Being an ex-smoker was associated with increased risk of gout in NZ East Polynesian and UK Biobank European subsets. This is consistent with previous evidence of increased risk of gout in ex-smokers [18, 19, 21, 40, 41], although reduced prevalence of gout in ex-smokers has also been reported [22, 42]. It is also consistent with evidence that ex-smokers have higher serum urate than never-smokers (reviewed in [20]) and that serum urate increases following smoking cessation [23, 43]. In our analysis of NZ Polynesian people and the large UK Biobank European cohort, no association was observed between self-reported current smoking and prevalence of gout in the overall cohorts. However, current smoking status was associated with increased risk of gout in women and decreased risk of gout in men of the UK Biobank European cohort. These opposing associations of current smoking with gout in men and women are also consistent with sex-specific differences reported previously where current smoking was found to be protective against developing gout in men, but not in women [14, 15]. Current smoking has been found to be associated with reduced incidence and prevalence of gout in multiple studies [14–16, 18, 19, 41, 44], although there have also been conflicting studies reporting increased risk of gout [21, 22] (reviewed in [20]). No association was observed between current- or ex-smoker status and gout in the UK Biobank South Asian population.
This study focused on identifying potential interactions of 5 loci with smoking that influence the risk of gout in a NZ study population of East and West Polynesian ancestry and in study populations of South Asian and European ancestry from the UK Biobank. One locus, TRIM46 (rs11264341), was found to interact with ex-smoker status and with current-smoker status to influence gout in the NZ Polynesian and UK Biobank South Asian sample sets, respectively. There was no interaction of TRIM46 (rs11264341) with smoking behaviour in the UK Biobank European population. We considered the potential nature of the interactions observed in the NZ Polynesian and UK Biobank Asian study populations by examining the effect of the TRIM46 (rs11264341) locus on prevalence of gout in smoking status subgroups. The TRIM46 (rs11264341) C allele was associated with higher prevalence of gout specifically in never-smokers, but not in ex-smoker or current-smoker status subgroups of the NZ Polynesian population. This suggests that the risk effect of TRIM46 is only evident in the absence of current- or past- smoking behaviours. Perhaps the risk effect of past-smoking overrides the effect of the TRIM46. In the UK Biobank South Asian sample set TRIM46 was specifically associated with gout in current-smokers, indicating that current-smokers of this ancestry may be specifically susceptible to the risk effect of the C-allele. This is consistent with a study of 4,332 individuals of Chinese ancestry where the TRIM46 locus C-allele was associated with higher serum urate in current-smokers [β coefficient for change in serum urate (β) = 7.541, false discovery rate corrected p value (pFDR) = 0.040], but not in ex-smokers (β = 0.240, pFDR = 0.975) or never-smokers (β = 18.860, pFDR = 0.053) [11]. However, no formal test of interaction (e.g. using a TRIM46*smoking-status interaction term versus serum urate) was done by Dong et al. [11].
Whether the differing interactions of TRIM46 with smoking behaviour and gout seen in the NZ Polynesian and South Asian populations is explainable through the same biological pathway remains unknown. It is possible that inter-population variation in linkage disequilibrium pattern of rs11264341, involving a nearby causal variant, is responsible for differences in interaction effects. This could also explain why the main effects of TRIM46 on gout risk are in opposing directions in the Polynesian and South Asian populations. Thus, subject to replication in other sample sets of Asian and Polynesian ancestry, we conclude evidence for a population-specific interaction of TRIM46 (rs11264341) with risk of gout, with opposing effects on gout in current- and ex-smokers.
Causal candidate genes have been identified at the TRIM46 (rs11264341) locus by colocalising the GWAS signal with cis expression quantitative trait locus (eQTL) analysis [45]. These include SHE, MUC1, GBAP1, and FAM189B, but not TRIM46 itself or PKLR which were previously suggested to be causal candidates by the less accurate Gene Relationships Across Implicated Loci (GRAIL) tool [3, 46]. Based on current knowledge of these candidate causal genes, MUC1 appears to be the most likely to interact with smoking behaviour to influence gout. MUC1 encodes a transmembrane mucin expressed on the apical surface of epithelial cells lining the mucosa of the lungs, and in haematopoietic cells (reviewed in [47]). In the lung it has a pivotal anti-inflammatory function at later stages of the inflammatory response to bacterial infections, which it mediates by inhibiting Toll-like receptor signalling. There is evidence that chronic exposure to cigarette smoke in mice leads to increased expression of Muc1 in lung epithelial cells and macrophages, and increased numbers of Muc1-expressing macrophages in the lungs [48]. The same study also found that treating human macrophage cell lines with cigarette smoke extract induced MUC1 protein expression. There is also evidence that cigarette smoke provokes aberrant glycosylation and subcellular redistribution of MUC1, causing loss of E-cadherin and basolateral adherens junction integrity, which in turn leads to initiation of epithelial–mesenchymal transition and lung cancer development [49, 50].
Limitations of this study include the relatively small number of participants and people with gout in the NZ Polynesian and South Asian sample sets, which means potential interactions affecting these ancestral population may go undetected. Additionally, we did not have data of when current- or ex-smokers commenced or ceased smoking to test how this related to onset of gout, therefore we could not exclude ex-smokers who quit smoking after gout onset or current-smokers who took up smoking after gout onset. Furthermore, we did not have information on pack year history or serum cotinine levels to assess how duration and dose of cigarette smoking may interact with genetic factors to impact gout risk. It is possible that the higher prevalence of gout in ex-smokers may be influenced by increased access to smoking-cessation support in participants receiving care for gout, compared to those without gout.