|Home | About | Journals | Submit | Contact Us | Français|
Joint instability is presumed to cause abnormality in cartilage contact mechanics, which accumulatively damages articular surface over years, leading to osteoarthritis. The purpose of this study was to clarify the effect of instability on dynamic cartilage contact mechanics. Using human ankle cadaver specimens, potentially unstable ankles were modeled by introducing a coronally-directed step-off incongruity of the anterior tibial surface and/or by transecting the anterior talofibular ligament. Specimens were subjected to a duty cycle with quasi-physiologic stance-phase motion and loading. AP tibial forces were modulated, causing a controlled, quantifiable ankle subluxation during the duty cycle. Instantaneous changes in local articular contact stresses in the ankle were continuously measured using a thin and flexible pressure transducer. Tests were repeated while varying the tibial surface condition (anatomic, 1mm step-off, and 2mm step-off), both before/after transection of the anterior talofibular ligament, with various AP force magnitudes, so that situations of various degrees of instability were created for each specimen. Instability events occurred in ankles when the step-off incongruity was introduced, with the abnormality in joint kinematics being greater after ligament transection. Contact stress data revealed that these instability events involved distinctly abrupt increases/decreases in local articular contact stresses, and that the degree of abruptness was correlated very nearly linearly with the abnormality in joint kinematics. The severity of cartilage contact stress aberration appeared to be correlated closely with the degree of instability. Given this linear relationship, even small instability events presumably involve appreciable abnormality in dynamic joint contact mechanics.
Joint instability has been implicated closely in the etiology of post-traumatic osteoarthritis (OA). Normally, articular surface geometry in concert with ligamentous restraints provides passive joint stability, while allowing appropriate laxity, guiding smooth joint motions throughout physiologic activities. Accordingly, any dysfunction of these geometric and ligamentous restraints can cause more or less abnormality in joint kinematics. Even minor abnormalities, if not compensated for, and if habitual, may cause cumulative damage to cartilage leading to gradual disease progression over years. However, the pathomechanical link between instability and cartilage degeneration is still unclear.
The most prominent example of instability-associated cartilage degeneration occurs in human knees with anterior cruciate ligament (ACL) deficiency. Although previous gait analyses have documented several kinds of abnormalities in those knees [1-4], instability events causing habitual cartilage damage have not yet been identified. Some investigators have hypothesized that joint looseness causes macroscopic spatial shift in the articular contact location to non-physiologic contact regions, initiating a pathomechanical cascade leading to cartilage degeneration [5-7]. Given that cartilage degeneration develops in ACL reconstructed knees [8-10], pathomechanical loading likely occurs even in joints with subtle looseness. Another possible explanation – not necessarily exclusive – is that relatively small (sub-measurable level) instability events (i.e., ‘microinstability’) cause habitual exposure of local cartilage to supra-physiologically high rate loading. Several studies have shown that articular cartilage is particularly sensitive to pathologic changes in loading rate (experimentally, cartilage loading with elevated rates causes cartilage damage and biosynthetic dysfunction) [11-16]. Therefore, it is plausible that instability-associated increases in cartilage loading rates play an appreciable, and perhaps predominant, role in the etiology of post-traumatic OA.
The purpose of the present study was to clarify the relationship between joint instability and aberrations in dynamic cartilage contact mechanics. The joint of interest was the human ankle. In this joint, intraarticular fractures pose a high risk of post-traumatic OA [17,18], whereas the prevalence of primary OA is low . Given that passive ankle stability is primarily dependent on articular surface geometry [20-22], it is logical that altered joint geometry, such as what might occur after an intraarticular fracture, can cause joint instability. A recent experimental study, using human cadaver ankles, has demonstrated that incongruity of the distal tibial articular surface resulted in elevated local articular contact stresses and elevated rates of change of local articular contact stresses . Based on the information from those studies, a testing system was developed using a human ankle cadaver model, configured such that dynamic cartilage contact mechanics could be explored under conditions of various degrees of instability. It was hypothesized that the severity of instability-associated dynamic contact stress aberrations would be correlated with the degree of joint instability.
Potentially unstable conditions in human cadaver ankles were created by impairing articular surface restraint and/or ligamentous restraint. The articular surface restraint was impaired by introducing a full-width coronally-directed osteotomy of the anterior one-third of the distal tibial surface (Figure 1A). The osteotomized fragment was secured rigidly in a controlled, proximally-displaced position (Figure 1B). The ligamentous restraint was impaired by transecting the anterior talofibular ligament (ATFL). Under these unstable conditions, a posteriorly directed force of sufficient magnitude applied to the tibia could subluxate the talus anteriorly under the tibia (Figure 1C, Movie 1).
