Heterogeneity of Diabetes: Integrated Muli-Omics to Identify Physiologic Subphenotypes and Evaluate Targeted Prevention - ABSTRACT T2D and prediabetes are defined by measures of glucose elevation, but the underlying physiology is complex and differs between individuals: this heterogeneity is likely to drive differences in the course of diabetes, including development of comorbidities and response to treatment. Thus, while recent advances have led to greater individualization of diabetes treatment, current approaches, based on clinical traits such as cardiovascular (CVD) or kidney disease or desire for weight loss, do not truly represent precision medicine, which entails identifying the most appropriate therapy for a given patient based on biology. Furthermore, while there is a general consensus that subtyping T2D would be beneficial and enable targeted treatment approaches, current clustering methods have included only a few clinical and laboratory biomarkers that lack stability or presence of complications identified in the medical record. We have shown that individuals with prediabetes and early T2D can be subclassified according to defects in one of four major physiologic processes that are present in relatively equal distribution, including insulin resistance (IR), β-cell function, incretin effect, and hepatic IR, as quantified by gold-standard physiologic tests. The study goal is thus to advance precision diabetes medicine via the following AIMS: 1) define subphenotypes according to underlying metabolic physiology in the prediabetic/early diabetic state, 2) evaluate heterogeneity of responses to three established interventions according to metabolic subphenotype, 3) use integrated multi-omics to identify a molecular signature for each physiologic subtype and assess both baseline predictors of response and pathways that change differentially with treatment according to metabolic phenotype. To accomplish these aims, we plan to enroll 200 individuals with HbA1c 5.7 to 7.0% and perform gold-standard metabolic tests at baseline for muscle IR, hepatic IR, β-cell defect, and incretin defect. Relative deviance from cohort means will be used to classify individuals according to their dominant or co-dominant metabolic phenotype. To determine whether clinical response to treatment varies by metabolic subphenotype, individuals will then undergo three sequential treatments of 4-mos duration (andom order with 3-mo washouts): Mediterranean diet, metformin, and liraglutide. The primary endpoint is Δ HbA1c. Secondary endpoints include CGM glycemic measures, CVD risk markers, and regional fat distribution. With completion of 150 participants (37 per subphenotype), the study has 80% power, α=0.05, to detect a difference in Δ HbA1c of 0.17% in at least one treatment arm for a given subphenotype, or in at least one subphenotype for a given treatment. OMICS: we will apply our novel multiomics approach with soft hierarchical clustering that enables us to sample thousands of biomolecules to define a signature for each metabolic subphenotype. We will then identify a targeted panel of omics/clinical data that correspond to gold-standard defined subphenotypes and develop a CLIA- certified test for clinical use. Lastly, we will use omics to determine whether baseline signatures predict treatment response and examine longitudinal changes to highlight differences in biological treatment responses according to metabolic phenotype. The proposed study, leveraging our unique expertise, skills, and resources, will enable us to advance the goal of subclassifying T2D according to underlying physiology, and determine if a precision approach is effective.
