The association between exposure to PM2.5 components from coal combustion and mortality in female breast cancer patients

PM2.5 components may promote the development of breast cancer and increase the risk of mortality. This study aims to investigate the associations between long-term exposure to PM2.5 components and multiple causes of mortality among women with breast cancer living in Inner Mongolia, China. We constructed an Inner Mongolia cohort of 33 952 breast cancer patients from 2012 to 2021 using data from the Inner Mongolia Regional Health Information Platform. We assessed each patient’s exposure to PM2.5 components using the Tracking Air Pollution in China database. Cox regression models were used to estimate adjusted hazard ratios and 95% confidence intervals (95% CIs). A total of 3295 deaths were identified. For each interquartile increase in concentration in the 5 years before diagnosis, the all-cause mortality increased significantly by 5% (HR: 1.05, 95%CI: 1.00–1.10) for black carbon and by 4% (HR: 1.04, 95%CI: 1.00–1.09) for sulfate (SO4 2−), and decreased by 7% (HR: 0.93, 95%CI: 0.88–0.98) for nitrate (NO3 −). An association between organic matter and an increased all-cause mortality was also observed. Similar results were found for associations with risk of death from breast cancer-specific causes, cardio-cerebrovascular disease (CCVD) causes, and respiratory causes. Stronger associations were observed in older age groups and in Han Chinese patients. Our results showed that long-term exposure to black carbon, organic matter, and SO4 2− were more responsible for the increased risk of death from all causes, breast cancer-specific causes, CCVD causes, and respiratory causes. This suggests that more effective measures to control coal combustion emissions in Inner Mongolia are urgently needed. The elderly and Han Chinese populations may be at high risk.


Introduction
Breast cancer is the most common malignancy in women worldwide, and its morbidity and mortality rank first among women's tumors [1].Mortality rates vary by country, with breast cancer survival rates significantly lower in low-and middle-income countries than in high-income countries [2,3].Breast cancer ranks the first in the number of new cancer cases (0.42 million/2.09million) and the fourth in the number of cancer deaths (0.12 million/1.18million) among Chinese women [4].Breast cancer imposes a significant health burden and economic cost on individuals and healthcare systems worldwide.Moreover, breast cancer patients often experience longer survival and exhibit better prognoses then other cancer patients.Yet, being a vulnerable group, they are more susceptible to the long-term effects of certain risk factors leading to death.Therefore, it is crucial to identify and study the risk factors and mechanisms associated with different causes of mortality in breast cancer patients, including all-cause mortality, breast cancer-specific mortality, and cardio-cerebrovascular disease (CCVD) mortality.Such research is vital for reducing the health and economic burden faced by breast cancer patients.
Although the breast cancer susceptibility protein (BRCA) genes have been shown to be associated with breast cancer morbidity and mortality, the variation in breast cancer morbidity and mortality in geographic distribution reminds us of the importance of environmental and other factors in breast cancer incidence and mortality [3,5,6].Previous studies have found that genetics, smoking, alcohol abuse, late childbearing, sedentary lifestyle, vitamin deficiency and low socioeconomic position are all risk factors for breast cancer incidence and mortality [3,7,8].The epidemiology of migration also suggests that environmental pollution is an important factor [9].
Fine particulate matter (particulate matter with an aerodynamic diameter of less than 2.5 µm, PM 2.5 ) is recognized as an environmental pollutant associated with a variety of non-communicable diseases, including meningitis, CCVD, respiratory disease, and cancer [10,11].In Inner Mongolia, the primary source of PM 2.5 is attributed to the combustion of biomass fuels, particularly coal [12].The burning of coal for various purposes, such as heating and cooking, releases fine particulate matter into the air, including PM 2.5 .The components of PM 2.5 are complex, including black carbon, ammonium (NH 4 + ), NO 3 − , organic matter and SO 4  2− , and different components have different physicochemical properties and toxicological characteristics.Previous studies have provided evidence linking long-term exposure to PM 2.5 and its components, such as black carbon and NO 3 − , to an increased risk of breast cancer [13][14][15].In addition, several cohort studies have found that exposure to PM 2.5 components, including black carbon, SO 4 2− , NO 3 − , NH 4 + , and so on, is associated with increased mortality from a variety of specific causes of death [16][17][18].Two cohort studies of older Americans have also found similar findings [19,20].Despite these findings, the specific impact of PM 2.5 components on different causes of mortality in the breast cancer population remain largely unexplored.Addressing this research gap is crucial for providing valuable insights into the role of PM 2.5 components in different causes of mortality in the breast cancer population.From the perspective of biological mechanisms, PM 2.5 components may enter the body through the respiratory tract and cause inflammation and oxidative stress, or affect the DNA methylation process in breast cancer patients, thus exacerbating breast cancer progression and even death [21][22][23][24].However, it is still unknown that which PM 2.5 components are responsible for the increased risk of different causes of mortality in breast cancer patients and which sources of pollution are most important to control.
In this study, we hypothesized that baseline exposure to higher concentration of PM 2.5 components were associated with an increased all-cause mortality, breast cancer-specific mortality, CCVD mortality, and respiratory mortality among female breast cancer patients.And we aimed to evaluate the associations between long-term (1, 3, and 5 years) exposure to PM 2.5 components, including black carbon, NH 4 + , NO 3 − , organic matter, and SO 4 2− , before breast cancer diagnosis and the different causes of mortality, and to describe the exposure-response relationship in a retrospective cohort population in the Inner Mongolia Autonomous Region, China.We also identified the susceptible population and examined the modified effect of age, ethnicity, and Charlson comorbidity index (CCI) on the associations.

