Summary
Background
Fine particulate matter (PM2.5) has been identified as a key risk factor of lung cancer incidence and survival. PM2.5 from wildfire smoke is an increasingly important contributor to total PM2.5 in the United States driven by climate change and potentially more toxic than non-wildfire PM2.5 due to its smaller particle size and higher toxic contents. However, its effects on survival of patients with lung cancer remain unclear.
Methods
This cohort study included patients ≥65 years with histologically confirmed primary small cell lung cancer (SCLC) or non-small cell lung cancer (NSCLC), as defined by the International Classification of Diseases for Oncology, Third Edition (ICD-O-3), from SEER-Medicare database (01/01/2006–31/12/2019), with residential zip code-based estimates of wildfire smoke and non-wildfire PM2.5. A time-varying Cox proportional hazards model was applied to estimate the long-term effects of three-year moving averages of wildfire smoke PM2.5 and non-wildfire PM2.5 concentrations on overall survival.
Findings
A total of 414,016 patients contributed 1,211,490 person-years of follow up during the analysis period (01/01/2008–31/12/2019), with a median follow-up of 2 years (interquartile range [IQR] 1–4 years). Of these, 208,468 (50.4%) were female and 205,548 (49.6%) were male. In terms of race/ethnicity, 356,356 (86.1%) were White, 36,914 (8.9%) were Black, 18,982 (4.6%) were Asian or Pacific Islander, and 1,269 (0.3%) were American Indian or Alaska Native; race was unknown for 495 participants (0.1%). The median concentrations of wildfire smoke PM2.5 and non-wildfire PM2.5 were 0.28 (IQR 0.21–0.39) and 8.11 (IQR 6.84–9.47) μg/m3, respectively. Each 1-SD increase in wildfire smoke PM2.5 and non-wildfire PM2.5 was associated with a 2.29% (HR 1.0229, 95% CI 1.0189–1.0270, p<0.0001) and 3.84% (HR 1.0384, 95% CI 1.0335–1.0432, p<0.0001) higher risk of death among lung cancer patients, respectively. Notably, the mortality risk per 1-μg/m3 increase was substantially greater for wildfire smoke PM2.5 (HR 1.0782, 95% CI 1.0641–1.0924, p<0.0001) than of non-wildfire PM2.5 (HR 1.0171, 95% CI 1.0149–1.0192, p<0.0001). Overall, approximately 700 and 3,470 average annual excess deaths in our cohort were attributable to wildfire smoke and non-wildfire PM2.5, respectively.
Interpretation
Wildfire smoke PM2.5 exhibits greater toxicity per unit exposure and is the fastest-growing contributor to ambient air pollution, underscoring the urgent need for updated wildfire management and mitigation strategies.
Funding
NIEHS; NCI; NIH; NCATS; CDC(NPCR); Stony Brook University.
Introduction
Lung cancer is among the most commonly diagnosed cancer and is the leading cause of cancer-related death worldwide.1 Globally, it accounts for nearly one in eight new cancer cases and one in five cancer deaths.1 In 2025, it is projected to cause 226,650 new cases and 124,730 deaths in the United States (US), corresponding to about 620 new cases and 340 deaths per day.2 Lung cancer primarily affects older adults, with approximately 75% of deaths occurring among those aged 65 years or older.3 Most cases present at an advanced stage, contributing to markedly high mortality. In most countries, the five-year survival generally ranges between 10% and 20%, and in the US it remains below 30%.4 Therefore, understanding the risk factors associated with mortality of lung cancer patients is crucial for developing effective strategies to improve their outcomes.
Outdoor air pollution, particularly fine particulate matter (PM2.5), is classified as a Group 1 carcinogen for lung cancer by the International Agency for Research on Cancer.5 With the changing climate, wildfire smoke is emerging as a major contributor to PM2.5 in the US, and reversing decades of air-quality improvements.6 Moreover, wildfire smoke PM2.5 pollution is projected to increase 2–3 times by 2050 under future climate scenarios relative to the past decade.6 Compared to PM2.5 not originating from wildfire smoke (non-wildfire PM2.5), toxicological studies have found stronger toxicity of wildfire smoke-derived PM2.5.7,8 However, large-scale epidemiological evidence remains limited, underscoring the need to quantify the attributable burden and inform wildfire and climate mitigation strategies.
A few studies have reported the impact of wildfire exposure on survival among patients with non-small cell lung cancer (NSCLC), using wildfire event indicators rather than quantitative smoke concentrations and with limited adjustment for key prognostic factors.9,10 Recently, advances in spatio-temporal prediction models have enabled more accurate estimation of wildfire smoke PM2.5 concentrations across time and space,11 facilitating the investigation of its health effects at a finer resolution and on a larger scale. However, these model-based estimates remain subject to exposure misclassification due to reliance on ambient concentrations as proxies for personal exposure.
In this study, we leveraged a cohort of 414,016 well-characterized lung cancer patients from the Surveillance, Epidemiology, and End Results (SEER)-Medicare database, linked with satellite-derived exposure data. We aimed to compare the effects of wildfire smoke PM2.5 versus non-wildfire PM2.5 on the survival of older lung cancer patients, quantify the attributable burden, characterize the exposure-response relationships, and identify vulnerable populations.
Methods
Study design and participants
This study included patients with primary lung cancer from the linked SEER-Medicare database, which records longitudinal medical information about Medicare beneficiaries with cancer residing in SEER-covered regions. SEER is a population-based cancer registry currently covering about 45.9% of the US population. It collects data on newly diagnosed primary cancer cases, including patients’ demographics (e.g., age at diagnosis, sex, and race), tumor information (e.g., site, stage, and histology), and first course of treatment, and is linked to the National Death Index for vital status. Medicare provides health insurance for ~97% of people aged ≥65 years in the US. Most Medicare beneficiaries are enrolled in both Part A (inpatient) and Part B (outpatient) coverage. Their longitudinal claims records include information on residential zip code, demographic characteristics, diagnosis codes for outpatient visits and hospitalizations, and date of death. The SEER-Medicare linkage represents one of the largest, population-based cohort of older cancer patients residing in the US.
Each participant was followed annually from diagnosis until death, loss to follow-up, or the end of the study in 2019, whichever came first. From 2006 to 2019, 659,805 participants diagnosed with incident primary lung cancer were identified. We excluded those aged <65 years at diagnosis, diagnosed at in situ stage, without continuous enrollment in Medicare parts A and B, or with missing exposure or covariate data. Moreover, we included only participants with histologically confirmed small cell lung cancer (SCLC) or NSCLC, the two major clinical classifications of primary lung cancer. NSCLC included adenocarcinoma, squamous cell carcinoma, large cell carcinoma, and other specified NSCLC subtypes, as defined by the International Classification of Diseases for Oncology, Third Edition (ICD-O-3) (detailed codes are provided in appendix pp 7). After applying these criteria, the study cohort consisted of 414,016 participants, representing 1,211,490 person-years of follow-up.
This study was approved by the Institutional Review Boards of the Icahn School of Medicine at Mount Sinai and the Harvard T.H. Chan School of Public Health and followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.
