Comprehensive Assessment of Smoking and Sex Related Effects in Publicly Available Gene Expression Data

Smoking greatly reduces life expectancy in both men and women, but with different patterns of morbidity. After adjusting for smoking history, women have higher risk of respiratory effects and diabetes from smoking, while men show greater mortality from smoking-related cancers. While many smoking-related sex differences have been documented, the underlying molecular mechanisms are not well understood. To date, identification of sex differences in response to smoking has been limited to a small number of studies and the resulting smoking-related effects require further validation. Publicly available gene expression data present a unique opportunity to examine molecular-level sex and smoking effects across many tissues and studies. We performed a systematic search to identify smoking-related studies from healthy tissue samples and found 31 separate studies as well as an additional group of overlapping studies that in total span 2,177 samples and 12 tissues. These samples and studies were overall male-biased. In smoking, while effects appeared to be somewhat tissue-specific and largely autosomal, we identified a small number of genes that were consistently differentially expressed across tissues, including AHRR and GZMH. We also identified one gene, AKR1C3, encoding an aldo-keto reductase, which showed strong opposite direction, smoking-related effects in blood and airway epithelium, with higher expression in airway epithelium and lower expression in blood of smokers versus non-smokers. By contrast, at similar significance thresholds, sex-related effects were entirely sex chromosomal and consistent across tissues, providing evidence of stronger effects of smoking than sex on autosomal expression. Due to sample size limitations, we only examined interaction effects in the largest study, where we identified 30 genes with sex differential effects in response to smoking, only one of which, CAPN9, replicated in a held-out analysis. Overall these results present a comprehensive analysis of smoking-related effects across tissues and an initial examination of sex differential smoking effects in public gene expression data.