Study population
The data for this study were obtained from three main sources: the medical insurance system, the chronic disease registration system, and the cause-ofdeath reporting system (CDRS) of Inner Mongolia Autonomous Region.These sources are all part of the Inner Mongolia Regional Health Information Platform, which covers a wide range of individuals including employed and retired workers in cities, as well as unemployed urban and rural residents, such as children, students, and older adults.
Within the platform, when a patient is diagnosed with breast cancer (ICD-10 code C50) by a healthcare professional in a hospital, their information is automatically recorded and managed in the system according to the guidelines [25].For our study, the health insurance system provided demographic information and diagnostic data for the participants.The chronic disease registration system offered a comprehensive chronic disease history for each participant based on hospital diagnoses.Additionally, information on death status and cause of death was obtained from the CDRS.
Our study included all newly diagnosed female (individuals who self-identify as woman) breast cancer patients in hospitals within Inner Mongolia from 2012 to 2021.By matching desensitized ID numbers, we were able to collect and integrate medical treatment records and other relevant information from the aforementioned data sources.This enabled us to establish a population-based cohort of breast cancer patients in Inner Mongolia Autonomous Region, China from 2012 to 2021.All Medicare data were anonymized to protect patient privacy and approved by the Institutional Review Board (IRB) of Ethics Committee of Inner Mongolia Autonomous Region Center for Disease Control and Prevention (IRB No. NMCDCIRB2021001).
Between January 2012 and March 2021, we initially identified a total of 35 508 breast cancer patients in Inner Mongolia.Subsequently, we excluded patients with missing or non-local address information (N = 1005), male patients (N = 481), and patients with missing important variable data (N = 70).As a result, our final cohort comprised 33 952 breast cancer patients with complete data (figure S1).

Main exposure
The primary exposures of this study were the concentrations of the PM 2.5 components 1, 3 and 5 years before the breast cancer diagnosis.We obtained daily PM 2.5 component exposure data from the Tracking Air Pollution in China (TAP) program [26], which provides estimates of PM 2.5 chemical component at high spatial resolution across mainland China (10 KM).The details about TAP have been previously described [27,28].We extracted the daily average concentration data of PM for Inner Mongolia, China from 1 January 2007 to 31 December 2021 from TAP.We assigned exposure concentration data to each patient in our cohort by utilizing the latitude and longitude coordinates associated with their first complete and accurate address.
This assignment followed a priority order, starting with the permanent address, followed by the hospital address where breast cancer was diagnosed, and finally, the address of the health insurance institution.By following this prioritization, we aimed to ensure the most accurate and relevant exposure for each patient.Subsequently, we calculated the 1, 3, and 5 year average exposure concentration of PM 2.5 components before the diagnosis of the breast cancer.

