Objective: To create a tool to predict probability of remission and low disease activity (LDA) in patients with RA being considered for anti-TNF treatment in clinical practice.
Methods: We analysed data from GO-MORE, an open-label, multinational, prospective study in biologic-naïve patients with active RA (DAS28-ESR ⩾3.2) despite DMARD therapy. Patients received 50 mg s.c. golimumab (GLM) once monthly for 6 months. In secondary analyses, regression models were used to determine the best set of baseline factors to predict remission (DAS28-ESR <2.6) at month 6 and LDA (DAS28-ESR ⩽3.2) at month 1.
Results: In 3280 efficacy-evaluable patients, of 12 factors included in initial regression models predicting remission or LDA, six were retained in final multivariable models. Greater likelihood of LDA and remission was associated with being male; younger age; lower HAQ, ESR (or CRP) and tender joint count (or swollen joint count) scores; and absence of comorbidities. In models predicting 1-, 3- and 6-month LDA or remission, area under the receiver operating curve was 0.648-0.809 (R(2) = 0.0397-0.1078). The models also predicted 6-month HAQ and EuroQoL-5-dimension scores. A series of matrices were developed to easily show predicted rates of remission and LDA.
Conclusion: A matrix tool was developed to show predicted GLM treatment outcomes in patients with RA, based on a combination of six baseline characteristics. The tool could help provide practical guidance in selection of candidates for anti-TNF therapy.
Keywords: biologic; predictors of response; remission; rheumatoid arthritis; tumour necrosis factor.
© The Author 2016. Published by Oxford University Press on behalf of the British Society for Rheumatology.