By continuing to browse this site you agree to us using cookies as described in About Cookies
Wiley Online Library is migrating to a new platform powered by Atypon, the leading provider of scholarly publishing platforms. The new Wiley Online Library will be migrated over the weekend of March 17 & 18. You should not experience any issues or loss of access during this time. For more information, please visit our migration page: http://www.wileyactual.com/WOLMigration/
Meniscal injuries are common,1 and every effort is made to restore normal meniscal function. Removal of major portions of menisci is associated with local chondral damage in the meniscectomized2, 3 and the adjacent tibiofemoral compartments,4, 5 and in the patellofemoral joint.6 While medial meniscectomies are common, lateral meniscectomy is associated with poorer outcomes and more progressive cartilage damage,2, 7 requiring follow-up surgeries more often than medial meniscectomy.8 Nevertheless, the cause of chondral damage with either meniscectomy is thought to be altered biomechanics at the joint surfaces resulting from the absence of the meniscus.
In vitro partial or complete meniscectomy profoundly alters tibiofemoral contact stress in the meniscectomized compartment,9, 10 and the general thought is that the location of peak contact stress corresponds to the location of chondral damage in vivo. Despite this intuitive assumption, no consensus exists that confirms if, or where, the peak contact stress location migrates following meniscectomy,9, 11, 12 likely because no in vivo data exist upon which to model the orientation and position of the femur and tibia when the joint is compressed to simulate weight-bearing. Without this knowledge, the challenge remains to relate and validate results from experimental and theoretical models to determine where and why cartilage damage actually occurs in vivo.
Three studies suggested that knee kinematics and kinetics are different following complete or partial meniscectomy in either medial or lateral compartments compared to healthy knees in humans.13–15 However, none of these studies examined how related changes in alignment and joint moments affect tibiofemoral surface interactions.
Our objectives were to quantify in vivo 3D stifle kinematics before and after lateral meniscectomy in a sheep model and to determine the potential relationship between altered kinematics, the location of minimum tibiofemoral distance, and the corresponding spatial distribution of chondral damage. We hypothesized that complete lateral meniscectomy causes an immediate increase in joint abduction and shifts the lateral compartment in vivo locations of minimum tibiofemoral distance laterally. We also hypothesized that the regions of greatest cartilage damage are qualitatively associated with the new locations of minimum tibiofemoral distance.
Subjects and Surgical Approach
Nine skeletally mature female Suffolk-cross sheep (80.5 ± 8 kg) were allocated randomly into two groups: sham (n = 4; 5 sheep underwent sham surgery, but data from one sheep were discarded due to technical errors) or lateral meniscectomy (n = 5). We chose to remove the lateral meniscus because it carries a larger proportion of load across the lateral compartment relative to the load borne directly by cartilage.16 Also, chondral changes in sheep following lateral meniscectomy have been well characterized.17 Surgeries were performed under general halothane anaesthesia. To remove the lateral meniscus, an incision was made along the midline of the patellar tendon, the skin retracted, and fat, fascia, and the lateral joint capsule dissected to access the anterior and posterior horns of the lateral meniscectomy. The meniscal horns and meniscofemoral ligament were transected, and the meniscus removed. Sham surgery involved the same approaches as the experimental group including a brief, controlled, medial dislocation of the patella and dissection of the medial joint capsule. Before closure, the site was flushed with saline and Penicillin G Sodium (Novopharm Limited, Toronto, ON, Canada). Deep layers were closed with 2–0 Vicryl sutures (Ethicon Inc., Markham, ON, Canada), and the skin was closed in two layers with 2-0 PDS II (Ethicon). The skin was sprayed with Gentocin (Schering-Plough Animal Health Corp, Union, NJ) and Op-site (Smith & Nephew, Hull, UK). All procedures were approved by the University's Animal Care Committee and comply with Canadian Council on Animal Care guidelines.
