Ligament creep recruits fibres at low stresses and can lead to modulus-reducing fibre damage at higher creep stresses: a study in rabbit medial collateral ligament model

ELSEVIER

Journal of Orthopaedic Research 20 (2002) 967-974

Journal of Orthopaedic Research www.elsevier.com/locdte/orthres

Ligament creep recruits fibres at low stresses and can lead to modulus-reducing fibre damage at higher creep stresses: a study in rabbit medial collateral ligament model G.M. Thornton, N.G. Shrive, C.B. Frank * McCuig Centre f b r Joint Injury und Arthritis Reseurclz, University of Culgury, 3330 Hospitul Drive N W, Culgary, Alherta, Cunudu T2N 4 N I

Abstract

Ligaments are subjected to a range of loads during different activities in vivo, suggesting that they must resist creep at various stresses. Cyclic and static creep tests of rabbit medial collateral ligament were used as a model to examine creep over a range of stresses in the toe- and linear-regions of the stress-strain curve: 4.1 MPa (n = 7), 7.1 MPa ( n = 6), 14 MPa ( n = 9) and 28 MPa ( n = 6). We quantified ligament creep behaviour to determine if, at low stresses, modulus would increase in a cyclic creep test and collagen fibres would be recruited in a static creep test. At higher creep stresses, a decrease in measured modulus was expected to be a potential marker of damage. The increase in modulus during cyclic creep and the increase in strain during static creep were similar between the three toe-region stresses (4.1, 7.1, 14 MPa). However, at the linear-region stress (28 MPa), both these parameters increased significantly compared to the increases at the three toe-region stresses. A concurrent crimp analysis revealed that collagen fibres were recruited during creep, evidenced by decreased area of crimped fibres at the end of the static creep test. Interestingly, a predominance of straightened fibres was observed at the end of the 28 MPa creep test, suggesting a limited potential for fibre recruitment at higher, linear-region stresses. An additional 28 MPa ( n = 6) group had mechanically detectable discontinuities in their stress-strain curves during creep that were related to reductions in modulus and suggested fibre damage. These data support the concept that collagen fibre recruitment is a mechanism by which ligaments resist creep at low stresses. At a higher creep stress, which was still only about a third of the failure capacity, damage to some ligaments occurred and was marked by a sudden reduction in modulus. In the cyclic tests, with continued cycling, the modulus increased back to original values obtained before the discontinuity suggesting that other fibres were being recruited to bear load. These results have important implications for our understanding of how fibre recruitment and stress redistribution act in normal ligament to minimize creep and restore modulus after fibre damage. 0 2002 Orthopaedic Research Society. Published by Elsevier Science Ltd. All rights reserved.

Introduction Holden et al. [4] have shown recently that ligaments are subjected to a range of loads depending on the activity performed in vivo. Ligaments will respond to this environment through creep (increase in strain due to sustained and repeated stress), and thus ligaments need to resist creep over a range of stresses. The stress-strain behaviours of ligaments are non-linear, comprised of a toe-region (increasing modulus) and a linear-region (constant modulus). Stresses due to normal daily activities are generally considered to be in the toe-region and possibly in the early portion of the linear-region [3,9]. * Corresponding author. Present address: Department of Surgery, University of Calgary, 3330 Hospital Drive NW, Calgary, Alberta, Canada T2N 4N1. Tel.: +1-403-220-6881; fax: +1-403-283-7742. E-mail address: [email protected] (C.B. Frank).

Furthermore, Viidik [ 181 descriptively related the stressstrain curve to crimp pattern. He proposed that the toeregion represents the straightening out of crimp as collagen fibres are sequentially recruited t o bear load, whereas the linear-region occurs after the completion of recruitment and involves the stretching of straightened fibres. With this interpretation, the level of crimp is an index of collagen fibre recruitment. An important functional relationship between creep and fibre recruitment may be revealed, therefore, if creep responses and crimp patterns are measured at stresses in different regions of the stress-strain curve. Others have previously speculated that the creep of tendon is the result of a combination of collagen fibre creep and collagen fibre recruitment. Cohen et al. [2] postulated that the activation energy for creep Of tendon (T-jump analysis) was between the high activation energy for collagen fibril creep and the low activation

0736-0266/02/$ - see front matter 0 2002 Orthopaedic Research Society. Published by Elsevier Science Ltd. All rights reserved. PII: SO736-0266(02)00028-1

