Abstract
Background
Although emerging studies link air pollution to mortality in patients with breast cancer, large-scale evidence remains limited. We aimed to evaluate associations between chronic exposures to three key regulated air pollutants, fine particulate matter, ozone, and nitrogen dioxide, and mortality in a nationwide cohort of older patients with breast cancer.
Methods
We constructed a cohort of patients with primary diagnosis of breast cancer aged 65 years or older between 2000 and 2016 using the Surveillance, Epidemiology, and End Results–Medicare database. High-resolution ambient concentrations of annual fine particulate matter, warm-season ozone, and annual nitrogen dioxide were estimated using hybrid models and linked to patients’ residential zip codes as proxy exposures. A 3-pollutant Cox model was fitted to estimate hazard ratios for mortality associated with each pollutant, adjusting for demographics, tumor characteristics, cancer treatments, comorbidities, lifestyle factors, meteorological variables, and neighborhood-level characteristics.
Results
Among 593 333 patients with breast cancer, a 1-µg/m3 increase in annual fine particulate matter, a 1-part per billion increase in warm-season ozone, and a 1-part per billion increase in annual nitrogen dioxide were associated with hazard ratios for mortality of 1.0048 (95% CI = 1.0026 to 1.0070), 1.0021 (95% CI = 1.0013 to 1.0029), and 1.0022 (95% CI = 1.0014 to 1.0030), respectively. This finding translated to 49 annual excess deaths attributable to fine particulate matter, 21 to ozone, and 22 to nitrogen dioxide within the cohort. Effects were substantially larger at low exposure levels. Fine particulate matter and nitrogen dioxide had greater effects in younger patients, individuals who received chemotherapy or radiation, and individuals with disease diagnosed at later stages.
Conclusion
Our findings identified air pollution as a risk factor for mortality in older patients with breast cancer. Protective measures and air pollution control strategies may help reduce exposure and improve outcomes.
Keywords: Breast cancer, mortality, air pollution, SEER-Medicare
Introduction
Breast cancer is the most common non–skin cancer and the second-leading cause of cancer death in women in the United States.1 As of 2022, more than 4 million women in the United States were living with breast cancer, and this number is projected to rise to 4.9 million by 2030.2 This rapidly growing population suggests a need for strategies to reduce mortality in patients with breast cancer; strategies to date have primarily focused on screening, early detection, and adjuvant treatment to reduce recurrence risk.3 More recently, a growing body of research has begun exploring the effects of air pollution on mortality in patients with breast cancer, with evidence suggesting its role in promoting tumor progression.4-7 In addition to the direct effect, air pollution may indirectly affect mortality in patients with breast cancer by exacerbating comorbidities such as hypertension, chronic obstructive pulmonary disease (COPD), and myocardial infarction (MI).8-10 These comorbidities are the leading causes of death in patients with breast cancer and have been linked to air pollution exposure.
Despite the growing interest, however, evidence of air pollution’s impact on breast cancer mortality remains limited.11,12 The lack of information about cancer-related confounders, including tumor characteristics, cancer treatments, and comorbidities, which are associated with both mortality and air pollution exposure through behavioral changes, is at least partially responsible for this gap.4 Large-scale studies are particularly scarce because few have linked large breast cancer cohorts with external air pollution estimates across time and space. Most studies are hampered by small samples or short study periods.13
We constructed an open cohort of 593 333 patients with breast cancer aged 65 years or older using the linked Surveillance, Epidemiology, and End Results (SEER)–Medicare database to evaluate the associations between chronic exposure to 3 major air pollutants (fine particulate matter (PM2.5), ozone, and nitrogen dioxide (NO2)) and mortality.
Methods
Study design and breast cancer cohort
We constructed an open cohort comprising Medicare beneficiaries aged 65 years and older with a primary diagnosis of breast cancer between 2000 and 2016, based on the linked SEER-Medicare database. The SEER-Medicare database integrated data from the SEER cancer registries and Medicare claims, which collectively provided longitudinal health records for Medicare beneficiaries diagnosed with cancer. The SEER component provided patient demographics (eg, age, sex, and marital status) and cancer-specific information, such as cancer diagnosis codes, date of diagnosis, histology, grade, stage at diagnosis, and first-course cancer treatments. Medicare is a federal health insurance program in the United States; the vast majority of Americans (about 97%) enroll in Medicare Part A (inpatient) and Part B (outpatient) when they first become eligible, typically at age 65 years. The Medicare component provided longitudinal records for each beneficiary, including residential zip code, demographics, diagnosis codes for all Part A and Part B claims, and date of death. It also summarized the presence of common chronic conditions based on diagnosis codes from Medicare claims, including common comorbidities in patients with breast cancer (eg, MI, Alzheimer disease, COPD, diabetes, and depression) and behavioral factors (eg, tobacco use, alcohol use disorder, drug use disorder). Because the identification of these conditions relied on claims records in Medicare parts A and B, we restricted our analysis to individuals with continuous enrollment in parts A and B throughout the follow-up period.14 Relaxing this requirement could introduce individuals with incomplete claims data, increasing the risk of missing diagnoses for those additional individuals (Figure 1). The SEER and Medicare components were linked longitudinally by individual identification codes.
Figure 1.
Schematic flowchart of the study cohort selection. SEER = Surveillance, Epidemiology, and End Results.
