1304-P: MODY Calculator Overestimates MODY Probability in Multiethnic Cohort with Youth-Onset Type 2 Diabetes Phenotype

Introduction & Objective: Identifying monogenic diabetes (MODY) remains a challenge for clinicians. Shields and colleagues developed a widely used MODY probability calculator (https://www.diabetesgenes.org/) based on clinical measures to assist in this decision. Because the calculator was developed in a predominantly adult White European population without any pediatric T2D cases, we sought to test its accuracy in ProDiGY, a multiethnic cohort with clinician-diagnosed T2D under age 20.

Methods: MODY calculator probabilities were computed for 615 youth with clinician-diagnosed T2D (n=147 in SEARCH, n=468 in TODAY; overall, >60% non-White), including 20 participants with MODY (3%). Given the longitudinal design of these studies, MODY probability was calculated for each participant at multiple time points, allowing maximum and median probabilities.

Results: Of the 20 individuals with MODY, 19 (95%) had a >60% median probability of having MODY. However, this calculator overestimated the probability in participants with no MODY variant detected. In fact, 85% of individuals without MODY had a maximum probability >60% of having MODY, and 67% had a median probability >60%. Family history of diabetes did not discriminate diabetes type with more T2D patients having a parent affected (64% v 55% in MODY). In contrast, HbA1c and BMI each had discriminatory capacity (ROC AUC >0.67).

Conclusion: In a group of multi-ethnic youth with diabetes, the MODY calculator correctly identified those with MODY but overestimated the probability for those with a clinical phenotype of T2D, likely due to young age at diagnosis and high proportion with positive family history of diabetes.