968

G.M. Thornton et al. I Journal of Orthopaedic Research 20 (2002) 967-974

energy for interfibrillar sliding due to straightening out of the wavy fibrils. In later studies [5,6], these investigators modelled this behaviour combining the activation energy for the toe-region (straightening out of wavy fibres) and the linear-region (fibril creep) based on different amounts of collagen fibre crimp. However, no morphological documentation of fibre crimp was made. We have previously found ligament creep behaviour to be significantly over-predicted if the inverse stress relaxation behaviour alone is used to predict creep [15]. Our speculation was that a stress-controlled test like creep would cause collagen fibre recruitment, and the resulting behaviour would be a combined effect of fibre creep and fibre recruitment. Thus, fibre recruitment acts to resist creep, effectively decreasing the creep response from the greater response of the fibres alone (inverse stress relaxation). We recently tested these assumptions in a structural model of ligament creep with results supporting this mechanism [16]. Therefore, our major purpose in this experimental investigation was to quantify both the creep behaviour and the changes in crimp patterns at stresses in different regions of the stress-strain curve of the rabbit medial collateral ligament (MCL) modeled above [ 161 with the goal of understanding the role of fibre recruitment during ligament creep. Our hypotheses were that collagen fibres would be recruited during a static creep test and that the area of crimped fibres would be decreased at the end of the test compared to the start of the test. Our second purpose was to examine the results of cyclic creep testing, carefully looking for evidence of fibre recruitment versus fibre damage over a range of stresses. Torp et al. [17] documented increased linear modulus of tendon in successive cycles during testing to lowstrain levels. Rigby [l 11 correlated similar modulus increases during cycling to improved orientation of the tendon structure using X-ray diffraction. In addition, Torp et al. [ 171 correlated decreased linear modulus with tendon damage in successive cycles during testing to high-strain levels. More recently, Wang and Ker [20] documented similar decreased modulus and tendon damage in high-stress creep test cycles. At lower stresses, the tendon resisted creep rupture. We were interested in examining if intact MCL would have similar cyclic properties to those found previously for the isolated tail tendon units described above. Our hypotheses were that ligaments would have an increase in cyclic modulus at lower stresses and that a decrease in measured modulus would be a potential marker of damage at higher stresses. Methods Creep test stresses were selected in the toe- and linear-regions of a typical normal skeletally mature rabbit MCL stress strain curve (Fig. 1): 4.1 and 7.1 MPa were in the toe-region of the curve, 14 MPa was at

120 100

4

/

0.05

0.00

0.10

0.15

0.20

0.25

Stram ( d m m )

Fig. I . Stress-strain curve of a typical normal rabbit MCL. Failure strength of sample shown was 104.9 MPa and average normal MCL failure strength was 95.4 =k 12.3 MPa (n = 15) [15].

the transition between the toe- and linear-regions and 28 MPa was in the linear-region (one-third of the average failure strength). Forty-two MCLs from 36 normal skeletally mature (one-year-old) female New Zealand white rabbits were used in this study; that is, only one MCL per rabbit was used from 30 animals and both MCLs were used from the remaining 6 animals (Figs. 2 and 3). These 42 MCLs were divided into three groups: creep analysis only (n = 26), crimp analysis only (n = 8) and creep analysis with subsequent crimp analysis (n = 8). In other words, 8 MCLs were common between the creep analysis (n = 34; Fig. 2) and the crimp analysis (n = 16; Fig. 3). For the creep analysis, the distribution was as follows: 4.1 MPa (n = 7), 7.1 MPa (n = 6), 14 MPa (n = 9) and 28 MPa (n = 6); with two sets of paired MCLs between the 7.1 and 28 MPa groups (Fig. 2). Another group of 6 MCLs tested at 28 MPa had mechanically detectable discontinuities during the creep test where there was a step increase in deformation with no change in the programmed force. Eight of the creep tested MCLs (2 MCLs from each group except 7.1 MPa) underwent crimp analysis following creep analysis: 4.1 MPa (n = 2), 14 MPa (n = 2), 28 MPa (n = 2), and 28 MPa with discontinuity ( n = 2). These samples were identified as “post-creep’’ samples in the crimp analysis (Fig. 3). For comparison, additional MCLs had crimp analysis at the start of the static creep test and were identified as “pre-creep” samples: 4.1 MPa (n = 2), 14 MPa ( n = 2), 28 MPa (n = 2). Crimp was also analyzed after establishment of “ligament zero” (n = 2) as a control. The 4.1 MPa pre-creep and 4.1 MPa postcreep samples were paired (Fig. 3). In addition, there was one pair

CREEP ANALYSIS MCLs (n=30 unpaired MCLs + n=4 paired MCLs) 4.1 MPa

*

7.1 MPa

n=4 unpaired MCLs + n=2 MC1.s paired with 28 MPa

1

14 MPa * n=9 unpaired M C l s 28 MPa *

n=4 unpaired MCLs + n=2 MCl,s paired with 7.1 MPa

Fig. 2. Creep analysis group designations. * n = 2 samples from each of these four groups underwent crimp analysis following creep analysis and were identified as “post-creep” samples in the crimp analysis groups (Fig. 3).

