Long-Term Exposure to Fine Particulate Matter and Incidence of Esophageal Cancer: A Prospective Study of 0.5 Million Chinese Adults

Abstract

Background & aims: Evidence is sparse and inconclusive on the association between long-term fine (≤2.5 μm) particulate matter (PM_2.5) exposure and esophageal cancer. We aimed to assess the association of PM_2.5 with esophageal cancer risk and compared the esophageal cancer risk attributable to PM_2.5 exposure and other established risk factors.

Methods: This study included 510,125 participants without esophageal cancer at baseline from China Kadoorie Biobank. A high-resolution (1 × 1 km) satellite-based model was used to estimate PM_2.5 exposure during the study period. Hazard ratios (HR) and 95% CIs of PM_2.5 with esophageal cancer incidence were estimated using Cox proportional hazard model. Population attributable fractions for PM_2.5 and other established risk factors were estimated.

Results: There was a linear concentration-response relationship between long-term PM_2.5 exposure and esophageal cancer. For each 10-μg/m³ increase in PM_2.5, the HR was 1.16 (95% CI, 1.04-1.30) for esophageal cancer incidence. Compared with the first quarter of PM_2.5 exposure, participants in the highest quarter had a 1.32-fold higher risk for esophageal cancer, with an HR of 1.32 (95% CI, 1.01-1.72). The population attributable risk because of annual average PM_2.5 concentration ≥35 μg/m³ was 23.3% (95% CI, 6.6%-40.0%), higher than the risks attributable to lifestyle risk factors.

Conclusions: This large prospective cohort study of Chinese adults found that long-term exposure to PM_2.5 was associated with an elevated risk of esophageal cancer. With stringent air pollution mitigation measures in China, a large reduction in the esophageal cancer disease burden can be expected.

Keywords: Ambient Air Pollution; Digestive Tract; Malignant Tumor; Observational Study.

Figure 1

Lipid-related metabolites that were significantly associated with the newly affluent southern pattern scores [β coefficient (95%CI)]. Further, these metabolites were all associated with risk of general obesity [OR (95%CI)]. Models were adjusted for age, sex, region, education, household income, occupation, marital status, tea-drinking habit, smoking status, alcohol intake, physical activity, self-rated health, fasting time, and groups of participants in the original case-control study. All analyses relating to one DP were adjusted for others. Black squares represented coefficients or ORs, while gray horizontal lines represented 95%CIs.

Figure 2. Lipoprotein subfractions that were significantly associated with the newly affluent southern pattern scores (β).

Further, these metabolites were all associated with risk of general obesity (OR). Models were adjusted for age, sex, region, education, household income, occupation, marital status, tea-drinking habit, smoking status, alcohol intake, physical activity, self-rated health, fasting time, and groups of participants in the original case-control study. All analyses relating to one DP were adjusted for others. Black squares represent coefficients or ORs, while gray horizontal lines represent 95%CIs.

Figure 3

Global comparisons of β coefficients for metabolites and newly affluent southern pattern vs. ORs for (A) general obesity, (B) overweight, (C) WC-defined central obesity, and (D) WHR-defined central obesity with these metabolites. Models were adjusted for age, sex, region, education, household income, occupation, marital status, tea-drinking habit, smoking status, alcohol intake, physical activity, self-rated health, fasting time, and groups of participants in the original case-control study. All analyses relating to one DP were adjusted for others. Blue dots represented metabolites associated with the risk of obesity but not with the newly affluent southern pattern, while red dots represented metabolites associated with both the newly affluent southern pattern and the risk of obesity, with overlapping dots darker in color. The gray horizontal line and vertical line represented 95%CIs of coefficients and ORs, respectively. Pearson correlations of coefficients and ORs were annotated in the lower right corner.