Traditionally, anterior cruciate ligament reconstructions have been performed with good-to-excellent short-term results in many patients. The operation is normally successful with regard to improving knee stability and allowing patients to return to pivoting sports. However, in some patients, symptoms of instability may persist. There are two primary clinical tests to evaluate the status of the anterior cruciate ligament. Historically, the Lachman test has been used to objectively evaluate the amount of anterior tibial translation at 20° to 30° of knee flexion1. This test can be quantified with use of the KT-1000 instrumented laxity testing device (MEDmetric, San Diego, California). However, numerous clinical studies have found no correlation between subjective clinical scores and KT-1000 measurements in patients with a reconstructed anterior cruciate ligament2-9. This suggests that the functional instability associated with a torn or reconstructed anterior cruciate ligament may be more complex than excessive anterior laxity alone.
The pivot shift test is also widely accepted as a clinically useful test to assess the stability of knees that have sustained an anterior cruciate ligament injury. In this test, a valgus moment is applied to the tibia (sometimes in combination with a slight internal tibial torque), causing an internal rotation of the tibia and a corresponding anterior subluxation of the lateral tibial plateau near full extension. As the examiner flexes the knee from 0° to 90°, the line of action of the passive iliotibial band tension changes, causing it to become a knee flexor. The resulting external tibial torque causes the tibia to rotate externally (with a spontaneous reduction of the anteriorly subluxated lateral tibial plateau), producing a pivot shift, which is assigned a clinical grade. The presence of a pivot shift sign has been shown to have a significant association with the functional outcome of patients undergoing anterior cruciate ligament reconstruction9.
Although these two clinical examinations have been used for decades, the relationship between the two tests has received limited experimental study10. It is not known whether an anterior cruciate ligament-deficient knee displaying a large amount of anterior tibial translation will also demonstrate a large pivot shift, or whether the difference in laxity between the injured and the normal knee (the injured-normal difference) is a better predictor of pivot shift behavior. Also unknown is how the pivot shift magnitude changes when the anteroposterior laxity of a knee with a reconstructed anterior cruciate ligament is increased, as might occur with progressive stretching of an anterior cruciate ligament graft.
The goals of this study were to determine the linear correlation of (1) pivot shift magnitude versus absolute anteroposterior laxity for a sample group of anterior cruciate ligament-reconstructed knee specimens with the grafts unfixed (the anterior cruciate ligament-deficient condition), and (2) the increase in pivot shift magnitude versus the increase in laxity (the difference between the laxity of the intact anterior cruciate ligament condition and the anterior cruciate ligament-deficient condition) for these knees when an anterior cruciate ligament graft was removed. A third goal was to plot pivot shift magnitude versus laxity for each reconstructed knee by incrementally loosening the anterior cruciate ligament graft.
Seventeen fresh-frozen unpaired cadaver knee specimens from male donors were used. The mean age (and standard deviation) of the donors at the time of death was 38.6 ± 14.7 years (range, sixteen to sixty-seven years). Anteroposterior laxity (between 100 N of applied anterior and posterior force) was recorded for the intact knee at 30° of flexion. A cap of bone containing the tibial insertion site of the anterior cruciate ligament was mechanically isolated with use of an appropriately sized cylindrical coring cutter (a 24 to 26-mm inside diameter). Small screws attached to the undersurface of the bone cap were cast into a cylindrical construct with use of polymethylmethacrylate acrylic; this potted acrylic construct also contained a threaded metal core for attachment to a load-cell mounted on a bracket attached to the tibia11. The bone cap remained in its precise anatomical position during all tests. Laxity testing was repeated with the bone cap in place to confirm that no changes occurred after the bone cap installation procedure.
The bone cap was disconnected from the load-cell (removing tension from the native anterior cruciate ligament) to simulate an anterior cruciate ligament-deficient condition. The knee was placed in a testing apparatus that applied a valgus moment to the tibia by suspending a weight (specific for each specimen) at a fixed distance from the joint line (Fig. 1). This caused the tibia to rotate internally and the lateral tibial plateau to subluxate anteriorly. A weight, applied to the iliotibial band tendon through a pulley system, simulated the force developed in the muscle as the knee was flexed. This caused the tibia to rotate externally. Knee flexion angle was measured by an inclinometer (with a resolution of 0.5°). Tibial rotation and varus-valgus rotation were measured by rotatory potentiometers with a resolution of 0.05°. Linear displacement of a point on the lateral tibial plateau (on the Gerdy tubercle) was measured by a line connected to the core of a linear variable-displacement transducer with a resolution of 0.02 mm. Zero positions of tibial rotation and lateral plateau displacement were defined with the intact knee in full extension (0° of flexion). Further details of the test apparatus are described in a prior publication12.