Kinematic Data Collection
In vivo kinematics were collected using a rigid removable bone-mounted plate-post-maker assembly.18 The reproducibility of the post and marker fit for all 9 sheep was 0.16 ± 0.09 mm. A 4 camera video-based system (Motion Analysis, Santa Rosa, CA) was used to record (120 Hz) the 3D position of retroflective markers over 100 gait cycles (from hoof strike to hoof strike). Hind limb kinematics were recorded prior to injury, and at 2, 4, and 20 weeks post-surgery. The average root mean square error of the video analysis was 0.41 ± 0.13 mm. At 20 weeks, animals were euthanized via intravenous injection of euthanyl (Euthanyl, Bimeda-MTC, Cambridge, ON, Canada). Both hind limbs were disarticulated and then dissected.
At dissection, tibiofemoral cartilage was graded at 10 standard locations (Fig. 1A) based on anatomical landmarks and subjective boundary delineation (Fig. 1B). For the femur and tibia, the mediolateral axes were based on collateral ligament insertions. Tibial surfaces were divided into anterior, central, and posterior thirds along this axis based on the location of the tibial eminences. Femoral surfaces were divided into anterior and posterior halves where the mediolateral axis passed between the cruciate ligament origins. Grading was performed by a single, experienced technician whose grades were validated by comparing gross scores with the corresponding histopathology (Fig. S1). We standardized the grading location to map and quantify spatial patterns of damage consistently between sheep. A 6-point scale ranging from 0 – no damage, to 5 – large areas of exposed bone spanning >10% of the surface was used19 (Table S1). To account for inter-subject variation and after ensuring that no significant difference existed between grades of contralateral stifles, scores were expressed as the difference from the un-injured contralateral stifle at each location. Scores from all locations were also summed to create a composite score, providing an overall magnitude of cartilage damage in each joint.
Ground Reaction Force
To assess if plate implantation or meniscectomy reduced hind limb peak vertical ground reaction force (vGRF), sheep were led across an embedded force platform (Kistler Instrumente, Winterthur, Switzerland) until 20 hind limb hoof strikes were recorded at 1,200 Hz. vGRF was determined prior to surgical plate implantation, and then serially prior to each kinematic data collection.
Coordinate System and Surface Reconstruction
Following cartilage grading and collection of histology samples, remaining cartilage was digested using a papain solution. The posts and marker assemblies were secured to their respective bone plates, and the tibia and femur were each mounted in a stereotaxic frame. A hand-held coordinate measuring machine (CMM) (Faro Technologies, Lake Mary, FL; ±0.025 mm precision) was used to digitize ligament anatomical landmarks. Femoral and tibial coordinate systems were constructed based on the landmarks.18 Lastly, the 3D coordinates of the tibiofemoral surfaces were traced using the hand-held CMM.
Video-based data were filtered using a cubic generalized cross-validation filter with a 6 Hz low pass cut-off, and kinematics were expressed in a joint coordinate system.20 Using in vivo kinematics as input, the CMM traces of the femoral condyles were oriented relative to the tibial plateau and then a thin-plate spline was fit to the traced surfaces. Vectors normal to the curvature of the plateau were calculated, and the distance from the plateau to the location at which the normal vector intersected the condyle was determined. x and y locations on the plateaus and condyles at which this distance was minimal were “points of minimum tibiofemoral distance.”
The video-based tracking system error and plate-post-marker assembly reproducibility was similar to that reported previously (0.3–0.4° degrees in angular kinematics and 0.7–1.0 mm in joint translations18). To assess what effect this error had on the points of minimum tibiofemoral distance, angular and positional errors of 0.35° and 0.85 mm were carried through the calculations of minimum tibiofemoral distance at each of the 6 points within an average gait cycle for one animal. The change was 0.72 ± 0.68 mm (for all 4 tibiofemoral surfaces). Similarly, the change in minimum tibiofemoral distance was −0.09 ± 0.56 mm.
These mean errors represent a worst-case scenario; the angular and translational errors assume that the change in tibiofemoral alignment was the result of each bone incurring the maximal error that was equal and in the opposite direction (2 × RMS error of the video-based tracking system). The location calculation assumes that this maximal error occurred simultaneously in all degrees of freedom when calculating locations of minimum tibiofemoral distance and proximity.
Points of minimum tibiofemoral distance were chosen rather than a centroid of a region within some threshold of minimal distance video because a centroid does not account for areas that may experience load via contact with the meniscus.21 Presumably, if cartilage–cartilage contact is occurring, some load transmission or surface interaction at these locations exists.
