POTENTIAL IMPACT OF DIABETES MELLITUS ON CARTILAGE THICKNESS AND COMPOSITION IN SUBJECTS WITH AND WITHOUT OSTEOARTHRITIS – A MATCHED CASE-CONTROL STUDY

INTRODUCTION

Diabetes mellitus (DM) and osteoarthritis (OA) are interconnected through metabolic and inflammatory pathways that independently contribute to joint pain and structural degeneration [1]. Elevated blood glucose can induce systemic inflammation and oxidative stress, promoting joint symptoms and cartilage damage. Also, DM is frequently associated with obesity, potentially increasing mechanical loading and cartilage wear, particularly in weight-bearing joints.

OBJECTIVE

To assess the association of DM with femorotibial cartilage morphology and composition (T2 relaxation time), compared with matched controls without DM. Matching included age, sex, obesity status, knee pain, and radiographic OA (ROA) status. Analyses were stratified by the presence or absence of ROA.

METHODS

Participants were selected from the Osteoarthritis Initiative (OAI) [2]. A total of 362 individuals with DM were identified based on the Charlson Comorbidity Index. Of those, 260 were successfully matched to DM-negative controls based on the same/similar sex, age (±5 years), BMI (±5 kg/m²), WOMAC pain score (±5 on a 0–100 scale), pain frequency (±1 on a 0–2 scale), body height (±10 cm), and Kellgren-Lawrence (KL) grade [2]. Femorotibial cartilage thickness was derived from sagittal DESSwe MRIs at 3T using fully automated segmentation methodology. This involved a deep-learning-based pipeline combining 2D U-Net segmentation of subchondral bone and cartilage with atlas-based post-processing for subchondral bone area reconstruction [3]. Laminar cartilage T2 (deep 50%, superficial 50%) were calculated from MESE MRI (7 echoes), also using automated segmentation [3]. Statistical comparisons between DM and non-DM subjects were performed using paired t-tests, without correction for multiple comparisons across joint regions. For cartilage thickness, analyses were stratified by ROA status (KLG 2–4 vs. KLG 0–1). T2 analysis was restricted to KLG 0–2, as laminar T2 becomes less interpretable once cartilage loss is present.

RESULTS

DM participants were 63.4 ± 8.9y old, 53% female, BMI 31.5±4.5 kg/m². A total of 244 matched pairs were available with cartilage data at baseline (234 with thickness, 222 with T2; 78x KLG0, 46 × 1, 62 × 2, 52 × 3, 6x KLG4). In non-arthritic participants, the medial cartilage thickness was 3.45 mm (95% CI: 3.35–3.55) in DM subjects and 3.43 mm (3.33–3.54) in controls. Lateral thickness was 3.90 mm (3.80–4.00) in DM vs. 3.87 mm (3.76–3.97) in controls. Among ROA cases, medial thickness was 3.16 mm (3.03–3.29) in DM vs. 3.30 mm (3.17–3.42) in controls; lateral thickness was 3.68 mm (3.53–3.83) vs. 3.76 mm (3.64–3.88), respectively. None of the DM vs. non-DM differences reached statistical significance. In the 170 matched pairs that were KLG 0–2, no significant differences in cartilage T2 were identified: In the medial superficial layer, T2 was 48.2 ms (47.7–48.7) in DM vs. 48.7 ms (48.2–49.3) in controls, and in the deep layer, 37.4 ms (37.0–37.7) vs. 37.7 ms (37.4–38.1). Laterally, superficial T2 was 47.0 ms (46.6–47.5) in DM vs. 47.5 ms (47.0–47.9) in controls, and in the deep layer, 36.3 ms (36.0–36.6) vs. 36.4 ms (36.1–36.7).

CONCLUSION

This study utilized state-of-the-art 3T MRI data from the OAI and fully automated, deep-learning-based methods for evaluating cartilage morphology and composition [3]. The findings suggest that DM status, when tightly matched for demographic and clinical factors (especially BMI and pain), is not substantially associated with reduced cartilage thickness or increased T2 relaxation times. Future analyses will have assessed the impact of less stringent matching—particularly for BMI and pain—and will explore longitudinal changes in these participants.