Associations between long-term exposure to ultraviolet radiation and site-specific cancer incidence: an ecological study in China between 2006 and 2016

Evidence regarding long-term ambient ultraviolet radiation (UVR) and multiple cancer outcomes remains limited in non-Western populations. We assessed associations between long-term ambient UVR availability and total and site-specific cancer incidence across mainland China. This nationwide ecological study included 4,010,117 incident cancer cases from 543 cancer registry areas during 2006–2016. Population-weighted ambient UVR metrics for lag10-lag50 were derived from ERA5 monthly surface downward UV radiation and WorldPop data. Poisson generalized nonlinear models with registry-area and year fixed effects estimated relative risks (RRs) per 1 W/m 2 increase in UVR, adjusting for temperature, cloud cover, per-capita GDP, hospital beds, population aged ≥ 65 years, PM2.5, O3, urban–rural status, and population size. Sex-stratified, exposure–response, FDR-corrected, warm-season ambient UVR, and stable-registry analyses were conducted. At lag30, ambient UVR availability was inversely associated with total cancer incidence (RR = 0.97, 95%CI: 0.95–0.99). Total-cancer estimates were also inverse at lag10 (0.94[0.93–0.96]) but positive at lag45 (1.05[1.03–1.08]) and lag50 (1.03[1.01–1.05]). FDR-significant positive lag30 associations were observed for liver (1.19[1.12–1.26]), thyroid (1.21[1.07–1.37]), brain (1.78[1.56–2.02]), stomach (1.27[1.19–1.36]), and testis cancers (2.31[1.13–4.71]). Inverse associations were observed for lymphoma (0.81[0.73–0.92]), lung (0.87[0.84–0.90]), female breast (0.74[0.69–0.78]), and prostate cancers (0.73[0.65–0.84]). Melanoma was positively associated only at lag50 (1.89[1.11–3.21]). Sex-specific estimates differed mainly in magnitude. Stable-registry analyses broadly supported the principal site-specific patterns. Long-term ambient UVR availability showed heterogeneous lag-, cancer-site-, and sex-specific ecological associations with cancer incidence. Individual-level studies are needed to clarify the underlying mechanisms and causal relevance.