Long-term exposure to particulate matter and survival among patients with gastric cancer: A hospital-based cohort study.

Evidence regarding the prognostic relevance of long-term particulate matter exposure among patients with gastric cancer remains limited. We evaluated the associations between historical pre-index particulate matter exposure and subsequent mortality and estimated model-based population-attributable fractions (PAFs) under hypothetical lower historical exposure distributions. We conducted a hospital-based cohort study of 4441 patients with stage I-III gastric cancer who underwent radical surgery between 2012 and 2022 in Zhejiang, China. Residential exposures to fine particulate matter (PM2.5), inhalable particulate matter (PM10), and coarse particulate matter (PM2.5-10) during the 36 months preceding surgery were estimated using the ChinaHighAirPollutants (CHAP) dataset. Cox proportional hazards models were used for all-cause mortality, whereas cause-specific Cox and Fine-Gray subdistribution hazard models were used for gastric cancer-specific mortality, treating confirmed non-gastric cancer deaths as competing events. Counterfactual analyses were further conducted to estimate PAFs under various exposure reduction scenarios. During a median follow-up of 3.83 years, 1318 patients died (29.7%), including 919 gastric cancer deaths, 268 deaths from other causes, and 131 deaths of unknown cause. Each 1-SD increase in PM2.5, PM10 and PM2.5-10 was associated with higher all-cause mortality (adjusted HRs 1.15 [95% CI: 1.06-1.25], 1.14 [1.06-1.24], and 1.12 [1.03-1.21], respectively). The corresponding subdistribution hazard ratios for gastric cancer-specific mortality were 1.31 (1.20-1.44), 1.28 (1.17-1.40), and 1.17 (1.07-1.28). Associations were generally consistent across most prespecified subgroups and alternative analytical specifications. In the exploratory counterfactual analyses, hypothetical historical exposure limits corresponding to the WHO 2021 annual Air Quality Guidelines yielded model-estimated all-cause mortality PAFs of 29.2% (95% uncertainty interval, 13.3%-42.5%) for PM2.5 and 27.5% (12.2%-40.2%) for PM10. Higher residential particulate matter exposure was associated with higher subsequent all-cause and gastric cancer-specific mortality. These findings support the potential prognostic relevance of historical particulate matter exposure.