To produce a simulated pivot-shift event, a valgus moment was applied, the knee was flexed to between 20° and 40°, and an iliotibial band force was then applied. With a proper combination of valgus moment and iliotibial band force, the anteriorly subluxated lateral tibial plateau reduced spontaneously as the iliotibial band force externally rotated the tibia. This reduction event (pivot shift) was observed in all knee specimens. The combination of valgus moment and iliotibial band force necessary to pivot the anterior cruciate ligament-deficient knee, and the knee flexion angle at which the pivot occurred, were determined by trial and error for each specimen. This was necessary because the pivot mechanics were unique for each knee because of specimen variability. If the same valgus moment had been applied to all specimens, some knees would not have initially demonstrated anterior subluxation of the lateral tibial plateau. Conversely, if the same iliotibial band force had been applied to all knees, the reduction event would not have occurred in some specimens. For the seventeen specimens tested, the mean valgus moment (and standard deviation) that was applied was 3.06 ± 1.32 Nm, the mean iliotibial band force was 25.65 ± 6.77 N, and the mean flexion angle at pivot was 27.8° ± 3.50°.
Internal-external rotation of the tibia and anterior-posterior displacement of the lateral tibial plateau were recorded with valgus moment alone (before the pivot), and with valgus moment and iliotibial band force (after the pivot). Total rotations and displacements during the pivot shift event were calculated by subtracting the values before the pivot from the values after the pivot. The bone cap was reattached to the load-cell (thereby "restoring" the native anterior cruciate ligament), and the same loading conditions were repeated on the intact knee.
The anterior cruciate ligament was cut, and the tibial bone cap was removed from the knee. The remaining fibers of the anterior cruciate ligament were dissected down to their tibial attachment, an outline of the tibial footprint of the ligament was marked on the bone cap surface, and the center of the footprint was marked. With use of an alginate casting material, a negative mold of the tibial bone cap was made and filled with polymethylmethacrylate, resulting in an acrylic replica (positive) of the original tibial bone cap. The outline and center of the tibial footprint were replicated on the acrylic cap.
A patellar tendon allograft (obtained from the Musculoskeletal Transplant Foundation, Edison, New Jersey) was prepared for each knee specimen. The tendon fibers inserting into the tibia were dissected away from the bone surface, leaving a free end. The free end of the graft tissue was trimmed to fit tightly within a 7-mm femoral tunnel. The patellar end of the graft contained tendon tissue that inserted into an appropriately sized patellar bone block. The central portion of the tibial footprint was cored out from the acrylic bone cap replica with a handheld rotary grinder, and the patellar bone block (approximately 8 × 8 × 10 mm in size) was potted with polymethylmethacrylate into the recess. The potted bone block was centralized within the tibial footprint.
The anterior cruciate ligament fibers in the femoral notch were scraped from the bone, and a 5-mm offset guide was used to drill a 7-mm tunnel at the eleven o'clock position (right knee). The acrylic replica, with potted bone block and attached patellar tendon tissue, was attached to the tibial load-cell. A single low-stretch, high-strength synthetic line (135-lb [61.2-kg] test Spectra Fiber; Izorline, Gardena, California) was whip-stitched into the tissue end of the graft. The lines connected to the end of the graft in the femoral tunnel passed out the femur and through a split-clamp cemented into the femoral potting acrylic. The graft was tensioned by pulling on the lines, and the lines were fixed by tightening the split-clamp.
Pivot shift testing was first performed with the graft untensioned (the anterior cruciate ligament-deficient condition). The graft was then tensioned to restore total anteroposterior laxity (between 100 N of applied anterior and posterior force) to within 1 mm of that of the intact knee at 30° of flexion, and pivot shift testing was repeated to confirm the absence of significant changes in pivot shift behavior from the intact knee condition. Fixation of the lines passing through the femoral split-clamp was then adjusted such that the anteroposterior laxity was increased incrementally by approximately 2 mm, and pivot shift testing was repeated. This protocol was repeated until the anterior cruciate ligament-deficient laxity was reached (the graft lines were unclamped). For each testing cycle, three anteroposterior tests were applied to precondition the graft.