To minimize inter-subject variability, kinematics were expressed as the change from the intact state. One-way repeated measures ANOVAs were used to assess changes in: flexion-extension (FE), ab-adduction (AA), internal-external rotation (ROT), medial-lateral translation (ML), anterioposterior translation (AP), and inferior-superior translation (IS). Eight points within the gait cycle were analyzed: hoof strike (HS), early weight bearing (EWB), full weight bearing (FWB), hoof-off (HO), early swing (ESW), peak swing (PSW), late swing (LSW), and peak extension (EXT) (Fig. S2). Points within the gait cycle were the within-subjects levels, while injury type was the between-subject factor, and time (intact, 2, 4, and 20 week) was the repeated measure. Planned linear contrasts were utilized to examine if kinematics changed over time.
A Mann–Whitney U-test was used to compare sham and experimental surgery based on composite cartilage scores. Differences in location-specific scores for gross morphology of meniscectomized and sham sheep were evaluated using a Kruskal–Wallis test, while Friedman exact tests were used to determine if significant differences existed in gross morphology scores across the standardized stifle locations within each group.
Two-way repeated measures ANOVA was used to assess differences in peak vGRF between groups and over time. Two-way repeated measures ANOVA was used to confirm significant changes in the location of points of minimum tibiofemoral distance at 6 points within the gait cycle: HS, EWB, FWB, HO, ESW, and LSW. The points were the within-subjects levels, injury status was one between-subjects factor, and time was the repeated measure and second between-subjects factor. Planned linear contrasts were also performed to assess changes over time. For all statistical tests, differences between sham and meniscectomy outcome measures were considered significant if p ≤ 0.05.
Hind Limb Ground Reaction Force
Peak vGRF significantly changed over time, but was variable within and between sheep. A reduction in hind right limb weight-bearing reached significance only in meniscectomized sheep immediately following surgery (−22 ± 36%), but by 4 weeks post-meniscectomy, vGRF had returned to normal (within 2 ± 37% of intact values). At the 2-week off-loaded time, the left limb saw only a 7 ± 30% increase in vGRF; given the large variation, the mean changes in the meniscectomized vGRFs were brief and increases in contralateral limb vGRF fell just outside of “normal” variation only at 2 weeks post-meniscectomy.
Stifle Kinematics are Altered by Injury
For all degrees of freedom, sham and meniscectomy kinematics were injury-specific. Resection of the lateral meniscus led to an abrupt, consistent, and permanent increase in abduction across the gait cycle (Fig. 2) among all meniscectomized sheep. Translational differences approached the resolution of our measurement system, but meniscectomy also led to an abrupt, consistent increase in medial tibial translation of 0.8 ± 0.05 mm (Fig. 3). Joint motion in the remaining degrees of freedom and across the 8 points within the gait cycle also differed significantly from intact; however, changes were minor (<6% of the intact range of motion) compared to the shift in medial translation and joint abduction (∼25% of the intact range of motion for each). Because we used each animal as its own control and could pool data from 30 strides from each animal, even small changes in joint motion following meniscectomy were significant. For this reason, joint motion was significantly abnormal in the remaining degrees of freedom, but did not necessarily represent “hallmarks” of meniscectomy-induced changes in joint kinematics.
Injury Changes the Locations of Points of Minimum Tibiofemoral Distance
The locations at which minimum tibiofemoral distance occurred were significantly different between meniscectomized and sham-operated stifles in both tibiofemoral compartments; the locations changed over time and the direction of change was unique to each group. In the lateral compartment, the points shifted laterally on both tibial and femoral surfaces immediately after meniscectomy, but then remained stable over time (Fig. 4A). Meniscectomy also resulted in a significant anterior shift in the locations in the lateral compartment (Fig. 4B). The peak lateral shift in location occurred at FWB for all meniscectomized sheep. When this dramatic shift was expressed as the percentage of the width of each sheep's joint surfaces, the shift corresponded to 45 ± 12% of the width of the lateral plateau and 40 ± 14% of the lateral condyle width. In shams, the peak medial shift in the medial compartment occurred at EWB (Fig. 5), corresponding to ±37 ± 24% and 30 ± 20% of each Sham's tibial and femoral surface widths, respectively.
