Source-informed air pollution mixture axes, proteomic response domains, and incident bladder cancer risk in the UK Biobank

Long-term ambient air pollution has been inconsistently associated with bladder cancer, partly because correlated pollutants are often examined separately rather than as components of broader exposure mixtures. We conducted a prospective cohort analysis in the UK Biobank. Two prespecified, source-informed air pollution mixture axes were constructed to summarize traffic/combustion-related pollution and particulate burden. Associations with incident bladder cancer were assessed using multivariable Cox models in the primary cohort, a stable-address subgroup, and a 5-year lag subgroup. In the Olink subcohort, we evaluated associations between the pollution axes and five predefined plasma proteomic response domains, and between these domains and subsequent bladder cancer risk. The traffic/combustion axis showed a more consistent association with incident bladder cancer than the particulate-burden axis, with hazard ratios per standard deviation increase of 1.060 (95% CI 0.988–1.138), 1.088 (95% CI 1.011–1.171), and 1.142 (95% CI 1.047–1.247) across the three analytic subsets. In proteomic analyses of 38,959 participants with 92 incident events, both pollution axes were associated with selective proteomic perturbations. Epithelial-remodeling and metabolism/stress domains were positively associated with subsequent bladder cancer risk, with epithelial remodeling showing the most consistent pattern across the proteomic analyses. Source-informed mixture summaries and domain-level proteomic profiling may help characterize associations between correlated ambient air pollutants and bladder cancer risk.