Kinematic Performance of Medial Pivot Total Knee Arthroplasty

Abstract

Background:

Total knee arthroplasty (TKA) implants have continued to evolve to accommodate new understandings of knee mechanics. The medial-pivot implant is a newer design, which is intended to limit anterior-posterior translation in the medial compartment while allowing lateral compartment translation. However, evidence for a generalized medial-pivot characteristic across all activities is limited. The purpose of the study was to quantify and compare in vivo knee joint kinematics using high-speed stereo radiography during activities of daily living in patients who have undergone a TKA with a cruciate sacrificing medial-pivot implant to age-matched and sex-matched native controls.

Methods:

Fifteen participants (7 patients, 4 women, mean age 70 years and 8 nonsymptomatic controls, 4 women, mean age 64 years) performed 6 functional tasks in high-speed stereo radiography: deep-knee lunge, chair rise, step down, gait, gait with 90° turn, and seated knee extension. Translational differences between groups (surgical versus control) were assessed for the medial and lateral condyle, while pivot location was normalized to subject-specific tibial plateau geometry.

Results:

The surgical cohort displayed a more constrained medial condyle that provided greater stability of the medial compartment and did not result in the paradoxical anterior translation at mid-flexion angles during weight-bearing activities, but was associated with less condylar translation than native knees. Additionally, the transverse tibial pivot location occurs most commonly in the middle third of the tibial plateau and secondarily on the medial third.

Conclusions:

Some variability in pivot location occurs between activities and is more in nonsymptomatic, native knee controls.

Total knee arthroplasty (TKA) is a surgical intervention that can relieve pain and restore function in severely disabled knee joints. One newer design is the medial-pivot implant, which is intended to provide a kinematic pattern that closely resembles native-knee kinematics [1-8] by limiting anterior-posterior translation in the medial compartment while allowing translation in the lateral compartment [9]. Several implant manufacturers have introduced medial-pivot designs; however, each medial-pivot design is unique and the in vivo kinematics of some medial-pivot implants is unexplored.

The common understanding is that the transverse axis of internal-external rotation, or pivot, of the tibio-femoral joint in a healthy knee is located on the medial side [1-3,10-12], yet the evidence for a generalized medial-pivot is limited. Medial pivot is commonly interpreted as any anterior-posterior translation of the lateral femoral condyle, relative to the tibial plateau, that exceeds translation of the medial femoral condyle [3,13]. An alternative method to describe pivot type is to quantify normalized tibio-femoral pivot location on the tibial plateau. Banks and Hodge [14] calculated tibio-femoral center-of-rotation location by solving a least-squares system of equations for the lines between the medial and lateral low point across the flexion range in patients who had a TKA. Pivot location was a continuous distribution of center-of-rotation location; on average, most occurred centralized to the tibial plateau [14]. Hamilton et al [15] performed a similar analysis on 53 nonsymptomatic, native adult knees and found comparable results with central-pivot to medial-pivot locations for leg press, knee extension, lunge, and gait. Ultimately, studying implanted kinematics across varying activities is necessary to understand pivot location.

The purpose of our study was to quantify in vivo knee-joint kinematics using high-speed stereo radiography (HSSR) during activities of daily living in patients who have undergone TKA with a cruciate sacrificing medial-pivot implant. We hypothesized that the pivot location of the medial-pivot TKA participants will be activity-dependent, similar to the native knee.

Methods

Patient Selection

Institutional Review Board approval was obtained with 2 groups assessed and compared. Surgical group: There were 7 patients (4 women; Supplementary Table 1) who underwent TKA with a cruciate sacrificing medial-pivot implant (Medtronic, Responsive Orthopedics, Minneapolis, MN). Surgical patients were selected from a convenience sample, who were willing to travel to participate in the study and were recruited as part of their clinical visits.

