siblings did not act as donors or recipients for one another. Distribution of Animals Five allograft ligament transplant animals were studied at each of five time intervals posttransplantation: 3, 6, 12, 24, and 48 weeks (total = 25 animals). Experimental Procedures Under halothane general anesthesia, all animals had their right knee shaved and prepared with iodine. An anteromedial incision was used to expose the MCL. Muscle and fascia were dissected from around both MCL insertion sites and a small circular saw was used to removed rectangles of insertional bone from both tibia and femur without disrupting ligament substance. The distal femoral and proximal tibia1 cuts were 5 mm from the joint line and were used as landmarks to control graft tension during reinsertion. Sharp dissection was used to separate the MCL from the underlying medial meniscus and the bone-MCL-bone complex was removed (Fig. 1). Corners of graft beds were rounded with a drill to minimize the risk of fractures. Donor MCL complexes were then rinsed in phosphatebuffered saline (PBS), placed in sterile 20 ml culture tubes without any cryoprotectants, and the tubes were then wrapped in two sterile towels. Similar to a freezing and storage protocol documented elsewhere (9), towels were placed in a freezer at - 70°C and frozen at approximately - l”C/min. They were
J Orthop Res, Vol. 8, N o . 1, 1990
P . SABISTON ET AL.
48
FIG. 1. Schematic drawing of allograft medial collateral ligament transplant procedure showing how bone-MCL-bone graft was removed from donor and subsequently reinserted into recipient using cortical screws.
then maintained at - 70°C for 1 week until implantation. Prior to implantation, donor MCL allograft complexes were thawed rapidly in sterile 37°C PBS. Grafts were fixed with 2.0 mm cortical screws, supplemented with a small washer over the femoral insertion, and tension was then adjusted by altering graft position if the joint was found to be abnormally lax or tight on manual valgus testing. In all cases, wounds were closed in layers using interrupted 4-0 nylon suture. A soft dressing was applied and animals were allowed to mobilize with unrestricted cage (65 x 45 x 30 cm) activity. Wounds were inspected on the third postoperative day and all animals were weighed regularly as a general screening index of health. Anteroposterior and lateral radiographs of operated knees were obtained 3 to 6 weeks postoperatively to monitor internal fixation and graft incorporation. Any animal with signs of infection, displaced grafts, or long bone fractures was excluded from further study and replaced by another animal. Controls
Left contralateral MCL complexes of each animal served as an internal control. We cannot refer to these contralaterals as “normal ligaments” de-
J Orthop Res, Vol. 8, N o . I , 1990
spite their being unoperated, due to the possibility of secondary effects from the treatments described. We will therefore refer to them as contralateral controls throughout. A separate series of five unoperated mature animals, sacrificed soon after supply, provided baseline external “time zero” controls of both fresh and frozen normal ligament biomechanical controls. This was done to establish baselines with which to compare the effect of contralateral ligament surgery and to test the effect of freezing on normal ligament properties. Five fresh control MCL complexes (right legs) were tested immediately while five frozen controls (left legs) were stored for 1 week at - 70°C and thawed in 37°C PBS prior to mechanical testing. No difference was found between these fresh and frozen controls for any parameter studied and there was no statistical difference between either of these normal controls and contralaterals. Contralaterals, however, were not static over time and showed some changes in mechanical properties between various healing intervals (see the Results section, Figs. 5-7). Specimen Preparation
