Summary

The accuracy of the three dimensional evaluation of the femoral bowing was higher than two dimensional evaluation, and as the grade of the osteoarthritis of the knee became high, the femoral lateral bowing became large

Abstract

Anatomically femoral bone has bowing. Many reports suggested that the femoral bowing affected the likelihood of bone cut error in total knee arthloplasty (TKA). Femoral bowing has been evaluated using two dimensional (2D) radiography, but accuracy of 2D evaluation seemed to be not high. Purpose of this study was to suggest a novel method to evaluate accurate three dimensional (3D) femoral bowing and compare 3D and 2D femoral bowing in patients with knee osteoarthritis (OA) to examine the possible superiority of 3D femoral bowing analysis.
Femoral CT images and full-leg anterior-posterior radiographic images of 74 patients with knee OA were enrolled in this study. The xyz-coordinate system was introduced in 3D femoral bone CAD-model. Reference plane of the coordinate system was defined with three bony landmarks, medial/lateral epicondyles and center of the femoral head.
The cross-sectional contours of femoral canal (cancellous/cortical border) were extracted along the Z axis. The range of the cross-sectional slicing was set between the lesser trochanter and the distal end of the epiphysis. For each extracted cross-sectional contour, a least-square fitted ellipse was calculated. A least-square line was fitted to the centers of the cross-sectional ellipses. Proximal and distal anatomical axes were calculated with proximal and distal half of the ellipse data, respectively. Angle between these two axes was measured and defined as total bowing of the femur. Directions of lateral/anterior bowing were defined as the plus directions. Proximal and distal axes were made to project to these YZ and XZ planes, and angles of coronal and sagittal bowing were examined. Next, the femoral bone on the full-leg radiograph films was divided into two parts by the center line of the femoral canal. The proximal and distal axes were set grossly, and the angle between these axes was measured. We compared the difference in femoral bowing between 2D and 3D.
The average of 2D coronal bowing was 0.6±3.2 degrees, and the absolute value of the difference between the 2D and 3D coronal bowing was 1.7±1.5 (0.1-6.4) degrees. The averages of 2D coronal, 3D coronal, and 3D sagittal bowing for the moderate OA group (Kellgren-Lawrence (K-L) grade 0~2, n=43) and severe OA group (K-L grade 3,4 n=31) are examined. Statistically, the severe OA group showed greater coronal bowing compared to the moderate OA group. The correlation coefficients between 2D coronal bowing and 3D coronal bowing, and between 2D coronal bowing and 3D sagittal bowing, were 0.70 (p<0.01) and 0.36 (p<0.01) in all OA knees, 0.72 (p<0.01) and 0.17 (p=0.28) in moderate OA knees, and 0.57 (p<0.01) and 0.42 (p=0.018) in severe OA knees.
The severe OA had greater coronal bowing compared to the moderate OA. The difference in femoral coronal bowing between 2D radiography and 3D CT was average 1.7 degrees, and can be as much as 6.4 degrees. 2D femoral bowing analysis has limited accuracy in severe knee OA and would lead to patient outliers in TKA. Thus, for precise TKA, we recommend preoperative planning and analysis of femoral bowing using 3D CT.