All surgeries were performed by the same board-certified, fellowship-trained orthopedic surgeon (C.L.B.) using a medial parapatellar approach. Standard instrumentation was used for the TKA procedure with intramedullary instrumentation on the femur and extramedullary instrumentation on the tibia, without robotics or other enhanced technology. Preoperatively, the surgical knees were neutral to varus with the goal to achieve mechanical TKA alignment. Soft-tissue gaps were balanced symmetrically versus a slightly looser lateral than medial compartment. The radius of the femoral geometry was visually comparable between the medial and lateral condyles. The geometric constraint was provided by the polyethylene insert with the medial compartment being more constrained than the lateral compartment. However, the medial compartment of the polyethylene insert was not overly constrained and allowed for independent anterior-posterior translations of the femoral component relative to the tibial tray. Knee Injury and Osteoarthritis Outcome Scores (KOOS) were measured by clinical staff presurgery and postsurgery. Control group: There were 8 native knees, nonsymptomatic age-matched and sex-matched controls (4 women; Supplementary Table 1) who participated in the study to provide representative native-knee kinematics. Controls were recruited locally and were part of a larger National Institutes of Health–funded study; they were selected due to their comparable demographics to the TKA patients post hoc.

Data Collection

A custom HSSR imaging system (Center for Orthopedic Biomechanics at the University of Denver, Denver, Colorado) [16] was used to quantify tibio-femoral knee kinematics that comprised 2 40-centimeter (cm) image intensifiers integrated with high-speed, high-definition (1,080 × 1,080) digital cameras placed at a 70° relative angle. Activities were captured using pulsed radiography at 50 frames/second for the high-flexion activities (ie, chair rise, lunge, and knee extension) and 100 frames/second during overground walking activities (ie, step down, gait, and gait with 90° turn). The rate at which participants moved through the field of view was more in the overground walking activities. The greater sampling rate of 100 frames/second allowed for a comparable number of absolute frames to be tracked, similar to the 50 frames/second activities, across the entire range of motion for each activity. Surface marker motion capture was collected (100 Hz; VICON, Centennial, CO) with ground forces measured from an indwelling force plate (1,000 Hz; Bertec, Columbus, OH).

The 6 functional tasks consisted of chair rise, lunge, step down, gait, gait with 90° turn, and knee extension (Figure 1). Chair rise had participants start seated on a 46-cm platform and rise to a standing position. During lunge, participants started in a standing split stance with the knee of interest forward and lower their hips until their back knee touched the ground. Step down required participants to stand on a 23-cm platform and step off in one fluid motion to a target mark and take several steps of normal gait. Gait trials were performed at each participant’s preferred comfortable pace with a target mark placed on the floor to align heel strike and weight acceptance. Gait with 90° turn had participants walk to the same mark and turn 90° away from the leg of interest (eg, turn left when studying right knee) and continue walking at their preferred pace. Knee extension was performed unweighted and seated with participants starting in full knee flexion and extending through their full range of motion.

Fig. 1.

Functional tasks used to quantify tibio-femoral kinematics included chair rise (A), lunge (B), step down (C), gait (D), gait with 90 ° turn (E), and knee extension (F). The high-flexion activities (chair rise, lunge, and knee extension) were performed in approximately 3 seconds across the largest range of motion possible. Overground walking activities (ie, step down, gait, and gait with 90 ° turn) were performed at each participant’s normal, preferred walking speed with the emphasis on recording heel strike to weight acceptance.

Maximal voluntary contractions (MVCs) of knee-extensor and knee-flexor strength were assessed at 70° of knee flexion for the surgical group. Participants were instructed to perform a ramped contraction from rest to maximum in 3 seconds; the maximal value was recorded as the average force of a 0.5-second window centered at peak force. At least 3 MVCs were performed, with no more than 5 total (2 trials within 5%), for each leg and task. Moment-arm length was recorded from the approximated knee rotational center to a cuff at the ankle where the force was measured by a load cell perpendicular to the shank. Strength was normalized to knee torque/body mass (Nm/kg).