G M Thornton et u1 I Journul of Orthopaedic Rescorch 20 (2002) 967-974

I

CRIMP ANALYSIS n=16 MCLs (n-8 unpaired MCLs + n-8 paired MCLs)

969

I 1

CONTROL

CREEP TESTED

4.1 MPa

pre-creep

post-creep

pre-creep

pre-creep n=2 MC1.s (n=l paired wih posl-crcep)

n=2 MCLs

n=2 MCLs (n=l paired h i t h prc-crccp)

post-creep

*

n-2 MCLs paircd a i i h pre-crecp

4

*

post-creep with discontinuity

*

0-2 MCLS

(n=I paired with prc-crecp)

Fig. 3. Crimp analysis group designations. * n

=2

samples from each of these four groups were included in the creep analysis groups (Fig. 2).

between 28 MPa pre-creep and 28 MPa post-creep and one pair between 28 MPa pre-creep and 28 MPa post-creep with discontinuity. After sacrifice with pentobarbital (Euthanyl, 1.5 mllanimal, MTC Pharmaceuticals, Cambridge, Ontario, Canada), the hindlimbs of the animal were disarticulated at the hip and ankle. Rapid dissection of muscle and fascia from the femur and tibia (leaving the menisci, collateral and cruciate ligaments) occurred before the bones were transected 3 cm from the MCL insertions. Tissues were kept moist by intermittent application of phosphate buffered saline. The tibia was cemented into the upper grip of our test system. The grip was then attached to the 500 N load cell o n the hydraulic actuator of our MTS system (MTS Systems Corporation, Minneapolis, MN). The MCL was aligned with the load axis of the actuator, and load was zeroed to account for the weight of the specimen. The femur was cemented into the lower grip with the knee at 70” flexion, and displacement was zeroed. The M C L was tested with the joint at 70” flexion because this joint position was previously defined as an “anatomic” position for the rabbit MCL, the joint position naturally attained before mechanical testing when all muscles were removed and all ligaments and menisci remained intact [7]. Matyas et al. [8] documented differences in the crimp pattern of rabbit MCL midsubstance under no load. In flexion (150-130” flexion), the anterior region of the MCL had reduced crimp pattern and the posterior region had a regular crimp pattern, suggesting that anterior fibres were more taut in flexion compared to the posterior fibres. In extension ( 6 0 4 0 ” flexion), the situation was reversed, as posterior fibres were more taut compared to anterior fibres. At 70” flexion, we observed a consistent crimp pattern throughout the complete midsubstance. This joint position may therefore represent an “average” crimp appearance avoiding the extremes of full flexion and extension. Using 70” flexion avoids the known nonuniform loading created by the extremes of joint position. The joint underwent two cycles between 5 N of compression and 2 N of tension at 1 mmlmin; after which menisci, cruciate and lateral collateral ligaments were removed, leaving the isolated MCL. Additional compression-tension cycles were performed in which the second cycle ended at 0.1 N of tension, establishing “ligament zero”. The MCL length was measured at the transition between periosteal and ligamentous tissue at both femoral and tibia1 insertions using Vernier calipers (accurate to 0.01 mm). For clearance of our area caliper [13], a small portion of the medial femoral condyle lateral to the M C L and distal to the femoral MCL insertion was removed. Then, 5 N of tension was maintained, and the caliper traversed the midsubstance of the ligament. The cross-sectional area was calculated by integrating the trace of tissue thickness versus caliper position (caliper accurate t o 0.0 I mm2). Our customized environment chamber (37 “C and 99% relative humidity) [22] was then installed, and compression-tension cycles were used to re-establish “ligament zero”. In each cyclic creep test, the ligament was loaded for 30 cycles at 1 Hz from “ligament zero” to the prescribed stress for that particular test. For each static creep test,

the ligament was then loaded immediately to the same stress used in the cyclic creep test and held in load control for 20 min. Creep stress was controlled by applying a load calculated from the midsubstance cross-sectional area for each ligament. Strain was defined as ligament deformation divided by the undeformed ligament length. Total creep strain was defined as the increase in strain from the peak of the first loading cycle in the cyclic creep test to the end of the 20-min static creep test. Thus, total creep strain included the increase in strain from both cyclic creep and static creep as well as the small strain increase that occurred during the 3 s that the MTS system required to switch from the cyclic command to the static command. Cyclic creep strain was defined as the increase in strain from the peak strain of the first cycle to the peak strain of the thirtieth cycle. Static creep strain was defined as the increase in strain from the beginning of the constant stress application to the end of the 20-min constant stress application. Multiple linear regression with random effects was used to compare the total creep strain at the different creep test stresses [14]. The main effect was creep test stress, which had four levels (4.1, 7. I , 14 and 28 MPa), and the random effects accounted for the two paired samples between 7.1 and 28 MPa groups. Cyclic creep strain and static creep strain were analyzed using the same methods as total creep strain. The cyclic creep test was further analyzed determining the cyclic modulus of the first and thirtieth cycle (i.e. the slope of the stress-strain plot). The tangent modulus of the loading curve was taken over the last 80% of the stress range ensuring that Y’ was greater than 0.99. Multiple linear regression with random effects was used to analyze the modulus data [14]. The main effects were creep test stress, which had four levels (4.1, 7.1, 14 and 28 MPa), and cycle, which had two levels (cycle 1 and cycle 30). The regression model included the interaction between these main effects (cycle and creep test stress). The model also accounted for the random effects of the two paired samples between 7.1 and 28 MPa groups and the repeated measures on the same sample for the different levels of the cycle effect. If the interaction was significant between the main effects, linear contrasts would be required for the multiple comparisons of interest between the different levels of the main effects. Linear contrasts provide an exact p-value for the specific comparison of interest. If there was a discontinuity in any cyclic creep curve, the average modulus of the two cycles before and the two cycles after the cycle involving the discontinuity were compared using a Student’s t-test for paired samples. For all statistical analyses. significance was defined at p < 0.05. Separate MCLs designated for crimp analysis underwent the creep protocol previously described (Fig. 3). Crimp was assessed at two test timepoints at each of the 4.1, 14, and 28 MPa creep stresses: at the start of static creep (pre-creep; n = 2 for each stress) and at the end of static creep (post-creep; n = 2 for each stress). Crimp was also assessed after establishment of “ligament zero” ( n = 2) as a control. At the desired test timepoint, MCLs were snap-frozen (washed for 20 s with liquid nitrogen) while mounted in the MTS. This facilitated rapid freezing of