Covariates
We obtained demographic and health data from the medical insurance system, chronic disease registration system and CDRS of the Inner Mongolia Autonomous Region.Demographic data included age at the year of diagnosis and ethnicity.Health data included breast cancer stage and CCI [29].The CCI is a well-established and widely utilized scoring system for assessing comorbidities.It quantifies the burden of comorbid conditions by considering the number and severity of diseases a patient may have.The CCI is a valuable tool for predicting the risk of mortality associated with various diseases, providing insights into the overall health status and potential impact on patient outcomes.We additionally included area-level socioeconomic variables from the 2020 Inner Mongolia Autonomous Region Census Yearbook, including the community deprivation index (CDI).The CDI [30,31] is a census level composite indicator used to assess socioeconomic position.It is derived using principal component analysis from the following census tract-level variables: percentage of persons with less than a high school education (25 years and older), percentage of persons living below the poverty line, percentage of households with at least one person over the age of 60, percentage of persons in management, science and arts occupations, number of rooms per capita, percentage of households without a car, percentage of unemployed persons (16-64 years of age in the labor force), and percentage of households without a home.The CDI was standardized to have a mean of 0 and a standard deviation (SD) of 1 by dividing the index by the square of the eigenvalue [32], with higher numbers indicating greater deprivation (lower socioeconomic position).

Outcomes
We modeled the data to analyze four primary outcomes among breast cancer patients: all-cause mortality, breast cancer-specific mortality, CCVD mortality, and respiratory disease mortality.Underlying causes of death were reported by the hospital and classified according to the International Classification of Diseases-10 (ICD-10).In this study, all-cause mortality encompassed mortality from any causes, including breast cancer-specific mortality, respiratory mortality, CCVD mortality, and accidental mortality, etc; breast cancer-specific mortality was defined using C50; CCVD mortality was defined using I00 to I99, which primarily encompass conditions such as hypertension, ischemic heart disease and intracerebral hemorrhage.Similarly, respiratory disease mortality was defined using J00 to J99, mainly encompassing conditions such as chronic lower respiratory tract disease and respiratory failure.

Statistical analysis
Descriptive analyses were conducted for all variables.Categorical and continuous variables were presented as frequency (percentage) and mean (SD) respectively.Differences in the distribution of baseline characteristics and PM 2.5 components exposure between surviving cases and deaths cases were evaluated using the Wilcoxon rank-sum test for continuous variables, and the chi-squared test for categorical variables.We used Cox regression models to estimate the association between long-term exposure to PM 2.5 components and the risk of death from all causes, breast cancer-specific causes, CCVD causes, and respiratory disease causes.Our study specified two models to adjust for a number of variables used to adjust for confounding.In the basic model (Model 1), we adjusted for age (<50, 50-60, or 60 + years), ethnicity (Han versus minorities), and CDI (<0 versus 0+).In the fully adjusted model (Model 2), we further adjusted for CCI (0 versus 0+).We conducted the spline regression analysis between PM 2.5 components and the all-cause mortality using restricted cubic splines with knots at the 10th, 50th, and 90th percentiles of the distribution of 1, 3, and 5 year PM 2.5 component exposure estimates before breast cancer diagnosis, adjusting for the same covariates as in the fully adjusted models.By this way, allows us to detect of a potential point of exposure with significant increased health risk (threshold).We conducted subgroup analyses to examine the potential modified effect of age, ethnicity, and CCI on the associations between 5 year PM 2.5 component exposure and the risk of all-cause death by including terms for the interaction between exposure and potential modifiers.We assessed the effect correction using the likelihood ratio tests for each potential effect modifier, comparing models that included an interaction term between 5 year PM 2.5 component exposure and the effect modifier with models without an interaction term.
Sensitive analyses were conducted to test the robustness of our results by excluding cases that died within one year of diagnosis and by additionally adjusting for breast cancer stage.In addition, assuming a clean window of 3 years, we cannot definitively determine whether the patients included between 2012 and 2015 were newly diagnosed or had readmission cancer.To address this uncertainty, we conducted a sensitivity analysis to assess the robustness of our findings by restricting the dataset to patients included between 2015 and 2021.In order to assess the robustness of our results to potential unmeasured confounding, we performed a sensitivity analysis using the E-value [33].The E-value is calculated by HR using a specific formula that has been described in the past [33].We utilized the E-value calculator [34] to estimate the E-value and evaluate the robustness of our findings to unmeasured confounding.
To allow comparison across pollutants, all HRs with corresponding 95% confidence intervals (CIs) associated with an interquartile (IQR) increase in exposure to PM 2.5 components were calculated for each outcome.

Study population characteristics and exposure levels
A total of 33 952 patients with breast cancer were included in this study.Of the 3295 deaths, 1858 deaths were due to breast cancer-specific deaths, 241 deaths were due to CCVD, and 104 deaths were due to respiratory disease.
The baseline age was 52.82 years for all patients.Approximately 16.4% of the patients were ethnic minorities (table 1).
The 1 year, 3 year, and 5 year exposure concentrations of black carbon, NH 4 + , organic matter and SO 4 2− before breast cancer diagnosis for death cases were higher than those for surviving cases, and the difference was significant except for the 5 year exposure to NH 4 + (table 2).In terms of spatial distribution, the PM 2.5 exposure of breast cancer patients in the five years before diagnosis was low on both sides and high in the middle.Higher exposure concentrations were concentrated in the south and southeast (figure 1).

