Risk factors for gastric cancers in the United States: Variation by anatomic site and race/ethnicity.

Abstract

189 Background: Gastric cancer (GC) is a high mortality cancer in the US. Differences in risk factors by anatomic location and race/ethnicity have been suggested but remain understudied in the US population. Methods: The Multiethnic Cohort (MEC) is a prospective cohort study that collected data on 5 racial/ethnic groups [Whites (W), Blacks (B), Latino (L), Japanese-American (JA), and Hawaiian (HA)] from Hawaii and Los Angeles in 1993-1996. Participants completed a detailed baseline survey and were followed for development of incident cancer. Cox regression models were used to calculate hazard ratios (HR) and 95% confidence intervals (CI) to identify GC risk factors by anatomic location (cardia, non-cardia) and by race/ethnicity. Results: Data from 192,626 participants was available. The cohort was 25% W, 17% B, 23% L, 28% JA, and 7% HA. During a median follow up of 20.3 years, 1,109 non-cardia and 201 cardia incident GCs were diagnosed. Older age (per year, non-cardia HR 1.08, 95% CI 1.07-1.92; cardia HR 1.06, 95% CI 1.05-1.09), male sex (non-cardia HR 1.6, 95% CI 1.4-1.8; cardia HR 3.0, 95% CI 2.1-4.4), and current (non-cardia HR 1.7, 95% CI 1.5-2.2, cardia HR 3.4, 95% CI 2.2-5.3) or former (non-cardia HR 1.3, 95% CI 1.1-1.5; cardia HR 2.0, 95% CI 1.3-2.9) smoking were associated with both cancer types. Notably, race/ethnicity (ref W: B HR 3.0, 95% CI 2.2-4.0; L HR 2.5, 95% CI 1.8-3.3; JA HR 3.9, 95% CI 3.0-5.1; HA HR 3.9, 95% CI 2.8-5.5), foreign-born (ref: self & parents US born: HR 1.3 95% CI 1.1-1.7), and family history of GC (OR 1.9, 95% CI 1.5-2.3) were associated with non-cardia GC. BMI ≥30 (HR 1.6, 95% CI 1.1-2.3), having ≥1 drink/week (HR 1.6, 95% CI 1.1-2.3), and being JA (ref W: HR 1.9, 95% CI 1.2-2.9) were associated with cardia GC. Risk factors other than age differed by race/ethnicity for non-cardia GC. Male sex was a risk factor for B, L and JA only. Having less than a high school education was a risk factor for B and JA only, smoking a risk factor for L and JA only, and having diabetes a risk factor for B only. Being in the highest sodium intake quartile was a risk factor among W and HA. A family history of GC was a risk factor for W, L, and JA. Having foreign-born parents was a risk factor for W and being foreign-born was a risk factor for JA. Conclusions: GC risk factors differ between subtypes and, for non-cardia, between race/ethnic groups. These differences provide an insight into the etiology of GC and the disproportionate incidence rates in high-risk groups, potentially aiding in the design of targeted intervention strategies.