Integrating epidemiologic, clinical, genomic and metabolomic profiles to predict pancreatic cancer risk in a multiethnic population

ABSTRACT Pancreatic cancer is a highly lethal malignancy that has a very poor prognosis in the United States. It has a 5- year survival rate of only 9% and is projected to become the second most common cancer death by 2030. Pancreatic cancer also has a disproportionate burden across race/ethnicity, with higher incidence rates observed

among minority groups, such African Americans, Japanese Americans, and Native Hawaiians. Past prediction models have been developed to identify high-risk individuals and improve the earlier detection of this disease. However, these models were designed in individuals of primarily European or Asian ancestry and have not been

validated in multiethnic populations. In addition, these models included mainly known epidemiologic risk factors and only a few incorporated data on genetic variants or health conditions. Thus, a model that employs more granular data, such as comorbidities/symptoms, genomics and metabolomics, for the prediction of pancreatic

cancer across multiple races/ethnicities does not exist. In this study, we seek to apply an integrative systems biology approach to enhance the prediction of pancreatic cancer risk using data from the Multiethnic Cohort (MEC) Study. The MEC is a long-standing prospective cohort of over 215,000 racially diverse individuals that

has comprehensive lifestyle, environmental, clinical, and genetic data. We will use data from existing resources of the MEC, including epidemiologic risk factors from questionnaires, clinical health conditions from Medicare claims, genetic data from a large biorepository of blood samples, and cancer incidence and mortality information

from SEER Cancer registries and state and national mortality databases. We will also generate new metabolomic

data for a subset of MEC participants. Our specific aims are: 1) to identify clusters or patterns of clinical conditions associated with pancreatic cancer risk; 2) to validate existing prediction models in a multiethnic population and develop an enhanced prediction model that incorporates epidemiologic, clinical and genomic data; 3) to identify

metabolites associated with pancreatic cancer in a multiethnic population; and 4) to integrate epidemiologic, clinical, genomic and metabolomic data to identify individuals at high risk of pancreatic cancer. Results from this study are expected to elucidate etiologic mechanisms and improve the prediction of pancreatic cancer risk for

heterogeneous populations. This will have significant implications for improving strategies for earlier detection and reducing the overwhelming burden of this fatal cancer.