Pivot shift magnitude was plotted against absolute knee laxity for all knees with the grafts unfixed (the anterior cruciate ligament-deficient condition). The increase in pivot shift magnitude was plotted against the increase in laxity (the difference between the laxity of the intact anterior cruciate ligament and the anterior cruciate ligament-deficient condition) for these knees when an anterior cruciate ligament graft was removed. The slope of the best-fit line and r2 (coefficient of determination) were determined for each plot.
For regression analysis of individual knee specimens with incremental loosening of the graft, the magnitude of the pivot shift (expressed either in degrees of tibial rotation or millimeters of lateral tibial plateau displacement) was plotted against anteroposterior laxity in a separate graph for each knee, and the slopes were determined. A one-way repeated-measures analysis of variance model was used to compare mean laxities, tibial rotations, and plateau displacements in the seventeen specimens for three laxity conditions: intact knee laxity, laxity at the end point of the linear range (as described in Results), and anterior cruciate ligament-deficient laxity. Pairwise comparisons were made with use of the Student-Neuman-Keuls procedure. Paired Student t tests were used to compare mean primary and secondary slopes (as described in Results) obtained from seventeen individual plots of pivot shift versus laxity with variable graft fixation.
Source of Funding
Sources of funding included NFL Charities. In addition, human tissues utilized for this study were provided by the Musculoskeletal Transplant Foundation. The funding sources did not play any other role in the investigation.
Correlations Between Pivot Shift and Laxity
Correlations between pivot shift magnitude and absolute laxity for all seventeen anterior cruciate ligament-deficient knees were fair. With pivot shift plotted as millimeters of lateral tibial plateau displacement, the mean slope was 1.2342 mm/mm (r2 = 0.4102) (Fig. 2). With pivot shift plotted as degrees of tibial rotation, the mean slope was 0.8877°/mm (r2 = 0.3422) (see Appendix).
Correlations Between Pivot Shift Change and Laxity Change
Fair-to-good correlations were found for change in pivot shift versus change in laxity (the difference between the laxity of the intact anterior cruciate ligament and the anterior cruciate ligament-deficient condition) for all seventeen knees. With pivot shift plotted as millimeters of lateral tibial plateau displacement, the mean slope was 2.0840 mm/mm (r2 = 0.7039) (Fig. 3). With pivot shift plotted as degrees of tibial rotation, the mean slope was 1.5319°/mm (r2 = 0.5345) (see Appendix).
Graphs of Pivot Shift Versus Laxity for Individual Knees with Incremental Loosening of the Graft
Each graph of pivot shift magnitude versus laxity, obtained by varying the femoral graft fixation for an individual knee, displayed two distinct linear regions (as shown in Figures 4 and 5 for a sample knee specimen). In Figure 4, pivot shift magnitude was plotted in millimeters of lateral tibial plateau displacement, whereas, in Figure 5, pivot shift was plotted as degrees of tibial rotation. A primary slope was calculated for each sample knee specimen between a laxity value for the intact knee and a laxity value corresponding to the end point of the linear range, at which point the slope abruptly changed. This laxity end point was different for each individual specimen and was determined graphically by examination of all plotted data points for a given knee. A secondary slope was then calculated from data points between the laxity at the linear end point and the anterior cruciate ligament-deficient laxity. The mean values for primary and secondary slopes for the seventeen knees were significantly different from each other (p < 0.0001) for both kinematic measurements of pivot shift. The mean r2 values of >0.98 for both primary slopes indicate an excellent linear fit to the data for individual knee specimens (Table I).
Mean laxity for the anterior cruciate ligament-deficient condition in the seventeen knees was 24.6 mm (Table II). Reducing mean laxity by 4.7 mm (to the linear end point) did not significantly change either kinematic measurement of pivot shift magnitude (Table II). Further reductions in mean laxity from the linear end-point laxity (19.9 mm) to intact knee laxity (10.0 mm) produced significant reductions in both kinematic measurements of pivot shift magnitude (p < 0.05); mean tibial rotation decreased 11.0° and plateau displacement decreased 12.2 mm (Table II).