We followed each patient in each year, starting from 2000 or the first year of breast cancer diagnosis, whichever occurred later, to ensure that the follow-up began after diagnosis. Each patient was followed up until the earliest occurrence of death, loss to follow-up, or the study’s end in 2016.
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.
Air pollution exposure measures
We estimated daily ambient fine particulate matter, ozone, and nitrogen dioxide concentrations at 1-km2 grid cells across the contiguous United States between 2000 and 2016 using hybrid models that relied on satellite-based measurements, land-use regression, and chemical transport models.15-17 Specifically, our models incorporated predictors of normalized difference vegetation index, surface reflectance, satellite measures of absorbing aerosol index, aerosol optical depth, elevation, road density, emission inventory, traffic density, and other spatial covariates. For each pollutant, we used a neural network, random forest, and gradient-boosting machine for their capacity to capture nonlinearity and interactions and averaged their predictions with a geographically weighted regression. The cross-validated R2 values for final predictions were 0.89 for fine particulate matter, 0.86 for ozone, and 0.84 for nitrogen dioxide.
These daily, grid cell–level estimates were aggregated to the zip code level by averaging predictions for grid cells, with centroids within the boundaries of general zip codes.18 For zip codes without polygonal representations, such as apartment buildings, military bases, or post offices, predictions from the nearest grid cell were assigned. Daily zip code–level estimates were then averaged to annual levels for fine particulate matter and nitrogen dioxide and to warm season (April-September) for ozone, following previous literature.19 These averages were assigned to each patient based on their residential zip code for each follow-up year as proxy exposure measures.
Meteorological variables and neighborhood characteristics
Previous studies have linked meteorological variables and neighborhood characteristics with breast cancer survival outcomes.20-22 Some of these factors may also influence air pollution levels and thus act as potential confounders. We obtained daily minimum and maximum surface temperatures at 1-km2 grid cells across continental North America from Daymet version 4, a data product developed by the Oak Ridge National Laboratory using a weighted multivariate regression that interpolated and extrapolated ground-based observations.23 Daily average temperatures were calculated by averaging the minimum and maximum values. These daily gridded averages were then aggregated to the zip code level. Summer (June-August) and winter (December of the previous year, January, February) averages were calculated for each year and linked to each patient by residential zip code.
For neighborhood characteristics, we included the proportion of individuals aged 65 years or older living below the federal poverty line, the proportion of individuals aged 65 years or older without a high school diploma, and median household income. These data were obtained from the US Census and the American Community Survey and were available in 2000, 2010, and from 2010 to 2016. Missing values for the years 2001 to 2009 were linearly interpolated.24-26 Finally, we obtained the population density at 1-km2 grid cells in each year from the National Aeronautics and Space Administration’s Socioeconomic Data and Applications Center,27 which were aggregated to zip codes and adjusted as a measure of rurality. The rurality was used to reflect potential differences in access to breast cancer care.
Statistical analysis
We fitted single-pollutant and 3-pollutant time-varying Cox proportional hazard models to estimate hazard ratios for mortality per unit increase in annual fine particulate matter, warm-season ozone, and annual nitrogen dioxide, using follow-up year as the time scale. In the 3-pollutant model, all 3 pollutants were included to mutually adjust for their potential confounding effects among each other. Both single-pollutant and 3-pollutant models adjusted for sex, race, age, marital status, stage at diagnosis, and whether the patient received radiation or chemotherapy. We also adjusted common comorbidities in patients with breast cancer (MI, Alzheimer disease, COPD, diabetes, and depression), health behaviors (tobacco, alcohol, and drug use), and meteorological variables and neighborhood characteristics detailed in the previous section. Further, to evaluate effects at low exposure levels, we performed analyses restricting to patients consistently exposed to air pollutant levels below 5 µg/m3 for annual fine particulate matter, 30.56 parts per billion (ppb) for warm-season ozone, or 5.31 ppb for annual nitrogen dioxide throughout the follow-up period. These thresholds were selected based on the World Health Organization air quality guidelines,28 which were substantially lower than current US standards.29 Results were reported as hazard ratios for mortality per 1-μg/m3 increase in annual fine particulate matter, 1-part per billion increase in warm-season ozone, and 1-part per billion increase in annual nitrogen dioxide. Corresponding annual excess deaths per 1-unit increase in each exposure were estimated using the following formula:
annual excess deaths = annual average person-years × average mortality rate × (hazard ratio ‒ 1)
The average mortality rate was calculated as the total number of deaths divided by the total person-years.
To evaluate whether certain subgroups experienced a greater risk from the air pollutants, we fitted separate regressions for subgroups defined by sex, race, age, marital status, stage, and cancer treatment received. We further stratified the cohort by the upper or lower quartile of neighborhood characteristics, including median household income, proportion living below the federal poverty level, proportion without a high school diploma, and population density.
Due to the limited sensitivity of first-course treatments of chemotherapy and radiation as well as the International Statistical Classification of Diseases coding in identifying tobacco, alcohol, and drug use, we conducted a sensitivity analysis that excluded these variables to assess the robustness of the main results. In addition, given the decreasing air pollution levels and improving survival rates for certain molecular subtypes (eg, ERBB2 [formerly HER2] positive) over the study period, we adjusted for calendar year and breast cancer subtypes (estrogen receptor/progesterone receptor/ERBB2) to further evaluate robustness; as breast cancer subtype information was available only after 2010, this analysis was restricted to the period 2010 to 2016.