There was no direct patient or public involvement in the design or conduct of this study, primarily because the SEER-Medicare dataset consists of deidentified administrative and clinical data collected for routine healthcare and cancer registry purposes.
Procedures
Daily estimates of ambient wildfire smoke PM2.5 and total PM2.5 across the contiguous US (2006–2019) were obtained from previously developed and validated models. Wildfire smoke PM2.5 estimates were derived from a satellite-based approach,11 while total PM2.5 was estimated using a spatio-temporal ensemble model.12 Estimates were spatially aggregated to the zip code level (see appendix pp 2–3 for details). Non-wildfire PM2.5 was calculated as the difference between total PM2.5 and wildfire smoke PM2.5, and negative values (about 1% of the data) were excluded. All exposures were matched to each participant based on longitudinal residential zip codes recorded in the SEER-Medicare database.
We addressed potential sources of bias by adjusting for relevant confounders selected based on prior literature,10,13,14 including individual-level prognostic factors, seasonal temperature, neighborhood-level characteristics, and calendar year, with the latter controlling for temporal variations. Individual-level prognostic factors included age at diagnosis, race, sex, residential area, cancer stage, histology, number of comorbidities, and first-course treatment. Seasonal temperature and neighborhood-level variables were defined as described in the appendix (pp 3–4).
Outcomes
The outcome of interest was all-cause mortality, defined as death from any cause during follow-up. The primary exposure of interest is the 3-year moving averages (current and previous two years) of wildfire smoke PM2.5 and non-wildfire PM2.5. Other metrics of wildfire smoke PM2.5, including duration and frequency,15 were also examined by 3-year moving averages of smoke day frequency and smoke wave frequency. A smoke day was defined as a day with smoke PM2.5 concentrations >0 μg/m3, and a smoke wave as ≥2 consecutive days with daily wildfire smoke PM2.5 concentrations exceeding the 95th percentile, which corresponds to 2.65 μg/m3 in this study. We chose 3-year moving average exposure window (lag 0–2) because the associations became statistically non-significant from lag 3 onwards in preliminary lag analyses (appendix pp 8).
Statistical analysis
A time-varying, stratified Cox proportional hazards model with follow-up year as the time scale was used to estimate hazard ratios for all-cause mortality associated with long-term exposure to wildfire smoke PM2.5 and non-wildfire PM2.5. Accordingly, a person-year dataset was constructed, assigning one observation to each participant for every calendar year (appendix pp 9). To measure their independent effect, wildfire smoke PM2.5 and non-wildfire PM2.5 were simultaneously included as continuous variables in the model, that is, a two-pollutant model (see appendix pp 4–5 for additional statistical details). A Wald test was used to compare the effect of wildfire smoke PM2.5 with those of non-wildfire PM2.5. In models assessing wildfire smoke PM2.5 exposure duration and frequency, non-wildfire PM2.5 concentrations was included as a continuous covariate. In the main analyses, 3-year moving averages (lag 0–2) were treated as exposure window, restricting the analytic period to 01/01/2008–31/12/2019. Given the narrower concentration range of wildfire smoke PM2.5 compared with non-wildfire PM2.5, associations were expressed as hazard ratios (HRs) with corresponding 95% confidence intervals (CIs) per 1-standard deviation (SD) increase, to reflect the relative magnitude of exposure variation in real-world settings. In addition, HRs with 95% CIs were also reported per 1-unit increase to facilitate direct comparison of the effect estimates. To quantify the overall mortality burden attributable to each pollutant, we estimated the average annual excess all-cause deaths by (HRΔx − 1)/HRΔx × total annual deaths within the study population, where HRΔx is evaluated the HR at the observed mean exposure level Δx 16.
To assess the potential nonlinear relationships of wildfire smoke PM2.5 and non-wildfire PM2.5 with all-cause mortality, exposure-response curves were fitted using restricted cubic splines, with knots at the 10th, 50th, and 90th percentiles, using 0 as the reference. Stratified analyses were conducted by age at diagnosis, sex, race, residential area, cancer stage, histology, number of comorbidities, tobacco use disorder status, and receipt of surgery, radiotherapy, or chemotherapy in the first course of treatment. P-values for interaction were obtained from models including an interaction term between exposure and the grouping variable in the whole study population.
In sensitivity analyses, we assessed associations using annual average exposures, single-pollutant models for wildfire smoke PM2.5 and non-wildfire PM2.5, and models replacing seasonal temperatures with annual average temperature. Moreover, we adjusted for tobacco, alcohol, and drug use disorders identified from Medicare parts A and B claims. These diagnoses primarily capture more severe behavioral disorders. Due to their limited sensitivity, these disorders were not adjusted for in main analyses. To assess potential measurement error from subtracting wildfire smoke PM2.5 from total PM2.5, negative values of non-wildfire PM2.5 were set to zero rather than excluded. All analyses were performed in R (version 4.4.1) using Minerva High Performance Computing system,17 and a two-sided P value <0.05 was considered as statistically significant.
Role of the funding source
The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Results
From 01/01/2008 to 31/12/2019, this study cohort included 414,016 lung cancer patients, with a median follow-up of 2 years (IQR 1–4), contributing a total of 1,211,490 person-years (Table 1). During the study period, 319,928 (77.3%) patients died. 208,468 (50.4%) patients were female, 356,356 (86.1%) were White, and 367,965 (88.7%) had NSCLC.
Table 1.
Characteristics of older lung cancer cohort, SEER-Medicare 2008–2019.
| Characteristics | N or n (%) |
|---|---|
| Number of individuals | 414,016 (100) |
| Number of person-years | 1,211,490 |
| Average years of follow-up | 2.9 |
| Number of deaths during follow up | 319,928 (77.3) |
| Age at diagnosis, y | |
| 65–74 | 217,959 (52.6) |
| 75–79 | 91,701 (22.1) |
| 80–84 | 63,649 (15.4) |
| ≥85 | 40,707 (9.8) |
| Sex | |
| Male | 205,548 (49.6) |
| Female | 208,468 (50.4) |
| Race and ethnicity | |
| White | 356,356 (86.1) |
| Black | 36,914 (8.9) |
| Asian or Pacific Islander | 18,982 (4.6) |
| American Indian or Alaska Native | 1,269 (0.3) |
| Unknown | 495 (0.1) |
| Residential area | |
| Urban | 359,248 (86.8) |
| Rural | 54,768 (13.2) |
| Stage at diagnosis | |
| Localized | 111,026 (26.8) |
| Regional | 99,502 (24.0) |
| Distant | 203,488 (49.1) |
| Histology | |
| SCLC | 46,687 (11.3) |
| NSCLC | 367,329 (88.7) |
| Adenocarcinoma | 201,183 (54.8) |
| Large Cell Carcinoma | 5,810 (1.6) |
| Squamous Cell Carcinoma | 96,817 (26.3) |
| Other NSCLC | 63,519 (17.3) |
| First course treatment | |
| Surgery | 76,978 (18.6) |
| Chemotherapy | 112,632 (27.2) |
| Radiotherapy | 94,732 (22.9) |
| Number of any of the 6 comorbidities a | |
| 0 | 179,523 (43.4) |
| 1 | 98,720 (23.8) |
| 2 | 70,341 (17.0) |
| ≥3 | 65,432 (15.8) |
| Comorbidity b | |
| Chronic obstructive pulmonary disease | 203,478 (49.1) |
| Chronic kidney disease | 130,251 (31.5) |
| Congestive heart failure | 120,982 (29.2) |
| Diabetes | 113,999 (27.5) |
| Stroke | 39,181 (9.5) |
| Acute myocardial infarction | 14,057 (3.4) |
| Behavioral disorders b | |
| Tobacco use disorder | 86650 (20.9) |
| Alcohol use disorder | 13,918 (3.4) |
| Drug use disorder | 11,879 (2.9) |
Note: SCLC, small cell lung cancer; NSCLC, non-small cell lung cancer.