All experimental and control MCL specimens were tested fresh. On the day of testing, animals were sacrificed with an overdose of intravenous barbiturate (275 mg/kg). Hindlimbs were disarticulated, the left limb (contralateral control) placed in a sealed container to prevent drying, while the right limb (transplant) was prepared for immediate testing. All soft tissues, including muscles and fascia, were removed from the femur and tibia, excluding the collateral ligaments, the cruciate ligaments, and the menisci. Bones were transected 4-5 cm from the joint line. To facilitate surface strain measurement, parallel dye lines were applied at approximately 5 mm intervals onto the surface of the medial collateral ligament, transverse to the longitudinal axis, with Verhoeff elastin stain. The tibia was then potted in a specially designed clamp using polymethylmethacrylate (PMMA) and this clamp was attached to the crosshead of a materials testing system (Model 1122, Instron Corp., Canton, MA, U.S.A.). The knee flexion angle was adjusted to approximately 60” and the cross-head lowered to place the femoral end of the specimen into a second clamp, where it was fixed with PMMA. The longitudinal axis of the MCL was aligned with the load axis of
ALLOGRAFT LIGAMENT TRANSPLANTATION (II) the testing system using a system of slide plates (Fig. 2). With the knee joint fixed in this position, the MCL was isolated by sequentially cutting and removing both menisci, lateral collateral, and both cruciate ligaments. The knee joint was taken through two cycles of loading and unloading at an extension rate of 1 mmlmin, first compressing the joint by 4 N , and then reversing to a tensile force of 2 N taken up by the ligament. This procedure was designated a compression-tension cycle. The crosshead position where the isolated MCL first registered a tensile load (tO.05 N) was defined as “ligament zero” and was used as the starting position for all subsequent mechanical tests on the ligament. Ligament laxity was determined by measuring the amount of cross-head displacement between this point at which the MCL took up tension and the point where articular surfaces were measurably compressed using maximal load scale sensitivity (compression of 0.05 N). A small portion of medial
49
femoral condyle was then removed, care being taken to avoid damage to the femoral ligament insertion, to allow in situ measurement of midsubstance MCL cross-sectional area with a specially designed cross-sectional area measuring device (28). An environment tank surrounding the clamping system was then flooded with 0.9% phosphatebuffered saline (PBS) at 35°C and pH 7.2. A television camera connected to a video dimension analyzer (VDA Model 303, Instrumentation f o r Physiology and Medicine Inc., San Diego, CA, U.S.A.) (34) was then positioned perpendicular to the longitudinal axis of the ligament to quantify separation of surface dye lines (32). The video output was recorded for later analysis. The analog load signal from the load cell was digitized at 50 Hz with a multiprogrammer (Model HP6944, HewlettPackard, Rockville, MD, U.S.A.) and stored in a hard disk for subsequent analysis (Fig. 3). Mechanical Test Protocol
The test sequence for each MCL complex began with cycling 30 times between “ligament zero” and a fixed deformation of 0.68 mm (approximately 3% average strain) at an extension rate of 10 mmlmin. This led to a duty cycle of about 4 s of loading time and 4 s of unloading time for each cycle. The mechanical state of the complex at the 30th cycle was taken to represent a relative steady state. This steady state was defined mathematically by identical hysteresis areas under two successive loaddeformation cycles (usually reached after about 15 cycles) as determined by the automatic Instron integrator. Immediately following this cyclic test, specimens were distracted at 10 mm/min to 0.68 mm from “ligament zero,” where they were held and allowed to load-relax for 1,200 s. Ligaments were then returned to “ligament zero” for about 10 min and loaded to failure at an extension rate of 20 mm/ min. Load, deformation, and midsubstance strain data were recorded during all three tests, and the modes of failure recorded from video replays. Data Analysis
FIG. 2. Clamping system showing a mounted test specimen with surface dye lines on ligament that were used for strain analysis.
The cyclic peak load in the 30th cycle was normalized to that in the 1st cycle and was used as one index of cyclic behavior. The difference between the final (at 1,200 s) and the initial load, divided by the initial load during the load-relaxation test, were used to determine the percentage of load relaxation.
J Orthop Res, Vol. 8, No. 1, 1990
P. SABISTON ET AL.
50
FIG. 3. Schematic representation of the mechanical testing measuring systems.