Main analysis
In the fully adjusted model (Model 2), for each IQR increase in the concentration of 5 year exposure to black carbon and SO 4 2− before diagnosis, the allcause mortality increased significantly by 5% (HR: 1.05, 95%CI: 1.00-1.10)and 4% (HR: 1.04, 95%CI: 1.00-1.09),respectively, and decreased significantly by 7% (HR: 0.93, 95%CI: 0.88-0.98)for NO 3 − .We also observed an association between black carbon, organic matter, and SO 4 2− and an increased risk of all-cause death, especially in the 3 years of exposure before the diagnosis of breast cancer (table 3).In the curve of the exposure-response relationship, we did not observe a threshold effect.The curves also showed that the risk of all-cause death increased with increasing exposure to the PM 2.5 components in the 5 years before breast cancer diagnosis (figure 2).The p values for the nonlinear spline components of black carbon, Abbreviations: PM2.5, particulate matter with an aerodynamic diameter ⩽2.5 µm; IQR, interquartile range.Note: Continuous variables are described as the mean ± standard deviation (SD).Differences in the distributions between surviving cases and deaths cases were tested using the Wilcoxon rank-sum test.IQR refers to the interquartile range of concentrations of PM2.5 components in the total sample.NH 4 + , NO 3 − , organic matter, SO 4 2− for all-cause mortality among breast cancer patients were all less than 0.001, indicating that the associations were nonlinear.Similar relationships were also found for exposure to PM 2.5 components 1 and 3 years before breast cancer diagnosis (figures S2 and S3).For each IQR increase in the concentration of 5 year black carbon and SO 4 2− exposure before diagnosis, the breast cancer-specific mortality was significantly increased by 8% (HR: 1.08, 95%CI: 1.02-1.15)and 7% (HR: 1.07, 95%CI: 1.01-1.14),respectively.Similar but higher increased risk can also be observed in CCVD outcomes.For each IQR increase in the concentration of 5 year exposure to all five PM 2.5 components before diagnosis, the respiratory mortality increased significantly by almost 200% (table 3).

Subgroup analysis
Stratified analysis showed that the association between 5 year exposure to PM 2.5 components before breast cancer diagnosis and the risk of all-cause death was higher among the elderly and Han Chinese.For  Note: All hazard ratios with corresponding 95% confidence intervals associated with one interquartile range (IQR) increase in exposure to PM2.5 components.
Model 1 was adjusted for age, ethnicity, and community deprivation index.Model 2 was further adjusted for Charlson comorbidity index.

Figure 2.
Exposure-response relationships between 5 year exposure and the all-cause mortality for each PM2.  each IQR increase in 5 year exposure to black carbon, the all-cause mortality increased by 9% (HR: 1.09, 95%CI: 1.02-1.16) in patients aged over 60 years, which was higher than the other two age groups.This study also observed a higher death risk for Han Chinese compared to ethnic minority populations, e.g. for black carbon, the all-cause mortality was increased by 7% (HR: 1.07, 95%CI: 1.02-1.13)for Han Chinese and reduced by 10% (HR: 0.90, 95%CI: 0.80-1.01)for ethnic minorities.Similar results were found in the stratified analyses of the associations between other PM 2.5 components and the risk of death from all causes, breast cancerspecific causes, CCVD causes, and respiratory causes.

Sensitive analysis
Sensitive analysis showed that the association between 5 year exposure to PM 2.5 components and all-cause mortality remained after excluding deaths within 1 year of breast cancer diagnosis.In addition, the associations between 5 year exposure to PM 2.5 components and all-cause mortality remained significant after additional adjustment for breast cancer stage.The results of restricting the dataset to patients enrolled between 2015 and 2021 suggest that our conclusions remain robust, assuming a clean window of three years to exclude readmission patients.The E-value suggest that our conclusions are overall robust to unmeasured confounding bias (table S4).