This variable was calculated according to the presence of six comorbidities, including chronic obstructive pulmonary disease, acute myocardial infarction, congestive heart failure, diabetes, stroke, and chronic kidney disease, at the time of diagnosis.
These data were defined as the presence at diagnosis and during follow-up.
The median concentrations of 3-year moving average wildfire smoke PM2.5 and non-wildfire PM2.5 were 0.28 μg/m3 (IQR 0.21–0.39) and 8.11 μg/m3 (IQR 6.84–9.47), respectively (Table 2). Over a 3-year exposure window, the median number of smoke days was 24.67 (IQR 19.00–32.33) and the median number of smoke waves was 2.67 (IQR 2.00–3.67).
Table 2.
Distributions of exposure levels and neighborhood-level characteristics.
| Variables | Min. | Mean (SD) | Max. | Percentile | ||||
|---|---|---|---|---|---|---|---|---|
| 5th | 25th | 50th | 75th | 95th | ||||
| Three-year moving averages | ||||||||
| Wildfire smoke PM2.5, μg/m3 | 0.01 | 0.35 (0.30) | 5.41 | 0.09 | 0.21 | 0.28 | 0.39 | 0.85 |
| Non-wildfire PM2.5, μg/m3 | 0.64 | 8.24 (2.23) | 20.52 | 4.77 | 6.84 | 8.11 | 9.47 | 12.17 |
| Number of smoke days, n a | 2.33 | 27.56 (13.91) | 132.67 | 11.00 | 19.00 | 24.67 | 32.33 | 57.33 |
| Number of smoke waves, n b | 0.00 | 2.95 (1.64) | 13.67 | 0.67 | 2.00 | 2.67 | 3.67 | 6.00 |
| Summer temperature, °C | 13.12 | 22.99 (2.76) | 35.45 | 18.42 | 21.28 | 22.96 | 24.45 | 27.82 |
| Winter temperature, °C | −14.87 | 4.22 (6.20) | 22.96 | −4.24 | −0.56 | 2.33 | 9.86 | 14.5 |
| Neighborhood-level annual characteristics | ||||||||
| Median property value, $ | 9,999 | 315,732 (236,450) | 2,000,001 | 79,864 | 146,900 | 255,400 | 410,000 | 777,300 |
| High school completion rate, % | 0.00 | 86.69 (8.53) | 100.00 | 70.29 | 82.14 | 88.90 | 93.03 | 96.56 |
| Percent of Black, % | 0.00 | 10.76 (16.86) | 97.46 | 0.44 | 1.56 | 4.02 | 11.46 | 47.72 |
| Percent of Hispanic, % | 0.00 | 15.52 (17.28) | 97.74 | 1.50 | 3.81 | 8.68 | 20.04 | 55.21 |
| Percent of age over 65 years old, % | 0.11 | 12.09 (4.86) | 75.34 | 6.61 | 9.38 | 11.52 | 13.76 | 18.76 |
3-year average number of days exposed to smoke PM2.5 exceeding 0 μg/m3 over a 3-year moving window.
A smoke wave was defined as 2 or more consecutive days with daily smoke PM2.5 concentration exceeding the 95th percentile of the distribution of daily smoke PM2.5 data (2.65 μg/m3).
In the two-pollutant model, 1-SD increases in 3-year moving average concentrations of wildfire smoke PM2.5 and non-wildfire PM2.5 were associated with higher risk of death, with HRs of 1.0229 (95% CI 1.0189–1.0270, p<0.0001) and 1.0384 (95% CI 1.0335–1.0432, p<0.0001), respectively (Table 3).
Table 3.
Hazard ratios for mortality (per 1-SD or 1-unit increase) and estimated annual excess deaths in the study population attributable to wildfire smoke PM2.5 and non-wildfire PM2.5 exposure.
| Three-year moving average exposure | Hazard ratio (95% CI) per 1-SD increase | Hazard ratio (95% CI) per 1-unit increase | Annual excess deaths (95% CI) c |
|---|---|---|---|
| Concentration, μg/m3 a | |||
| Wildfire smoke PM2.5 | 1.0229 (1.0189, 1.0270) | 1.0782 (1.0641, 1.0924) | 696 (576, 814) |
| Non-wildfire PM2.5 | 1.0384 (1.0335, 1.0432) | 1.0171 (1.0149, 1.0192) | 3,465 (3,061, 3,864) |
| Smoke-event duration and frequency, day/event b | |||
| Number of smoke days | 1.0489 (1.0443, 1.0535) | 1.0034 (1.0031, 1.0037) | - |
| Number of smoke waves | 1.0293 (1.0249, 1.0338) | 1.0177 (1.0151, 1.0204) | - |
Note: All model was adjusted by calendar year, age at diagnosis, sex, race, residential area, cancer stage, histology, summer temperature, winter temperature, median property value, high school completion rate, percentage of Hispanic, percentage of Black, percentage of age over 65 years old, number of comorbidities, and whether the patient received surgery, radiotherapy or chemotherapy.
The 3-year moving average of smoke PM2.5 and non-wildfire PM2.5 were both included in the model to mutually adjust for their potential confounding effects between each other. A total of 319,928 death events were recorded across 1,211,490 person-years. The Wald test showed a significant difference between the effects of wildfire smoke PM2.5 and non-wildfire PM2.5 per 1-μg/m3 increase (χ2 =76.75, P< 0.001).
Non-wildfire PM2.5 was included in the model as a continuous covariate. A smoke day is any day on which wildfire smoke PM2.5 concentrations exceeds 0 μg/m3. A smoke wave was defined as at least two consecutive days with daily wildfire smoke PM2.5 concentrations exceeding the 95th percentile of the daily smoke PM2.5 distribution, which corresponds to 2.65 μg/m3 in this study.
Annual excess all-cause deaths were estimated by (HR(x)-1)/HR(x) × total annual deaths, where represents the hazard ratio at the observed mean exposure level x.
