P077 Can we predict who responds to intra-articular corticosteroid injection in knee arthritis? A clinical trial comparing response in inflammatory arthritis and osteoarthritis

Abstract

Background/Aims The global burden of knee arthritis is significant and rising. Patients can experience pain and disability over a long duration. Intra-articular corticosteroid injections (IACI) can reduce inflammation and pain in arthritis. However, there is disparity in pain reduction and efficacy duration following IACI. We aimed to identify factors that influence IACI response in knee osteoarthritis (OA) and inflammatory arthritis (IA), to aid stratification of future patients. Methods In this multi-centre prospective study, participants with knee OA and IA (rheumatoid, psoriatic or mixed arthritis on standard treatment) were recruited. At baseline, demographics, DAS28, PsARC, painDETECT and Western Ontario and McMasters Universities Arthritis Index (WOMAC) for pain, stiffness and function were completed. We classified painDETECT scores with ≥19 “sensitised” and ≤18 “non-sensitised” for pain. Sonographic assessment and IACI was performed, with synovitis/effusion categorised dichotomously (present/absent). Two musculoskeletal radiology consultants graded knee radiographs as low Kellgren-Lawrence grade 0-2 (LKLG) or high grade 3-4 (HKLG). At 3 months, all questionnaires were repeated. IACI response was defined as 20% improvement from baseline WOMAC pain. Differences between responders/non-responders were assessed using Wilcoxon (continuous variables) and Fisher’s exact tests (categorical variables). Analysis was conducted in R. Results There were 129/136 patients recruited who had IACI: 89 were analysed (62 OA, 27 IA) (Table 1), with 40 lost to follow-up. Irrespective of arthritis pathology, odds of IACI response among sensitised versus non-sensitised participants was 0.33(p = 0.051). In the LKLG-OA cohort, responders had significantly lower baseline painDETECT scores than non-responders (p = 0.015); 10/14 (71%) non-sensitised responded to IACI, compared with 0/6 (0%) of the sensitised. No difference in baseline painDETECT scores between responders and non-responders in the HKLG-OA cohort(p = 0.50) or in either inflammatory cohort(LKLG p = 0.86, HKLG p = 0.77) was detected. In the OA-synovitis/effusion cohort, responders had lower baseline painDETECT scores; while this was not statistically significant (p = 0.13), 0/6 (0%) of sensitised participants responded, versus 17/33 (52%) non-sensitised patients. Conclusion Our study suggests imaging and clinical measures can predict IACI response, with pain sensitisation conferring lower likelihood of response, particularly in OA with LKLG. Following validation studies, design of a predictive clinical model could allow patient stratification for personalised treatment, reducing IACI waiting times and optimising resource allocation. Disclosure A. Paterson: None. K. Feather: None. A. Lambarth: None. M. Siebachmeyer: None. V. Ejindu: None. F. Howe: None. A. Rudnicka: None. A. Ezeonyeji: None. R. Ramsden: None. N. Sofat: None.