G. M . Thornton el al. I Journal

970

01Orthopedic

Reseurcii 20 (2002J 967-974

3.0

$+

2.5

1 ,

0 cyclic creep strain Ncommand switch static creep strain

I

I

2.0 Type I Crimp

Type I1 Crimp

Type 111 Crimp

%ihstdntla~Crimp

Intermediate Crimp

Minimal Crimp

Fig 4 Crimp analysis cdtegories Image sues 140 x 140 pm2

the ligament in the loaded state to obtain a “snapshot” of crimp at the desired test timepoint; however, only the midsubstance section could be harvested. The midsubstance was then quickly embedded in frozen sectioning media (TissueTek O.C.T. Compound, Sakura Finetek, Torrance, CA). Ligament tissue was cut in 10 pm thick sagittal sections and stained using hematoxylin and Sirius Red. Midsubstance sections were completely analyzed in two dimensions (anterior-posterior and superior-inferior). Generally, only midsubstance sections from the middle of the M C L (moving from medial to lateral) are analyzed in histological protocols from this laboratory because of the consistency of serial sagittal section appearance. Around 20-30 midsubstance sagittal sections per ligament were prepared. All sections were examined and those with artifacts from preparation were removed. Of the remainder, one section was chosen randomly and analyzed for percent crimped area. All sections were examined in a blinded fashion using polarized light microscopy and VIDAS image analysis software (VIDAS 2.1, Kontron Electronik GmbH, Eching, Germany). Percent crimped areas were measured based on three categories (Fig. 4). Type I crimp or substantial crimp was characterized by a regular banding with a zigzag waveform. Type I 1 crimp or intermediate crimp had a reduced crimp angle and period. Type 111 crimp or minimal crimp was characterized by straightened fibres. The striking difference between the categories was adequate to distinguish the crimped areas visually. Nonetheless, Types I and I1 crimp were further classified: Type 1 had a crimp period range of 70 20 pm and crimp angle range of 35-20” and Type I1 crimp had crimp angle less than 20” and period less than 20 pin. One section per Iigament was analyzed by examining a grid of fields of view that covered the complete section in two dimensions. For each field, areas containing Type 1 and then Type I11 were defined. Next, the area of the total section in the analysis field was defined. Finally, the area of Type I1 was calculated as the difference between the area of the total section from the combined area of Types I and 111. The values of the three types of crimp for the complete section were calculated by summation of the values for all the analysis fields required to cover the complete section. The percent area of the complete section corresponding to the three categories was then calculated. Percent crimped areas were compared using Student’s (-tests for paired samples at 4.1 MPd and Student’s t-tests for independent samples at 14 and 28 MPa. We required n = 2 independent samples to detect a percent area difference of 30% between pre-creep to post-creep, assuming a 10% standard deviation (standardized difference = 0.33, power = 0.80 and alpha = 0.05). At 4.1 MPa, where we predicted smaller differences in crimp pattern, we required n = 2 paired samples to detect a percent area difference of IS’% between pre-creep to post-creep, assuming a 10% standard deviation of the changes (standardized difference = 0.33, power = 0.80 and alpha = 0.05).

Results The total creep strain of normal MCLs tested at 4.1 and 7.1 MPa (toe-region stresses) were not significantly different (Fig. 5). At 14 MPa, the stress at the transition between the toe- and linear-regions of the stress-strain curve, the total creep strain was greater than at 4.1 and 7.1 MPa (p < 0.03) and was significantly less than at 28

7.1 14 Creep Test Stress (MPa)

4.1

28

Fig. 5. Three components of creep strain (total, static and cyclic) of normal MCLs at various creep test stresses. Total creep strain includes the strain increase from both cyclic and static creep as well as the small strain increase that occurred during the 3 s that the MTS system required to switch from the cyclic command t o the static command. “s.d.” indicates standard deviation. (a) All creep components (total, static and cyclic) greater than at the three lower stresses ( p < 0.01). (b) Cyclic creep 0, < 0.002) and total creep (p < 0.03) greater than at the two lower stresses. Note, however, that the static creep strains at 4. I , 7.1 and 14 MPa were not significantly different.

