An Immunochip-based interaction study of contrasting interaction effects with smoking in ACPA-positive versus ACPA-negative rheumatoid arthritis.

Jiang X, Källberg H, Chen Z, Ärlestig L, Rantapää-Dahlqvist S, Davila S, Klareskog L, Padyukov L, Alfredsson L

Rheumatology (Oxford) 55 (1) 149-155 [2016-01-00; online 2015-08-15]

To investigate the gene-environment interaction between smoking and single nucleotide polymorphisms (SNPs), using Immunochip material, on the risk of developing either of two serologically defined subsets of RA. Interaction between smoking and 133,648 genetic markers from the Immunochip was examined for two RA subsets, defined by the presence or absence of ACPA. A total of 1590 ACPA-positive and 891 ACPA-negative cases were compared with 1856 controls in the Swedish Epidemiological Investigation of RA (EIRA) case-control study. Logistic regression models were used to determine the presence of interaction. The proportion attributable to interaction was calculated for each smoking-SNP pair. Replication was carried out in an independent dataset from northern Sweden. To further validate and extend the results, interaction analysis was also performed using genome-wide association studies data on EIRA individuals. In ACPA-positive RA, 102 SNPs interacted significantly with smoking, after Bonferroni correction. All 102 SNPs were located in the HLA region, mainly within the HLA class II region, 51 of which were replicated. No additional loci outside chromosome 6 were identified in the genome-wide association studies validation. After adjusting for HLA-DRB1 shared epitope, 15 smoking-SNP pairs remained significant for ACPA-positive RA, with 8 of these replicated (loci: BTNL2, HLA-DRA, HLA-DRB5, HLA-DQA1, HLA-DOB and TAP2). For ACPA-negative RA, no smoking-SNP pairs passed the threshold for significance. Our study presents extended gene variation patterns involved in gene-smoking interaction in ACPA-positive, but not ACPA-negative, RA. Notably, variants in HLA-DRB1 and those in additional genes within the MHC class II region, but not in any other gene regions, showed interaction with smoking.

Bioinformatics Support for Computational Resources [Service]

NGI Uppsala (SNP&SEQ Technology Platform) [Service]

National Genomics Infrastructure [Service]

PubMed 26272072

DOI 10.1093/rheumatology/kev285

Crossref 10.1093/rheumatology/kev285

pii: kev285