SUN-132 Investigating the Accuracy of Risk Factor Calculators in Diagnosing Maturity Onset Diabetes of the Young (MODY)

Abstract

MODY is a rare form of diabetes caused by impaired insulin production attributable to autosomal dominant monogenic mutations. It is an important diagnosis as it results in a treatment change for 25% of patients, allows for the screening of 1st degree relatives and identifies patients who have MODY subtypes associated with reduced rates of micro- and macro-vascular complications. Diagnosing MODY is challenging and several MODY risk factor calculators have been proposed to guide rational genetic testing. The aim of this study was to retrospectively apply the MODY risk factor calculators to patients with diagnosed MODY who attended the diabetes service at a tertiary hospital in the West of Ireland to identify the most accurate calculator in predicting MODY in this patient cohort. MODY risk factor calculators in the literature focus on young age at diagnosis, low BMI and low HbA1c as defining features of MODY. Ellard, Bellanne-Chantelot et al. (2008) proposed age at diagnosis ≤25, BMI <30 and HbA1c ≤58.8 mmol/L as the recommended criteria for genetic testing in MODY, and this captured 6/21 (28.57%) patients with MODY in our cohort. Thanabalasingham, Pal et al. (2012) proposed age at diagnosis ≤30, BMI <30 and HbA1c ≤63.9 mmol/L, which accounted for 8/21 (38.1%) of our MODY patients. The MODY Probability Calculator proposed by Shields et al. (2012) was developed using logistic regression to determine the probability of MODY. Variables include current age, age at diagnosis, sex, treatment with insulin, time to insulin treatment, BMI, HbA1c and parental history of diabetes. The MODY Probability Calculator cannot be directly compared to the other calculators, as it does not define clinical criteria in a binary fashion but assigns a weighting to each factor. Patients have to be aged ≤35 at diagnosis for this calculator to be applied, which resulted in the exclusion of 3/22 patients in our cohort. 9/22 (40.9%) patients had a positive predictive value (PPV) of >50%, which conferred a 0.47/100 false negative rate and a 3/100 false positive rate if used as the basis for referring for genetic testing. 5/22 (22.73%) patients had a PPV of >75%, which conferred a 0.9/100 false negative rate and a 1.2/100 false positive rate, which is highly sensitive and specific. 12/22 (54.55%) patients had a PPV of >20%, which conferred a 0.14/100 false negative rate and a 16.8/100 false positive rate. This work demonstrates that the MODY Probability Calculator currently achieves the best balance of sensitivity / specificity in diagnosing patients with MODY. Nevertheless, a significant portion (10/22) of known MODY patients were not identified. Limitations include excluding patients aged >35 at diagnosis and excluding patients with a strong family history of diabetes who do not have an affected parent. These calculators are diagnostic adjuncts in the Endocrinologist’s toolbox to stratify MODY risk but do not replace clinical judgement.