Novel statistical methodology reveals that hip shape is associated with incident radiographic hip osteoarthritis among African American women

Abstract

Introduction: Hip shape is a risk factor for the development of hip osteoarthritis (OA), and current methods to assess hip shape from radiographs are limited; therefore this study explored current and novel methods to assess hip shape.

Methods: Data from a prior case-control study nested in the Johnston County OA Project were used, including 382 hips (from 342 individuals). Hips were classified by radiographic hip OA (RHOA) status as RHOA cases (baseline Kellgren Lawrence grade [KLG] 0 or 1, follow-up [mean 6 years] KLG ≥ 2) or controls (KLG = 0 or 1 at both baseline and follow-up). Proximal femur shape was assessed using a 60-point model as previously described. The current analysis explored commonly used principal component analysis (PCA), as well as novel statistical methodologies suited to high dimension low sample size settings (Distance Weighted Discrimination [DWD] and Distance Projection Permutation [DiProPerm] hypothesis testing) to assess differences between cases and controls.

Results: Using these novel methodologies, we were able to better characterize morphologic differences by sex and race. In particular, the proximal femurs of African American women demonstrated significantly different shapes between cases and controls, implying an important role for sex and race in the development of RHOA. Notably, discrimination was improved with the use of DWD and DiProPerm compared to PCA.

Conclusions: DWD with DiProPerm significance testing provides improved discrimination of variation in hip morphology between groups, and enables subgroup analyses even under small sample sizes.

Keywords: Hip morphology; Hip osteoarthritis; Linear discriminant analysis; Principal component analysis; Racial differences.

Figure 1

Mean femur curves of participants by sex and by race. Red curves correspond to the mean shape for incident RHOA cases and blue curves correspond to the mean shape for incident RHOA controls.

Figure 2. DiProPerm test results on African American females for the difference between incident RHOA cases and controls

The black, red and blue curves show the smoothed histograms of the entire set of African American females, incident RHOA cases, and controls, respectively. The black dots in the right panel represent the simulated mean difference (MD) statistics. The black curve shown is the smoothed histogram of the MD values, so the area under the curve equals 1. The green vertical line shown on the right side of the black curve represents the observed between-group MD value, while the p-value is the proportion of simulated MD values exceeding the observed value.

Figure 3. ROC curves for the DWD direction and the PC directions among African American women. (a) shows the ROC curve of the discrimination rule based on the projections on the DWD direction. (b), (c), (d) and (e) show the ROC curves of the discrimination rule based on the projections on the 1st, 2nd, 3rd and 4th PC directions, respectively

The AUC is reported on the right side of each figure.

Figure 4. Proximal femur shape variation between the incident RHOA cases and controls. (a) shows femur shape variation on the DWD direction (p=0.03); (b) shows femur shape variation on the 3rd principal direction (p=0.052)

The mean shape is shown in green, incident RHOA cases in red, and incident RHOA controls in blue.

Figure 5

DiProPerm test result for the difference between the males (red) and females (blue) among incident RHOA cases.