Post Processing

The surgical femoral component and tibial tray implants were oriented in accordance with Grood and Suntay [17] decomposition (Figure 2A) with origins retained from the manufacturer-provided computer-automated design drawings. Control group native-knee bony geometries were segmented (Simpleware ScanIP, version 02,018.12, Sunnyvale, CA) from knee-specific computed tomography scans to generate subject-specific femur and tibia segments. Femoral geometries were aligned using a fitted cylinder method with the tibiae made coincident in terminal knee extension (Figure 2B [18]). Geometries were manually tracked (XROMM Autoscoper, Brown University, Providence, RI) in 2 planes to recreate movement in 3-dimensional space. The Grood and Suntay convention was used to calculate tibio-femoral kinematics in 6 degrees-of-freedom and reported as range of motion for each activity. Medial and lateral low points of the femoral component were identified as the lowest point relative to the z-axis of the tibial tray (Figure 3A and B), or plateau, and were used to calculate transverse tibia pivot location during the 6 functional tasks (Figure 3C [15]). Low-point kinematics were represented relative to knee-flexion angle for chair rise, lunge, and knee extension with step-down, gait, and gait with 90° turn normalized to heel strike to toe off (0% = heel strike; 100% = toe off).

Fig. 2.

Coordinate system assignment. (A) Surgical group origin location and local coordinate system assignment of the femoral component and tibial tray in accordance with grood and suntay decomposition. (B) Control group origin location and local coordinate system assignment with the tibia coincident to the femur. Surgical group. Illustrative low-point kinematics for the surgical group during chair rise, lunge, step down, gait, gaith with 90° turn, and knee extension. Red denotes medial low points and blue indicates lateral low points.

Fig. 3.

Representative low-point kinematics and center-of-rotation location during chair rise. (A) Red dots represent medial low point and blue dots represent lateral low points with absolute translation across the flexion range reported above each, respectively. Color-coded lines depict knee flexion angle across the range of motion. (B) Anterior-posterior (AP) low-point location across the flexion range: red = medial; blue = lateral. (C) Center-of-rotation location for chair rise activity reported in A and B. Pivot location was normalized to subject-specific tibial plateau geometry from tibial origin (0) to a maximal anterior landmark (+100; Anterior), maximal posterior landmark (−100; Posterior), maximal medial landmark (−100; Medial), and maximal lateral landmark (+100; Lateral). The red ‘X’s signify pivot location in 30° discrete flexion ranges across the range of motion. Medial-lateral pivot location was quantified as medial (−100 < X < −33), central (−33 ≤ X ≤ 33), lateral (33 < X < 100), or no pivot (X ≤ −100 or X ≥ 100). Control group. Illustrative low-point kinematics for the surgical group during chair rise, lunge, and knee extension. Red denotes medial low points and blue indicates lateral low points.

Data Analyses

Age-matched and sex-matched controls were used to quantify implant motion relative to nonsymptomatic, native kinematics. Range of motion for the 6 degrees-of-freedom about the knee were calculated with anterior-posterior translations of the medial and lateral condyles compared between the surgical and control group. Also, we quantified normalized tibio-femoral pivot location on the tibial plateau for each group during 6 functional activities. Normality of data was assessed using the Shapiro-Wilk test and verified visually with box plots and quantile-quantile plots (theoretical versus sample quantiles). T-test and case-wise resampling across 5,000 iterations of linear regression models were performed to provide group difference to compare age, height, weight, and body mass index. Differences between operated and nonoperated leg MVCs were assessed with a paired t-test and verified with a bootstrapping procedure to provide a 95% confidence interval of the estimate (Table 1). Range of motion was assessed by case-wise resampling the linear-regression coefficient between the surgical and control group for each activity and degree of freedom (Supplementary Tables 2 through 4). Differences in anterior-posterior translations between the medial and lateral condyle were examined using linear mixed-effects models controlling for random effects (subject) and the interactions between fixed effects (knee angle or stance phase, and side) for each group. Translational differences between groups (surgical versus control) were assessed for the medial and lateral condyle, using linear regression models while controlling for the interaction between knee angle and group. Pivot location was normalized to subject-specific tibial plateau geometry with the sum of all pivots normalized to percent of total location (‘medial’, ‘central’, ‘lateral’, ‘no pivot’; Figure 3C) for each activity. Statistical procedures were performed using R (version 3.6.1, The R Project, Auckland, NZ) with α = 0.05. All regression equations and results are available as supplemental materials.