In this study, anterior cruciate ligament-deficient knees that demonstrated a large amount of absolute anteroposterior laxity did not necessarily demonstrate a correspondingly large pivot shift, since the two tests were poorly correlated. In contrast, when the unfixed anterior cruciate ligament grafts were tensioned to match the anteroposterior laxity of the intact knee, there were fair-to-good correlations between the change in pivot shift magnitude and the change in anteroposterior laxity. During tests on individual specimens with incremental loosening of the graft, we found that reducing anteroposterior laxity an average of 4.7 mm from the anterior cruciate ligament-deficient state had a negligible effect on the magnitude of the pivot shift. These findings may help to explain some confusing aspects of the clinical literature related to anterior cruciate ligament reconstruction.
The Lachman test and pivot shift test are standard clinical tools used to assess the stability of knees before and after an anterior cruciate ligament reconstruction. However, each test measures a different aspect of knee stability. Ideally, both tests should be positive for an anterior cruciate ligament-deficient knee and negative for an intact or successfully reconstructed knee. However, this is not always the case, and improving the side-to-side difference in anterior translation laxity after anterior cruciate ligament reconstruction does not always eliminate the pivot shift. Aglietti et al.13 reported that 12% of patients with a reconstructed anterior cruciate ligament with a mean side-to-side difference in anterior laxity of 1.98 mm still exhibited a pivot-glide after fourteen months. Lee et al.14 reported thirteen of 137 patients receiving a quadriceps tendon autograft exhibited a negative Lachman test and positive pivot-shift test after a minimum two-year follow-up period.
Clinical follow-up studies have suggested that mean instrumented measurements of anteroposterior laxity increase after anterior cruciate ligament reconstruction but then stabilize within three to six months postoperatively15-17. The expected increase in pivot shift magnitude corresponding to this laxity increase is unknown. In the past, anterior cruciate ligament grafts have been tensioned to match the anteroposterior laxity of the contralateral knee. It has been shown that 92% of patients with normal knees have a left knee-right knee difference in anterior displacement of no more than 2 mm18. Thus, the goal of anterior cruciate ligament reconstruction has been to restore normal anteroposterior laxity to approximate that of the contralateral knee19. In our study, we used intact knee laxity as the baseline; however, this is not possible in clinical practice, and the contralateral knee is commonly used to determine normal laxity. If a criterion related to the amount of pivot shift in the contralateral knee should also be considered, it is necessary to know the relationship between anteroposterior laxity and pivot shift in the reconstructed knee. This is especially important if instrumented pivot-shift tests are developed and applied in a clinical setting.
Lie et al.10 measured changes in anterior laxity and tibial rotation during a simulated pivot shift in eight specimens as the tension of an anterior cruciate ligament graft was increased from 0 to 60 N. They found that increasing the graft tension reduced the anteroposterior subluxation-reduction during the pivot shift, but rotational motions were not restored to normal. They did not investigate the relationships between anteroposterior laxity and pivot shift as graft tension was increased. Our study measured the relationships between the magnitude of a pivot shift and knee laxity, and these findings may help to explain inconsistencies between these two tests that are sometimes observed in patients with an anterior cruciate ligament-deficient knee during a clinical examination.
In contrast to the results for each individual knee with variable graft tensions, which had a nearly perfect linear correlation between pivot shift and anteroposterior laxity, correlations between pivot shift and laxity for all knees in the anterior cruciate ligament-deficient condition were relatively weak (r2 = 0.3422 for tibial rotation and r2 = 0.4102 for plateau displacement). Clinically, this weaker correlation may help to explain why an anterior cruciate ligament-deficient knee that demonstrates a large amount of absolute anteroposterior laxity (without comparison with the contralateral side) may not necessarily demonstrate a large pivot shift. We believe that this finding is most likely related to anatomical variations. The ability to elicit a pivot shift may be influenced not only by the absence of the anterior cruciate ligament but also by other factors such as the shape of the femoral condyles, posterior slope of the lateral tibial plateau, geometry of the menisci, overall ligamentous laxity of the knee, and integrity of the secondary restraints (including the iliotibial band).
When unfixed grafts of all knees were tensioned to restore intact knee laxity, the change in pivot shift was better correlated with the change in anteroposterior laxity (r2 = 0.5345 for tibial rotation, and r2 = 0.7039 for plateau displacement). Clinically, this suggests that patients with a unilateral anterior cruciate ligament deficiency who demonstrate a large side-to-side laxity difference should also demonstrate a large side-to-side difference in pivot shift magnitude. This assumes that both knees had the same laxity and pivot shift before the injury occurred.
