Quantification of bone shape and cartilage thickness using 3D MRI and their interrelationship in osteoarthritic knees of a Chinese cohort

Abstract

Purpose: Bone geometry plays an important role in joint biomechanics and pathogenesis of osteoarthritis. To date, no studies have evaluated directly the association between cartilage thickness and bone shape. The goals of this present study are tow-fold: 1) To investigate the distribution of cartilage thickness and bone shape among different Kellgren-Lawrence (KeL) gradings; and 2) to explore the correlation between bone shape and cartilage thickness. Methods: 50 patients (54.92 ± 11.62 years, BMI 24.37 ± 2.96 Kg/m 2 , 25 females) were scanned by 3.0 T MRI at our hospital (10 knees for each KeL grading from 0e4, matched by age, gender and BMI between KL grades). MRI imaging protocols included sagittal T1 fast spin echo (FSE) and sagittal fat-suppressed 3D fast spoiled gradient echo (FSPGR). Cartilage was segmented semi-automatically on FSPGR images into lateral/medial femoral condyle (LF/MF), lateral/medial tibia (LT/MT), patella (P) and trochlea (TRO) using in-house developed software. The average thickness was calculated for each compartment. The tibia and femur were segmented semi-automatically on FSE images, and the bone shape was quantified using statistical shape modeling (SSM) as developed in-house previously. Principal component analysis (PCA) of automatically defined and matched 3D landmarks was used to extract a compact signature of the shape. The first 20 femur and tibia modes were analyzed (T1eT20, F1eF20). The difference of cartilage thickness and bone shape modes among the 5 groups were analyzed by using ANOVA, respectively. Partial Pearson Correlation was calculated to explore the correlation between cartilage thickness and bone shape, adjusted for age, gender, BMI and K-L grading. The level of significance was P < 0.01. Results: Significance was observed in the cartilage thickness of patella and MT among the 5 KL groups (Figure 1A), and in one femur (F5) and two tibia (T6, T7) bone shape modes (Figure 1B). The area of tibia plateau increased from KeL 2 to K-L 4 (T7); posterior slope to tibia plateau increased from KeL 1 to K-L 2 (T6); The contour of medial femoral condyle differed contrarily between from KeL 0 to KeL 2 and from KeL 2 to KeL 4 (F5). There are correlations between cartilage thickness and bone shape in three femur (F10, F11, F16) and two tibia (T5, T18) modes (Table 1). Increasing medially of MF (F10) and increasing posteriorly of LF (F11) as well as increasing superiorly of intercondylar eminence (T18) are associated with more cartilage degeneration at LT (thinner cartilage). Increasing inferiorly of trochlea groove (F11) is associated with more cartilage degeneration at femoral trochlea. Increasing of tibia plateau is associated with more cartilage denegation at MT. Figure 2 illustrates the physical meaning of the bone shape modes with significance. Conclusions: Statistical shape model and 3 D MRI are powerful tools for quantifying bone shapes. Significant correlation was found between bone shapes, especially increasing medially of MF, increasing posteriorly of LF, increasing super...