Six fresh-frozen human ankle specimens were obtained at autopsy (donors ranging from 44 to 92 years old). After thawing at room temperature, each specimen was dissected free of soft tissue at the ankle, other than keeping all major supporting ligaments intact. (Note: the joint capsule on both anterior and posterior aspects was also removed, allowing insertion of a sensor sheet for contact stress measurement.) Gross inspection revealed no osseous or articular morphological abnormalities, and every specimen had a motion arc of at least 15° dorsiflexion to 20° plantar flexion. The mid-shaft of the tibia/fibula, the calcaneus, and the distal phalanges were secured in three separate polymethylmethacrylate blocks.
A custom ankle loading fixture, described elsewhere , was utilized (Figure 2). This fixture was mounted in a servo-hydraulic materials testing machine (Bionix®, Model 858.20, MTS Inc., Eden Prairie, MN), which applied quasi-physiologic stance-phase ankle motion to a specimen under a prescribed compressive load. Coronal and transverse plane rotations were unrestricted. The tibial fixture was equipped with a pneumatic force actuator (Airpot Corp., Norwalk, CT) that could deliver prescribed AP forces. Force histories were measured by a load cell (Omega Inc., Stanford, CT). A linear potentiometer recorded AP motion of the proximal tibia with respect to the loading device.
Local cartilage contact stresses in the ankle were measured by custom designed real-time contact stress sensors (TekScan® model #5033, Tekscan Inc., Boston, MA), described in detail elsewhere . The sensors were very thin (0.1 mm) and flexible, and conformed precisely to the curved joint surfaces with minimal migration during testing . The active area measured 27 mm (medial-lateral) by 39 mm (anterior-posterior) and incorporated a uniformly distributed 32 by 46 array of sensing elements (sensels), yielding contact stress data from 1472 distinct sites, with a spatial resolution of 0.694 mm2 per sensel. Sensor outputs from each sensel were collected at 132 Hz. The data were converted into contact stress values using a sensel-by-sensel-based calibration technique, described elsewhere .
Each specimen was initially tested intact. During a duty cycle, specimens were held under a constant 300 N compressive force, while simulated stance phase ankle motion (flexion angle: 10° → 15° → -10° → 10°, where positive values denote plantar and negative values dorsiflexion) was applied at a rate of 0.5 Hz (Figure 3). The AP tibial loading protocol consisted of a constant anteriorly-directed “baseline” force and a superimposed posteriorly-directed force pulse for potentially subluxating the talus anteriorly under the tibia. The anteriorly-directed baseline force (-30 N) was applied to prevent spontaneous subluxation of the talus under incongruous conditions. The posteriorly-directed subluxation force pulse was applied between the 45% and 70 % time-points of the duty cycle (corresponding to the heel-off phase). Each series of tests on every specimen started with an initial test conducted with the baseline force alone (no subluxation pulse). Subsequent tests were then conducted, varying the subluxation pulse magnitude randomly between 0 N and 120 N, in 20 N increments. After intact conditions were tested, specimens were osteotomized and an identical series of tests were repeated with the fragment reduced in the anatomic position, as well as with the fragment displaced 1.0 and 2.0 mm proximally, again with randomized application of the subluxation force pulse. Finally, the ATF ligament was sectioned, and tests again were repeated for the three tibial fragment positions (anatomic, 1.0 mm, and 2.0 mm stepoff).
Departures from normal ankle kinematics were quantified in terms of differences in AP tibial motion. The time point when the peak magnitude of the subluxation force pulse occurred was coincident (at 56% duty cycle) among all specimens under all testing conditions regardless of pulse magnitude. Therefore, the position of the tibia at this instant in the baseline-force-alone test was defined as the baseline position for each individual specimen under each testing condition. Changes in AP tibial motion in subsequent tests within that series were then measured as positional difference with respect to the baseline position.
Changes in instantaneous local contact stresses were analyzed using the data between the 40% and 90% time points (one second total duration) of the duty cycle, since changes in joint kinematics associated with load modulation occurred only during this period. Each data set consisted of contact stress scalar values reported from 1472 sites (the 32 × 46 grid), measured at 132 time frames (1 second), yielding 194,304 stress values in total. To minimize the effect of low-amplitude noise, stress values lower than 0.25 MPa were filtered out. Rates of change of local contact stresses were calculated using a Lagrange four-point central differencing formula . Specifically, stress-change rate for each site at each time frame was computed from contact stress data from the previous two and subsequent two time frames at that site. For example, a stress-change-rate value at a given site (x, y) at a given frame (n) is calculated from contact stress scalar values S (x, y, n-2), S (x, y, n-1), S (x, y, n+1), S (x, y, n+2), and the time length per frame (1 / 132 seconds), using the following formula:
(Note that this formula could not be applied to the very first two or the very last two time frames.) Positive and negative values of this parameter indicate increasing and decreasing cartilage contact stress, respectively. Data post-processing was performed with a custom computer program implemented in MATLAB® (Version 7.0, MathWorks, Inc., Natick, MA).
