Utility of Polygenic Scores for Differentiating Diabetes Diagnosis Among Patients With Atypical Phenotypes of Diabetes.

Context Misclassification of diabetes type occurs in people with atypical presentations of type 1 diabetes (T1D) or type 2 diabetes (T2D). Although current clinical guidelines suggest clinical variables and treatment response as ways to help differentiate diabetes type, they remain insufficient for people with atypical presentations. Objective This work aimed to assess the clinical utility of 2 polygenic scores (PGSs) in differentiating between T1D and T2D. Methods Patients diagnosed with diabetes in the UK Biobank were studied (N = 41 787), including 464 (1%) and 15 923 (38%) who met the criteria for classic T1D and T2D, respectively, and 25 400 (61%) atypical diabetes. The validity of 2 published PGSs for T1D (PGS_T1D) and T2D (PGS_T2D) in differentiating classic T1D or T2D was assessed using C statistic. The utility of genetic probability for T1D based on PGSs (GenProb-T1D) was evaluated in atypical diabetes patients. Results The joint performance of PGS_T1D and PGS_T2D for differentiating classic T1D or T2D was outstanding (C statistic = 0.91), significantly higher than that of PGS_T1D alone (0.88) and PGS_T2D alone (0.70), both P less than.001. Using an optimal cutoff of GenProb-T1D, 23% of patients with atypical diabetes had a higher probability of T1D and its validity was independently supported by clinical presentations that are characteristic of T1D. Conclusion PGS_T1D and PGS_T2D can be used to discriminate classic T1D and T2D and have potential clinical utility for differentiating these 2 types of diseases among patients with atypical diabetes. [ABSTRACT FROM AUTHOR]