Wildfire smoke and cancer risk in the United States: Evidence from the PLCO Trial

Background: Wildfires are becoming more frequent and severe, reversing decades of progress in reducing fine particulate matter (PM 2.5 ) in the United States. Although wildfire smoke (WFS) contains numerous carcinogens and toxicants, its association with cancer incidence remains unclear. Methods: We examined associations between WFS exposure and cancer incidence in the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial. The PLCO trial prospectively assessed cancer incidence from enrollment (1993-2001) through 2018. WFS exposure was quantified monthly at participants' residences using near-ground WFS PM 2.5 , WFS black carbon (BC), and satellite-derived WFS plume-day counts from 2006 until first cancer diagnosis or last contact. Guided by evidence that three years of air-pollution exposure can influence the development of EGFR-positive lung adenocarcinoma, WFS exposure was modeled as a time-varying variable using 36-month moving averages preceding each month. Hazard ratios (HRs) were estimated using Cox proportional hazards models stratified by study center, with proportional hazards assumptions verified. Restricted cubic splines were applied to evaluate dose-response relationships. Covariates included age, sex, race and ethnicity, education, smoking history, body mass index, and trial arm. Results : Among 91,460 PLCO participants with linked WFS exposure data, we identified incident cases of 1,758 lung, 800 colorectal, 1,739 breast, 242 ovarian, 896 bladder, and 1,696 hematopoietic cancers, and 1,127 melanoma during 2006-2018. Median (range) 36-month moving-averages were 0.37 (0.0083-1.72) µg/m 3 for WFS PM 2.5 , 0.0083 (0.00-0.21) µg/m 3 for WFS BC, and 1.94 (0.097-7.18) days for monthly WFS plume-day counts. Restricted cubic splines indicated statistically significant associations (P<0.05) of WFS exposure with increased risks of lung, colorectal, breast, bladder, and hematopoietic cancers, with an approximate linear dose-response observed for most analyses. Moreover, each 1 µg/m 3 increase in the 36-month moving-average of WFS PM 2.5 was associated with higher risks of lung (HR=1.92; 95% CI: 1.18, 3.15), colorectal (2.31; 1.11, 4.81), breast (2.09; 1.34, 3.26), bladder (3.49; 1.66, 7.34), and hematopoietic (1.63; 1.02, 2.60) cancers. No associations were found for ovarian cancer or melanoma. Results for WFS plume-day counts were generally consistent with those for WFS PM 2.5 , whereas associations for WFS BC exposure were observed only for breast and bladder cancers. Conclusion: WFS exposure was associated with the risk of lung, colorectal, breast, bladder, and hematopoietic cancers. Findings need to be replicated in additional cohorts.