State-dependent crossover between subchondral bone density distribution and cartilage thickness in healthy and osteoarthritic knees: a CT-osteoabsorptiometry and 3D-MRI study

Subchondral bone density distribution, quantified by CT-osteoabsorptiometry as the medial ratio, reflects mediolateral load. We examined whether it tracks cartilage thickness, whether it outperforms alignment, and whether this differs between healthy and osteoarthritic (OA) knees.

This retrospective study analyzed 83 OA and 16 non-OA knees. The medial ratio was measured by CT-osteoabsorptiometry, the hip-knee-ankle (HKA) angle radiographically, and cartilage thickness on 3.0-T MRI, as the mediolateral difference and in nine subregions.

The medial ratio explained the thickness difference no better than HKA (R² = 0.21 and 0.24) with no independent association. The relationship reversed with joint state: in non-OA knees a more medial bone-density distribution accompanied thicker medial cartilage (ρ = +0.68), in OA knees thinner cartilage (ρ = -0.17; Fisher z P = .001). The reversal held against varus knees alone (P = .004) and was absent for alignment (P = .090). Regionally, the medial ratio correlated with thickness in only two subregions, mid-central and mid-external (ρ = -0.40 and - 0.38, Bonferroni-corrected P < .005). This regional association persisted within varus knees, where compartment-level metrics did not.

The bone density-cartilage association was positive in healthy knees and negative in OA knees, a reversal not seen for alignment. Within varus knees the medial ratio was related to cartilage thickness in two subregions of the medial plateau, where alignment and compartment-level measures were not. It may therefore be a useful indicator of the state of the joint and of the regional distribution of cartilage loss within it.