Discussion
In this retrospective cohort study, long-term exposure to PM 2.5 components including black carbon and SO 4 2− , NH 4 + , NO 3 − , and organic matter, was associated with increased all-cause mortality, breast cancer-specific mortality, CCVD mortality, and respiratory mortality among this cohort of individuals with breast cancer diagnoses.These associations were more pronounced in older age groups and Han Chinese patients.Although the association for allcause mortality was non-linear and no threshold effect was observed, there was an overall upward trend.Our findings highlight the importance of improving air quality and reducing PM 2.5 component concentrations to mitigate the disease burden of breast cancer, particularly in areas with high air pollution.
When comparing our findings with the metaanalysis by Karimi and Samadi regarding the general population [35], our study found an increased risk of respiratory death and CCVD death in the breast cancer population with long-term exposure to black carbon, organic matter and SO 4 2− .However, and the risk for all-cause death was similar.The higher risk of CCVD death among the breast cancer population may be attributed to the additional inclusion of cerebrovascular disease for CCVD in this study compared to the review.The significantly increased risk of respiratory death may be attributable to additional respiratory mortality caused by Corona Virus Disease 2019, which may need to be verified in future studies.In addition, this discrepancy could also potentially be explained by the breast cancer population's higher propensity for comorbidities and elevated inflammation and oxidative stress levels [36], making them more prone to respiratory and CCVD mortality.Research has demonstrated that air pollution can disrupt the body's immune function and influence the process of DNA methylation in patients with breast cancer.Such interference may hasten the progression of the disease and, in severe cases, even lead to increased mortality rates [23,24].
Our study found a positive association between long-term exposure to black carbon and SO 4 2− and increased risk of all-cause death, CCVD death and respiratory death, confirming the findings of most previous studies [19,35,[37][38][39].The underlying mechanisms behind these associations are still controversial.Hypotheses from previous studies include black carbon's involvement in reactive oxygen species mediated DNA damage [40], cell apoptosis, and upregulation of the BRCA1 gene [41].Additionally, black carbon has been linked to trigger inflammatory reactions, evident by elevated cellular and molecular markers of inflammation upon inhalation [42].The small particle size of certain black carbon particles from traffic source (<100 nm) may facilitate their penetration into the alveoli and bloodstream [43].Furthermore, SO 4 2− has the potential to create an acidic environment in the human microcirculation, which may lead to airway hyperresponsiveness, changes in lung function, and enhanced absorption of other elements and particulate matter [44,45].
We also observed an association between organic matter and an increased death risk.Organic matter, generated from biomass and fossil fuels combustion and the oxidation of organic gases [46], contains various compounds, including polycyclic aromatic hydrocarbons (PAHs) and polychlorinated biphenyls (PCBs) [47].PAHs are recognized carcinogens that can accumulate in the adipose tissue of the breast and form PAH-DNA adducts.This leads to alterations in DNA structure and impact DNA damage repair mechanisms [48][49][50].Furthermore, studies have demonstrated that PAHs and PCBs can disrupt both human immune system and endocrine system [51][52][53].This combined influence on immune system, endocrine system and DNA damage repair mechanisms could potentially reduce treatment efficacy in the breast cancer population, thereby elevating the risk of mortality.
In contrast to the majority of previous studies, some of the results of this study indicated that NO 3 − was negatively associated with all-cause mortality.A recent meta-analysis showed that NO 3 − was significantly associated with increased all-cause mortality and cardiovascular mortality [35].Similar results were found in the systematic review by Yang et al and in cohorts from the USA, China and Denmark [16,19,39,54].There are several potential reasons that could explain the different results observed in our study compared to previous research, including differences in study design, exposure assessment techniques, characteristics of air pollution mixtures, population susceptibility, and statistical models [39].Additionally, it is possible that other factors (NO, NO 2 ) [55] contributing to all-cause mortality and CCVD mortality decreased when NO 3 − increased.The subgroup analysis revealed interesting findings, showing that the associations between PM 2.5 components and death risk were more pronounced among the elderly population aged over 60 years and among Han Chinese individuals.The vulnerability of the elderly to the adverse effects of long-term exposure to PM 2.5 components can be attributed to factors such as reduced physical function, weakened immunity, and a higher prevalence of underlying diseases.There is limited research on the health effects of air pollutant exposure in China that explores ethnic differences.However, a multi-ethnic cohort study in China demonstrated a stronger association between ambient air pollution exposure and cardiometabolic abnormalities in ethnic minorities compared to Han Chinese individuals [56].Another study conducted in Sichuan Province found ethnically diverse associations between long-term PM 2.5 exposure and hypertension prevalence among the Han, Tibetan, and Yi ethnic groups [57].In our study, the ethnic minorities in Inner Mongolia, primarily Mongolians, exhibited lower mortality risk compared to the Han Chinese.This difference could be attributed to variations in living environment, pollutant exposure levels, work patterns, dietary habits, and genetic factors between the Han Chinese and the Inner Mongolian minorities, particularly those maintaining nomadic lifestyles in grassland and pastoral areas, far from urban centers.
This study has several advantages.First, we filled the gap by investigating the associations between long-term exposure to PM 2.5 components and the risk of various causes of mortality in a breast cancer cohort, and our findings underscore the importance of mitigating exposure to PM 2.5 components, a key factor contributing to the risk of death in this population.Second, our cohort included 16.4% ethnic minorities, which paid attention to the health problems of ethnic minorities in Inner Mongolia.We also found that the Han Chinese population is at higher risk of mortality than the ethnic minority population.
This study also has several limitations.First, the exposure assessment in this study has measurement bias.We assigned exposure concentration data to each patient in our cohort by utilizing the latitude and longitude coordinates associated with their first complete and accurate address.However, people do not spend all of their time at one address.In their daily lives, patients may spend part of their time in other cities, so the exposure assessment in this study has a measurement bias.Due to the lack of data on patient movement, we do not currently have a better solution to this problem in this study.Second, our study was unable to adjust for some behavioral risk factors, such as smoking, alcohol consumption, obesity, parity, and hormone therapy, which could potentially influence our findings.To evaluate the potential impact of these unmeasured confounding factors, we performed an E-value analysis.The results suggest that our study's conclusions are generally robust and resistive to potential confounding bias from these unmeasured factors.In addition to the limitations, it is important to note that our study did not include the distinction between pre-and postmenopausal breast cancer diagnoses in the analysis due to the lack of available data on the menopausal status in our study cohort.

