QUANTITATIVE MEDIAL SUBCHONDRAL BONE ATTRITION FROM MRI IS ASSOCIATED WITH KNEE OSTEOARTHRITIS SEVERITY AND SYMPTOMS.

Authors

Abstract

Subchondral bone plays a key role in load transmission in knee OA, yet existing semi-quantitative (SQ) OARSI attrition scores are subjective and may be limited in detecting early structural changes. A quantitative measure of subchondral bone attrition (qSBA) may provide a more sensitive and objective characterization of these alterations. 1) To develop an automated MRI-based method to quantify qSBA. 2) To evaluate the cross-sectional associations of qSBA with radiographic knee OA severity and knee symptoms. This study included 91 knees from the OAI baseline visit, a longitudinal cohort of adults aged 45-79 years at risk of or with knee OA. We included all knees with baseline medial compartment SQ attrition score of 1-2 (n=33). A comparison group with SQ attrition score 0 was randomly sampled in each stratum of subcohort (incidence, progression and control) and Kellgren-Lawrence grade (KLG) (n=58). Sagittal knee MRI scans were acquired using a 3D DESS WE sequence (in-plane spatial resolution 0.365 × 0.456 mm; slice thickness 0.7 mm). Tibial masks were obtained using a deep learning segmentation model (nnU-Net) with subsequent manual correction where required. An automated pipeline was developed to quantify qSBA on sagittal slices of the medial tibial plateau. The medial compartment was identified using anatomical subdivision rules consistent with the Whole-organ MRI score (WORMS). Within each medial sagittal slice, a reference line was defined based on tibial axis (see Figure footnote). Attrition depth was computed at each anterior-posterior location as the perpendicular distance between the observed bone surface and the reference line. Slice-level measurements were then aggregated across medial slices to derive summary metrics, including maximum and mean depth (mm), and total attrition area (mm²) (Figure). Construct validity was assessed using Spearman correlation between qSBA and SQ medial attrition scores. Associations between qSBA and knee outcomes (KLG, WOMAC and KOOS) were evaluated using multivariable linear regression adjusting for age, sex, and BMI, with additional models further adjusting for KLG and SQ attrition scores. Multiple comparison was adjusted using Benjamini-Hochberg method. In the analytic sample, participants had a mean age of 64.3 (SD 9.3) years and BMI of 28.9 (SD 4.8); 49% were female and 84% were White. Mean qSBA showed a moderate positive correlation with SQ attrition scores (Spearman ρ=0.51, 95% CI 0.32-0.67). In adjusted models, higher KLG was associated with greater qSBA, with KLG3 showing 0.47 mm (p=0.062) and KLG4 showing 0.88 mm (p<0.001) larger mean attrition compared to KLG0. Greater qSBA was associated with worse symptoms across multiple domains (Table). In models adjusted for age, sex, and BMI, higher qSBA was associated with worse WOMAC and KOOS. After additional adjustment for KLG and SQ attrition, associations were attenuated but remained evident for KOOS symptoms, suggesting that qSBA captures aspects of symptom variation not fully reflected by visual semi-quantitative grading. These findings support the hypothesis that qSBA reflects structural features of attrition validated against established semi-quantitative scoring systems. The observed associations with knee symptoms, particularly after accounting for radiographic severity and SQ attrition scores, suggest that qSBA may capture clinically relevant variation not fully represented by existing measures. As a quantitative and spatially resolved metric, qSBA may complement current scoring approaches and facilitate more detailed characterization of subchondral bone changes. Future work will extend this approach to the lateral tibia and will examine whether changes in qSBA can reliably and validly track disease progression.

Tags

Topics