MPa ( p = 0.001). The cyclic and static components of creep at 14 MPa account for this behaviour. The static creep strain at 14 MPa was not different to the static creep strain at 4.1 and 7.1 MPa. The cyclic creep strain at 4.1 and 7.1 MPa were not significantly different; however, the cyclic creep strain at 14 MPa was increased over that which accrued at the two lower stresses ( p < 0.002). At 28 MPa, the linear-region stress, all of the components of creep (total, static and cyclic) were greater than at the three lower stresses 0, < 0.01). The cyclic modulus increased from cycle 1 to cycle 30 at all creep test stresses 0,< 0.001; Table 1). The increase in modulus was similar at the three lower stresses (4.1, 7.1 and 14 MPa), but significantly increased at Table 1 Cyclic modulus of normal MCLs at various creep test stresses Creep test stress (MPa)

n

Cyclic modulus at cycle Id (MPa)

Cyclic modulus at cycle 30h (MPa)

Increase in modulus from cycle 1 to cycle 30 (MPa)

4.1 7.1 14 28

7 6 9 6

363f35 512i38 579+51 810f88

399 f 37 550 i49 627 f 52 907 f 77

36& 12 39fll 4 8 i 12 97 i47d

*

Data are shown as mean standard deviation. “Cyclic modulus at cycle 1 was different for each norma\ group compared to all other normal groups (p < 0.001). bCyclic modulus at cycle 30 was different for each normal group compared to all other normal groups ( p < 0.001). ‘Cyclic modulus at cycle 30 was greater than at cycle 1 for all groups (p < 0,001). Increase in modulus at 28 MPa was greater than for all three lower stresses 0, < 0.001).

G. M . Thornton et id. I Journrtl of’ Ortliopnedic Rrseordi 20 (2002) 967-974 32 0

(4

(b)

I Modulus IMPd) I

0 00

0 02

0 04

0.06 0 08 Stiain (rnnlimm)

0.10

0 I2

0.00

0.02

0.04

006 0.08 Stram (mdmm)

0.10

0.12

Fig. 6. (a) Cyclic plot of a representative normal MCL creep tested at 28 MPa and (b) cyclic plot of normal MCL creep tested at 28 MPa with discontinuity at cycle 6 (C6). The modulus value increased after the discontinuity such that by the end of the test the modulus was similar to the modulus before the discontinuity.

28 MPa (p < 0.001). None of the normal MCLs tested at the lower stresses had discontinuities in the creep curves whereas 6 of 12 MCLs tested at 28 MPa had discontinuities (Fig. 6). Of the 6 MCLs creep tested at 28 MPa

Fig. 7. Crimp images and percent crimped areas of MCLs creep tested at 4.1 and 14 MPa. Image sizes are 280 x 280 pm’. representing approximately 1% of the complete section. Percent crimped areas are shown as mean +/- standard deviation. (a) Increased area of straightened fibres (Type 111 crimp) and decreased area of crimped fibres (Types I and 11 crimp) compared to pre-creep ( n = 2; p = 0.05). (b) Increased area of straightened fibres (Type 111crimp) and decreased area of crimped fibres (Types I and I1 crimp) compared to pre-creep ( n = 2; p = 0.03).

97 I

that had discontinuities in the creep curves, three of these occurred during cyclic creep. The cyclic discontinuity was followed by a significant reduction in modulus, 4.6 41 2.9% (p < 0.06; Fig. 6b). In all three cases, the discontinuity occurred before cycle 15, and the modulus at cycle 30 was similar to the modulus before the discontinuity occurred; hence, examining the modulus at only the first and thirtieth cycles would not have revealed the discontinuity. Crimp analysis was performed at the start and end of the static creep tests with the MCL held at the creep test stress. Thus, fibre recruitment during creep was quantified comparing crimp distributions pre-creep and postcreep. The crimp pattern of the control MCLs harvested at “ligament zero” (0.1 N of tension) had very few straightened fibres, 1.7 41 1.6%)(Type 111 crimp). Interestingly, significant increases in straightened fibres (Type 111 crimp) were observed post-creep at both 4.1 MPa (22.5 i0.1‘%) and 14 MPa (75.8 & 9.0%) (Fig. 7). At the linear-region stress, 28 MPa, an almost extinguished crimp pattern was observed post-creep, 93.5 f 0.2‘%) straightened fibres (Type I11 crimp; Fig. 8). Thus, all groups had significant decreases in crimped fibres (Types I and I1 crimp) and significant increases in straightened fibres (Type I11 crimp) post-creep (p < 0.05). All groups, therefore, exhibited fibre recruitment

Fig. 8. Crimp images and percent crimped areas of MCLs creep tested at 28 MPa with and without discontinuity. Image sizes are 280 x 280 pm’, representing approximately 1% of the complete section. Percent crimped areas are shown as mean +/- standard deviation. (a) Increased area of straightened fibres (Type 111 crimp) and decreased area of crimped fibres (Types 1 and 11 crimp) compared to pre-creep ( n = 2; p = 0.01). (b) Decreased area of straightened fibres (Type I11 crimp) and increased area of crimped fibres (Types I and I1 crimp) compared to post-creep without discontinuity ( n = 2; p = 0.03).

