The Genetic Biomarkers to Predicting Response of TNF Inhibitors Treatment in Rheumatoid Arthritis

Abstract

Background/Purpose: Although pharmacogenetic studies of TNF inhibitors (TNFi) response presented the estimates of high heritability, only few loci with suggestive weak association as biomarkers for TNFi response have been identified. We aimed to identify optimal phenotype for drug response using heritability estimates (h2) and new predictive biomarkers of response to TNFi using genome-wide association studies (GWAS) in the Korean population.

Methods: Disease Activity Scores based on 28 joint counts (DAS28) and Clinical Disease Activity Index (CDAI) were assessed at baseline, and after 6 months in 370 Korean RA patients who started TNFi due to moderate or high disease activity from Hanyang university hospital. Genotypes were generated on the Illumina HumanOmni2.5Exome array (2.5 million variants). Quality control (QC) procedures were applied using the PLINK 1.9 and R 3.2.2 software. We estimated heritability by a linear mixed effect modeling approach (GCTA) for TNFi response using changes (Δ) in DAS28 and CDAI. To identify clinical and genetic variables that influence response to TNFi, a multivariate generalized linear model (GLM) analysis was performed. We also conducted a gene-based analysis [optimal sequence kernel association test (SKAT-O)] of rare variants.

Results: We identified that clinical factors seem to influence the therapeutic good response of TNFi including male, high disease activity score at baseline, BMI. The heritability estimates were found for ΔDAS28 h2=0.44, ΔCDAI h2=0.62, Δprovider global assessment of disease activity (PrGA) h2=0.66, Δswollen joint count (SJC) h2=0.66, Δtender joint count (TJC) h2=0.59, Δpatient global assessment of disease activity (PtGA) h2=0.58, and ΔESR h2=0.49. We identified two novel significant functional SNPs [rs117811759 (UTR3 of SAP18), rs17279819 (exon of SKA3)] associated with response to TNFi, surpassing genome-wide significant threshold (P < 5.0×10−8). Using a gene-based approach, we also identified two genes (SAP18 and SKA3) with significant burden signals after correction for multiple comparisons.

Conclusion: The optimal phenotype based on heritability suggests the use of changes in clinical disease activity index (CDAI) including provider global assessment than DAS28 in pharmacogenetic study. Our study suggests that SAP18 and SKA3 associated with response to TNFi therapy may serve as the useful genetic biomarker in RA patients of Koreans.