When effects were expressed per 1-μg/m3 increase using the same increment for both pollutants to allow direct comparison of their relative toxicity, the HR for death was substantially higher for wildfire smoke PM2.5 (HR 1.0782, 95% CI 1.0641–1.0924, p<0.0001) than for non-wildfire PM2.5 (HR 1.0171, 95% CI 1.0149–1.0192, p<0.0001) (p<0.001). Overall, an estimated 696 (95% CI 576–814) and 3,465 (95% CI 3,061–3,864) annual excess all-cause deaths in our cohort were attributable to wildfire smoke PM2.5 and non-wildfire PM2.5, respectively.
For wildfire smoke PM2.5 exposure duration and frequency, the 3-year averages of smoke days and smoke waves were positively associated with mortality risk, with HRs of 1.0489 (95% CI 1.0443–1.0535, p<0.0001, SD 13.91 days) and 1.0293 (95% CI: 1.0249–1.0338, p<0.0001, SD 1.64 events) per 1-SD increase, respectively (Table 3). In the sensitivity analyses, all associations remained robust (appendix pp 10–11).
The exposure-response curves for the associations of wildfire smoke PM2.5 concentrations, non-wildfire PM2.5 concentrations, and the number of smoke days with the risk of death exhibited linear trends (all P for nonlinearity > 0.05) (Figure 1). The slope for wildfire smoke PM2.5 appeared clearly steeper than that for non-wildfire PM2.5, consistent with its larger HR of death per 1-μg/m3 increase. A J-shaped association was observed between the number of smoke waves and mortality risk (P for nonlinearity < 0.0001).
Figure 1. Exposure-response curves for 3-year moving averages of smoke PM2.5 and non-wildfire PM2.5 concentrations (A), number of smoke days (B), and number of smoke waves (C) and the risk of death.
A smoke day corresponds to a day with wildfire smoke PM2.5 concentrations greater than 0 μg/m3. A smoke wave was defined as at least two consecutive days with daily wildfire smoke PM2.5 concentrations exceeding the 95th percentile of the daily smoke PM2.5 distribution, which corresponds to 2.65 μg/m3 in this study. Estimates at higher concentrations should be interpreted cautiously due to sparse data.
In stratified analyses, the association between wildfire smoke PM2.5 and mortality risk did not significantly differ by age, race, cancer stage, histology, number of comorbidities, residential area, status of tobacco use disorder, or whether received radiotherapy during initial treatment (all P for interaction >0.05), but was modified by sex and whether received surgery or chemotherapy, with higher mortality risk among women and those who did not receive surgery or chemotherapy. The effect of wildfire smoke PM2.5 was significantly greater than that of non-wildfire PM2.5 across most subgroups (all P for difference < 0.05), except among Black individuals and those receiving surgery during initial treatment. Furthermore, greater mortality risk from non-wildfire PM2.5 was observed among younger older patients, Black individuals, patients diagnosed at a localized stage or with squamous cell carcinoma, rural residents, individuals with more comorbidities, and untreated patients.
Discussion
Based on a large-scale lung cancer cohort of elderly Medicare beneficiaries in the US, this study comprehensively explored the effects of chronic exposures to wildfire smoke PM2.5 and non-wildfire PM2.5 on overall survival of lung cancer patients. Three-year moving average concentrations of both wildfire smoke PM2.5 and non-wildfire PM2.5 were linearly associated with increased mortality risk. Wildfire smoke PM2.5 shows substantially greater toxicity, as shown by the effects per unit increase in exposure compared with non-wildfire PM2.5. Overall, wildfire smoke PM2.5 and non-wildfire PM2.5 contributed to approximately 700 and 3,470 average annual excess deaths, respectively, in our cohort. Notably, although the average concentration of non-wildfire PM2.5 was ~23.5 times higher than wildfire smoke PM2.5, the corresponding mortality burden was only about ~5 times greater. Higher wildfire smoke PM2.5 exposure duration and frequency were also positively associated with risk of death, with clear upward trends, particularly at higher exposure levels. Considering wildfire smoke PM2.5’s high toxicity and its status as the fastest-growing contributor to ambient air pollution, our findings suggest the need to reduce wildfire smoke PM2.5 exposure in clinical care settings for lung cancer patients, especially during wildfire episodes, for example by using high-efficiency particulate air filters, wearing protective masks, and improving indoor ventilation system in both home and hospitals.
Given the high incidence and poor prognosis of lung cancer, identifying modifiable risk factors that influence survival among patients is a critical public health priority. To the best of our knowledge, only two studies to date have examined the impact of general wildfire events on survival among lung cancer patients. In a cohort of 466,912 NSCLC patients who underwent surgery, Zhang et al. found that exposure to wildfire events within the first postoperative year was associated with shorter survival compared to unexposed patients.10 Similarly, among 202,935 adults with inoperable Stage III NSCLC receiving radiation therapy, Lichter et al. reported that exposure to wildfire events during treatment was related to poorer survival.9 Consistent with these findings, we also observed that the duration and frequency of exposure to wildfire smoke PM2.5 was associated with higher risk of dying. However, the two studies did not fully account for key prognostic factors and relied on general wildfire events, rather than specific smoke PM2.5 concentrations, as the exposure metric.
In this study, we found that long-term exposure to wildfire smoke PM2.5 was associated with a 2.29% higher risk of death per 0.30 μg/m3 increase. A UK Biobank cohort reported a 0.4% increase in all-cause mortality per 10-μg/m3 increase in 3-year cumulative wildfire smoke PM2.5 concentrations in the general population.18 Although the two populations differ in ways that preclude direct comparison, these findings suggest greater vulnerability among older lung cancer patients. The heightened susceptibility may reflect impaired lung function, weakened immune response, and elevated oxidative stress commonly observed in lung cancer patients, which can amplify the adverse effects of wildfire smoke PM2.5 exposure. PM2.5 can penetrate the respiratory tract, and in vitro studies have shown that exposure to wildfire particulate matter can induce oxidative stress and reactive oxygen species production, trigger inflammatory responses characterized by cytokine release (e.g., IL-6, IL-8, and TNF-ɑ), and cause cytotoxic and genotoxic effects such as reduced cell viability and DNA damage.19,20 Moreover, exposure to wildfire particulate matter at environmentally relevant doses has been observed to impair phagocytosis in THP-1 lung macrophages.19 These biological effects may compromise pulmonary immune defense and contribute to a microenvironment that facilitates tumor progression and adverse clinical outcomes.
Moreover, wildfire smoke PM2.5 exhibited greater toxicity than non-wildfire PM2.5 per 1-μg/m3 increase in exposure. Although they share many chemical components, their relative composition varies with fuel type, combustion conditions, and atmospheric processing. The greater carcinogenic potential of wildfire smoke PM2.5 may be partly attributed to its higher proportion of smaller particles and elevated concentrations of known human carcinogens (such as polycyclic aromatic hydrocarbons and heavy metals) and polar organic compounds.21 These polar compounds may further enhance the toxicity of wildfire smoke PM2.5 by promoting oxidative stress and inflammation.22 For instance, exposure to polycyclic aromatic hydrocarbons-bound PM2.5 has been shown to promote lung cancer cell metastasis by inducing epithelial-mesenchymal transition and extracellular matrix degradation via increased reactive oxygen species.23
Importantly, both wildfire smoke PM2.5 and non-wildfire PM2.5 showed increasing exposure-response relationships with risk of death, suggesting the absence of a safe threshold for their long-term impact on survival of lung cancer patients. Emissions from wildfires can travel thousands of miles, crossing borders and continents. Therefore, under a rapidly changing climate, addressing the adverse health effects of wildfire smoke requires coordinated efforts beyond national borders.