V WINDOWS
*
DISPLACEMENT
TV MONITOR
I
STRAIN
Area under the failure curve up to the point of maximum load was defined as failure energy. Engineering stress was obtained by dividing the load by the measured ligament cross-sectional area (measured at midsubstance) and engineering strain calculated from the Video Dimension Analyzer (VDA). A chord modulus was computed from the slope of the stress-strain curve between 6 and 8% strain. Methods of Statistical Analysis Differences between time intervals for allograft and contralateral ligaments were tested using oneway analysis of variance with significant differences being further tested using a least square difference multiple comparison procedure with an overall 01 level of 0.05. Comparisons between operated and contralateral ligaments were made at each time interval using paired f tests. For the purposes of discussion, comparisons between allografts and the autografts published elsewhere (26) were carried out using similar procedures. RESULTS Allograft Ligament Laxity Allograft ligaments show a tendency to be tighter at all intervals of healing in comparison to their contralateral ligaments. When results for laxity were pooled for all intervals, it was shown that allografts had a mean laxity of 0.07 0.02 mm compared to 0.15 ? 0.03 mm for pooled contralaterals (Table 1). With the exception of the 3 week interval, no statistical differences were found when laxity was examined at each healing interval. At 3 weeks, the allografts were significantly tighter (p < 0.05).
*
J Orthop Res, Vol. 8, N o . I , 1990
There was no subsequent trend statistically for the MCL allografts either to tighten or loosen even though the mean values remained lower than the contralateral mean values. Allograft Structural Properties
Prefailure The normalized cyclic peak loads of allograft ligaments were not statistically different from the contralateral ligaments at all intervals studied. While these normalized cyclic peak loads tended to increase between 12 and 48 weeks of healing (Fig. 4), this change was not statistically significant. Normalized load relaxation results, on the other hand, showed that allografts relaxed significantly (p < 0.05) more than contralaterals at most intervals of healing and showed some recovery toward contralateral values at the longest interval studied (Fig. 5). No statistical difference was observed at 3 and 48 weeks due to the variability in the measured values at those intervals.
Failure Failure testing demonstrated significantly lower failure loads and energy absorbed to failure of all TABLE 1. Ligament laxity (expressed in mrn 2 SEM) Weeks of healing
Allograft (n = 5)
3 6 12
0.05 i 0.02 0.08 t 0.04 0.07 t 0.02
24 48
0.06 i 0.03 0.06 f 0.05
Contralateral (n = 5) 0.27 0.15 0.13 0.20 0.27
t 0.09" 2 0.09 f 0.03 f
0.06
2 0.11
Normal control MCL laxity is 0.34 2 0.19 mm ( n Denotes p < 0.05.
=
8).
ALLOGRAFT LIGAMENT TRANSPLANTATION (II)
51
500
Y Q
w n 0
65
., 0.75
W
0.50
LT
-
3
*
cl"ATERAL (' pc0.05)
v
20.251
0
-
I
10
-
I
2 0
.
I
t
3 0
40
- I
0
50
1 0
WEEKS POST-TRANSPLANTATION FIG. 4. Normalized cyclic peak loads (30th vs. 1st cycles) of allograft vs. contralateral MCLs at various healing intervals (mean t SEM).
allograft ligaments, compared with contralaterals (Figs. 6 and 7). Allograft complex strength and energy absorbing capacity was found to be highest at the earliest (3 week) healing interval. Between 12 and 48 weeks posttransplantation, the failure load in the allograft ligament complex was between 5040% of contralateral values. In the same time period, the energy absorbed to failure was about 32% of contralaterals. Examination of the mechanism of failure revealed that the majority of contralateral ligaments (80%) failed in substance. Early (3 to 6 weeks) allograft ligament complexes most commonly failed through the screw fixation at either femoral or tibia1 host graft sites. From 12 weeks posttransplantation onward, all failures occurred within the transplanted complex either in the substance of the ligament or through one of the ligament insertion sites. By 24 weeks, the most common mode of failure was avul-
=P
Allograft Material Properties Allograft ligaments had larger mean crosssectional areas than contralaterals at all healing intervals (Table 2). These differences contributed to a marked decrease in the failure stress of all allograft ligaments, reaching a plateau of about 30% 10% of contralaterals at 12, 24, and 48 weeks. These failures can be subdivided by failure modes for better comparison of mechanical properties of different parts of MCL complexes (Table 3). Figure 10 depicts the composite stress-strain curves for allograft and contralateral ligaments up to 8% strain. Comparison of the chord moduli between 6 and 8% strain showed a significant difference between contralateral and allograft ligaments
*
1000 h
*z (3
50
a I
5 0
40
sion of the femoral insertion and this site was even more definite at 48 weeks (Fig. 8). The composite load-deformation curves (Fig. 9) show similar strengths and stiffnesses of allografts between 12 and 48 weeks of healing.