Results
Table 1 shows demographics of the study cohort, which consisted of 593 333 patients with breast cancer with 3 769 645 person-years of follow-up. During the follow-up, 173 253 (29.20%) patients died. Among all patients, most (85.34%) were White and aged 65 to 74 years (64.59%) at study entry. Approximately half the patients were married at the time of breast cancer diagnosis. More than 70% of patients were diagnosed at early stages (in situ or localized), and most (99.67%) had surgery after diagnosis. Compared with patients who survived, patients who died were older; more likely to be diagnosed at a later stage; and had higher rates of comorbidities, tobacco use, alcohol use disorder, and drug use disorder. Patients exposed to low levels of air pollutants had generally similar individual-level and neighborhood-level characteristics to patients in the full cohort (Tables S1 and S2).
Table 1.
Characteristics of patients with primary a diagnosis of breast cancer in the SEER-Medicare database.
| Characteristic | Total | Deceased | Survived |
|---|---|---|---|
| Individuals, No. | 593 333 | 173 253 | 420 080 |
| Person-years | 3 769 645 | 1 046 083 | 2 723 562 |
| Average follow-up, y | 6.35 | 6.04 | 6.48 |
| Sex, No. (%) | |||
| Male | 5059 (0.85) | 2140 (1.24) | 2919 (0.69) |
| Female | 588 274 (99.15) | 171 113 (98.76) | 417 161 (99.31) |
| Race, No. (%) | |||
| American Indian or Alaska Native | 2011 (0.34) | 539 (0.31) | 1472 (0.35) |
| Asian or Pacific Islander | 28 156 (4.75) | 4598 (2.65) | 23 558 (5.61) |
| Black | 52 852 (8.91) | 16 845 (9.72) | 36 007 (8.57) |
| White | 506 339 (85.34) | 150 457 (86.84) | 355 882 (84.72) |
| Other/unspecified | 3975 (0.66) | 814 (0.47) | 3161 (0.75) |
| Age group at study entry, No. (%) | |||
| 65-74 y | 383 208 (64.59) | 66 500 (38.38) | 316 708 (75.39) |
| 75–84 y | 141 513 (23.85) | 66 619 (38.45) | 74 894 (17.83) |
| >84 y | 68 612 (11.56) | 40 134 (23.16) | 28 478 (6.78) |
| Marital status at diagnosis, No. (%) | |||
| Single | 56 347 (9.50) | 15 274 (8.82) | 41 073 (9.78) |
| Married | 299 818 (50.53) | 64 031 (36.96) | 235 787 (56.13) |
| Formerly married | 204 874 (34.53) | 84 141 (48.57) | 120 733 (28.74) |
| Unknown | 32 294 (5.44) | 9807 (5.66) | 22 487 (5.35) |
| Stage at diagnosis, No. (%) | |||
| In situ | 102 990 (17.36) | 17 868 (10.31) | 85 122 (20.26) |
| Localized | 324 453 (54.68) | 83 108 (47.97) | 241 345 (57.45) |
| Reginal | 130 311 (21.96) | 47 884 (27.64) | 82 427 (19.62) |
| Distant | 24 209 (4.08) | 17 691 (10.21) | 6518 (1.55) |
| Unknown | 11 370 (1.92) | 6702 (3.87) | 4668 (1.11) |
| First-course treatment, No. (%) | |||
| Surgery | 591 360 (99.67) | 172 588 (99.62) | 418 772 (99.69) |
| Chemotherapy | 132 568 (22.34) | 33 361 (19.26) | 99 207 (23.62) |
| Radiation | 264 787 (44.63) | 55 842 (32.23) | 208 945 (49.74) |
| Comorbidity during follow-up, No. (%) | |||
| MI | 16 018 (2.70) | 10 235 (5.91) | 5783 (1.38) |
| Alzheimer disease | 42 211 (7.11) | 27 723 (16.00) | 14 488 (3.45) |
| COPD | 107 658 (18.14) | 54 269 (31.32) | 53 389 (12.71) |
| Diabetes | 161 392 (27.21) | 60 323 (34.82) | 101 069 (24.06) |
| Depression | 153 159 (25.81) | 59 194 (34.17) | 93 965 (22.37) |
| Behavioral factors during follow-up, No. (%) | |||
| Tobacco use | 35 156 (5.93) | 14 860 (8.58) | 20 296 (4.83) |
| Alcohol use disorder | 9384 (1.58) | 3967 (2.29) | 5417 (1.29) |
| Drug use disorder | 11 239 (1.89) | 4882 (2.82) | 6357 (1.51) |
Abbreviations: COPD = chronic obstructive pulmonary disease; MI = myocardial infarction; SEER = Surveillance, Epidemiology, and End Results.
Table 2 shows the descriptive statistics for average concentrations of annual fine particulate matter, warm-season ozone, and annual nitrogen dioxide as well as meteorological variables and neighborhood-level characteristics. During the study period, the annual average concentrations were 9.42 μg/m3 for fine particulate matter, 45.35 ppb for warm-season ozone, and 20.29 ppb for nitrogen dioxide, with weak to moderate correlations among them (Table S3). The distributions of these pollutants and other covariates were approximately normal.
