Clinico-genetic model to predict methotrexate intolerance in rheumatoid arthritis.

INTRODUCTION

Methotrexate is the gold-standard DMARD in rheumatoid arthritis but is often associated with "mild" adverse effects like intolerance or laboratory abnormalities. Although non-life threatening, they are responsible for drug discontinuation in 17-50%. There is limited data on clinical and genetic markers that predict their occurrence.

METHODS

This prospective study enrolled patients with active rheumatoid arthritis. They were started on methotrexate at a weekly dose of 15 mg, escalated gradually to reach 25 mg which was continued till the end of the study. Intolerance (symptomatic adverse effects) was ascertained by a questionnaire at 4, 8, 16, and 24 weeks. Laboratory testing for occurrence of cytopenia and/or transaminitis was done at the same study visits. Seven SNPs in four genes involved in methotrexate handling were genotyped using real-time polymerase chain reaction.

RESULTS

This study included 110 patients with rheumatoid arthritis who received methotrexate for 24 weeks; the final mean weekly methotrexate dose was 22.0 ± 4.0 mg. Methotrexate intolerance occurred in 40 (37%), common being nausea (and vomiting) in 29 and anxiety (and dizziness) in 9. It was associated with lower BMI at baseline (21.5 ± 3.7, 23.8 ± 4.6 kg/m, p = 0.01). FPGS rs10106 was significantly associated with intolerance with an allelic odds ratio (95% CI) of 2.02 (1.14-3.57) and the recessive genetic model (AA+AG versus GG) with an odds ratio of 3.8 (95% CI 1.5-9.6, p = 0.004). A model including both BMI and FPGS rs10106 could modestly predict methotrexate intolerance with an accuracy of 66.3%.

CONCLUSIONS

A clinical-genetic model including BMI and SNP FPGS 10101 was found to have a modest prediction ability for methotrexate intolerance.Key Points• Methotrexate intolerance (symptomatic adverse effects) was common and occurred in 37% patients over 6 months.• SNP FPGS rs10106 and low body mass index were associated with methotrexate intolerance.• Clinico-genetic model had a modest ability of 66% for predicting intolerance.