Smoking and sex specific proteomic markers of lung cancer: A prospective cohort study

Background While circulating proteins hold promise for elucidating lung cancer (LC) etiology, large-scale studies comprehensively assessing their associations with LC risk are limited. Crucially, the roles of these proteins across different sexes and smoking statuses, particularly in females and non-smokers who constitute a significant proportion of LC cases with distinct etiology, remain poorly characterized. Methods We integrated observational analyses from the UK Biobank (UKB) cohort (N=49,370 cancer-free at baseline; 444 incident LC cases; median follow-up 11 years) with two-sample Mendelian randomization (MR). Plasma levels of 2,920 proteins were measured at baseline. Cox regression identified proteins associated with incident LC, with stratified analyses by sex and smoking status. MR used cis-pQTLs from UKB-PPP and deCODE studies as instruments for proteins and LC GWAS data from FinnGen, TRICL, and Pan-UKBB. Robust causal evidence was defined by a significant MR result (IVW P<0.05) combined with either a colocalization posterior probability (PPH4) > 0.75 or a significant SMR result (HEIDI P>0.01). Results Sex-stratified analysis revealed 532 significant proteins in males and 375 in females. Among these, 271 were male-specific (e.g., C1R, HR=5.48, 95%CI: 2.23-13.50) and 114 were female-specific (e.g., SMAD5, HR=4.53, 95%CI: 2.39-8.58). Only 261 proteins were shared between sexes. Smoking-stratified analysis identified 683 significant proteins in smokers and 39 in non-smokers. A vast majority (657 proteins) were smoker-specific, including top risk protein SERPINA1 (HR=24.62, 95%CI: 5.65-107.39) and top protective protein GSN (HR=0.15, 95%CI: 0.09-0.27). In contrast, only 13 proteins were non-smoker-specific, such as risk protein CYTL1 (HR=3.46, 95%CI: 1.11-10.82) and protective protein CNTN4 (HR=0.17, 95%CI: 0.07-0.45). MR analysis provided genetic support for IL19 for LC among smokers (OR=1.18, 95%CI: 1.03-1.35 from cis-MR; OR=1.12, 95%CI: 1.01-1.23 from SMR). Conclusion This study unveils strikingly distinct proteomic signatures of LC risk by sex and smoking status, predominantly in males and smokers, with genetic validation for key candidates like IL19. These findings underscore etiological heterogeneity and support stratified approaches to LC risk prediction and prevention, pending external validation in diverse populations.