Table 2.
Distributions of annual average air pollutant exposure levels and covariates at zip codes.
| Variable | Minimum | 5th percentile | 25th percentile | Median | Mean | 75th percentile | 95th percentile | Maximum |
|---|---|---|---|---|---|---|---|---|
| Fine particulate matter, µg/m3 | 0.01 | 4.56 | 7.54 | 9.18 | 9.42 | 11.13 | 14.79 | 29.53 |
| Warm-season ozone, ppb | 19.70 | 32.04 | 41.30 | 45.01 | 45.35 | 49.39 | 59.72 | 80.76 |
| Nitrogen dioxide, ppb | 0.01 | 6.85 | 12.62 | 19.58 | 20.29 | 26.70 | 36.73 | 121.06 |
| Summer temperature, °C | 9.94 | 17.45 | 21.16 | 23.08 | 23.05 | 24.93 | 28.03 | 35.38 |
| Winter temperature, °C | −19.14 | −4.14 | 1.27 | 7.77 | 6.57 | 11.89 | 15.09 | 23.71 |
| Population density, persons/km2 | 0.00 | 15.66 | 143.61 | 716.78 | 1247.33 | 1699.21 | 4155.28 | 58 371.95 |
| Percentage below the federal poverty level | 0.00 | 3.43 | 6.47 | 10.34 | 12.41 | 16.70 | 27.71 | 92.82 |
| Median household income, $ | 670.23 | 30 990.97 | 44 201.19 | 58 118.30 | 63 217.19 | 76 750.85 | 112 315.00 | 330 555.15 |
| High school completion rate, % | 9.43 | 67.92 | 80.88 | 88.17 | 85.70 | 92.57 | 96.30 | 100 |
Abbreviation: ppb = parts per billion.
Table 3 shows the hazard ratios for mortality and the estimated excess deaths associated with a unit increase in each pollutant. In the 3-pollutant analysis, per 1-µg/m3 increase in annual fine particulate matter, a 1-ppb increase in warm-season ozone, and a 1-ppb increase in annual nitrogen dioxide were associated with hazard ratios for mortality of 1.0048 (95% CI = 1.0026 to 1.0070), 1.0021 (95% CI = 1.0013 to 1.0029), and 1.0022 (95% CI = 1.0014 to 1.0030), respectively. This finding translated to approximately 49 (95% CI = 26 to 71), 21 (95% CI = 13 to 30), and 22 (95% CI = 14 to 31) excess deaths per year within the cohort attributable to fine particulate matter, ozone, and nitrogen dioxide, respectively. The effect sizes became larger in the single-pollutant analysis. In the low-level analysis restricted to patients consistently exposed to low air pollutant levels throughout the follow-up period, the effects became substantially larger, with hazard ratios of 1.2574 (95% CI = 1.1889 to 1.3299) for fine particulate matter, 1.1300 (95% CI = 1.0691 to 1.1944) for ozone, and 1.0695 (95% CI = 0.9059 to 1.2625) for nitrogen dioxide. The results remained consistent when excluding first-course cancer treatments of chemotherapy and radiation as well as behavioral factors (Table S4) and additionally adjusting for calendar year and breast cancer subtype (Table S5).
Table 3.
Hazard ratios (95% CIs) for mortality and excess deaths (95% CI) associated with 1-µg/m3 increase in annual fine particulate matter, 1-ppb increase in warm-season ozone, or 1-ppb increase in annual nitrogen dioxide.
| Model | Exposure | Hazard ratio (95% CI) | Excess deaths, No. (95% CI) |
|---|---|---|---|
| 3-pollutant analysis | Annual fine particulate matter | 1.0048 (1.0026 to 1.0070) | 49 (26-71) |
| Warm-season ozone | 1.0021 (1.0013 to 1.0029) | 21 (13-30) | |
| Annual nitrogen dioxide | 1.0022 (1.0014 to 1.0030) | 22 (14-31) | |
| Single-pollutant analysis | Annual fine particulate matter | 1.0090 (1.0070 to 1.0109) | 91 (71-111) |
| Warm-season ozone | 1.0035 (1.0027 to 1.0042) | 36 (28-43) | |
| Annual nitrogen dioxide | 1.0036 (1.0030 to 1.0043) | 37 (31-44) | |
| Low-level analysisa | Annual fine particulate matter | 1.2574 (1.1889 to 1.3299) | 58 (43-75) |
| Warm-season ozone | 1.1300 (1.0691 to 1.1944) | 7 (4-11) | |
| Annual nitrogen dioxide | 1.0695 (0.9059 to 1.2625) | 2 (-3-7) |
Abbreviation: ppb = parts per billion.
Low-level analysis was conducted by restricting to patients consistently exposed to air pollutant levels below 5 µg/m3 for annual fine particulate matter, 30.56 ppb for warm-season ozone, or 5.31 ppb for annual nitrogen dioxide throughout the follow-up period.
Figure 2 showed the effect estimates for subgroups defined by individual-level and neighborhood-level characteristics. We found larger effects for fine particulate matter and nitrogen dioxide among patients diagnosed with breast cancer at younger ages (≤75 years), patients who received chemotherapy or radiation for treatment, and patients diagnosed at later stages. For warm-season ozone, we found reversed differences between subgroups, although the sizes of these differences were relatively smaller. The differences between subgroups defined by neighborhood characteristics were small or varied across pollutants.