For assessing whole-joint, time-cumulative changes in dynamic cartilage contact mechanics, stress-change rate data were subjected to distribution analysis. For each individual test (with a certain combination of specimen/load conditions), data from contact sites, collected throughout the duration of interest, were gathered and plotted in a histogram (Figure 4). In every case, the histogram showed a bell shape symmetrical about the zero-rate axis. Given the data distribution to be nearly normal, expansion of the width of the bell shape was assumed to indicate that abrupt increase/decrease of contact stresses occurred more frequently and/or over a larger area. To quantify such changes, the width of the 95% distribution range (absolute difference between the positive and negative 95th percentile values) was calculated for each data set.
Differences in joint kinematics across specimen conditions were assessed by a one-way ANOVA. Pairwise comparisons were performed only when this test was positive. P-values less than 0.05 were regarded as indicating statistically significant differences. Interaction between joint kinematics and stress-change rate was assessed by linear regression analysis.
In all specimens, a visually distinct transient subluxation occurred once the step-off incongruity was introduced. In contrast, under intact conditions or with an anatomic reduction of the osteotomized fragment, subluxation never occurred regardless of ATFL condition or magnitude of the subluxation force pulse. Under intact conditions, tibial displacement increased linearly with increasing magnitude of the subluxation force pulse (R2 > 0.99 for all specimens) (Figure 4). (Note: the reported displacement values include the effects of both subtalar motion and flexure of the tibial shaft; hence, this measure does not directly indicate tibial-talar displacement.) Nearly identical pulse-displacement relationships were found under conditions with anatomic reduction of the osteotomized fragment, either with or without ligament transection. Under step-off incongruity conditions, the pulse-displacement relationship became sigmoidal (Figure 5). With low pulse magnitudes, the force-displacement relationship remained linear. However, once a specimen-specific subluxation pulse magnitude was exceeded (defined as the instability threshold, ranging between 20 N and 100 N, depending on specimen and conditions), tibial anterior displacement increased rapidly. Since the AP tibial-talar relationship remained nearly unchanged throughout intact tests regardless of posterior pulse magnitude (as observed during testing), the deviation of the force-displacement relationship from the linear ‘intact’ relationship corresponded to the distance of subluxation (anterior displacement of the talus with respect to the distal tibia) (Figure 5).
In Figure 6, the subluxation distance data under all experimental conditions with a 120 N subluxation force showed that significant alteration in joint kinematics occurred with a 1mm step-off incongruity (p < 0.001), and that the subluxation distance increased as the step-off height increased to 2.0 mm (p = 0.002). The effect of step-off incongruity on subluxation distance was enhanced by ATFL transection (p < 0.05, for both step-off heights). In contrast, under conditions with anatomic reduction of the osteotomized fragment, ATFL transection resulted in minimal tibial-talar subluxation.
Representative sequential contour plots of instantaneous local contact stresses and the rate of stress change, comparing stable to unstable motion, are depicted in Figure 7, and in Movies 2, 3, 4, and 5. Contact stresses during stable motion (Figure 7A-1 and Movie 2) were relatively uniformly distributed over a broad area. The contact patch maintained its general shape throughout the motion, without abrupt changes. Correspondingly, rates of contact stress change were low-magnitude, indicating that no sudden changes occurred in articular contact stresses (Figure 7A-2 and Movie 3).
The contact features changed when subluxation occurred under step-off conditions (Figure 7B-1 and Movie 4). Immediately prior to subluxation, contact stresses concentrated along the osteotomized border of the intact posterior tibial surface (as shown in the first three frames in Figure 7B-1). As the talus subluxated, a sharp focus of contact stress abruptly arose on the anterior fragment, indicating impingement of the subluxated talus onto the anterior tibial “step-off” surface (as in the next two frames). Subsequently, as the talus reduced into its original posterior position, the anterior focus of peak contact stress on the fragment surface disappeared, and stress was redistributed onto the posterior region. Corresponding contour plots of stress-change rates demonstrate dramatic foci of adjacent positive and negative changes during subluxation (Figure 7B-2 and Movie 5). As the talus subluxated anteriorly, sudden decreases in local contact stresses occurred along the posterior osteotomy border, with corresponding sudden increases on the anterior displaced fragment. A reversal of this process occurred when the talus reduced to its original position later in the duty cycle, with sudden decreases and increases in local contact stresses on the anterior fragment and on the intact posterior tibial surface, respectively.
