Neighborhood Environment in Relation to Racial and Ethnic Colorectal Cancer Disparities in the Multiethnic Cohort Study: Consideration for Downstream Individual-Level Behavior and Microbial Factors

Abstract

Background: Racial and ethnic disparities in colorectal cancer (CRC) incidence and mortality persist in the U.S., even after accounting for differences in screening, access to care, and other socioeconomic factors. To better understand the multifactorial and multilevel risk factors for disparities, this dissertation leverages data from a large multiethnic cohort to study associations of neighborhood environment with gut microbiota and CRC risk and mortality and the mediating role of health behaviors. Discovering the mechanisms that underlie how the neighborhood environment contributes to racial and ethnic disparities in CRC can translate to targeted community-level and behavioral interventions aimed at addressing disparities. Methods: Data came from the Multiethnic Cohort (MEC) Study, a prospective cohort of about ~200,000 adults from five racial and ethnic groups in California and Hawaii which began recruitment in 1993 and followed individuals through 2019. Study variables included 1) five objectively-measured neighborhood exposures based on geocoded residential addresses linked to census, business, street, and traffic data; 2) health behaviors (diet quality, alcohol intake, and moderate to vigorous physical activity) measured in 5-year intervals via validated questionnaires; 3) Gut microbiota alpha and beta diversity and bacterial genera abundance assessed using 16S rRNA gene sequencing; and 4) CRC risk and CRC-specific mortality after diagnosis tracked via linkage to tumor registries and death certificates. In the first chapter, we used linear and zero-inflated Gaussian mixed models with correction for the false discovery rate (FDR) and confounder adjustment to examine associations of neighborhood environment with gut bacterial diversity and bacterial taxa abundance in a subset of MEC Study participants in 2013-2016 (n = 1,589). In the second chapter, we used confounder-adjusted stratified multivariable Cox regression models to estimate hazard ratios (HR) and 95% confidence intervals (CI) for CRC risk (n = 185,874) and mortality after CRC diagnosis (n = 5,812) overall and by race and ethnicity with cluster-robust standard errors. In the third chapter, we interrogated the mediating role of health behaviors by testing associations between neighborhood environment, health behaviors, and gut microbiota with confounder-adjusted mixed models (n = 1,548). Results: In Chapter 1, lower neighborhood socioeconomic status was associated with abundance of multiple bacterial genera, including higher abundance of Paraprevotella, Catenobacterium, Desulfovibrio, Lactobacillus, and Victivallis and lower abundance of Defluviitaleaceae UCG-011 and Clostridium methylpentosum group (FDR-adjusted p<0.05). In Chapter 2, CRC risk was 23% higher among Japanese Americans living in neighborhoods in the highest quartile of alcohol outlet density compared to Japanese Americans living in neighborhoods with no alcohol outlets (95% CI = 0.99,1.53; p-trend = 0.007). Among African American participants with CRC, residing in a neighborhood with lower street connectivity was associated with greater CRC-specific mortality (HRQuintile1vs.5 = 1.47; 95% CI = 1.04, 2.08; p-trend = 0.03). In Chapter 3, neighborhood environment was not associated with health behaviors, but diet quality was associated with gut microbiota. Conclusion: Neighborhood socioeconomic status and diet quality were related to gut microbial profiles, neighborhood characteristics of participants differed by race and ethnicity, and alcohol outlet density may pose different risks for CRC across race and ethnicity. We present compelling evidence for the role of neighborhood environment and health behaviors in shaping gut microbiota, but additional research is needed to clarify whether neighborhood-based changes to the gut microbiota can explain disparities in CRC risk and prognosis.