Meniscectomy led to significantly more cartilage damage than sham surgery in all sheep: composite score = 8.6 ± 2.7 versus 0.3 ± 1.0 for sham. The worst damage in the meniscectomized stifles was focal, subject-, and location-specific, occurring in anterior and posterior regions of the lateral plateau and condyle. Meanwhile, the central region of the lateral plateau showed mild damage in 4 of 5 meniscectomized sheep, spanning a larger area that was more consistent between sheep (Fig. 6). All meniscectomized sheep also experienced some cartilage damage to the medial aspect of the medial femur and/or tibial eminence (Fig. 6).
Corresponding Locations of Minimum Tibiofemoral Distance and Cartilage Damage
In the meniscectomized compartment, the post-meniscectomy locations of minimum tibiofemoral distance corresponded to the subject-specific locations of most severe cartilage damage, while the medial compartment locations remained the same. However, proximity plots showed a “wedging effect” whereby the area where the medial tibial eminence and the medial femoral condyle are in close proximity is reduced following meniscectomy (Fig. 7). The location of focal damage over the medial eminence corresponds to the region of reduced proximity, but the regions of damage over the condyle correspond to regions of increased proximity during gait.
We sought to determine if dynamic in vivo 3D stifle kinematics would be altered by meniscectomy and if so, what relation exists between altered kinematics, the location of minimum tibiofemoral distance, and the corresponding spatial distribution of chondral damage. Our results support our hypotheses that lateral meniscectomy alters dynamic joint alignment, shifting the location at which tibiofemoral distance is at a minimum, and that the new locations of minimum distance qualitatively correspond to locations of chondral damage.
As expected based on the ovine meniscectomy model,17, 22 all meniscectomized sheep exhibited significant cartilage damage. In vitro studies9, 23–26 and analytical models have16, 27–29 shown that meniscectomy increases contact stress because of the reduced contact area. This in turn focuses the joint contact force over the locations undergoing contact in the meniscectomized compartment. Therefore, it would seem plausible that in the absence of a meniscus, in vivo locations of cartilage contact in the meniscectomized compartment likely undergo increased contact stress at locations where the distance between surfaces is minimal (i.e., locations of contact). Our results support this hypothesis; cartilage was damaged at the locations of minimum tibiofemoral distance, suggesting that increased contact stress is likely a primary driver of cartilage damage in this model.
An unexpected finding was that sham-operated sheep also exhibited mechanical effects of surgery. However, these sheep developed only minor tibial cartilage damage at locations unrelated to the new locations of minimum tibiofemoral distance. One explanation for this damage is that arthrotomy alone induced inflammation-mediated mild degradative changes. Medial joint capsule dissection may have also contributed to the slight medial shift in the locations of minimum tibiofemoral distance. Lastly, neuromuscular adaptation, pain, and scar tissue formation over the medial capsule could also contribute to the change in sham kinematics.
We also used the change in sham kinematics to strengthen our working hypothesis that contact stress, concentrated over regions of direct contact in the meniscectomized compartment,24, 27 could be the primary driver of cartilage damage in the meniscectomy model. Although our data cannot directly confirm this hypothesis, in the case of a meniscus-competent compartment we can infer that a change in tibiofemoral proximity and its location alters the joint contact stress in some way.21, 30 As expected, sham surgery did not cause the median tibiofemoral proximity to decrease from the intact state. Meanwhile, in the meniscectomized compartment, surface plots of the proximity, the location of reduced proximity, and location of focal cartilage damage suggest that the location of hypothesized increased contact stress qualitatively corresponds to locations of more severe damage (Fig. 8). The additional damage to the posterior aspect of the meniscectomized compartment likely reflects locations of tibiofemoral contact during activities other than walking. For example, when sheep rise from a prone position, the stifle is initially hyperflexed with contact occurring between the posterior aspects of the knee.31
Conversely, the meniscectomy-induced changes to the actual locations of minimum tibiofemoral distance are different from locations of peak contact stress predicted in vitro. For example, Cottrell et al.12 used in vivo kinematics from an intact sheep stifle (which were originally measured by our lab group) to drive a knee simulator. Their results predicted that lateral meniscectomy moves the location of peak contact stress medially towards the lateral tibial eminence in the meniscectomized compartment, which is opposite to our in vivo results. Thus our findings emphasize that future in vitro and modeling studies must incorporate meniscectomy-induced changes in joint alignment to predict accurately changes to both contact stress magnitude and location.
