Background/Purpose: The RA-MAP consortium is a UK flagship industry-academic partnership investigating clinical and biological predictors of disease outcome in patients with rheumatoid arthritis (RA).

Methods: One RA-MAP workstream investigated clinical predictors of remission by collation of patient-level data from non-biologic arms of academic- and industry-sponsored clinical trials, completed since 2002, evaluating therapies in early inflammatory arthritis and RA. Anonymised data on key baseline and follow-up variables were requested and amalgamated. Logistic regression was used to identify predictors of remission at 6 months, as defined by DAS28-ESR. Initial models considered separate effects of potential predictors after adjusting for age, gender, disease duration, race and treatment arm. Multivariate models were built with variables important at initial screen and with low percentage of missing values.  Latent class mixed models were used to characterise disease activity over time, adjust for potential predictors and introduce random effects, and to cluster trajectories that may identify clinically important sub-populations. Both Akaike and Bayesian information criteria were used to decide on number of clusters.

Results: 19 trials were received, comprising 3555 patients with over 30,000 baseline and follow-up records. The non-biologic arms included treatment with placebo, intramuscular steroids, methotrexate or other DMARDs, or methotrexate in combination with other drugs. 78% of patients were female, 86% white and 72% rheumatoid factor positive. Mean age at onset (SD) was 46.7 (13.9) years. Mean age (SD) and median disease duration (IQR) at entry were 52.8 (12.8) years and 3 (1-9) years respectively. Mean baseline DAS28-ESR, 28-swollen and tender joint counts (SDs) were 6.4 (1.1), 12 (6) and 14.7 (7.2). Remission rate, 6 months from entry, was 11% (95%CI 9.8%-12.3%). In the final multivariate logistic model, remission at 6 months was predicted by being white, male, younger, never smoked, RF negative, methotrexate naïve at entry, randomised to methotrexate singly or in combination, better functional health and lower levels of baseline disease activity (Table 1).  Latent class mixed modelling suggested patients could be clustered into sub-populations that included fast improver and non-responder groups and one or two intermediate classes – a slower improver group and a possible sub-population that improved in the first 6 months then showed no further improvement.

Conclusion: Through this novel partnership, the RA-MAP project has confirmed several known factors related to non-biologic induced remission. The remission rate of 11% provides a benchmark for non-biologic arms of future RA trials. The potential to stratify RA patients into distinct disease activity trajectories was demonstrated and may become the basis of personalising treatment choices.

Table 1: Logistic regression for remission

« Back to 2015 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/evaluation-of-patients-from-the-non-biologic-arms-of-inflammatory-arthritis-clinical-trials-identifies-several-predictors-of-remission-as-well-as-distinct-responder-subgroups/