Abstract
Purpose
Identifying knee osteoarthritis patient phenotypes is relevant to assessing treatment efficacy. Biomechanics have not been applied to phenotyping, yet features may be related to total knee arthroplasty (TKA) outcomes, an inherently mechanical surgery. This study aimed to identify biomechanical phenotypes among TKA candidates based on demographic and gait mechanic similarities, and compare objective gait improvements between phenotypes post-TKA.
Methods
Patients scheduled for TKA underwent 3D gait analysis one-week pre (n=134) and one-year post-TKA (n=105). Principal Component Analysis was applied to frontal and sagittal knee angle and moment gait waveforms, extracting the major patterns of gait variability. Demographics (age, gender, BMI), gait speed, and frontal and sagittal pre-TKA gait angle and moment PC scores previously found to differentiate gender, osteoarthritis severity, and symptoms of TKA recipients were standardized (mean=0, SD=1). Multidimensional scaling (2D) and hierarchical clustering were applied to the feature set [134×15]. Number of clusters was assessed by silhouette coefficients, s, and stability by Adjusted Rand Indices (ARI). Clusters were validated by examining inter-cluster differences at baseline, and inter-cluster gait changes (PostPCscore–PrePCscore, n=105) by k-way Chi-Squared, Kruskal-Wallace, ANOVA and Tukey's HSD. P-values <0.05 were considered significant.
Results
Four (k=4) TKA candidate groups yielded optimum clustering metrics (s=0.37, ARI=0.57). Cluster 1 was a compact (n=7) male cluster, walking with faster gait speeds (1.20.2m/s, 3<2<1,4, P<0.001) and higher adduction moment magnitudes (PC1, 3,4<2,1, P<0.001). Cluster 1 had the most dynamic kinematic (stance-phase flexion angle range PC4, 3,4,2<1, P<0.001) and kinetic (flexion moment range PC2, 3<2<4<1, P<0.001; adduction moment range PC2, 3,2<4<1, P<0.001 and PC3, 3,2<1, P=0.001) loading/un-loading range patterns among the clusters. Cluster 1 represented a higher-functioning (less “stiff-kneed”) male subset, most resembling asymptomatic patterns. Cluster 2 was also mostly males (44/47), demonstrating adduction moment magnitudes (PC1) comparable to Cluster 1. However, Cluster 2 was older (67.07.4years, 1,4<2, P=006), walking with slower gait speeds (0.80.2m/s), and less flexion moment (PC2) and adduction moment (PC2) range; representing an older, “stiff-kneed” male subset. Cluster 3 was mostly females (32/34) with the slowest gait speeds (0.70.1m/s), the lowest overall flexion angle magnitudes (PC1, 3<2,4,1, P<0.001), stance-to-swing flexion angle (PC2, 3<2,1, P=0.004) and flexion moment range (PC2). Cluster 3 captured a slow female subset, with the “stiffest-kneed” gait among the clusters. Cluster 4 was mostly females (43/46) with faster gait speeds (1.00.1m/s) and less stiff kinematic and kinetic patterns relative to Clusters 2–3, representing a higher-functioning female phenotype. Post-TKA, higher-functioning clusters demonstrated less dynamic gait improvement (flexion angle ΔPC2, 1,4,2<3, P<0.001; flexion moment ΔPC2, 4<2,3, P=0.009; adduction moment ΔPC2, 1<3, P=0.01), with some sagittal range patterns decreasing post-operatively.
Conclusions
TKA candidates were characterized by four clusters, differing by demographics and biomechanical severity features. Pre-TKA, stiff-kneed clusters (2 and 3) had less dynamic loading/un-loading kinetics. Post-TKA functional gains were cluster-specific; stiff-kneed clusters experienced more improvement, while higher-functioning clusters demonstrated some functional decline. Results suggest the presence of cohorts who may not benefit functionally from TKA. Cluster profiling may aid in triaging and developing osteoarthritis management and surgical strategies that meet individual or group-level function needs.