THU0220 Establishment of a Prediction Model for Hepatotoxicity of Methotrexate by Using Dmet Microarray Profiling in Japanese Patients with Rheumatoid Arthritis

Background Efficacy and toxicity of methotrexate (MTX) differs among individual patients, suggesting the influence of genetic variations in enzymes associated with MTX metabolism and folate metabolic pathway. Recent advances in pharmacogenetics (PGx) may permit the prediction of an individual patient’s response to MTX. Objectives In order to create a prediction model for adverse events of MTX, we investigated the polymorphisms relevant to the hepatotoxicity using the DMET microarray and direct-sequencing (DS) method for detection of 1,971 variants in enzymes concerned in drug-metabolism. Methods: Study subjects A total of 200 RA patients treated with MTX were retrospectively recruited from the Center for Rheumatic Diseases, Shinko Hospital, and 33 patients with hepatotoxicity (AST or ALT > 40 IU/L with MTX ≤ 8mg/week) were sorted. Roughly age- and sex-matched 38 patients without hepatotoxicity (AST and ALT ≤ 40 IU/L with MTX ≥ 8mg/week) were also selected. Genotyping DNA samples were extracted from whole blood samples. We used DMET Plus array (Affymetrix, Inc.) for genotyping 1,934 variants in 231 genes related to pharmacokinetics. Additionally, the other 37 variants in 17 genes, formerly reported to be associated with MTX efficacy or toxicity, were genotyped by DS method 1, 2) . Statistical analysis We firstly compared frequencies of genotypes in the two groups by Fisher’s exact test, and the variants with p Results Among 37 variants in 17 genes reported formerly to be associated with MTX efficacy/toxicity, only 2 SNPs in 2 enzymes (MTRR and ADORA2A) were shown to be significantly associated with hepatotoxicity (p Next, frequency of genotypes in 1,971 variants of 246 genes for patients with and without hepatotoxicity was compared by Fisher’s exact test. 49 SNPs were extracted from a total of 1,971 variants with a significance level of p Finally, multiple logistic regression analysis was used to construct the prediction model for hepatotoxicity. By stepwise selection procedure, we obtained a prediction model in whicha combination of 12 SNPs could discriminate patients between with and without hepatotoxicity. The ROC analysis showed an AUC of 0.9857 with 97.0% sensitivity and 89.5% specificity. Conclusions We found a combination of 12 SNPs in genes related to drug metabolism/transport that could predict hepatotoxicity of MTX with high sensitivity and specificity. After validation in other cohort, this model may lead to appropriate usage of MTX with PGx. References Ranganathan P, McLeod HL: Methotrexate Pharmacogenetics. The Fist Step toward Individualized Therapy in Rheumatoid Arthritis. Arthritis Rheum. 54:1366-1377, 2006. Wessels JAM, et al: A Clinical Pharmacogenetic Model to Predict the Efficacy of Methotrexate Monotherapy in Recent-Onset Rheumatoid Arthritis. Arthritis Rheum. 56:1765-1775, 2007. Acknowledgements We are grateful to Ms. K. Otsuji and Ms. M. Yasuda for their excellent technical assistance. Disclosure of Interest S. Kumagai Grant/research support from: Sysmex Co. (Donation Lab.), G. Tsuji: None Declared, S. Sendo: None Declared, Y. Ichise: None Declared, R. Umeda: None Declared, Y. Hagihara: None Declared, H. Uga: None Declared, H. Kurata: None Declared