100
a
a 9
30
FIG. 6. Failure load of allograft vs. contralateral MCLs at various healing intervals (mean t SEM).
E E
0
2 0
WEEKS POST-TRANSPLANTATION
I
W z W W
.pF
500
a
0
10
20
30
40
50
WEEKS POST-TRANSPLANTATION FIG. 5. Normalized load-relaxation (percent load relaxed at 1,200 s vs. initial peak load) of allograft vs. contralateral MCLs at various healing intervals (mean i SEM).
2a LL
(" pc0.05)
0 0
10
20
30
5 0
4 0
WEEKS POST-TRANSPLANTATION FIG. 7. Energy absorbed to failure of allograft vs. contralatera1 MCLs at various healing intervals (mean i SEM).
J Orthop Res, Vol. 8, No. 1 1990 ~
52
P . SABISTON ET A L . TABLE 2. Midsubstance ligament cross-sectional area (expressed in mm2 SEM)
*
Weeks of healing 3 6 12 24 48
Allograft (n = 5 ) 7.0 t 10.0 2 8.3 2 12.8 2 8.2 2
1.8 0.5 1.2 2.0 0.5
Contralateral (n = 5) 3.7 4.3 3.3 2.8 3.3
0.2 0.8" s 0.2" +- 0.2" 2 0.5" 2 2
..
CONT
3
6
24
12
Normal control MCL cross-sectional area is 3.2
4 8
WEEKS POST-TRANSPLANTATION FIG. 8. Bar graph representing different modes of failure of allografts at each healing interval vs. pooled contralateral MCLs. Note the increasing trend for insertional (mainly femoral) failures at 12 weeks and beyond. Also note, however, that some allograft complexes failed in midsubstance at all intervals.
beyond 6 weeks posttransplantation, with allografts having about 3040% of contralateral values at 12, 24, and 48 weeks (Fig. 11). DISCUSSION
Allograft Behavior Like previous biomechanical studies on replacement of ligaments with various allograft tissues (7,16,18,27,29), present results confirm that allografted ligament complexes do not return to normal within the first year after transplantation. While some properties are recovering slowly during that period, others are not, suggesting either an extremely slow recovery process, or the possibility of a permanent mechanical deficiency. A detailed look at individual parameters will shed some light on these possibilities. I
400 I
0
1
2
3
4
5
DEFORMATION (mm) FIG. 9. Composite, nearly identical, load-deformation curves of allografts at 12 and 48 week healing intervals vs. composite load-deformation curve of contralaterals (mean 2 SEM). Results demonstrate an apparent plateau in allograft recovery.
J Orthop Res, Vol. 8 , No. 1, 1990
-t
0.2 mm2
( n = 8). a
Denotes p < 0.05.
Similar to all other models (7,16,27,29), allograft complexes were structurally weaker than contralaterals at all intervals studied. Like allografted fascia lata (7), these results show that complexes got weaker during the first 12 weeks after transplantation. After that time, similar to Shino (27) but unlike the improvement suggested by Nikolau (18) and Webster (31), a plateau in complex strength at about 60% of contralateral loads was observed between 12 and 48 weeks posttransplantation. This is below the 88% plateau reported by Nikolau (18) but is similar to the 24 week results of Curtis (7). All other replacements of ACLs have been found to have less than 35% of control strength at various intervals after implantation (16,27,29). Comparison of such failure percentages alone, however, is misleading. For example, the failure loads alone do not reflect changes in failure modes that occur during the healing processes and instead simply indicate the strength of the "weakest link" in the bone-ligament-bone complex. The current model showed a shift in failure modes (Fig. 8), suggesting that tissues are definitely changing over time. Most allografts fractured through grafts or through screw holes during the first 6 weeks of healing at over 80% of control values. After 6 weeks, there was a dramatic increase in