Conclusions
Our study establishes a significant association between long-term exposure to PM 2.5 components and higher risk of all-cause death, breast cancerspecific mortality, respiratory mortality, and CCVD mortality in a breast cancer cohort.Implementing effective strategies, such as stringent regulation of coal combustion emissions, can mitigate the harmful impacts of long-term PM 2.5 exposure on diverse causes of mortality among breast cancer patients in Inner Mongolia, China.
A 2-sided p-value <0.05 was considered statistically significant.Stata version 16.0 for Mac (Stata Corp, College Station, TX, USA) and R Studio version 1.2.5042(The R Project for Statistical Computing, Vienna, Austria) were used for the statistical analyses.

Figure 1 .
Figure 1.The location of Inner Mongolia in China (a) and the spatial distribution of average concentrations in the 5 years before breast cancer diagnosis by tertile of each PM2.5 component; Black Carbon (b), NH4 + (c), NO3 −

3 )
Figure 2. Exposure-response relationships between 5 year exposure and the all-cause mortality for each PM2.5 component: Black Carbon (a), NH4 + (b), NO3 − (c), Organic matter (d), and SO4 2− (e).Note: The exposure-response curve was calculated using restricted cubic splines with knots at the 10th, 50th, and 90th percentiles of the distribution of 5 year PM2.5 component concentrations before breast cancer diagnosis.The reference exposure level was set at the 25th percentile of the distribution of 5 year black carbon (1.38 µg m −3 ), NH4 + (3.19 µg m −3 ), NO3 − (4.30 µg m −3 ), organic matter (5.93 µg m −3 ), and SO4 2− (4.71 µg m −3 ) concentrations.Hazard ratios were adjusted for age, ethnicity, community deprivation index and Charlson comorbidity index.The solid line indicates the estimated hazard ratio values, and the dashed lines indicate their 95% confidence intervals.The bars are histograms (dependent on the right y-axis) and indicate the distribution of the corresponding PM2.5 component concentration data.The red dot indicates a hazard ratio of 1 when the corresponding PM2.5 component concentration is the reference exposure level.Abbreviations: PM2.5, particulate matter with an aerodynamic diameter ⩽2.5 µm.

Table 1 .
Characteristics of breast cancer patients.Continuous variables are described as the mean ± standard deviation (SD), and categorical variables are expressed as counts and percentages.Differences in the distribution between surviving cases and deaths cases were tested using the Wilcox rank-sum test for continuous variables and the chi-squared test for categorical variables.

Table 2 .
Comparison of exposure concentrations of PM2.5 components between surviving cases and deaths cases.

Table 3 .
Association between long-term exposure to PM2.5 components and risk of death.

Table 4 .
Subgroup analysis of associations between 5 year exposure to PM2.5 components before diagnosis and risk of all-cause death.