Abstract
Osteoarthritic (OA) changes to the bone morphology of the proximal tibia may exhibit load transfer patterns during total knee arthroplasty not predicted in models based on normal tibias. Prior work highlighted increased bone density in transverse sections of OA knees in the proximal-most 10mm tibial cancellous bone. Little is known about coronal plane differences, which could help inform load transfer from the tibial plateau to the tibial metaphysis. Therefore, we compared the cancellous bone density in OA and cadaveric (non-OA) subjects along a common coronal plane.
This study included nine OA patients (five women, average age 59.1 ± 9.4 years) and 18 cadaver subjects (four women, average age 39.5 ± 14.4 years). Patients (eight with medial OA and one with lateral OA) received pre-operative CT scans as standard-of-care for a unicompartmental knee replacement. Cadavers were scanned at our institution and had no history of OA which was confirmed by gross inspection during dissection.
3D reconstructions of each proximal tibia were made and an ellipse was drawn on the medial and lateral plateau using a previously published method. A coronal section (Figure 1) to standardize the cohort was created using the medial ellipse center, lateral ellipse center, and the tibial shaft center 71.5mm from the tibial spine. On this section, profile lines were drawn from the medial and lateral ellipse centers, with data collected from the first subchondral bone pixel to a length of 20mm. The Hounsfield Units (HU) along each profile line was recorded for each tibia; a representative graphical distribution is shown in Figure 2. The Area Under the Curve (AUC) was calculated for the medial and lateral sides, which loosely described the stiffness profile through the region of interest. To determine differences between the medial and lateral subchondral bone density, the ratio AUC[medial] / AUC[lateral] was compared between the OA and cadaver cohorts using a two-sample t-test. Data from the sole lateral OA patient was mirror-imaged to be included in the OA cohort.
The majority of the OA patients appeared to have higher subchondral bone density on the affected side. Figure 3 compares the medial and laterals sides of each group using the AUC ratio method described above. For the cadaver group the AUC was 1.2 +/− 0.22, with a median of 1.1 [0.9 1.6], smaller than the mean AUC for the OA group, which was 1.4 +/− 0.39, with a median of 1.6 [0.93 2.1]. The p-value was 0.06.
The increased density observed in OA patients is consistent with asymmetric loading towards the affected plateau, resulting in localized remodeling of cancellous bone from the epiphysis to metaphysis. From the coronal plane, bone was often observed in OA patients bridging the medial plateau to the metaphyseal cortex. Although the cadaver subjects were normal from history and gross inspection, some subjects exhibited early bone density changes consistent with OA. Future work looks to review more OA scans, extend the work to the distal femur, and convert the HU values to bone elastic moduli for use in finite element modelling.