insertional failures (ligament pulling out with small pieces of insertional bone) usually at the femoral end. In essence, this demonstrates that screw holes are no longer the weakest link in the rabbit MCL complex after 6 weeks of healing, since the femoral insertion then becomes even weaker (at 60% of contralateral complex strength). Just as importantly, however, it was found that some samples failed in ligament substance at all intervals, but at quite different loads. When these substance failures are compared (Table 3), it is clear that allograft ligament substance weakened during the first few months of healing and reached a plateau that ranged between 60-75% of
ALLOGRAFT LIGAMENT TRANSPLANTATION (II)
53
TABLE 3. Ligament complex failures allowing comparison by mode of failure Allograft Weeks of healing
n
Substance Fracture Insertion Substance Fracture Insertion Substance Fracture Insertion Substance Fracture Insertion Substance Fracture Insertion
3
6 12 24 48
2 3 0 1 4 0 1 0 4 1 0 4 2 0 3
Contralateral
Peak load -+ SEM (N)
Peak stress i SEM (MPa)
295 2 7 239 2 35
75 f 4 31 2 8 30 24 i 3 23 25 i 8 40 13 2 2 26 20? 1
-
326 234 2 20 -
172 192 i 36 25 1 179 2 6 193 i 1 173 2 16
contralateral substance failure values. Collectively, this suggests early structural weakness of perifixational bone in this model, with subsequent deterioration of both insertional bone and ligament substance. Femoral insertional bone was obviously the biggest weak link in longer-term samples (Fig. 8). This is consistent with other investigations, which showed various combinations of substance (7,27, 29,31), insertional (16,18,27,31), and bony failures (16) but it is the first suggestion that there may be a sequence of structural changes in such composite allografts. While degradation or removal of allograft bone and ligament could explain this deterioration, it seems more likely that some “creeping substitution” by host tissue is taking place. This possibility 80
cn
n
Peak load f SEM (N)
Peak stress 2 SEM (MPa)
368 i 12 314 ? 13 369 i 23 368 f 7 263 2 13 333 f 8 313 f 14
1 1 1 t 18 82 6 93 i 5 90 2 38 81 ? 5
3 2 0 3 2 0
5 0 0 5 0 0
5 0
*
-
122 i 3 105 7 -
*
-
-
0
is supported by evidence that the mass and the cross-sectional area of the allograft complex is increased at four intervals (Table 2) and extensive information, provided elsewhere (25), that both matrix and insertions have shown both morphological and some biochemical evidence of cellular ingrowth, revascularization, and return toward normality. If allograft bone is being replaced by host bone (3,12,14,21), it is not surprising that the much thinner cortical bone of the femoral insertion in this rabbit model would be affected first. Deterioration of midsubstance stress-strain properties of MCL allograft complexes further supports the observation that ligament substance, as a material, is also being altered. The failure stress at midsubstance of the allografts was 74% of contralateral
I
I
3 J 15 3
n
40-
0
[r
0
2
4
6
8
1 0
STRAIN (%) FIG. 10. Composite stress-strain curves of 12 and 48 week healing allograft vs. contralateral MCLs (mean 5 SEM), shown up to 8% strain. Note the inferiority of allografts. Also note no statistical difference between these two allograft intervals.
I
lo
g
5
4w
o GOUT
3
6
12
24
48
WEEKS POST-TRANSPLANTATION FIG. 11. Bar graph representing midsubstance chord modulus (mean i SEM) between 68% strain of allografts at various healing intervals vs. pooled contralaterals. Inferiority of allograft midsubstance is statistically significant at 6 and 48 week intervals.
J Orthop Res, Vol. 8, No. I , 1990
54
P . SABISTON ET A L .
