Understanding Diabetes Using a Large Data Repository

In celebration of its 15-year anniversary, the Mass General Brigham Biobank — a bank of blood samples and data available to researchers studying the causes of common diseases — is running a series of articles highlighting innovative research using Biobank samples, data, and more.

The Mass General Brigham Biobank was established in 2009 with the goal to dramatically speed up the pace of research. The Biobank provides researchers with samples and data, including genomic data, that are linked to the electronic health record. More than 150,000 patients have provided their consent to participate in the Biobank. It continues to grow to serve the research needs of our community. 

To date, the Biobank has distributed 250,000 samples and genomic data for 65,000 participants to nearly 600 unique studies. Researchers have published more than 400 peer-reviewed articles on studies that use Biobank samples and/or genomic data. Their research covers a wide array of disease areas, including cancer, diabetes, auto-immune diseases, psychiatric disorders, and more.

This article in the series focuses on diabetes research done with Biobank samples and data.

Understanding how genetic, environmental, and socioeconomic factors cause the disease

Miriam Udler, MD, PhD, director of the Diabetes Genetics Clinic and an investigator in the Center for Genomic Medicine, both at Massachusetts General Hospital, frequently uses the Mass General Brigham Biobank for her research on diabetes. Diabetes is a disease characterized by high blood sugar levels. With the help of clinical and genetic data from the Biobank, she and her team have been able to study how genetic, environmental, and socioeconomic factors cause the disease. As a practicing clinician, she often forms hypotheses in the data from her patients and tests them using Biobank data.

Type 2 diabetes is the most common form of diabetes in adults. It is a complex disease that is linked to a variety of factors, including genetic, environmental, and socioeconomic factors. In one study led by Sara Cromer, MD, in the Udler Lab, people in the Biobank were classified according to their genetic risk of type 2 diabetes and obesity. Then, using address data from the electronic health record (EHR), they were able to show how neighborhood-level measures of socioeconomic factors play a role across genetic risk groups. Educational attainment levels in the areas where people lived impacted diabetes risk within genetic risk groups. Those with high genetic risk and high socioeconomic risk had the highest risk of type 2 diabetes and obesity. Importantly, people living in low socioeconomic risk areas had drastically reduced disease risk even if they had high genetic risk. Their work illustrates that providing care for people at risk for type 2 diabetes and obesity requires a whole-person approach, with attention not only to medical risk factors, but also to the environment in which an individual lives.

Dr. Udler’s team also developed approaches to identify genetic pathways that lead to type 2 diabetes using machine learning techniques applied to large genetic datasets. For each disease pathway, they developed a genetic risk score that capture a person’s genetic chance of developing type 2 diabetes via that specific pathway. They applied these genetic risk scores to large datasets like the Biobank and demonstrated these genetic pathways can help explain why people with diabetes vary in their clinical presentations. For example, work led by Kirk Smith, MS, and Aaron Deutsch, MD, both part of the Udler lab, has helped explain why people of East Asian genetic ancestry develop diabetes at a lower body mass index (BMI) compared to individuals of European genetic ancestry. Dr. Udler collaborates with other researchers to apply this technique across a wide variety of diseases and conditions including, coronary artery disease, obesity, fatty liver disease, and Alzheimer’s. “The genetic mechanisms contributing to these conditions provide exciting opportunities to improve our understanding of diseases and may also lead to new therapies,” she explains.

Studying atypical diabetes

Dr. Udler’s team is also studying unusual forms of diabetes in the Biobank. There are many different forms of diabetes like gestational diabetes or type 1, but some cases do not fit with known forms of diabetes. Identifying patients with atypical diabetes is often a complex process requiring manual medical chart review. Using the EHR data of Biobank participants, Dr. Cromer created a series of algorithms using clinical data to classify an individual’s diabetes subtype as atypical. One immediate goal of this effort is to enhance enrollment in a large study funded by the National Institutes of Health aimed at improving understanding of atypical diabetes called the Rare and Atypical Diabetes study (RADIANT). Dr. Udler is one of the RADIANT study’s principal investigators, and Dr. Cromer leads the study recruitment effort at Mass General, one of many RADIANT clinical study sites located across the country.

As part of the study, participants receive results of clinical-grade whole genome sequencing, which is performed to try to identify the cause of their diabetes. By applying the algorithms that Dr. Cromer developed in the Biobank, the research team has found they can identify a greater number of people with atypical diabetes, and that these individuals are not as represented in biomedical research than those recommended for the study by other sources, such as a doctor’s recommendation. Therefore, their research suggests that using EHR algorithms to support recruitment in research studies could be an effective way to improve participant inclusion, especially for populations that have historically been underrepresented in biomedical research.