Table 1.

Surgical Group Maximal Voluntary Contraction (MVC) Knee Torque.

MVC Torque	Nonoperated	Operated	Difference	P Value
Knee Extensors (Nm/kg)	0.95 ± 0.16 (0.73, 1.40)	0.90 ± 0.08 (0.79, 1.14)	−0.05 ± 0.10 (−0.27, 0.11)	.6257
Knee Flexors (Nm/kg)	0.61 ± 0.12 (0.44, 0.94)	0.52 ± 0.07 (0.39, 0.68)	−0.09 ± 0.07 (−0.23, 0.04)	.2988

Mean ± standard deviation (95% confidence interval).

Study Sample

The surgical (n = 7; Supplementary Table 1) and control group (n = 8; Supplementary Table 1) were similar in age (difference = 5.2 ± 3.3 years [−2.6, 10.8]; P = .1788), height (0.05 ± 0.06 m [−0.06, 0.16]; P=.4554), weight (−0.2 ± 9.7 kg [−25.0, 15.1]; P = .9846), and body mass index (−0.5 ± 2.0 [−5.0, 2.8]; P = .8094). Presurgical average KOOS values were 62.2 ± 7.4 (47.5, 67.2) and 90.2 ± 9.8 (77.8, 96.6) at that time of kinematic evaluation, 18 months (range, 10 to 37) postsurgery. Similar normalized knee torques were identified for the surgical group between the operated and nonoperated limb during maximal effort knee extension and knee flexion (Table 1; Supplementary Figure 1). These findings highlight a measurable postsurgery recovery with the identified native knee, nonsymptomatic participants providing a realistic control.

In general, the control group had a greater, or similar, range of motion for the 6 degrees-of-freedom about the knee compared to the surgical group (Figure 4). The only condition in which the surgical group had a greater range of motion was in internal-external rotation during gait (3.5° [1.8, 6.2]; Figure 4C; Supplementary Tables 2 through 4). Range of motion was most similar during the overground walking tasks and most dissimilar during the high-flexion activities (Figure 4). Knee extension was the only activity in which the control group had a greater range of motion for all 6 degrees-of-freedom.

Fig. 4.

Range of motion difference between the surgical and control groups for 6° of freedom for chair rise, lunge, step down, gait, gait with 90° turn, and knee extension. A positive value indicates greater range of motion in the surgical group. (A) Flexion-Extension (°); (B) Varus-Valgus (°); (C) Internal-External (°); (D) Medial-Lateral (mm); (E) Anterior-Posterior (mm); (F) Superior-Inferior (mm). * indicates statistical significance (P < .05).

Results

Low-Point Kinematics

The anterior-posterior translation of the medial condyle was limited during weight-bearing activities for the surgical group (Figure 4A). Anterior translation of the medial low point for surgical implants was less than 1 mm across 100° knee flexion for lunge (Equation 5a) and the entire stance phase during gait (Equation 11a), and 2.4 mm across 100° knee flexion for knee extension (Equation 13a; Figure 4A). Despite directional differences, the translation of the lateral low point exceeded the medial low point translations for all activities (Equations 1, 5, 11b, 10d, 12b, and 13b), except knee extension (Equation 13).

For controls, increased knee flexion was associated with posterior translation of the medial condyle during chair rise (Equation 2a) and anterior translation during lunge (Equation 6a and Figure 5B). Alternatively, posterior translation of the lateral condyle occurred with increased knee flexion for all activities (Equations 2, 6, and 14b). Similar to the surgical group, lateral low-point translations exceeded medial translations during chair rise (Equation 2) and knee extension (Equation 14), but produced greater medial translations during lunge (Equation 6).

Fig. 5.

Anterior-posterior low-point translation. Anterior-posterior translation of the medial (red) and lateral (blue) low point for the surgical (A) and control (B) group. Solid lines indicate least-squares regression fits for the chair rise, lunge, step down, gait, gait with 90° with turn, and knee extension activities. Dotted lines represent ±1 standard error of the least-squares regression fit. Boxplots are reported in 20° increments across the flexion range (chair rise, lunge, and knee extension) and in 20% increments across the stance phase (step down, gait, and gait with 90° turn).

