Investigation of the interaction between smoking and TRAF1/C5 on radiological severity of inflammatory polyarthritis

Abstract

Background/Purpose: A number of factors have been identified as contributing to the severity of rheumatoid arthritis (RA), including SNP markers at the TRAF1/C5 region (rs2900180 and rs10760130). The role of smoking in RA disease severity is debated in the literature. We investigated whether smoking status conditions the association of TRAF1/C5 severity variants in patients with recent onset inflammatory polyarthritis (IP). Methods: Data were collated from the Norfolk Arthritis Register, a prospective inception cohort for recent onset IP recruited from both primary and secondary care. X-rays were taken on patients who satisfied the ACR criteria for RA at one year and on all patients at 5 years follow-up. DNA and X-ray data were available for 1860 patients and X-ray with smoking data for 1789 patients. The interaction between smoking status at baseline and the carriage of polymorphisms at the TRAF1/C5 region with radiological damage, as measured using the Larsen score or the presence of erosions, was investigated longitudinally to include repeat measurements in the same individual at different time points. Interaction was defined as significant departure from a multiplicative model of the individual odds ratios (OR). Results: Smoking status was not associated with erosions, whilst the carriage of rs2900180 alleles was significantly associated with erosions (OR: 1.82 (95% CI 1.1, 3.0), p-value 0.019). An association was found in ever smokers (OR 3.39 (95% CI 1.7, 6.8), p-value 0.001); when stratified into current smokers and previous smokers, this association was significant in previous smokers (OR: 4.09 (95% CI 1.7, 9.6), p-value 0.001). The OR followed the same direction for current smokers (OR 2.25 (95% CI 0.7, 7.7), p-value 0.196), but not for non-smokers (OR: 0.61 (95% CI 0.2, 1.6), p-value 0.307). The interaction between ever smoker status and the carriage of rs2900180 departed significantly from multiplicativity (interaction OR: 5.56 (95% CI 1.8, 17.6), p value 0.004). The interaction between previous smoker status and the carriage of rs2900180 departed significantly from the multiplicative model (interaction OR: 6.83 (95% CI 2.0, 23.7), p-value 0.003); however the interaction was not significant in current smokers. Consistent results and a significant interaction were also found when using the Larsen score as outcome variable in longitudinal modelling. Total smoking duration and the length of time between smoking cessation and disease onset were, independently, highly significantly correlated with erosions in previous smokers - if the smoking duration is longer, there is a higher the risk of erosive disease. Analysis of rs10760130 showed similar patterns of associations and interactions. Conclusion: These results demonstrate an interaction between ever smoker status and the carriage of markers at the TRAF1/C5 region on erosive disease in IP/RA patients. These findings support a model of RA pathogenesis in which smoking prevalence and time prior to disease onset affects the development of erosive disease in genetically predisposed individuals and underlines the importance of gene-environment interactions. We are currently validating these findings in another large data set.