Background Erosive osteoarthritis (OA), a subset of hand OA with high disease burden, is hypothesized to develop in distinct anatomical phases. These can radiographically be identified as loss of joint space (pre-erosive phase), followed by erosions and remodelling (E- and R-phase, according to the Verbruggen-Veys (VV) system). It is unknown which disease processes play a role in this development, although synovitis and bone marrow lesions (BMLs) on MRI have shown to be associated with radiographic progression in OA finger joints. Therefore synovitis and BMLs could be a target for therapy.
Objectives To investigate the presence of synovitis and BMLs in erosive compared to non-erosive joints, and to study the effect of TNF inhibition on these MRI abnormalities.
Methods Analyses were performed in a subset of patients participating in a double-blind trial (NTR 1192), comparing etanercept (ETN, 25–50mg/week) to placebo. Patients with symptomatic erosive OA with clinical and ultrasonographic signs of inflammation in ≥1 interphalangeal joint (IPJ) were included. This analysis concerned 20 patients who underwent contrast-enhanced MRI of the 8 distal and proximal IPJs of one hand at baseline and 1 year. Images were scored for synovitis and BMLs (0–3 per joint, total score 0–24), blinded for patient characteristics, according to the OMERACT hand OA MRI score (HOAMRIS). Radiographs of the same hand were scored according to the VV system. Logistic regression with generalised estimating equations (GEE) were used to associate the presence of an MRI feature in a joint with being in an erosive (E/R) versus non-erosive anatomical phase (N=normal/S=stationary) (adjusted for age and sex, excluding IPJs from the ETN group at 1 year). A treatment effect of ETN on the change in MRI features over 1 year on joint level was also estimated with GEE.
Results The total synovitis score was similar at baseline and 1 year in both the placebo (median (range): 10.5 (7–16) and 10.3 (6–14) respectively) and ETN group (7.0 (3–17) and 6.9 (3–14)). For BMLs, the total score was 5.4 (2–9) and 7.0 (0–9) in the placebo, versus 4.5 (3–9) and 3.7 (0–8) in the ETN group. The presence of BMLs was associated with being in E-phase (OR 12.5, 95% CI 2.1–74.7) and R-phase (OR 3.9, 1.2–12.0). Synovitis was not associated with these anatomical phases. Synovitis did not change in either group, whereas BMLs decreased more in the ETN compared to the placebo group (table). The same effect was observed in analyses on patient level (total score and number of IPJs with a feature, data not shown). Besides, an interaction was observed between ETN treatment and the presence of grade 2/3 synovitis at baseline in that IPJ (table).
Conclusions The presence of BMLs, but not synovitis, was associated with the presence of an erosive anatomical phase in a joint. Treatment with ETN appeared effective in inhibiting these lesions, suggesting a role for TNF in the pathophysiology of erosive OA, possibly through an effect on subchondral bone. The inhibitory effect of ETN on BMLs was more pronounced in IPJs with severe synovitis at baseline, suggesting that inflamed synovial tissue could be a source of TNF production in erosive OA.
Disclosure of Interest None declared