Diabetes and liver cancer risk: A stronger effect in Whites than Blacks?

BACKGROUND Both diabetes and liver cancer are overrepresented among African Americans, but limited information is available on the interrelationship of these two diseases among African Americans. We examined the association of diabetes with the incidence of liver cancer and whether this varied by participant self-reported race/ethnicity. METHODS Using the Southern Community Cohort Study, we conducted a cancer follow up (2002-2016) of a cohort of mostly low-income participants aged 40-79 with diabetes (n = 15,879) and without diabetes (n = 59,077) at study baseline. Cox regression was used to compute Hazard Ratios (HR) and 95% CIs for the risk of incident liver cancer. RESULTS With 790,132 person years of follow up, 320 incident cases of liver cancer were identified. In analyses controlling for age, sex, race, BMI, current and former smoking, total alcohol consumption, family history of liver cancer, any hepatitis infection, hyperlipidemia and socioeconomic factors, the association between diabetes and risk of liver cancer differed significantly (pinteraction = 0.0001) between participants identifying as Black/African American (AA) or White/European American (EA). Diabetes was associated with 5.3-fold increased cancer risk among EAs (HR 5.4, 95% CI 3.2-9.3) vs an 80% increase (HR 1.8, 95% CI 1.3-2.5) among AAs. Furthermore, controlling for diabetes greatly attenuated the higher risk of liver cancer among AAs; indeed, while the cancer risk among those without diabetes was twice as high among AAs than EAs (HR = 2.0, 95% CI = 1.4-2.9), no excess in AAs was observed among those with diabetes (HR = 0.7, 95% CI = 0.4-1.1). CONCLUSION While liver cancer risk in general is greater in AAs than EAs and diabetes increases this risk in both racial/ethnic groups, diabetes appears to impact liver cancer to a much greater extent among EAs. The findings raise the possibility of racially different mechanisms and impacts of diabetes on this often fatal cancer among AAs and EAs.