Direct comparison of anterior-posterior translation between the 2 groups indicated a general trend of greater translation for the control group during the high-flexion activities. The control group exhibited greater posterior translation with increased knee flexion of the medial (Equation 3) and lateral low point (Equation 4) during chair rise. A similar amount of posterior translation was demonstrated during lunge (Equation 8) on the lateral condyle between groups, but the native-knee controls presented a greater anterior translation of the medial compartment (Equation 7). During knee extension, the control group again exhibited greater posterior translation of the medial (Equation 15) and lateral low point (Equation 16) with increased knee flexion.

Center of Rotation

The primary pivot location was identified to be projected over the central aspect of the tibial tray for the surgical group during chair rise, lunge, gait, and gait with 90° turn (Figure 6A). Step down was the only activity that had a primary medial pivot, whereas medial was the second most common pivot location during chair rise, lunge, gait, and gait with 90° turn. Interestingly, knee extension for the surgical group exhibited a central-pivot to lateral-pivot paradigm. A central-pivot to medial-pivot paradigm emerged for the controls during chair rise, gait with 90° turn, and knee extension (Figure 6B). Lunge had a central-pivot to lateral-pivot while step down had a medial-pivot to lateral-pivot paradigm. Surprisingly, during gait, 50% of our control participants exhibited no pivot.

Fig. 6.

Transverse tibia center-of-rotation location. Center-of-rotation location for the surgical (A) and nonsurgical group (B) for the chair rise, lunge, step down, gait, gait with 90° turn, and knee extension activities. Locations are identified as medial (red; medial third), central (yellow; central third), lateral (blue; lateral third), and off the tibial plateau (black). Percentage of location and number of observations is color-coded for each group and activity.

Discussion

The objective of this study was to quantify in vivo knee-joint kinematics of a medial-pivot TKA implant and compare the performance with nonsymptomatic, native-knee controls during 6 functional activities. The range of motion about the knee was most similar during overground walking with the controls exhibiting a greater range of motion in the majority of the degrees-of-freedom in high-flexion activities. Consistent with previously published research in medial-pivot TKA, the surgical group produced greater lateral femoral condyle rollback than anterior-posterior medial compartment translation during weight-bearing activities [7,8,19,20]. Controls produced similar kinematic trends during chair rise and lunge, but generated much greater low-point anterior-posterior translations across similar flexion ranges (Figure 5). The pivot location for the surgical group during weight-bearing activities was central-pivot to medial-pivot, with step down being the only activity that yielded medial-pivot as the most common pivot location (Figure 6A). Controls also produced a central-pivot to medial-pivot paradigm, but varied more in locations across activities (Figure 6B).

One notable finding was that TKA patients who had received the medial-pivot implant were able to move through a similar range of motion as the controls during overground activities that require less than 25° of knee flexion (Supplementary Tables 2 through 4). In addition, the patients produced a greater range of internal-external rotation during comparable flexion-extension and anterior-posterior ranges of motion as the controls (Supplementary Table 4). This highlights the implant design constraints to allow pivoting motion while retaining anterior-posterior stability at mid-flexion ranges [8,21 ].

The range-of-motion differences between the surgical and control groups are evident during activities that require more than 75° of knee flexion. Despite similar flexion-extension ranges between groups during chair rise and lunge, the range of motion for the remaining degrees-of-freedom are markedly less in the surgical group (Supplementary Tables 2 through 4). This is more noticeable when the low-point translations are compared between the surgical and control groups. The predicted translations from 0 to 100° knee flexion during chair rise for the surgical group were 0.4 mm anterior (Equation 3b) and 3.4 mm posterior in the medial and lateral condyles, respectively. Controls produced posterior translation on both condyles for the same predicted range but at a much greater magnitude (medial: 5.9 mm posterior; lateral: 14.5 mm posterior). The reduced magnitudes in posterior translation are consistent with reported translations for other medial-pivot designs [22,23].