Analysis of individual knees with incremental loosening of the graft showed that when the anterior cruciate ligament graft was tensioned to restore anteroposterior laxity of the intact knee at 30° of knee flexion, all knees pivoted to some extent. Therefore, there was no threshold level of laxity at which a pivot shift suddenly appeared. However, the mean pivot magnitudes shown in Table II for anterior cruciate ligament grafts tensioned to intact knee laxity (10.7° of external tibial rotation and 8.1 mm of lateral plateau displacement) could possibly still be graded as normal during a clinical examination. It should be noted that our testing apparatus was very sensitive to even subtle changes in rotational laxity. As such, there was no way to correlate our pivot data with a commonly used clinical grading scale that can only discriminate between much larger changes in rotational laxity.
For each individual knee specimen with an anterior cruciate ligament reconstruction, pivot shift magnitude increased linearly from the intact knee laxity up to a laxity corresponding to the end point of the linear range, at which point the slope abruptly decreased. Between this linear end-point laxity and anterior cruciate ligament-deficient laxity, the mean increase in pivot shift magnitude was relatively small (0.5° of external tibial rotation and 1.76 mm of plateau displacement), increases that would not likely be detected during pivot shift testing in a clinical examination. Therefore, reducing mean laxity by 4.7 mm from the anterior cruciate ligament-deficient condition had a negligible effect on reducing the pivot shift.
Clinically, this may help to explain why an insufficiently tensioned anterior cruciate ligament graft could substantially reduce anteroposterior laxity from the anterior cruciate ligament-deficient condition and still leave the patient with a virtually unchanged pivot shift. The mean values from Table II show that if the anterior cruciate ligament grafts were tensioned to restore anteroposterior laxity to the intact condition and the grafts were then to stretch out so that laxity increased by approximately 10 mm (to the linear end point), the mean pivot-shift magnitudes would increase by 11.0° (tibial rotation) and 12.2 mm (lateral plateau displacement), changes that should be readily apparent in a clinical pivot-shift examination.
There are several limitations to this study. The patellar tendon grafts used were sized to fill a 7-mm femoral tunnel drilled at an eleven o'clock orientation in the femoral notch (right knee). Clinically, the femoral tunnel for an anterior cruciate ligament graft is typically 10 to 11 mm in diameter, and some surgeons choose to drill the tunnel at a more oblique ten o'clock position in the notch. We do not believe use of a smaller diameter graft for this study influenced the results because the graft was tensioned to restore intact knee laxity for each specimen. Thus, we theorized the laxity with a 10-mm-diameter graft would be identical to that with our 7-mm-diameter graft. A recent study from our laboratory comparing pivot shift magnitudes with eleven o'clock and ten o'clock femoral tunnels found no significant differences between tunnel locations20. Therefore, we do not believe femoral tunnel orientation would substantially affect the findings of this study. Finally, this is a cadaver study that could not replicate the role of passive knee musculature during the pivot shift examination. It is likely that the pivot shift magnitudes recorded in the present study represent the maximum that would occur because knee musculature would be expected to reduce the pivot magnitudes in vivo.
We are unaware of other studies that have directly measured the relationship between pivot shift and anteroposterior laxity in a laboratory setting. However, the question remains as to how our laboratory measurements of pivot shift correlate with the pivot shift grade as determined by a clinical examination. This is difficult to determine since the clinical pivot-shift test is somewhat subjective and difficult to quantify. The exact level of valgus moment applied during the test varies between examiners, and the precise levels of tibial reduction during the pivot shift event (as measured by degrees of tibial rotation or millimeters of plateau displacement) that the examiner judges to be abnormal are unknown.
In conclusion, the injured-normal difference in knee laxity was a good predictor of the injured-normal difference in pivot shift magnitude. However, laxity of an anterior cruciate ligament-deficient knee alone was a poor predictor of pivot shift magnitude. Reducing anteroposterior laxity of anterior cruciate ligament-deficient knees by an average of 4.7 mm had a negligible effect on pivot shift magnitude. However, when the grafts were tensioned to restore intact knee laxity, the pivot shift magnitudes were restored to intact knee levels. As future in vivo techniques and instrumented devices are developed to better quantify the pivot shift examination, this study provides a potential baseline reference for future clinical studies of patients before and after anterior cruciate ligament graft reconstruction.