Background/Purpose: Methotrexate (MTX) constitutes first-line therapy in rheumatoid arthritis (RA). However, up to 40% of RA patients do not benefit from MTX therapy. To restrain joint damage, prediction of insufficient response to MTX prior to treatment is desirable to initiate alternative treatment options. Previously, we developed a prognostic multivariable prediction model for insufficient clinical response to MTX at 3 months in the “treatment in the Rotterdam early arthritis cohort” (tREACH, ISRCTN26791028). This model was replicated in another Rotterdam cohort (MTX-R), however, information on two predictors: smoking and BMI was absent. The purpose of the current study was to replicate the complete prediction model in an independent, non-Rotterdam early RA cohort (U-Act early), which included data on smoking and BMI.

Methods: Clinical, laboratory and life style parameters were assessed from 91 early RA patients of the MTX treatment arm of the U-Act early cohort (NCT01034137). 92% of the subjects met the 2010 ACR/EULAR classification criteria for RA. Insufficient response was defined as DAS28-ESR >3.2 after 3 months of MTX treatment. The following predictors and the corresponding cut-off values were defined in the prediction model for insufficient response: baseline DAS28 >5.1, baseline HAQ >0.6, baseline erythrocyte-folate < 750 nmol/L, Adenosine triphosphate Binding Cassette transporter (ABC) family B member 1 (ABCB1) rs1045642 genotype, ABCC3 rs4793665 genotype, current smoking and BMI >25kg/m2.  Univariable and multivariable logistic regression models were used to assess the associations between the predictors and insufficient response to MTX in the current cohort, which was reported as odds ratio (OR) and 95% confidence intervals (CI). To assess the predictive power of the model, a receiver operating characteristic (ROC) curve was constructed from the predicted probabilities, with its corresponding area under the curve (AUC) and 95% CI. The model in the current study (U-Act early) was considered to be replicated when the 95% CI  overlapped with that of the tREACH model (AUC: 0.80, 95% CI: 0.73 – 0.86).

Results: In the U-Act early, the AUC of the complete prediction model was 0.75 (95% CI: 0.64 – 0.85), which resembles the 95% CI of the tREACH (P >0.05). The strongest predictors were baseline DAS28 >5.1 (univariable OR=4.5, 95% CI: 1.7 – 11.6) and baseline HAQ >0.6 (univariate OR=3.0, 95% CI: 1.2 – 7.5), which together constructed a ROC with an AUC of 0.70 (95% CI: 0.59 – 0.80). With the exception of current smoking, all predictors contributed to the improvement of the prediction model in this study. For these predictors the AUC was also 0.75 (95% CI: 0.64 – 0.85).

Conclusion: We successfully replicated our previously published prognostic prediction model of insufficient MTX response constructed of genetic, clinical and life style variables in an independent cohort, where 75% of insufficient responders were correctly classified. A cost-benefit analysis will be decisive on which predictors to use when implementing the model in a clinical setting.

« Back to 2019 ACR/ARP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/replication-of-a-prognostic-multivariable-prediction-model-for-insufficient-clinical-response-to-methotrexate-in-early-rheumatoid-arthritis/