by 3 weeks after transplantation (Table 3) and decreased to about 30 & 10% at all subsequent intervals. The moduli of allograft midsubstances chord (slopes of stress-strain curves taken between 6-8% of tensile strain) were similarly inferior (Fig. 11) and demonstrated an early deterioration to a similar plateau. These normalized values give strong evidence that allograft substance is being altered during the first few weeks after transplantation but then reaches a relative equilibrium, which lasts for at least 1 year in the rabbit. Another unique observation in the present study is the quantification of MCL complex laxity (Table 1). Results suggest that, unlike virtually all other studies of allograft replacements, which show some degree of graft laxity (2,16,29), MCL allograft complexes are, if anything, tighter than contralateral ligaments. This is most likely due to graft surgical placement but it could also be due to graft contraction during the early healing process (8). A third possibility is that there may have been an undetectable difference in joint surfaces, such as undetectable early swelling of articular cartilage, that could have secondarily decreased our measurement of ligament laxity. As with the speculation regarding rapid ligament contraction, this seems unlikely. The possibility of a relaxation difference of allografts during the measurement of laxity can be ruled out, since both the cyclic and the load relaxation behavior (Figs. 4 and 5 ) show that, if anything, allografts should relax more quickly than controls. While prefailure data support failure results in many ways, with allograft values consistently below normal, there was one important difference. Unlike failure data, which reached a plateau below normal levels, both cyclic and load relaxation behavior recovered. In the latter case, relaxation was significantly greater in allografts from 6 to 24 weeks, but became indistinct statistically from contralaterals at 48 weeks of healing. The cyclic peak load was even less affected, being indistinguishable from contralaterals at any interval (Fig. 4). Assuming that these tests have some functional significance in vivo, these results suggest that allograft complexes are either maintained or are remodeling for the cyclic loads of everyday life, rather than the highly unusual loads of failure. The apparent trend for late recovery of these properties also suggests, importantly, that complexes may be changing for many months after transplantation in ways that may be too subtle to be seen during failure tests.
J Orthop Res, Vol. 8, No. 1. 1990
Comparison of Allograft with Autograft Failure Behavior It is beyond the scope of this paper to present all autograft properties. Those properties are presented elsewhere (26) for review. For the purpose of this paper, it is most important to compare allografts with autografts in terms of behavior, where important similarities and differences can be seen. This comparison allows a better understanding of the effects of the model (bone-MCL-bone free graft) on MCL properties as distinct from the variables involved with an allograft complex. Whereas allografts showed an early deterioration in structural strength and a subsequent plateau at about 60% of contralateral failure loads between 12 and 48 weeks, autografts were, on average, better. At 3 weeks, like allografts, autografts failed at approximately 75% of the strength of contralaterals. By 48 weeks, however, in contrast to the allografts, the autografts had reached over 90% of contralaterals in terms of strength. Failure modes of allografts were also different from that of the autografts in longer-term healing groups. Like allografts, autografts failed either in substance or through the femoral screw hole area for the first 6 weeks. Unlike allografts, however, autografts then showed some recovery of both the substance and the insertions between 24 and 48 weeks of healing, with the majority of failures occurring through bone away from insertions at 48 weeks. The major weakening of the insertions seen in allografts (73% were insertional failures between 12 and 48 weeks) was not seen in autografts (27% were insertional failures over the same period). Reasons for these differences between allografts and autografts remain to be defined. We speculate that explanations could be either “biological” (e.g., immune-mediated changes, or changes in graft cellularity/graft cell behavior) or ‘‘mechanical.” The former is the most common explanation for differences in any type of allograft performance and has been studied extensively in other tissues such as bone (3,14,21,23,24).The likelihood that cell behavior does influence graft maintenance or recovery, combined with the fundamental difference between allografts and autograft cell biology (immunologically), and the fact that autografts were fresh and thereby very cellular, while allografts were frozed thawed and thereby less cellular (9), makes this an
ALLOGRAFT LIGAMENT TRANSPLANTATION (II) appealing explanation for slight allograft “inferiority. The latter possibility, that differences may also relate to differences in mechanical loading of grafts, is somewhat unique. This speculation relates to the qualitative observation that allografts were not the “ideal fit” seen in autografts. Despite being from age- and sex-matched donors, allograft bone plugs varied slightly in size and shape and were likely loaded slightly differently than original host tissues. These possibilities require further investigation.
