ObjectivesTo identify, by a systematic literature review, predictors of clinical response to methotrexate (MTX) treatment in rheumatoid arthritis (RA) patients, which would facilitate personalised treatment.MethodsPubMed and Embase databases were searched for original articles. Additionally, congress abstracts of European League Against Rheumatism (EULAR) and American College of Rheumatology (ACR) annual meetings of the past 2 years were screened. Articles describing baseline predictors of clinical response to MTX after 3 to 6 months were included, since this reflects the time span used to determine treatment effectiveness and decide on treatment changes in treat-to-target recommendations.Results30 articles were included, containing 100 different predictors and 11 predictive models, but only 19 predictors and 2 predictive models were studied in multiple similar RA cohorts. Of these 19 predictors, 4 were described in multiple articles sharing details on statistical analyses and response outcomes, allowing pooling of the predictive values of these 4 factors (Figure 1). Conflicting non-statistically significant associations between age and response were found (Figure 1a).1-4 Female gender was found to be a predictor of non-response in two studies (OR (odds ratio) 0.55 and 0.54), but these findings could not be replicated in two other studies (Figure 1b).1-4 In two studies, smoking predicted non-response (adjusted OR 0.35 and 0.60), but this was inconsistent over all response criteria assessed (Figure 1c).3,5 Higher disease activity score was a predictor of response in two studies (adjusted OR 1.12 and 2.7), conflicting and non-statistically significant predictive values were found in two other studies (Figure 1d).2-4,6 Rheumatoid factor positivity predicted non-response in two studies (adjusted hazard ratio 0.61, adjusted OR 0.4), although this was not found in three other studies.2-4,7,8 Heterogeneity between studies prohibited pooling of predictive values. Additionally, a validated epigenetic model was found: area under the receiver operating characteristic curve 0.90 in the development cohort and 0.91 in a validation cohort.9ConclusionNo predictors were identified reliably predicting clinical response to MTX after 3 to 6 months in the individual patient: clinical predictors were weak. However, a promising epigenetic model was found that needs further validation. Future studies on prediction of response to MTX should focus on combining clinical characteristics with genetic and other laboratory biomarkers and on predicting outcome after 3 to 6 months of treatment, to get closer to personalised medicine.References[1] Kavanaugh, et al. Ann Rheum Dis2017;76:822-3.[2] Ponchel, et al. Ann Rheum Dis2014;73:2047-53.[3] Saevarsdottir, et al. Ann Rheum Dis2011;70:469-75.[4] Wessels, et al. Arthritis Rheum2006;54:2830-9.[5] Saevarsdottir, et al. Arthritis Rheum2011;63:26-36.[6] Tan, et al. Int J Rheum Dis2016;19:482-9.[7] Bugatti, et al. Ann Rheum Dis2017;76:790-1.[8] Stühlmuller, et al. Clin Immunol2016;171:50-61.[9] Carini,...