A new method to measure anatomic knee alignment: a tool for large studies of OA?

Abstract

Purpose: Knee malalignment alters load distribution across the articular surface and is a potent predictor of knee osteoarthritis (OA) progression. The hip-knee-ankle angle (HKA) provides the most direct measurement of biomechanical leg alignment, but requires a long-limb radiograph. The femorotibial alignment angle (FTA) of the knee joint is an indirect measure of mechanical alignment but can be performed using a standard knee radiograph. Traditional methods of measuring the FTA require a reader to identify landmarks in the tibial spines, which can be difficult and unreliable for abnormally positioned knee images. We have developed and validated a new software method to rapidly determine FTA that may expedite the measurement of knee radiographs from large studies such as the Osteoarthritis Initiative (OAI). Methods: The method is based on the coordinate system developed for location specific joint space width (JSW) measurements, previously used to demonstrate that quantitative radiographic measures are comparable with MRI for detecting OA progression. The coordinate system directly determines the orientation of the femur (Figure 1) and also defines a central point of the knee. The axis is perpendicular to a line tangent to the base of the femoral condyles, and centered between the outer aspects of the condyles. Four points on margins of the tibial shaft define the axis of the tibia. Two points are marked 10 cm from the center point and two are placed 1 cm from the base of the tibial plateau. The study used subjects from the Progression Cohort of the Osteoarthritis Initiative (OAI) (OAI Datasets 0.1.1, 0.B.1, and 1.B.1) where HKA measurements were available from the OAI. We excluded images where the center of the knee was less than 10 cm from the lower edge of the image, leaving 266 total knees from 142 subjects for the final analysis. The study used two readers. Reader 1 read the data twice with the new procedure and two times using a traditional method. Reader 2 performed a single reading with each of the two methods. Linear regression analysis was used to assess reader reproducibility, and correlation with the HKA measurement. Results: Table 1 provides the results comparing the two methods. The new method of measuring FTA was related to HKA by FTA = 0.52HKA - 4.7 degrees. The traditional method was related to HKA by FTA = 0.65HKA + 1.5 degrees. The additional reader time was less than 30 seconds per image once an image had been processed for the location-specific JSW measurement. Conclusions: We found improved inter and intra reader reproducibility compared with the traditional method and higher correlation with the HKA. With the new method, very little difference was observed between the inter and intra reader precision suggesting a robust technique. The fast reader time implies that assessment of large numbers of subjects is feasible. Furthermore, the method does not require landmarks placed in the tibial spines; it can be performed by a relatively unskilled reader whose task is limited to marking points on the shaft of the tibia and the femoral condyles. A limitation of the method is that it relies on the coordinate system set up for location specific JSW; additional time is necessary if this step has not already been performed. However, setting up the coordinate system is substantially automated and could be performed independent of measuring JSW if necessary. In summary, we have documented a new technique that can provide a precise, accurate, rapid and reproducible measurement of radiographic knee anatomic alignment, with significant implications for efficiency and cost-effectiveness for very large studies of knee OA.