Subgroup analyses found that women were more susceptible to wildfire smoke PM2.5 than men. Sex-related biological differences, including lower DNA repair capacity, smaller airway and lung volume, and enhanced inflammatory reactivity in women,24,25 may contribute to this increased vulnerability. Compared to patients who received first-course treatment, those who did not receive surgery, chemotherapy or radiotherapy were more susceptible to both wildfire smoke PM2.5 and non-wildfire PM2.5. One possible explanation is that untreated patients were more likely to have advanced-stage cancer or poor physical condition, increasing susceptibility to environmental stressors. For example, first-course surgical treatment may mitigate PM2.5-associated mortality, as mortality rate in the non-surgical group was approximately three times that of the surgical group in our study. The associations were not modified by status of tobacco use disorder, suggesting that PM2.5 acts as an independent risk factor irrespective of tobacco use disorder.
Consistent with previous studies,21,26 we observed higher vulnerability to non-wildfire PM2.5 among younger beneficiaries and Black individuals. Interestingly, patients with NSCLC other than squamous cell carcinoma had lower non-wildfire PM2.5-related mortality risks than those with SCLC. Both squamous cell carcinoma and SCLC are strongly associated with heavy smoking,27 and smoking-related pulmonary impairment may amplify the adverse effects of non-wildfire PM2.5 exposure. In addition, the highly aggressive nature of SCLC means mortality is often dominated by tumor progression, potentially reducing the relative contribution of non-wildfire PM2.5. Similarly, patients diagnosed at distant stage appeared less susceptible than those with localized or regional disease. Individuals with more comorbidities had higher mortality risk, probably because that non-wildfire PM2.5 may contribute to the development or worsening of cardiopulmonary diseases such as chronic obstructive pulmonary disease, further exacerbating mortality risk.
Our study has a few strengths. First, to our knowledge, this is one of the largest longitudinal cohort studies of lung cancer patients in the US, with detailed sociodemographic, clinical, cancer treatment, and comorbidity profiles from the SEER-Medicare database. Second, we comprehensively assessed the impact of wildfire smoke PM2.5 on overall survival of lung cancer patients by including multiple exposure indicators (i.e., concentration, duration, and frequency), estimating the overall mortality burden, modeling exposure-response relationships, and identifying susceptible subgroups. Third, we directly compared the effect of wildfire smoke PM2.5 with those of non-wildfire PM2.5, offering strong evidence of the greater toxicity of wildfire-derived PM2.5 than non-wildfire PM2.5.
Several limitations should also be acknowledged. First, the predictive model for wildfire smoke PM2.5, with suboptimal performance and a relatively coarse spatial resolution (10 × 10 km), may not capture fine-scale differences in individual exposure, potentially introducing exposure measurement error that biases results toward the null.28 Second, the study population consisted of older lung cancer patients. Thus, caution is warranted when generalizing our findings to younger patients. Third, restricted by available data sources, smoking status and history were not considered in the main analysis. However, we additionally adjusted for behavioral disorders in the sensitivity analyses, and the results changed minimally. Subgroup analyses also showed that exposure to both wildfire smoke and non-wildfire PM2.5 was associated with lower survival, regardless of whether patients had tobacco use disorders. In addition, some unmeasured confounders may remain, such as proximity to major highways or power plants.29,30 Finally, variables for first-course treatments of chemotherapy and radiotherapy in the SEER-Medicare database showed only moderate sensitivity, raising the potential for nondifferential misclassification of treatment status. Although we adjusted for calendar year to partially account for temporal trends in survival,4 unmeasured contemporary therapies, such as immunotherapy or targeted agents, may represent residual confounding.
In conclusion, long-term exposure to both wildfire smoke PM2.5 and non-wildfire PM2.5 was associated with increased risk of death in older lung cancer patients, with stronger per-unit effect observed for wildfire smoke PM2.5. Notably, although the average concentration of non-wildfire PM2.5 being ~23.5 times higher than that of wildfire smoke PM2.5, the corresponding mortality burden was only about 5 times greater. Our findings highlight the importance of wildfire management and exposure reduction strategies in clinical settings to mitigate the growing health impacts of wildfire smoke, especially for vulnerable lung cancer patients.
Supplementary Material
Table 4.
Hazard ratios for mortality associated with 1-SD or 1-μg/m3 increase in wildfire smoke PM2.5 and non-wildfire PM2.5 for each subgroup of individual-level characteristics.
| Characteristic | No. of death events/ No. of person-years | Wildfire smoke PM2.5 | Non-wildfire PM2.5 | P for difference a | ||||
|---|---|---|---|---|---|---|---|---|
| Hazard ratio (95% CI) per 1-SD increase | Hazard ratio (95% CI) per 1-μg/m3 increase | P for interaction | Hazard ratio (95% CI) per 1-SD increase | Hazard ratio (95% CI) per 1-μg/m3 increase | P for interaction | |||
