Background: Previous studies have reported associations between breast cancer incidence and ambient air pollution, notably exposure to nitrogen dioxide (NO2) and particulate matter <2.5µm in diameter (PM2.5), with weaker evidence for particulate matter <10µm in diameter (PM10). Most studies, however, included predominantly post-menopausal women. We used data from the Generations Study, a prospective cohort of women across the UK, to evaluate these associations across a wide age range.
Objectives: To investigate how air pollutant exposure levels (NO2, PM2.5, PM10) at the household level are associated with breast cancer incidence, and how these associations vary by menopausal status and tumour characteristics.
Methods:  Participant records were spatially linked to annual average concentrations of NO2, PM2.5, and PM10 via their home addresses at baseline and during follow up, using Expanse air pollution surfaces from 2004-2019.  Cox proportional hazards models were fitted to estimate hazard ratios (HR) and confidence intervals (95% CI) for the mean exposure of each pollutant accounting for home address changes throughout follow-up. Interaction analyses investigated the association between air pollution and breast cancer by menopausal status as a time updated covariate, using likelihood ratio tests (LRT) for interaction. Differences in HR estimates by oestrogen receptor (ER) type were evaluated using the Lunn McNeil method and tested using Wald tests.
Results: Study population included 104,576 women (58% pre-menopausal) at study entry. Within the total follow up of 1,219,106 person-years, 3,643 women developed breast cancer. Models adjusted for age and year of entry showed a significant (p<0.05) 11% increase in breast cancer risk for PM2.5 [HR(95%CI)=1.11 (1.01, 1.22) per 5 µg/m³, p-value=0.024], and weaker or no evidence of risk associations for PM10 [1.06 (0.99, 1.12) per 5 µg/m³, p-value=0.089] and NO2 [1.03 (0.98, 1.08) per 10 µg/m³, p-value = 0.249]. Adjustment by potential confounders at the individual level (body mass index, alcohol consumption, smoking, education, reproductive history) and small area level (deprivation index) had minimal impact on HR estimates. Interaction analysis showed similar risk associations for pre-menopausal and post-menopausal women [LRT p-value for interaction= 0.345 for NO2, 0.817 for PM2.5, and 0.406 for PM10]. Additionally, subgroup analysis showed similar risk associations for ER positive and negative breast cancers [Wald p-value= 0.652 for NO2, 0.453 for PM2.5, and 0.352 for PM10].
Conclusion: Our analyses are consistent with prior evidence for a weak breast cancer risk association with exposure to air pollutants, particularly PM2.5, and indicate that risk associations are similar for pre- and post-menopausal women and by ER status. Further studies should evaluate risk associations for individual pollutants among particulate matter, as well as combinations of air pollutants.