LAUSR

Background: Recent reconsideration of the classification of diabetes identified novel unique diabetes phenotypes using data on traditional diabetes characteristics. Our objective was to assess whether unique diabetes phenotypes are present in a multi-ethnic U.S. sample and, if so, are differentially associated with incident complications. Methods: We included 1201 MESA participants who were identified as having diabetes at exams 1 (2000-02) or 2 (2002-04) via any of the following: self-report of previous diabetes diagnosis, diabetes medication use, fasting glucose ≥126 mg/dL, random glucose ≥200 mg/dL, and glycated hemoglobin (HbA1c) ≥6.5%. We used an iterative data-driven method (k-means clustering) to partition the sex-stratified sample into unique groups based on data for each of the following: age at diabetes diagnosis, body mass index, waist circumference, HbA1c, and years of insulin use. We estimated the association between diabetes subtype with risk for CVD, death, chronic kidney disease, and retinopathy from 2002 through 2015. Results: We identified four unique subtypes of diabetes related to aging (AR; 52% of sample), insulin use (IU; 3%), severe obesity (SO; 32%), and severe hyperglycemia (SH; 14%). Compared to the AR group, risk for death was higher for the SH and IU groups and risk for retinopathy was higher for SH (Table). Conclusions: Our work supports consideration of unique diabetes subtypes that have differential risk for complications. Disclosure M. Bancks: None. M.R. Carnethon: None. H. Chen: None. M. Cotch: None. B.E.K. Klein: None. R. Klein: None. M. Szklo: None. A. Bertoni: Consultant; Self; Merck & Co., Inc. Funding National Heart, Lung, and Blood Institute (HHSN268201500003I)