| Age at diagnosis | ||||||||
| 65–74 years old | 158821/701120 | 1.0235 (1.0177, 1.0293) | 1.0802 (1.0599, 1.1008) | ref | 1.0413 (1.0343, 1.0483) | 1.0183 (1.0153, 1.0214) | ref | <0.0001 |
| 75–79 years old | 71943/263936 | 1.0267 (1.0184, 1.0351) | 1.0914 (1.0623, 1.1212) | 0.61 | 1.0464 (1.0363, 1.0566) | 1.0206 (1.0161, 1.0250) | 0.56 | <0.0001 |
| 80–84 years old | 53206/162642 | 1.0191 (1.0093, 1.0291) | 1.0650 (1.0313, 1.0997) | 0.40 | 1.0317 (1.0200, 1.0435) | 1.0141 (1.0089, 1.0193) | 0.0049 | 0.0027 |
| ≥ 85 years old | 35958/83792 | 1.0203 (1.0085, 1.0323) | 1.0690 (1.0284, 1.1112) | 0.64 | 1.0208 (1.0065, 1.0352) | 1.0093 (1.0029, 1.0156) | <0.0001 | 0.0034 |
| Sex | ||||||||
| Male | 165805/571056 | 1.0178 (1.0121, 1.0235) | 1.0604 (1.0408, 1.0803) | ref | 1.0372 (1.0305, 1.0440) | 1.0165 (1.0136, 1.0195) | ref | <0.0001 |
| Female | 154123/640434 | 1.0281 (1.0224, 1.0339) | 1.0964 (1.0763, 1.1169) | 0.030 | 1.0397 (1.0327, 1.0468) | 1.0176 (1.0145, 1.0207) | 0.62 | <0.0001 |
| Race | ||||||||
| White | 276167/1049097 | 1.0232 (1.0190, 1.0274) | 1.0789 (1.0643, 1.0938) | ref | 1.0380 (1.0329, 1.0432) | 1.0169 (1.0147, 1.0192) | ref | <0.0001 |
| Black | 29315/101114 | 0.9994 (0.9790, 1.0202) | 0.9980 (0.9320, 1.0687) | 0.064 | 1.0710 (1.0489, 1.0935) | 1.0313 (1.0217, 1.0409) | 0.020 | 0.35 |
| Others | 14446/61279 | 1.0295 (1.0087, 1.0506) | 1.1012 (1.0293, 1.1781) | 0.77 | 0.9923 (0.9653, 1.0200) | 0.9965 (0.9843, 1.0089) | 0.030 | 0.0035 |
| Cancer stage | ||||||||
| Localized | 62385/506509 | 1.0237 (1.0153, 1.0323) | 1.0810 (1.0516, 1.1111) | ref | 1.0517 (1.0401, 1.0635) | 1.0229 (1.0178, 1.0280) | ref | <0.0001 |
| Regional | 72779/336658 | 1.0283 (1.0197, 1.0371) | 1.0972 (1.0668, 1.1285) | 0.59 | 1.0613 (1.0505, 1.0722) | 1.0271 (1.0223, 1.0318) | 0.13 | <0.0001 |
| Distant | 184764/368323 | 1.0211 (1.0157, 1.0265) | 1.0716 (1.0529, 1.0907) | 0.75 | 1.0290 (1.0228, 1.0351) | 1.0129 (1.0102, 1.0156) | <0.0001 | <0.0001 |
| Histology | ||||||||
| SCLC | 42332/86978 | 1.0197 (1.0071, 1.0325) | 1.0671 (1.0239, 1.1121) | ref | 1.0360 (1.0218, 1.0504) | 1.0160 (1.0097, 1.0223) | ref | 0.020 |
| NSCLC | 277596/1124512 | 1.0232 (1.0190, 1.0275) | 1.0792 (1.0644, 1.0942) | - | 1.0388 (1.0336, 1.0440) | 1.0172 (1.0150, 1.0195) | - | <0.0001 |
| Adenocarcinoma | 677077/143176 | 1.0261 (1.0202, 1.0321) | 1.0893 (1.0686, 1.1105) | 0.46 | 1.0326 (1.0252, 1.0401) | 1.0145 (1.0112, 1.0178) | 0.069 | <0.0001 |
| Large Cell Carcinoma | 4931/15617 | 1.0131 (0.9575, 1.0719) | 1.0442 (0.8658, 1.2593) | 0.89 | 1.0044 (0.9512, 1.0606) | 1.0020 (0.9778, 1.0268) | 0.63 | 0.66 |
| Squamous Cell Carcinoma | 76000/280808 | 1.0204 (1.0116, 1.0294) | 1.0695 (1.0390, 1.1009) | 0.98 | 1.0574 (1.0465, 1.0683) | 1.0254 (1.0206, 1.0303) | 0.038 | 0.0042 |
| Other NSCLC | 53489/151010 | 1.0161 (1.0052, 1.0272) | 1.0546 (1.0172, 1.0933) | 0.59 | 1.0271 (1.0156, 1.0389) | 1.0121 (1.0070, 1.0173) | 0.023 | 0.024 |
| Number of any of the 6 comorbidities | ||||||||
| 0 | 125994/561430 | 1.0247 (1.0184, 1.0310) | 1.0842 (1.0624, 1.1065) | ref | 1.0303 (1.0229, 1.0377) | 1.0135 (1.0102, 1.0168) | ref | <0.0001 |
| 1 | 66587/278525 | 1.0223 (1.0136, 1.0312) | 1.0761 (1.0458, 1.1074) | 0.79 | 1.0401 (1.0294, 1.0508) | 1.0178 (1.0131, 1.0225) | 0.63 | 0.00012 |
| 2 | 58718/193978 | 1.0352 (1.0255, 1.0451) | 1.1219 (1.0872, 1.1576) | 0.052 | 1.0399 (1.0283, 1.0517) | 1.0177 (1.0126, 1.0229) | 0.021 | <0.0001 |
| ≥3 | 68629/177557 | 1.0139 (1.0049, 1.0230) | 1.0469 (1.0164, 1.0784) | 0.10 | 1.0431 (1.0319, 1.0545) | 1.0191 (1.0142, 1.0241) | 0.00081 | 0.074 |
| Surgery | ||||||||
| Yes | 41785/379577 | 1.0089 (0.9990, 1.0189) | 1.0298 (0.9966, 1.0640) | ref | 1.0102 (0.9979, 1.0227) | 1.0046 (0.9991, 1.0101) | ref | 0.13 |
| No/unknown | 278143/831913 | 1.0248 (1.0204, 1.0292) | 1.0847 (1.0692, 1.1004) | 0.047 | 1.0421 (1.0368, 1.0474) | 1.0187 (1.0164, 1.0210) | <0.0001 | <0.0001 |
| Chemotherapy | ||||||||
| Yes | 90604/300088 | 1.0081 (1.0006, 1.0156) | 1.0271 (1.0021, 1.0526) | ref | 0.9993 (0.9911, 1.0075) | 0.9997 (0.9960, 1.0034) | ref | 0.030 |
| No/unknown | 229324/911402 | 1.0257 (1.0209, 1.0305) | 1.0878 (1.0710, 1.1048) | 0.026 | 1.0520 (1.0460, 1.0580) | 1.0230 (1.0204, 1.0257) | <0.0001 | <0.0001 |
| Radiotherapy | ||||||||
| Yes | 82169/281537 | 1.0089 (1.0003, 1.0175) | 1.0297 (1.0011, 1.0592) | ref | 0.9995 (0.9908, 1.0083) | 0.9998 (0.9959, 1.0037) | ref | 0.040 |
| No/unknown | 237759/929953 | 1.0241 (1.0195, 1.0287) | 1.0823 (1.0664, 1.0985) | 0.34 | 1.0476 (1.0418, 1.0535) | 1.0211 (1.0186, 1.0237) | <0.0001 | <0.0001 |
| Residential area | ||||||||
| Rural | 44101/147493 | 1.0171 (1.0070, 1.0274) | 1.0580 (1.0233, 1.0938) | ref | 1.0539 (1.0367, 1.0713) | 1.0238 (1.0163, 1.0314) | ref | 0.050 |
| Urban | 275827/1063997 | 1.0233 (1.0188, 1.0277) | 1.0794 (1.0640, 1.0950) | 0.47 | 1.0347 (1.0295, 1.0399) | 1.0154 (1.0131, 1.0177) | <0.0001 | <0.0001 |
| Tobacco use disorder | ||||||||
| No | 254739/1017321 | 1.0205 (1.0159, 1.0251) | 1.0696 (1.0536, 1.0859) | ref | 1.0368 (1.0314, 1.0423) | 1.0164 (1.0140, 1.0188) | ref | <0.0001 |
| Yes | 65189/194169 | 1.0266 (1.0173, 1.0360) | 1.0912 (1.0588, 1.1246) | 0.18 | 1.0463 (1.0342, 1.0585) | 1.0205 (1.0152, 1.0259) | 0.14 | <0.0001 |
Note: Each subgroup-specific model was adjusted for the same set of covariates as the primary multivariable model, except for the stratification variable.
