Clinical care and other categories posters: Type 2 diabetes

Aim: Recent research suggests high rates of misclassification in adults with type 1 diabetes who are islet autoantibody negative. We aimed to assess the utility of a type 1 diabetes diagnostic model, to classify autoantibody negative adults clinically diagnosed with type 1 diabetes. Methods: We studied 152 adults from the DARE population cohort, diagnosed with diabetes from age 18 y, and insulin treated from diagnosis. We assessed the ability of a previously developed multivariable clinical diagnostic model, incorporating a type 1 diabetes genetic risk score (T1DGRS) and clinical features (age at diagnosis and BMI), to identify islet autoantibody negative adults with type 1 diabetes defined by random nonfasted cpeptide <200pmol l1 (>3y postdiagnosis). Results: 39% (60/152) were autoantibody negative with 45%(27/60) having type 1 diabetes as defined by cpeptide <200 pmol l1, median(IQR) age at diagnosis 34(26– 50)y, duration 18(11– 29)y, BMI 28 (25– 31). T1DGRS was discriminative of type 1 diabetes (ROCAUC 0.78, 95%CI 0.66, 0.90). The combined model (T1DGRS + clinical features) was highly discriminative of type 1 diabetes (ROCAUC 0.88, 95%CI 0.79, 0.97): Median model probability of type 1 diabetes for people with cpeptide <200 pmol l1 was 0.651 (IQR 0.361– 0.840) 0.028 (IQR 0.007– 0.108) in those with cpeptide concentration >200pmol l1. Conclusion: A type 1 diabetes diagnostic model including T1DGRS and clinical features is highly discriminative of autoantibody negative insulin deficiency defined type 1 diabetes. It is therefore a useful tool in classifying diabetes type in adults with a clinical diagnosis of type 1 diabetes with negative autoantibodies.