55
REFERENCES
”
The Extra-Articular Model Finally, there are both negative and positive aspects to this extraarticular ligament model, which should be discussed. One negative aspect of this model is that the MCL complex is small, making it technically difficult to handle. In addition, this small tissue mass and thin bone may be much more easily influenced by degradative and reparative processes than larger complexes. Another important consideration is the relatively sedentary nature of the rabbit, possibly inhibiting graft incorporation during healing. Conversely, unlike studies utilizing immobilization (18,27,3I), allowance of early cage activity could have damaged graft fixation. Positive aspects of this model are its previously documented normal and healing mechanical baseline ( l O , l l ) , its relative anatomic simplicity, and its presumably “advantaged extra-articular, vascular environment. As such, it therefore represents a quantifiable model that may well represent the “best case” for examining the processes and possibly “the fastest” rates of healing of a boneligament-bone allograft. As with ACL models, however, each of these factors will have to be examined in more detail in order to understand their relevance to human ligament allografting. Acknowledgment: The authors gratefully acknowledge the financial support of the Medical Research Council of Canada (#MA 9858), the CORE/ACORE Award from the Canadian Orthopaedic Foundation, the Alberta Heritage Foundation for Medical Research, and the Foundation of the Alberta Children’s Hospital. We would also like to acknowledge the assistance of Mr. E. Damson, Dr. M. Andersen, Ms. P. Edwards, Mr. J. Matyas, Mrs. D. McDonald, Mrs. J. Swainger, Judy Crawford, and Mrs. B. Walley in the completion of this work.
1. Alm A, Ekstrom H, Gillquist J, Stromberg B: The anterior cruciate ligament. A clinical and experimental study on tensile strength, morphology and replacement by patellar ligament. Acta Chir Scand 14O(suppl445):4-49, 1974 2. Arnoczky SP, Warren RF, Ashlock MA: Replacement of the anterior cruciate ligament using a patellar tendon allograft. J Bone Joint Surg [Am] 68:37&385, 1986 3. Bos GD, Goldberg VM, Powell AE, Heiple KG, Zika JM: The effect of histocompatibility matching on canine frozen bone allografts. J Bone Joint Surg [Am] 65:89-96, 1983 4 . Butler DL, Noyes FR, Waltz KA, Gibbons MJ: Biomechanics of human allograft treatment. Trans Orthop Res Soc 12:128,1987 5. Cabaud HE, Feagin JA, Rodkey WG: Acute anterior cruciate ligament injury and augmented repair. A m J Sports Med 8:395401, 1980 6. Cho KO:Reconstruction of the anterior cruciate ligament by semitendinosus tenodesis. J Bone Joint Surg [Am] 57:608612, 1975 7. Curtis RJ,Delee JC, Drez DJ: Reconstruction of the anterior cruciate ligament with freeze dried fascia lata allografts in dogs. Am J Sports Med 13:408414,1985 8. Dahners LE, Wilson CJ: An examination of the mechanism of ligament and joint contracture. Trans Orthop Res Soc 12:44, 1987 9. Frank C, Edwards P, McDonald D, Bodies D, Sabiston P: Viability of ligaments after freezing: an experimental study in a rabbit model. J Orthop Res 1:95-102, 1988 10. Frank C, Schachar N, Dittrich D, Shrive N, Edward GE: Normal ligament healing in a rabbit model. A morphological, histological, biochemical and biomechanical study. Trans Orthop Res Soc 6:197, 1981 11. Frank C, Woo SL-Y, Amiel D, Gomez MA, Hanvood FL, Akeson WH: Medial collateral ligament healing: a rnultidisciplinary assessment in rabbits. A m J Sports Med 11:379389, 1983 12. Goldberg VM, Lance EM: Revascularization and accretion in transplantation: quantitative studies of the role of the allograft barrier. J Bone Joint Surg [Am] 54:807-816, 1972 13. Grood ES, Butler DL, Noyes FR: Models of ligament repairs and grafts. In: AAOS Symposium on Sports Medicine: The Knee, ed by G Finerman, St. Louis, CV Mosby Co., 1985,pp 169-181 14. Heiple KG, Chase SW, Herndon CH: A comparative study of the healing process following different types of bone transplantation. J Bone Joint Surg [Am] 45:1593-1616, 1963 15. Hollis JM, Lyon RM, Marcin JP, Horibe S, Lee EB, Woo SL-Y: Effect of age and loading axis on the failure properties of the human ACL. Trans Orthop Res Soc 13:83, 1988 16. Jackson DW, Grood ES, Arnoczky SP, Butler DL, Simon TM: Freeze dried anterior cruciate ligament allografts: preliminary studies using a goat model. A m J Sports Med 4:295303, 1987 17. Jackson DW, Grood ES, Wilcox P, Butler DL, Simon TM, Holden JP: The effects of processing techniques on the mechanical properties of bone-anterior cruciate ligament-bone allografts: an experimental study in goats. A m J Sports Med 2:lOl-105, 1988 18. Nikolaou PK, Seaber AV, Glisson R, Ribbeck BM, Bassett FH: Anterior cruciate ligament allograft transplantation. Long-term function, histology, revascularization, and operative technique. A m J Sports Med 14:341360, 1986 19. Noyes FR, Butler DL, Grood ES, Zernicke, RF, Hefzy MS: Biomechanical analysis of human ligament grafts used in
J Orthop Res, Vol. 8, No. I , 1990
P . SABISTON ET AL.