A Wald test was used to compare the effect of wildfire smoke PM2.5 with those of non-wildfire PM2.5 per 1-μg/m3 increase.
Research in context.
Evidence before this study
We searched PubMed for literature published in English between database inception and May 20, 2025, using the following terms: (fine particulate matter[Title/Abstract] OR PM2.5[Title/Abstract]) AND (lung cancer[Title/Abstract] OR NSCLC[Title/Abstract] OR SCLC[Title/Abstract]) AND (mortality[Title/Abstract] OR survival[Title/Abstract]). The 238 identified studies only focus on total PM2.5, without distinguishing between PM2.5 from wildfire and non-wildfire sources. To our knowledge, only two studies have reported the impact of wildfire exposure on survival among patients with non-small cell lung cancer (NSCLC), using general wildfire events rather than specific smoke concentrations as the exposure metric, and with incomplete adjustment for key prognostic cancer-related factors.
Added value of this study
This study, the most detailed to date, compares the effects of wildfire smoke PM2.5 versus non-wildfire PM2.5 on the survival of older lung cancer patients (n=414,016, 1,211,490 person-years). By including multiple exposure indicators (i.e., concentration, duration, and frequency), estimating the number of attributable deaths, plotting exposure-response curves, and identifying susceptible subgroups, we comprehensively assessed the impact of wildfire smoke PM2.5 on overall survival of lung cancer patients. Our findings highlight a substantially higher toxicity associated with wildfire smoke PM2.5 compared with non-wildfire PM2.5.
Implications of all the available evidence
Our findings underscore the potential benefits of targeted policies and interventions to reduce exposure to wildfire smoke PM2.5, with important implications for survival among lung cancer patients. Given its higher toxicity and its rapidly increasing contribution to overall air pollution, greater emphasis should be placed on mitigation strategies to reduce wildfire smoke exposure, particularly for lung cancer patients, who may be especially vulnerable.
Acknowledgments
Research reported in this publication was also supported by the National Institute of Environmental Health Sciences (NIEHS) under award number R01ES036566, P30ES023515, and P30ES019776; the National Cancer Institute (NCI) under award number P30CA138292; and the Office of Research Infrastructure of the National Institutes of Health under award number S10OD026880 and S10OD030463. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. This work was also supported in part through the Minerva computational and data resources and staff expertise provided by Scientific Computing and Data at the Icahn School of Medicine at Mount Sinai and supported by the Clinical and Translational Science Awards (CTSA) grant UL1TR004419 from the National Center for Advancing Translational Sciences. The collection of cancer incidence data used in this study was supported by the California Department of Public Health pursuant to California Health and Safety Code Section 103885; Centers for Disease Control and Prevention’s (CDC) National Program of Cancer Registries, under cooperative agreement NU58DP007156; the National Cancer Institute’s Surveillance, Epidemiology and End Results Program under contract HHSN261201800032I awarded to the University of California, San Francisco, contract HHSN261201800015I awarded to the University of Southern California, and contract HHSN261201800009I awarded to the Public Health Institute. The ideas and opinions expressed herein are those of the author(s) and do not necessarily reflect the opinions of the State of California, Department of Public Health, the National Cancer Institute, and the Centers for Disease Control and Prevention or their Contractors and Subcontractors. MQ acknowledges support from the Minghua Zhang faculty career catalyst award and climate change seed grant from Stony Brook University.
Footnotes
Declaration of interests
JPW reports research support from Sanofi, Boehringer Ingelheim, AstraZeneca, and Regeneron unrelated to the current work; consulting honoraria from PPD, Banook, Sanofi, and the American Medical Association; and serves as Chair of the Health Equality and Diversity Committee of the American Thoracic Society (unpaid). All other authors declare no competing interests.
Disclaimer
This study used the linked SEER-Medicare database. The interpretation and reporting of these data are the sole responsibility of the authors. The authors acknowledge the efforts of the National Cancer Institute; Information Management Services (IMS), Inc.; and the Surveillance, Epidemiology, and End Results (SEER) Program tumor registries in the creation of the SEER-Medicare database (https://healthcaredelivery.cancer.gov/seermedicare/obtain/use.html).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Contributor Information
Min Zhang, Department of Environmental Medicine, Icahn School of Medicine at Mount Sinai, New York, NY, USA.
Juan P. Wisnivesky, Department of Medicine, Icahn School of Medicine at Mount Sinai, New York, NY, USA.
Minghao Qiu, School of Marine and Atmospheric Sciences, Stony Brook University, Stony Brook, NY, USA; Program in Public Health, Stony Brook University, Stony Brook, NY, USA.
Mahdieh Danesh Yazdi, Program in Public Health, Stony Brook University, Stony Brook, NY, USA; Department of Family, Population, and Preventive Medicine, Stony Brook University, Stony Brook, NY, USA.
Kanhua Yin, Department of Surgery, University of Missouri-Kansas City School of Medicine, Kansas City, MO, USA.
Rosalind J. Wright, Department of Environmental Medicine, Icahn School of Medicine at Mount Sinai, New York, NY, USA; Department of Public Health, Icahn School of Medicine at Mount Sinai, New York, NY, USA.
Joel D. Schwartz, Department of Environmental Health, Harvard T.H. Chan School of Public Health, Boston, MA, USA; Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA, USA.
Christine C. Ekenga, Gangarosa Department of Environmental Health, Rollins School of Public Health, Emory University, Atlanta, GA, USA.
Robert O. Wright, Department of Environmental Medicine, Icahn School of Medicine at Mount Sinai, New York, NY, USA.
Yaguang Wei, Department of Environmental Medicine, Icahn School of Medicine at Mount Sinai, New York, NY, USA.
Data Sharing
We cannot provide SEER-Medicare datasets to any other individual or investigator; however, the datasets are available upon approval of a research protocol by the National Cancer Institute (https://healthcaredelivery.cancer.gov/seermedicare/obtain/). The local-level wildfire-specific PM2.5 data are available at https://github.com/echolab-stanford/daily-10km-smokePM. The originally predicted, grid level PM2.5 from all-sourced data are available at https://www.earthdata.nasa.gov/data/catalog/sedac-ciesin-sedac-aqdh-dapm25-us-1km-1.0. Other data are publicly available with sources described in the manuscript.
