A Geographically Sensitive Neighborhood Exposome-Wide Association Study for Breast Cancer Survival

Growing literature recognizes the importance of contextual influences on outcomes across the cancer continuum, beyond those of individual biological and lifestyle factors. These contextual influences represent a neighborhood exposome consisting of socioeconomic, demographic, and built environmental factors.

To identify which neighborhood-level factors are most relevant for cancer outcomes and determine the geographic scale of their operation.

This cohort study of women with a primary diagnosis of breast cancer developed a geographically sensitive neighborhood exposome-wide association study (gs-NWAS) framework to determine specific neighborhood-level factors that may be associated with outcomes beyond individual-level factors. The study was performed at the University of Virginia Comprehensive Cancer Center from January 2014 to December 2024.

An analysis framework that assigned many neighborhood exposome components at 3 geographic scales: census tract and 1- and 5-km area-weighted buffer. Individual-level confounders (race and ethnicity, alcohol use, tobacco use, and body mass index) were also considered.

Survival among women diagnosed with breast cancer. The gs-NWAS chose each component's most relevant scale of operation, determined factors adjusted for multiple comparisons associated with survival in adjusted Cox proportional hazards regression models, and used machine learning methods (elastic net regression) to validate and condense these components. Hazard ratios (HRs), 95% CIs, and P values were reported.

Participants included 2727 women diagnosed with breast cancer (median age, 66 [IQR, 57-75] years) followed up for a median of 1697 (IQR, 945-2536) days (330 Black [12%], 2227 White [82%], and 170 other or unknown race [6%]); known cancer stage at diagnosis included I in 1668 (61%), II in 741 (27%), and III in 301 (11%). Four neighborhood exposome factors were associated with shorter breast cancer survival: high housing cost among low-income households (HR, 1.02; 95% CI, 1.00-1.03; P = .02 [tract]), people who recently moved to their current residence (HR, 1.06; 95% CI, 1.02-1.10; P = .001 [5-km buffer]), people renting in crowded households (HR, 1.03; 95% CI, 1.02-1.05; P < .001 [tract]), and the number of public preschoolers (HR, 1.06; 95% CI, 1.02-1.10; P = .001 [1-km buffer]).

In this cohort study of women diagnosed with breast cancer, a gs-NWAS was developed and applied to determine key components of the neighborhood exposome and their geographic scales of operation, identifying neighborhood-level factors in housing, economic, and demographic domains associated with breast cancer survival. Future studies should validate these findings in different populations and tailor this framework to their respective populations.