972

G M Thornton et ul I Journul of Orthopedic Research 20 (2002) 967-974

and fibre straightening during creep, though recruitment was limited at 28 MPa given the predominance of fibre straightening: 40% of fibres already straightened at the start of creep and 93% straightened by the end of the creep test. The static creep strain of MCLs tested at 28 MPa with discontinuity in the creep curve (3.34 f 1.74%))was significantly greater those tested at 28 MPa without discontinuities (1.22 i0.34%) (p = 0.02). It is interesting to note that the crimp patterns of the MCLs with discontinuity had more crimped fibres and less straightened fibres than the MCLs from the tests that had no discontinuities (Fig. 8), presumably due to a “recoiling” of some fibres after their rupture.

Discussion These results reveal five important findings. First, these results show that normal ligaments exhibited creep behaviour with relative insensitivity to creep test stress at stresses in the toe-region of the stress-strain curve. Using total ligament cross-sectional area to calculate stress, a threefold increase in stress had minimal effects on creep behaviour: the increase in modulus over 30 cycles was similar from 4.1 to 14 MPa and the static creep strain was similar over these same stresses. Static creep strain and increase in cyclic modulus were significantly larger at a linear-region stress, 28 MPa. This insensitivity to increased creep stress is atypical of viscoelastic materials. For linear viscoelastic materials, one expects that if the creep test stress was doubled, the creep strain would double (constant ratio between stress and strain at a given time), or, for non-linear viscoelastic materials, the creep strain would increase with each increase in stress. Hence, ligaments have an elegant structural response that minimizes creep at toe-region stresses. Second, results showed that fibre recruitment (as measured by crimp) could well have an important role in this minimization of creep at toe-region stresses. The concurrent crimp analysis also provided a potential explanation for the increase in static creep at a stress in the linear-region of the stress-strain curve, 28 MPa. At the lower stresses, the crimp analysis showed significant increases in straightened fibres and decreases in crimped fibres post-creep, indicating that collagen fibres initially loaded were straightened and others were recruited during creep. Creep will be minimized as the load is distributed over a progressively increasing load-bearing area involving the fibres initially loaded and those progressively recruited. Additionally, the stress on the initially loaded fibres will be reduced as the load-bearing area increases, hence reducing the likelihood of fibre rupture. At the higher stress (28 MPa), straightening of fibres predominated with the crimp pattern being almost

extinguished post-creep, indicating that at this stress (in the linear-region) there was a limited capability for fibre recruitment. If recruitment is no longer available, creep will increase because the load can no longer be shared over an increasing load-bearing area. These results suggest that collagen fibre recruitment is a mechanism for minimizing creep at lower (working) stresses and such recruitment is limited at higher (linear-region) stresses. Further, the reduced capacity for stress redistribution at 28 MPa may increase the likelihood of fibre rupture. Indeed, the third important finding was that, in addition to an increased creep response, damage can occur to normal rabbit MCLs exposed to repetitive and sustained 28 MPa creep test stress. Of the 12 samples creep tested at 28 MPa, six had mechanically detectable discontinuities in the measured creep strain (three in cyclic creep and three in static creep). The discontinuity caused a step increase in deformation and a concomitant reduction in modulus. Changes in water content or in binding of water to proteoglycans in the ground substance are unlikely to happen so abruptly and dramatically. Changes in connections between fibres and the ground substance, changes in fibre interconnections and fibre rupture may all be mechanisms of the observed damage. However, such a rapid and large change in tissue elongation is most likely a result of fibre rupture. Interestingly, this discontinuity caused a sudden decrease in modulus, and eventually the modulus increased back to values that were measured before the discontinuity. This is likely explained by immediate load transfer from broken fibres to surrounding unbroken fibres and the subsequent load sharing with other fibres through creep by the mechanism described earlier; a very functional “compensatory” mechanism to redistribute stress and minimize overall tissue creep. Viidik [ 181 interpreted the toe-region of the stress-strain curve, characterized by increasing modulus with increasing strain, to represent the gradually increasing recruitment of collagen fibres. The corollary therefore is that a reduction in the number of fibres carrying load will result in a reduction in modulus. Also, Yahia et al. 1231 correlated rupture of thick collagen fibres to mechanically detectable discontinuities in the stress-strain curve of adolescent rabbit MCLs. In that case, the discontinuity was a rapid increase in strain with a corresponding decrease in stress after which the modulus returned to a positive but smaller magnitude than before the discontinuity. Given the similarity with our results, fibre rupture is therefore the most likely cause of the discontinuity we observed, with further subsequent fibre stretching and load sharing providing the subsequent increase in modulus. MCLs with mechanically detectable discontinuities in the 28 MPa static creep tests had greater percentage of crimp than MCLs with no discontinuities which may be due to “recoiling” of fibres after rupture. Crimp of fibres after rupture were easily distinguished from actual