Correlation of stress-change rates with the degree of instability was assessed by interpreting the 95% distribution width values. As addressed above, the degree of instability was affected by both stepoff height and ATFL integrity (Figures 5 and and6),6), with more severe instability occurring with higher subluxation pulse magnitudes (Figure 5). Accordingly, data compiled from the four “unstable” specimen conditions (in which distinct instability events occurred) created a data set with a wide spectrum of instability magnitude. In the data set compiled for each individual specimen, the 95% distribution width varied nearly linearly with the subluxation distance (Figure 8A), suggesting that the rate of contact stress change was highly correlated with the degree of instability. The series average of single-specimen R2-value was 0.92 (ranging from 0.79 to 0.97, n = 6). Furthermore, gathering data from the whole series of specimens, it was found that the trends were very consistent across specimens (Figure 8B). The R2 value of the series-wide correlation was 0.87.
In this model, joint instability was induced by modulating external forces on surgically modified human cadaver ankles. The 300 N axial load, while sub-physiologic, was still substantial. The AP forces, modulated for creating various levels of instability in each specimen setting, did not mimic any particular physiologic situation, though the range of modulation was still physiologic . Ankles were rendered potentially unstable either by altering tibial surface geometry, by transecting the anterior talofibural ligament (ATFL), or by a combination of the two. Both of these patho-anatomic conditions (articular incongruity and ligamentous disruption) are typical post-traumatic residuals, and specifically in this model mimic the clinical scenario of a pilon fracture. Removal of the joint capsule (for inserting a contact stress sensor) may have affected joint stability, but capsule rupture often accompanies clinical intraarticular fractures. Experimentally, in a related previous study with an identical ankle preparation , the effect of joint capsule removal on joint kinematics was undetectable. Taken together, the force/motion protocol and pathoanatomic conditions used in this experiment constitute a physiologically relevant model of ankle instability. Likewise, the associated dynamic cartilage contact stress aberrations are likely relevant for instability in general.
Ankle instability in this experiment resulted primarily from loss of contact between the anterior distal tibial surface and the talar dome, demonstrating that the anterior malleolar surface plays a predominant mechanical role in maintaining AP ankle stability. This observation is consistent with previous experimental studies that showed that AP ankle stability was primarily dependent on articular surface restraint (ex vivo) [20,22], that the anterior tibial surface played a leading role in restricting anterior talar displacement (ex vivo) , and that considerable talar contact on the anterior tibial surface occurred during the mid to late stance phase (in vivo) . Likewise, clinical observations have shown that the risk of ankle instability after ATFL injury is higher in ankles with reduced AP tibial coverage over the talar dome [32,33]. In contrast, distinct anterior talar subluxation did not occur with isolated ATFL transection, although such transection enhanced unstable motion that occurred with altered joint surface geometry. It appears that the ATF ligament worked secondarily, rather than primarily, as a restraint to anterior talar displacement, as previously described . It must be emphasized that this experiment was designed to create various levels of instability with minimal changes in extrinsic loading, and the findings were not meant to apply to any specific clinical condition regarding the ankle. The goal of the study was to document pathomechanical changes at the joint level that were caused by unstable motion in general, using the ankle as a model. However, the importance of articular surface geometry to ankle stability should be kept in mind when treating displaced pilon fractures or ankle OA associated with distal tibial deformity.
Instability-associated changes in dynamic cartilage contact mechanics in this study were characterized by elevated rates of contact stress change, with the degree of that elevation being correlated linearly with instability severity. These increases/decreases in local contact stresses occurred over a very short duration, typically within five to eight data sampling intervals (0.038 to 0.060 sec). It is of course possible that the temporal resolution of the sensors was inadequate to detect the very most rapid changes in the rate data. However, the stress-change rate data were distributed normally under all conditions (Figure 4), suggesting that the statistical analyses of the 95th-percentile values are likely valid.
The most striking feature of this study was the linear correlation between the rates of contact stress change and the level of instability (Figure 8). This suggests that even small instability events, some of which may not be detectable as kinematic abnormality, presumably involve some degree of dynamic contact stress aberration. Such “microinstability” events may occur in joints with subtle instability resulting from mild residual incongruity or malalignment after an articular fracture, or with residual ligamentous laxity after a ligament injury or ligamentous reconstructive surgery. These microinstability events could plausibly cause localized regions of increased rates of contact stress change in a joint, which may accumulate over years leading to cartilage degeneration. In experimental settings, measurement of contact stress transients might be helpful for detecting microinstability conditions that can damage articular cartilage.
Financial support was provided by NIH grant 5 P50 AR048939 and CDC grant R49 CCR721745. Drs. Annunziato Amendola and Joseph A. Buckwalter provided helpful suggestions.