Figure 2.
Hazard ratio (and 95% CI) for mortality associated with 1-µg/m3 increase in annual fine particulate matter, 1 part per billion (ppb) increase in warm-season ozone, or 1 ppb increase in annual nitrogen dioxide for each subgroup of individual-level and neighborhood-level characteristics. Subgroups of neighborhood-level characteristics were divided by upper or lower quartiles.
Discussion
We used a large, well-characterized breast cancer cohort to examine the associations between chronic exposures to fine particulate matter, ozone, and nitrogen dioxide and mortality after diagnosis in older patients enrolled in Medicare. We found that chronic exposures to the 3 pollutants were independently associated with increased risk of breast cancer mortality, adjusting for a wide range of cancer-related confounders. The low-level analysis results suggested that even in areas where air quality was considered satisfactory, small amounts of pollutants could have substantial health impacts on patients with breast cancer. We further identified that patients with breast cancer diagnosed at younger ages, individuals who received chemotherapy or radiation, and patients diagnosed at later stages were more susceptible to the effects of fine particulate matter and nitrogen dioxide.
Importantly, based on single-pollutant analyses, our mortality effect estimates for fine particulate matter (0.90%) and ozone (0.35%) among Medicare-enrolled patients with breast cancer were larger than those reported in the general Medicare population (0.81% for fine particulate matter and 0.23% for ozone).30 Results from single-pollutant analyses are more comparable because examining multiple pollutants simultaneously often reduces effect sizes because co-pollutants can partially explain each other’s effects.31 Nationwide survival analysis of nitrogen dioxide and mortality in the general Medicare population is lacking; therefore, we could not make a direct comparison. This greater susceptibility to air pollution in patients with breast cancer may be due to treatment-related side effects and comorbidities, such as immune deficiencies and hypertension, which make these individuals more sensitive to the external environment.9,32
Our findings of increased mortality associated with fine particulate matter and nitrogen dioxide exposure among patients with breast cancer are consistent with prior studies.6,7 Hu et al.33 reported that each 1-μg/m3 increase in fine particulate matter was associated with a 13% increase in mortality risk among 255 128 patients with breast cancer in California, based on SEER data between 1999 and 2009. More recently, Cheng et al.8 used clinical data for 3089 patients with breast cancer in California and found a 1.75% increase in mortality per 1-μg/m3 increase in fine particulate matter and a 0.75% increase per 1-ppb increase in nitrogen dioxide. To our knowledge, this is the first study to specifically evaluate the effect of ozone on mortality in patients with breast cancer. In the low-level analysis, in which we restricted analysis to patients consistently exposed to air pollution levels below the World Health Organization guidelines and well below current US standards, we observed substantially larger effects on mortality among older patients with breast cancer. We compared individual-level and neighborhood-level characteristics between the full cohort and the cohorts for the low-level analysis and found them to be generally similar, suggesting that the larger effects at low exposure levels may not be explained by those factors. Instead, it suggests a nonlinear exposure-response relationship and that even small amounts of pollutants could have substantial health impacts. Together with previous air pollution studies on mortality and other outcomes, which report larger effects at low exposure levels,30,34 our findings support that further reductions in air pollution could obtain more substantial health benefits.
Air pollution may contribute to mortality in patients with breast cancer directly by promoting tumor progression and indirectly by exacerbating comorbidities. Directly, fine particulate matter contains mutagenic and estrogen-disrupting components such as carbonaceous compounds and heavy metals,35,36 and radioactivity from radon attached to fine particulate matter may contribute to gene variations and oxidative stress in breast tissue.37 As a traffic-related pollutant, nitrogen dioxide serves as a marker for broader traffic emissions, such as diesel exhaust, which has been identified as a carcinogen that can influence estrogenic activity.38 Ozone’s role is less clear, but recent evidence has suggested its effects on hormone levels.4 Indirectly, air pollution is a well-established risk factor that exacerbates numerous diseases, including hypertension, COPD, MI, diabetes, liver disease, kidney disease, heart failure, and dementia,10 many of which are common comorbidities in patients with breast cancer and directly related to mortality.9 Given the limited existing evidence, it will be insightful for further research to identify the most affected pathway, whether directly through breast cancer progression or indirectly through comorbidities, to inform more targeted management strategies.
The identified susceptible subgroups provided foundations for developing targeted interventions. Consistent with other air pollution studies,34,39 we found that patients 75 years of age or older at diagnosis were less susceptible to the effects of air pollution, possibly because more vulnerable patients died earlier and thus were excluded from the older group.40 Another possible reason is that worse prognosis in older patients may reduce the relative importance of air pollution for mortality compared with other factors, such as tumor histology and comorbidities.41,42 We also found increased susceptibility of patients with late-stage breast cancer, probably because they were more sensitive to environmental stressors.43 Similarly, patients receiving chemotherapy or radiation (commonly used for late-stage disease) showed higher sensitivity, likely due to treatment side effects that enhance environmental susceptibility.32 Further, we found that male patients were more susceptible to fine particulate matter and ozone. Considering the biological differences in breast cancer between men and women,44 this finding suggests that air pollution may pose a greater risk to male patients with breast cancer. The underlying mechanisms remain unclear, however, and further research is warranted.