G. M. Thornton ct crl. I Jvurnal

of

Orthopcierlic Rwenrcli 20 (2002) 967-974

functional crimp. Although creating a new crimp category for crimp after rupture would have been instructive, these samples were analyzed using the three categories described previously with the following modifications: Type TI crimp had superficial and irregular banding with somewhat larger crimp period and Type I crimp occurred only in a particular fibre (not across a group of fibres) because individual fibres were separated from adjacent fibres. For example, crimp analysis of an MCL stretched to failure revealed 51.6% Type I crimp and 48.4‘1/0 Type TI crimp using these definitions. Interestingly, Viidik [ 181 had previously documented a slight waviness to tendon collagen fibres that had ruptured at low load levels (one-third of the maximum load); in addition, fibres were noted to glide apart after rupture. This evidence further supports that mechanically detectable discontinuities and subsequent modulus reduction during creep testing are apparently due to damage causing fibre ruptures. The fourth interesting finding is the importance of stress distribution in ligaments. The creep test stress at which discontinuities occurred was 28 MPa based on total cross-sectional area, only one-third of the failure strength of the MCL. This suggests that different fibres in the cross-section were at different stresses, including stresses high enough to cause fibre rupture. Area measurements based on total ligament cross-section therefore clearly oversimplify stress distribution and underestimate the peak stress on some recruited fibres. Furthermore, although some of the 28 MPa cyclic creep curves had discontinuities, none resulted in total ligament failure. Rather, the modulus increased back to the value that was obtained before the discontinuity. Such “microfailures”, if they do occur in functioning ligaments in vivo, must induce some repair. After the rupture of one fibre, stress on the other fibres will be increased causing increased deformation (and recruitment, if available) of additional fibres to carry the load. Thus, the stress would be effectively reduced on the nonruptured fibres in the original load-bearing area. As noted above, this evidence gives a new perspective on the ability of a normal ligament to redistribute stress after a fibre rupture and regain normal creep behaviour. Fifth, all these results (mechanical discontinuities, reduction in modulus and damage) suggest that there is some relationship between creep and fatigue at high stresses in ligament. Conventional definitions for creep and fatigue were developed studying engineering metals where the two could be easily separated: creep resulting from sustained loading and fatigue resulting from load fluctuating about a mean of zero load. For viscoelastic materials, these definitions are not as easily applied. Based on the above definitions, the cyclic creep test used here would combine both creep and fatigue. In tendon, Wang et al. [21] found through similar cyclic tests that tendon failure was not an isolated result of either creep

913

or fatigue. Recently, both have been recognized to cause damage and have a complex interaction in bone [l], tendon [20,21] and fibre-reinforced composites [lo, 121. Clearly, the new evidence presented here suggests that there is a complex interplay between creep and fatigue in ligament. This study had some limitations. We tested at only one joint angle, which had an “average” crimp pattern between the extremes of flexion and extension; however, we would still expect to see increased fibre recruitment (decreased crimped area) after static creep testing at any joint angle. We completely analyzed midsubtance sections in only two dimensions. However, due to the consistency of serial sagittal section appearance, the principle of increased fibre recruitment during creep would likely also apply throughout the ligament in the third dimension (medial-lateral or surface-to-deep). We documented crimp after fibre rupture but did not create a priori a Type IV crimp category. Since the goal of this study was to compare crimp before and after creep testing, we required a “snapshot” of crimp immediately upon reaching the desired creep stress. Although our post-creep observations of crimp are consistent with Viidik’s [ 181 description of crimp in the different regions of the stress-strain curve, our pre-creep observations at the same stresses have more crimp than that suggested by his description. More recently, Viidik [19] indicated that the transition between structural events and corresponding regions of the stress-strain curve were gradual. Our rapid loading and capture of crimp that was required for the current study were clearly faster than Viidik’s gradual straining method [18], one likely reason for the differences observed. Equally important, while we tested intact ligament, Viidik [18] used tendons that were cut parallel to fibre bundles obtaining specimens with fibre bundles visible on the surface. The fibre recruitment of an intact ligament could clearly be different to that of a sectioned tendon. In conclusion, this is the first work to quantify creep behaviour of knee ligaments over a range of stresses. Likewise, it is the first work to semi-quantify crimp distributions in an intact ligament. These new findings taken together show that fibre recruitment is a mechanism for minimizing creep of ligaments at low, toeregion stresses. This is a cleverly designed functional response that likely serves to prevent joint instability (by minimizing ligament elongation) and ligament fibre rupture (by progressively reducing the stress on the initially loaded fibres) during normal activity. Fibre recruitment and, correspondingly, resistance to creep was limited at higher, linear-region stresses. At these higher stresses, mechanically detectable discontinuities were related with morphological changes. Thus, a sudden reduction in modulus appears to be a valid indicator for damage in ligament. Creep testing at higher stresses also revealed that using total cross-sectional

