UCSF

Abstract

Type 2 diabetes is one of the top-10 causes of morbidity and mortality in the US, and disproportionally affects racial/ethnic minorities. Obesity is a well-known cause of type 2 diabetes, and in the past three decades, the obesity epidemic has contributed significantly to the drastic increase in type 2 diabetes in the US. Despite this clear relationship, it is unclear if the association between obesity measures (e.g. body mass index [BMI]) and type 2 diabetes risk varies by race/ethnicity. Identifying if race/ethnic-specific BMI thresholds should be used for risk-stratification can have important clinical implications as these can be incorporated into screening guidelines. Further, the precise mechanisms that connect obesity and type 2 diabetes remain unclear. It is hypothesized that one of these mechanisms is via the development of non-alcoholic fatty liver disease (NAFLD), which exacerbates hepatic insulin resistance and promotes the onset of type 2 diabetes. Understanding the possible mediating role that NAFLD has on the obesity-type 2 diabetes relationship may be of great interest as NAFLD prevention or management could be a promising target to reduce the obesity-related burden on type 2 diabetes.

This dissertation applies advanced epidemiologic methods to answer three distinct questions utilizing data from the well-characterized Multi-Ethnic Study of Atherosclerosis cohort (2000-2011) of 6,814 White, African American, Hispanic and Chinese American adults ages 45-84 years of age. The first chapter evaluated the usefulness of the BMI as a race/ethnic-specific predictor of type 2 diabetes risk to predict the 10-year risk of type 2 diabetes according to race/ethnicity at different BMI points. We found that African American, Hispanic and Chinese American adults had similar type 2 diabetes risk at lower BMI values compared to White adults, suggesting that racial/ethnic minorities should be screened for type 2 diabetes at lower BMI cut-points. In the second chapter I applied a causal mediation analysis using marginal structural models to estimate the overall association between obesity on risk of type 2 diabetes, decomposing this into the portion of the relationship mediated, and not mediated, by the degree of liver fat accumulation (i.e. indirect and direct effects, respectively). We found that NAFLD accounted for approximately 30% of the association between obesity and type 2 diabetes risk, underscoring the importance of this mechanism as a possible target for prevention of type 2 diabetes. And the third chapter developed a practical scoring tool for predicting NAFLD using participant demographic, medical history, anthropometry and laboratory data. We found that our prediction tool was simple but highly predictive and can aid clinicians identify adults at high NAFLD risk.

Together, these projects highlight the usefulness of generalized and ectopic obesity measures that can be used to identify high-risk adults in whom screening and possible interventions can help reduce the risk of or delay the onset of type 2 diabetes.