Our study’s strengths include accounting for a wide range of cancer-related confounders, long-term follow-up with a large cohort of patients with breast cancer, and identification of susceptible subgroups based on both individual-level and neighborhood-level characteristics. The study also has limitations. First, restricted by available data sources, we could not fully capture mortality-related factors in patients with breast cancer, such as parity, hormone use, and individual-level socioeconomic status. Second, the cohort mainly consisted of older women, who are most likely postmenopausal, which limited the generalizability of our findings. Breast cancer diagnosed in premenopausal women is often more aggressive and associated with higher mortality rates.41 Finally, due to limited sensitivity, the first-course treatments of chemotherapy and radiation as well as behavioral factors were subject to nondifferential misclassification. Because sensitivity analysis (Table S4) that excluded these confounders provided results largely consistent with those from the main analysis with “partial control,” however, the true results with “perfect control” would be expected to remain consistent.45
The collective findings of this study identified air pollution as a risk factor for mortality for older patients with breast cancer. The effect was larger in areas where air quality was considered satisfactory. Younger patients, individuals who received extensive treatments, and patients diagnosed at later stages were more susceptible. Unlike individual-level interventions, such as cancer treatments, identifying environmental factors offers a unique opportunity for scalable interventions that could have a broader impact. General strategies, such as the use of filtered masks and air purifiers, should be implemented to protect older patients with breast cancer from the health effects of air pollution and improve survival. In addition, established strategies for reducing traffic and industrial emissions should continue to further improve air quality.
Supplementary Material
Acknowledgments
The funding agency, the National Institutes of Health, had no role in the study design, data analysis, interpretation of results, or writing of the manuscript. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
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, Inc; and the SEER Program tumor registries in the creation of the SEER-Medicare database. 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; the Centers for Disease Control and Prevention’s National Program of Cancer Registries, under cooperative agreement 1NU58DP007156; and the National Cancer Institute’s SEER Program under contract No. HHSN261201800032I awarded to the University of California San Francisco, contract No. HHSN261201800015I, awarded to the University of Southern California, and contract No. HHSN261201800009I awarded to the Public Health Institute. The ideas and opinions expressed herein are those of the authors 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.
Contributor Information
Yaguang Wei, Department of Environmental Medicine, Icahn School of Medicine at Mount Sinai, New York, NY, United States.
Edgar Castro, Department of Environmental Health, Harvard T.H. Chan School of Public Health, Boston, MA, United States.
Kanhua Yin, Department of Surgery, University of Missouri-Kansas City School of Medicine, Kansas City, MO, United States.
Min Zhang, Department of Environmental Medicine, Icahn School of Medicine at Mount Sinai, New York, NY, United States.
Hannah Thompson, Department of Environmental Medicine, Icahn School of Medicine at Mount Sinai, New York, NY, United States.
Brent A Coull, Department of Environmental Health, Harvard T.H. Chan School of Public Health, Boston, MA, United States; Department of Biostatistics, Harvard T.H. Chan School of Public Health, Boston, MA, United States.
Joseph A Sparano, Department of Medicine, Division of Hematology and Medical Oncology, Tisch Cancer Institute, Icahn School of Medicine at Mount Sinai, New York, NY, United States.
Susan L Teitelbaum, Department of Environmental Medicine, Icahn School of Medicine at Mount Sinai, New York, NY, United States.
Robert O Wright, Department of Environmental Medicine, Icahn School of Medicine at Mount Sinai, New York, NY, United States.
Joel D Schwartz, Department of Environmental Health, Harvard T.H. Chan School of Public Health, Boston, MA, United States; Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA, United States.
Author contributions
Yaguang Wei, PhD (Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Validation, Visualization, Writing—original draft, Writing—review & editing), Edgar Castro, PhD (Data curation, Software, Writing—review & editing), Kanhua Yin, MD, MPH (Conceptualization, Writing—review & editing), Min Zhang, PhD (Data curation, Writing—review & editing), Hannah Thompson, MD (Writing—review & editing), Brent A. Coull, PhD (Writing—review & editing), Joseph A. Sparano, MD (Writing—review & editing), Susan L. Teitelbaum, PhD (Writing—review & editing), Robert O. Wright, MD (Writing—review & editing), and Joel D. Schwartz, PhD (Conceptualization, Funding acquisition, Writing—review & editing)
Supplementary material
Supplementary material is available at JNCI: Journal of the National Cancer Institute online.
Funding
This publication was funded by the National Institutes of Health grants UL1TR004419, P30ES023515, and P30ES000002. The funding agency, National Institutes of Health, had no role in the study design, data analysis, interpretation of results, or writing of the manuscript. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Conflicts of interest
None declared.
Data availability
The SEER-Medicare data are available upon request to the National Cancer Institute. The grid-level and zip code–level fine particulate matter, ozone, and nitrogen dioxide are publicly available from the National Aeronautics and Space Administration Socioeconomic Data and Applications Center.15-18 Other covariate data are publicly available, with sources described in the manuscript.