G. M. Thornton e l at. 1 Journal of’ Ortlzopurdic Researdi 20 (2002) 967-974

974

area to calculate stress oversimplifies the actual stress distribution and that stress redistribution may aid in preventing total ligament failure. These observations have important implications for our understanding of how ligaments resist creep and how stress is redistributed in ligaments through fibre recruitment. Acknowledgements

This work was supported by the Canadian Institutes of Health Research, the Canadian Arthritis Society, the Alberta Heritage Foundation for Medical Research and the McCaig Fund. The authors gratefully acknowledge the assistance of V. Stagg for the statistical analysis. References Caler WE. Carter D R . Bone creep-fatigue damage accumulation. J Biomech 1989;22:625-35. Cohen RE, Hooley CJ. McCrum NG. Viscoelastic creep of collagenous tissue. J Biomech I976;9: 175-84. Haut RC. The mechanical and viscoelastic properties of the anterior cruciate ligament and of ACL fascicles. In: Jackson DW, Arnoczky SP, Woo SL-Y. Frank CB, Simon T M , editors. The Anterior Cruciate Ligament: Current and Future Concepts. New York: Raven Press; 1993. p. 63-73. Holdeii JP, Grood ES, Korvick DL, Cummings JF. Butler DL, Bylski-Austrow DI. In vivo forces in the anterior cruciate ligament: direct measurement during walking and trotting in a quadruped. J Biomech 1994;27:517-26. Hooley CJ, Cohen RE. A model for the creep behaviour of tendon. Int J Macromol 1979;l: 123-32. Hooley CJ, McCrum NG, Cohen RE. The viscoelastic deforniation of tendon. J Biomech 1980;13:521-5. Lam T. The mechanical properties of the maturing medial collateral ligament. PhD Thesis, University of Calgary, 1988. Matyas JR, Chowdhury P, Frank CB. Crimp as an index of ligament strain. In: Proceedings o f t h e 22 rd Annual Meeting of the Canadian Orthopedic Research Society, Ottawa, Ontario. June 1988. p. 113. Noyes FR, Butler DL, Grood ES, Zernicke RF. Hefzy MS. Biomechanical analysis of human ligament grafts used in knee-

ligament repairs and reconstructions. J Bone Joint Surg [Am] 1984;66:344-52. [IO] Perreux D, Joseph E. The effect of frequency on the fatigue performance of filament-wound pipes under biaxial loading experimental results and damage model. Composites Science and Technology 1997;57:353-64. [ l l ] Rigby BJ. Effect of cyclic extension in the physical properties of tendon collagen and its possible relation to biological ageing of collagen. Nature 1964;202:1072-4. [I21 Shenoi RA, Allen HG. Clark SD. Cyclic creep and creepfatigue interaction in sandwich beams. J Strain Analysis 1997;32: 1-18. [13] Shrive NG, Lam TC, Damson E, Frank CB. A new method of measuring the cross-sectional area ol‘connective tissue structures. J Biomech Eng 1988;110:104-9. [I41 Stata Statistical Software: Release 6.0. Stata Corporation: College Station, 1999. [I51 Thornton G M , Oliynyk A, Frank CB, Shrive NG. Ligament creep cannot be predicted from stress relaxation at low stress: a biomechanical study of the rabbit medial collateral ligament. J Orthop Res 1997;15:652-6. [I61 Thornton G M , Frank CB, Shrive NG. Ligament creep behaviour can be predicted from stress relaxation by incorporating fibre recruitment. J Rheology 2001 ;45:493-507. [I71 Torp S , Arridge RGC, Armeniades CD, Baer E. Structureproperty relationships in tendon as a function of age. In: Atkins EDT, Keller A. editors. Structure of Fibrous Biopolymers. Butterworths, London: The Colston Research Society: 1975. p. 197--221. [I 81 Viidik A. Simultaneous mechanical and light microscopic studies of collagen fibres. Z Anat Entwicklungsgesch 1972;136:204 12. [I91 Viidik A. Structure and function of normal and healing tendons and ligaments. 111: Mow VC, Ratcliffe A, Woo SL-Y. editors. Biomechanics of Diarthrodial Joints. vol. 1. New York: Springer Verlag; 1990. p. 3-38. [20] Wang XT, Ker RF. Creep rupture of wallaby tail tendons. J Exp Biol 1995;198:83145. [21] Wang XT, Ker RF, Alexander RM. Fatigue rupture of wallaby tail tendons. J Exp Biol 1995;198:847--52. [22] Wilson A, Shrive N , Damson E, Frank C, Leask G , Mikalsoii I, et al. An environment chamber for soft tissue testing. In: Proceedings of the 1995 Bioengineering Conference ASMEBED. vol. 29, 1995. p. 255-6. [23] Yahia L-H, Brunet J , Labelle S, Rivard C-H. A scanning electron microscopic study of rabbit ligaments under strain. Matrix 1990; 10:58-64.