56
20. 21. 22.
23. 24. 25.
26. 27.
knee ligaments repairs and reconstructions. J Bone Joint Surg [Am] 66:344-351, 1984 O'Donoghue, Rockwood CA, Frank GR, Jack SC, Kenyon R: Repair of the anterior cruciate ligament in dogs. J Bone Joint Surg [Am] 48:503-519, 1966 Pappas AM, Beisaw NE: Bone transplantation: correlation of physical and histologic aspects of graft incorporation. Clin Orthop 61:79-91, 1968 Paulos L, France E, Rosenberg T, et al: Comparative material properties of allograft tissues for ligament replacement: effects of type, age, sterilization and preservation. Trans Orthop Res Soc 12:129, 1987 Peer LA: Transplantation of tendons in animals. In: Transplantation ofTissues, ed by LA Peer, Baltimore, Williams & Wilkins, 1955, pp 227-295 Peer LA: Transplantation of tendons in humans. In: Transplantation ofli'ssues, ed by LA Peer, Baltimore, Williams & Wilkins, 1955, pp 296-334 Sabiston P, Frank C, Lam T, Shrive N: Allograft ligament transplantation: a morphological and biochemical evaluation of a medial collateral ligament complex in a rabbit model. Am J Sports Med (in press, 1989) Sabiston P, Frank C, Lam T, Shrive N: A biomechanical study of an extraarticular ligament autograft-an experimental study in a rabbit model. J Orthop Res (in press) Shino K, Kawasaki T, Hirose H, Botoh I, Inoue M, Ono K: Replacement of the anterior cruciate ligament by an allo-
J Orthop Res, Vol. 8, No. I , 1990
28. 29.
30.
31. 32.
33. 34.
genic tendon graft: an experimental study in the dog. J Bone Joint Surg [Br] 66:672481, 1984 Shrive NG, Lam TC, Damson E, Frank CB: A new method of measuring the cross-sectional area of connective tissue structures. J Biomech Eng 2:104-109, 1988 Thorson EP, Rodrigo JJ, Vasseur HA, Heitter DO: Comparison of frozen allograft versus fresh autogenous anterior cruciate ligament replacement in the dog. Trans Orthop Res Soc 12:65, 1987 Turner WD, Vasseur P, Gorek JE, Rodrigo JJ, Wedell JR: An in vitro study of the structural properties of deep-frozen versus freeze-dried, ethylene oxide-sterilized canine anterior cruciate ligament bone-ligament-bone preparations. Clin Orthop 230:251-256, 1988 Webster DA, Werner FW: Freeze-dried flexor tendons in anterior cruciate ligament reconstruction. Clin Orthop 181:238-243, 1983 Woo SL-Y, Gomez MA, Seguchi Y, Endo CM, Akeson WH: Measurement of mechanical properties of ligament substance from a bone-ligament-bone preparation. J Orthop Res 1:22-29, 1983 Woo SL-Y, Hollis JM, Roux RD, et al: Effects of knee flexion on the structural properties of the rabbit femurACLtibia complex (FATC). J Biomech 637-563, 1987 Yin FCP, Tompkins WR, Peterson KL, Intaglietta M: A video dimension analyzer. IEEE Trans Biomed Eng 19:37& 381, 1972