References
- 1. Giaquinto AN, Sung H, Miller KD, et al. Breast cancer statistics, 2022. CA Cancer J Clin. 2022;72:524-541. 10.3322/caac.21754. [DOI] [PubMed] [Google Scholar]
- 2. Miller KD, Nogueira L, Devasia T, et al. Cancer treatment and survivorship statistics, 2022. CA Cancer J Clin. 2022;72:409-436. 10.3322/caac.21731. [DOI] [PubMed] [Google Scholar]
- 3. Escala-Garcia M, Morra A, Canisius S, et al. Breast cancer risk factors and their effects on survival: a Mendelian randomisation study. BMC Medicine. 2020;18:327. 10.1186/s12916-020-01797-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Wu AH, Wu J, Tseng C, et al. Air pollution and breast cancer incidence in the multiethnic cohort study. J Clin Oncol. 2025;43:273-284. 10.1200/JCO.24.00418 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. White AJ, Fisher JA, Sweeney MR, et al. Ambient fine particulate matter and breast cancer incidence in a large prospective US cohort. JNCI J Natl Cancer Inst. 2024;116:53-60. 10.1093/jnci/djad170 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Tagliabue G, Borgini A, Tittarelli A, et al. Atmospheric fine particulate matter and breast cancer mortality: a population-based cohort study. BMJ Open. 2016;6:e012580. 10.1136/bmjopen-2016-012580 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Hung L-J, Tsai S-S, Chen P-S, et al. Traffic air pollution and risk of death from breast cancer in Taiwan: fine particulate matter (PM2.5) as a proxy marker. Aerosol Air Qual Res. 2012;12:275-282. 10.4209/aaqr.2011.09.0155 [DOI] [PubMed] [Google Scholar]
- 8. Cheng I, Yang J, Tseng C, et al. Outdoor ambient air pollution and breast cancer survival among California participants of the Multiethnic Cohort Study. Environ Int. 2022;161:107088. 10.1016/j.envint.2022.107088. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Zullig LL, Sung AD, Khouri MG, et al. Cardiometabolic comorbidities in cancer survivors. JACC CardioOncol. 2022;4:149-165. 10.1016/j.jaccao.2022.03.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Manisalidis I, Stavropoulou E, Stavropoulos A, Bezirtzoglou E. Environmental and health impacts of air pollution: a review. Front Public Health. 2020;8:14. 10.3389/fpubh.2020.00014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Zhang Z, Yan W, Chen Q, Zhou N, Xu Y. The relationship between exposure to particulate matter and breast cancer incidence and mortality: a meta-analysis. Medicine (Baltimore) 2019;98:e18349. https://journals.lww.com/md-journal/fulltext/2019/12130/the_relationship_between_exposure_to_particulate.75.aspx [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. White AJ. Invited perspective: air pollution and breast cancer risk: current state of the evidence and next steps. Environ Health Perspect. 2021;129:51302. 10.1289/ehp9466 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. White AJ, Bradshaw PT, Hamra GB. Air pollution and breast cancer: a review. Curr Epidemiol Rep. 2018;5:92-100. 10.1007/s40471-018-0143-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Ma T, Yazdi MD, Schwartz J, et al. Long-term air pollution exposure and incident stroke in American older adults: a national cohort study. Global Epidemiology. 2022;4:100073. 10.1016/j.gloepi.2022.100073. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Di Q, Wei Y, Shtein A, et al. Daily and Annual NO2 Concentrations for the Contiguous United States, 1-km Grids, Version 1.10 (2000-2016). NASA Socioeconomic Data and Applications Center (SEDAC; ); 2024. [Google Scholar]
- 16. Di Q, Wei Y, Shtein A, et al. Daily and Annual PM2.5 Concentrations for the Contiguous United States, 1-km Grids, Version 1.10 (2000-2016). NASA Socioeconomic Data and Applications Center (SEDAC; ); 2024 [Google Scholar]
- 17. Requia WJ, Wei Y, Shtein A, et al. Daily 8-Hour Maximum and Annual O3 Concentrations for the Contiguous United States, 1-km Grids, Version 1.10 (2000-2016). NASA Socioeconomic Data and Applications Center (SEDAC; ); 2024. [Google Scholar]
- 18. Wei Y, Xing X, Shtein A, et al. Daily and Annual PM2.5, O3, and NO2 Concentrations at ZIP Codes for the Contiguous U.S., 2000-2016, v1.0. NASA Socioeconomic Data and Applications Center (SEDAC; ); 2022. [Google Scholar]
- 19. Wei Y, Yazdi MD, Di Q, et al. Emulating causal dose-response relations between air pollutants and mortality in the Medicare population. Environ Health. 2021;20:53. 10.1186/s12940-021-00742-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Gupta A, Roy AM, Gupta K, et al. Impact of environmental temperature on the survival outcomes of breast cancer: a SEER-based study. Breast Cancer Res Treat. 2024;207:383-392. 10.1007/s10549-024-07369-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Hiatt RA, Brody JG. Environmental determinants of breast cancer. Annu Rev Public Health. 2018;39:113-133. 10.1146/annurev-publhealth-040617-014101 [DOI] [PubMed] [Google Scholar]
- 22. Barber LE, Maliniak ML, Moubadder L, et al. Neighborhood deprivation and breast cancer mortality among Black and White women. JAMA Network Open. 2024;7:e2416499. 10.1001/jamanetworkopen.2024.16499 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Thornton M, Shrestha R, Wei Y, Thornton P, Kao S, Wilson B. Daymet: Daily Surface Weather Data on a 1-Km Grid for North America, Version 4. ORNL DAAC; 2020. [Google Scholar]