References
- 1.Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA: a cancer journal for clinicians 2024; 74(3): 229–63. [DOI] [PubMed] [Google Scholar]
- 2.Siegel RL, Kratzer TB, Giaquinto AN, Sung H, Jemal A. Cancer statistics, 2025. CA: a cancer journal for clinicians 2025; 75(1): 10–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Venuta F, Diso D, Onorati I, Anile M, Mantovani S, Rendina EA. Lung cancer in elderly patients. Journal of thoracic disease 2016; 8(Suppl 11): S908–s14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Allemani C, Matsuda T, Di Carlo V, et al. Global surveillance of trends in cancer survival 2000–14 (CONCORD-3): analysis of individual records for 37 513 025 patients diagnosed with one of 18 cancers from 322 population-based registries in 71 countries. Lancet (London, England) 2018; 391(10125): 1023–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Loomis D, Grosse Y, Lauby-Secretan B, et al. The carcinogenicity of outdoor air pollution. The Lancet Oncology 2013; 14(13): 1262–3. [DOI] [PubMed] [Google Scholar]
- 6.Qiu M, Li J, Gould CF, et al. Wildfire smoke exposure and mortality burden in the USA under climate change. Nature 2025. [DOI] [PubMed] [Google Scholar]
- 7.Kim YH, Tong H, Daniels M, et al. Cardiopulmonary toxicity of peat wildfire particulate matter and the predictive utility of precision cut lung slices. Particle and fibre toxicology 2014; 11: 29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Wegesser TC, Pinkerton KE, Last JA. California wildfires of 2008: coarse and fine particulate matter toxicity. Environmental health perspectives 2009; 117(6): 893–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Lichter KE, Larson B, Pagadala M, Mohamad O, Nogueira L. Impact of declared wildfire disasters on survival of lung cancer patients undergoing radiation. Cancer causes & control : CCC 2025; 36(6): 561–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Zhang D, Xi Y, Boffa DJ, Liu Y, Nogueira LM. Association of Wildfire Exposure While Recovering From Lung Cancer Surgery With Overall Survival. JAMA oncology 2023; 9(9): 1214–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Childs ML, Li J, Wen J, et al. Daily Local-Level Estimates of Ambient Wildfire Smoke PM(2.5) for the Contiguous US. Environmental science & technology 2022; 56(19): 13607–21. [DOI] [PubMed] [Google Scholar]
- 12.Di Q, Amini H, Shi L, et al. An ensemble-based model of PM(2.5) concentration across the contiguous United States with high spatiotemporal resolution. Environment international 2019; 130: 104909. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Li J, Lu X, Liu F, et al. Chronic Effects of High Fine Particulate Matter Exposure on Lung Cancer in China. American journal of respiratory and critical care medicine 2020; 202(11): 1551–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Zhu M, Han Y, Mou Y, et al. Effect of Long-Term Fine Particulate Matter Exposure on Lung Cancer Incidence and Mortality in Chinese Nonsmokers. American journal of respiratory and critical care medicine 2025; 211(4): 600–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Casey JA, Kioumourtzoglou MA, Padula A, et al. Measuring long-term exposure to wildfire PM(2.5) in California: Time-varying inequities in environmental burden. Proceedings of the National Academy of Sciences of the United States of America 2024; 121(8): e2306729121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Wei Y, Castro E, Yin K, et al. Ambient air pollution and mortality in older patients with breast cancer. JNCI: Journal of the National Cancer Institute 2025. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Kovatch P, Gai L, Cho HM, Fluder E, Jiang D. Optimizing High-Performance Computing Systems for Biomedical Workloads. IEEE International Symposium on Parallel & Distributed Processing, Workshops and Phd Forum : [proceedings] IEEE International Symposium on Parallel & Distributed Processing, Workshops and Phd F 2020; 2020: 183–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Gao Y, Huang W, Xu R, et al. Association between long-term exposure to wildfire-related PM(2.5) and mortality: A longitudinal analysis of the UK Biobank. Journal of hazardous materials 2023; 457: 131779. [DOI] [PubMed] [Google Scholar]
- 19.Bazina L, Deloid G, Fritzky L, Lizonova D, Vaze N, Demokritou P. Impact of Canadian Wildfire-Emitted Particulate Matter on THP-1 Lung Macrophage Health and Function. Environmental science & technology 2025; 59(8): 3869–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Heibati B, Renz H, Lacy P. Wildfire and wood smoke effects on human airway epithelial cells: A scoping review. Environmental research 2025; 272: 121153. [DOI] [PubMed] [Google Scholar]
- 21.Zhang M, Castro E, Qiu M, et al. Cardiopulmonary hospitalization risks from wildfire and non-wildfire PM(2.5) in 20 US states. medRxiv : the preprint server for health sciences 2025. [Google Scholar]
- 22.Williams KM, Franzi LM, Last JA. Cell-specific oxidative stress and cytotoxicity after wildfire coarse particulate matter instillation into mouse lung. Toxicology and applied pharmacology 2013; 266(1): 48–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Yue H, Yun Y, Gao R, Li G, Sang N. Winter Polycyclic Aromatic Hydrocarbon-Bound Particulate Matter from Peri-urban North China Promotes Lung Cancer Cell Metastasis. Environmental science & technology 2015; 49(24): 14484–93. [DOI] [PubMed] [Google Scholar]
- 24.Liao M, Braunstein Z, Rao X. Sex differences in particulate air pollution-related cardiovascular diseases: A review of human and animal evidence. The Science of the total environment 2023; 884: 163803. [DOI] [PubMed] [Google Scholar]
- 25.Mosleh B, Sarova P, Zehetmayer S, et al. Sex-based differences in lung cancer susceptibility and molecular genetics in the 2020s. Heliyon 2025; 11(2): e42089. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Wei Y, Feng Y, Danesh Yazdi M, et al. Exposure-response associations between chronic exposure to fine particulate matter and risks of hospital admission for major cardiovascular diseases: population based cohort study. BMJ (Clinical research ed) 2024; 384: e076939. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Morabia A, Wynder EL. Cigarette smoking and lung cancer cell types. Cancer 1991; 68(9): 2074–8. [DOI] [PubMed] [Google Scholar]
- 28.Wei Y, Qiu X, Yazdi MD, et al. The Impact of Exposure Measurement Error on the Estimated Concentration-Response Relationship between Long-Term Exposure to PM2.5 and Mortality. Environmental health perspectives 2022; 130(7): 77006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Alwadi Y, Alahmad B, Vieira CLZ, et al. National analysis of cancer mortality and proximity to nuclear power plants in the United States. Nature communications 2026; 17(1). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Bidoli E, Pappagallo M, Birri S, Frova L, Zanier L, Serraino D. Residential Proximity to Major Roadways and Lung Cancer Mortality. Italy, 1990–2010: An Observational Study. International journal of environmental research and public health 2016; 13(2): 191. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
We cannot provide SEER-Medicare datasets to any other individual or investigator; however, the datasets are available upon approval of a research protocol by the National Cancer Institute (https://healthcaredelivery.cancer.gov/seermedicare/obtain/). The local-level wildfire-specific PM2.5 data are available at https://github.com/echolab-stanford/daily-10km-smokePM. The originally predicted, grid level PM2.5 from all-sourced data are available at https://www.earthdata.nasa.gov/data/catalog/sedac-ciesin-sedac-aqdh-dapm25-us-1km-1.0. Other data are publicly available with sources described in the manuscript.