Lateral translations are not the only contributing factor to kinematic differences during weight-bearing activities. Predicted values during lunge from 20 to 120° knee flexion for the surgical group found 0.6 mm anterior translation for the medial condyle and 4.2 mm posterior translation for the lateral condyle compared with 8.4 mm anterior medial translation and 3.6 mm posterior lateral translation for controls. Our reported translations during lunge differ from other medial implant types. Shimmin et al [8] found a 3.2 and 5.6 mm posterior translation for the medial and lateral condyles with increased knee flexion during lunge. Scott et al [7] similarly reported 2 and 8 mm posterior translation across a 100° flexion range with Alesi et al [22] reporting 5.3 ± 0.9 mm for the medial and 10.9 ± 0.7 mm for the lateral condyle. None of these studies reported anterior translation of the medial condyle. However, performance of the studied implant during lunge produced translation patterns that more closely matched the nonsymptomatic, native-knee controls. Despite these differences, the currently studied implant supports the observation that medial and lateral low points are posterior to the anterior-posterior tibial midline, or tibial sulcus, during weight-bearing activities in a medial-pivot implant [8,24].

The emerging central-pivot to medial-pivot paradigm for the surgical group (Figure 6A) is reported for various implants [14] and for nonsymptomatic, native knees [15]. The central-pivot to lateral-pivot during knee extension for the surgical group is understandable in the context of the anterior translation of the medial low point with the nontranslating lateral low point (Figure 5A). This further highlights the difference in kinematics between weighted and unweighted conditions in medial-pivot implants and suggests the constraint of the medial compartment only occurs when the knee is weight-bearing. The pivot location of the controls was more varied across activities compared with Hamilton et al [15]; this discrepancy is likely due to differing sample sizes with Hamilton et al [15] providing data reported for 53 middle-aged and older adults (26 women, 50.8 ± 7.0 years).

This study presented several challenges. The main focus of the study was to initially quantify and validate the kinematic pattern of a single medial-pivot implant, and may not apply to all medially stabilized, congruent implant designs. Kinematics of the surgical group was not assessed preoperatively due to the nature of the HSSR data collection; the repeated X-ray exposure would have limited the number of activities and measures. As a result, no associations were identified between KOOS and kinematic outcomes at the time of kinematic assessment. The data from this study were intended to answer questions pertaining to kinematic performance after sufficient healing. Due to the convenience sample, all patients reported satisfaction with regained function and were observed with improved KOOS and strength assessments at the time of kinematic evaluation. We are unable to answer how performance changed over time as a function of healing or speak directly to thresholds in patient-reported outcomes. However, we were able to report high patient satisfaction despite reduced range of motion for the surgical cohort.

Data collection began in the fall of 2019 with the expectation to be completed by summer of 2020 but was ultimately delayed and limited due to the COVID-19 pandemic. The initial study sample was determined a priori to be powered at 10 participants. Subsequent power analyses determined that, given the kinematic trends of the provided sample of 7, the additional 3 participants would not significantly add to the expected power. Post-hoc power analyses showed that 7 surgical participants provided sufficient observed power to differentiate between medial and lateral translations in the activities (power ranging from 0.93 to 0.98). Due to this limited sample, statistical comparisons to quantify kinematic differences within groups were not performed. However, contribution of individual variability can be quantified by the linear mixed-effects equations reported in the paper (Marginal and Conditional R²; Supplemental Regression Equations), but kinematic trends can differ within groups ([25]; Supplementary Figure 1). Furthermore, we did not collect force-plate data for the controls to normalize overground walking to stance phase, limiting anterior-posterior condylar translation comparisons during gait trials. Despite this, transverse tibial pivot was quantified using the same methodology for the surgical and control groups.

In conclusion, the more constrained medial condyle of the medial-pivot implant design provides greater stability of the medial compartment and does not result in the paradoxical anterior translation at mid-flexion angles during weight-bearing activities, but is associated with less condylar translation than native knees. The transverse tibial pivot location occurs most commonly in the middle third of the tibial plateau and secondarily on the medial third. Slight variability in pivot location does occur between activities and is more in nonsymptomatic, native knee controls.