- 24. US Census Bureau. Census 2000 Summary File 3 Technical Documentation. US Census Bureau; 2002. [Google Scholar]
- 25. US Census Bureau. 2010 census summary file 1—United States. 2011. Accessed August 27, 2025. https://www.census.gov/data/datasets/2010/dec/summary-file-1.html
- 26. US Census Bureau. American community survey 1-year data (2011. ‐2018). Accessed August 27, 2025. https://www.census.gov/data/developers/data-sets/acs-5year/2010.html
- 27. Center for International Earth Science Information Network CCU. Gridded Population of the World, Version 4 (GPWv4): Population Density, Revision 11. NASA Socioeconomic Data and Applications Center (SEDAC; ); 2018. [Google Scholar]
- 28. Burki T. WHO introduces ambitious new air quality guidelines. Lancet 2021;398:1117. [DOI] [PubMed] [Google Scholar]
- 29. USEP Agency. National ambient air quality standards table. Accessed August 27, 2025. https://www.epa.gov/criteria-air-pollutants/naaqs-table
- 30. Di Q, Wang Y, Zanobetti A, et al. Air pollution and mortality in the Medicare population. N Engl J Med. 2017;376:2513-2522. 10.1056/NEJMoa1702747 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Johns DO, Stanek LW, Walker K, et al. Practical advancement of multipollutant scientific and risk assessment approaches for ambient air pollution. Environmental Health Perspectives. 2012;120:1238-1242. 10.1289/ehp.1204939 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Tommasi C, Balsano R, Corianò M, et al. Long-term effects of breast cancer therapy and care: calm after the storm? J Clin Med, 2022;11:7239. 10.3390/jcm11237239 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Hu H, Dailey AB, Kan H, Xu X. The effect of atmospheric particulate matter on survival of breast cancer among US females. Breast Cancer Res Treat. 2013;139:217-226. 10.1007/s10549-013-2527-9 [DOI] [PubMed] [Google Scholar]
- 34. Wei Y, Qiu X, Sabath MB, et al. Air pollutants and asthma hospitalization in the Medicaid population. Am J Respir Crit Care Med. 2022;205:1075-1083. 10.1164/rccm.202107-1596OC [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Kresovich JK, Erdal S, Chen HY, Gann PH, Argos M, Rauscher GH. Metallic air pollutants and breast cancer heterogeneity. Environ Res. 2019;177:108639. 10.1016/j.envres.2019.108639. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. White AJ, Weinberg CR, O'Meara ES, Sandler DP, Sprague BL. Airborne metals and polycyclic aromatic hydrocarbons in relation to mammographic breast density. Breast Cancer Res. 2019;21:24. 10.1186/s13058-019-1110-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. White AJ, Gregoire AM, Fisher JA, et al. Exposure to particle radioactivity and breast cancer risk in the sister study: a U.S.-wide prospective cohort. Environ Health Perspect. 2022;130:47701. 10.1289/EHP10288 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Pedersen JE, Strandberg-Larsen K, Andersson M, Hansen J. Breast cancer among Danish women occupationally exposed to diesel exhaust and polycyclic aromatic hydrocarbons, 1964–2016. Scandinavian Journal of Work, Environment & Health. 2021;47:154-162. 10.5271/sjweh.3923 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Wei Y, Feng Y, Yazdi MD, 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 2024;384:e076939. 10.1136/bmj-2023-076939 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Brown DM, Picciotto S, Costello S, et al. The healthy worker survivor effect: target parameters and target populations. Curr Environ Health Rep. 2017;4:364-372. 10.1007/s40572-017-0156-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Jackson EB, Gondara L, Speers C, et al. Does age affect outcome with breast cancer? Breast 2023;70:25-31. 10.1016/j.breast.2023.06.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Yancik R, Wesley MN, Ries LAG, Havlik RJ, Edwards BK, Yates JW. Effect of age and comorbidity in postmenopausal breast cancer patients aged 55 years and older. JAMA. 2001;285:885-892. 10.1001/jama.285.7.885 [DOI] [PubMed] [Google Scholar]
- 43. Planchat E, Abrial C, Thivat E, et al. Late lines of treatment benefit survival in metastatic breast cancer in current practice? Breast. 2011;20:574-578. 10.1016/j.breast.2011.07.010 [DOI] [PubMed] [Google Scholar]
- 44. Silva SN, Gomes BC, André S, Félix A, Rodrigues AS, Rueff J. Male and female breast cancer: the two faces of the same genetic susceptibility coin. Breast Cancer Res Treat. 2021;188:295-305. 10.1007/s10549-021-06159-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Savitz DA, Barón AE. Estimating and correcting for confounder misclassification. Am J Epidemiol. 1989;129:1062-1071. 10.1093/oxfordjournals.aje.a115210 [DOI] [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
The SEER-Medicare data are available upon request to the National Cancer Institute. The grid-level and zip code–level fine particulate matter, ozone, and nitrogen dioxide are publicly available from the National Aeronautics and Space Administration Socioeconomic Data and Applications Center.15-18 Other covariate data are publicly available, with sources described in the manuscript.