In the adjacent medial compartment of the meniscectomized stifle, our second hypothesis—that regions of chondral damage qualitatively coincide with post-meniscectomy locations of minimum tibiofemoral distance—held true for the locations of tibial but not femoral damage. Median change in proximity in the medial compartment was not different following meniscectomy; however, the area of close proximity was concentrated over the medial tibial eminence. If there is an approximation of two articulating surfaces coupled with a reduction in the area over which contact occurs (Fig. 7A), we could speculate that contact stress over the eminence may be greater and could cause damage. Conversely, the adjacent region of speculated increased contact stress on the medial condyle did not correspond to the central and peripheral regions of chondral damage on that condyle. Rather, regions of medial femoral damage more often coincided with regions where there was an abrupt decrease in tibiofemoral proximity (i.e., an increase in joint space) (Fig. 7B). Based on the spatial distribution of changes in tibiofemoral proximity in the medial femoral cartilage, an alternative hypothesis is that reduced contact stress32 may be causing cartilage atrophy.
The strengths of this study are: kinematics were subject-specific and were expressed as the change from a known, intact state; joint surface interactions were recorded with sub-millimeter accuracy; and locations of chondral damage were individually mapped and related to subject-specific changes in joint surface interactions.
In terms of limitations, sample sizes were small. However, we mitigated inter-subject variability by quantifying kinematics as the change from the intact state for each sheep, and we had enough animals to quantify the kinematic effects of meniscectomy statistically and to describe how those changes related to locations of chondral damage. Similarly for cartilage gross morphology, scores were expressed as the difference from intact, contralateral scores. Despite small sample sizes, statistical power for detecting between-group differences was 0.8–1 for all analyses. Further, our technique for measuring 3D stifle kinematics is invasive compared to the more recently developed dynamic model-based tracking technique.33 However, we showed good reproducibility in un-injured sheep,18 demonstrating that despite the invasiveness, kinematics are reproducible and changes over time are not an artifact of the methodology.
As a final limitation, we used subchondral bone surfaces and not actual cartilage surfaces in our calculations of minimum tibiofemoral distance. Changes in the magnitude of minimum tibiofemoral distance, because of either inclusions or exclusion of tibiofemoral cartilages in the outcome measure's calculations, would be most influenced in regions where cartilage is thickest. However, the locations of minimum tibiofemoral distance in the intact state were still centered over tibial and femoral regions where cartilage is thickest.22, 34 In vivo human studies also suggest that tibiofemoral contact occurs over regions where cartilage is thickest.35 In sheep, the cartilage is thickest over the tibial eminences (∼1.2–1.3 mm)22 and is more consistent across the condylar surfaces (∼1 mm).34 In an intact, un-injured state, we would expect locations of minimum tibiofemoral distance to coincide with locations of maximum cartilage thickness. This was in fact what we found: at full weight-bearing, the locations of minimum tibiofemoral distance of all 9 sheep were located over the tibial eminences and adjacent regions over the femoral condyles prior to surgical intervention. Therefore, we believe that the effect that meniscectomy had on the change in locations of minimum tibiofemoral distance was not affected by the exclusion of the relatively thin tibiofemoral cartilages in our calculations.
In summary, we have, for the first time in vivo, confirmed that meniscectomy significantly changes joint alignment and surface interactions in a sheep model, and that these changes are related to locations of subject-specific focal chondral damage. This new knowledge can be used to complement in vitro and modeling studies of joint contact stress so that they may represent the in vivo situation more accurately.
The authors gratefully acknowledge the following funding sources: the Canadian Institutes of Health, Alberta Innovates Health Solutions, the Canadian Arthritis Network, The Arthritis Society, the Natural Sciences and Engineering Council of Canada, and the Alberta Heritage Foundation for Medical Research ITG in OA. We also thank the following individuals for their technical contributions to this manuscript: C. Sutherland, L. Jacques, J. Brown, A. Isfeld and J. Ronsky. None of the authors have received financial support from affiliations that may be perceived as having biased the presentation of the data.