Design: Diagnostic utility analysis of routine referral data
Population: 1129 insulin-treated people referred for MODY gene panel testing
Primary outcome: Estimated type 1 diabetes prevalence and triage performance of the genetic risk score
Effect: ~20% type 1 overall; 16.1% of unsolved MODY flagged; AUC 0.60 overall, 0.83 under age 10

A Diabetes Care analysis studied 1129 insulin-treated people referred from routine practice for MODY gene panel testing. The panel also carried 10 variants used to calculate a type 1 diabetes genetic risk score.

Comparing score distributions with reference populations, about 20% (95% CI 14.9–25.2%) of referrals were estimated to have type 1 diabetes. Among those whose genetic test found no MODY variant, the estimate rose to 26%, and to 76% in children. Age-specific thresholds labelled 16.1% of genetically unsolved cases as probable type 1. As a triage tool before testing, the score discriminated poorly overall (AUC 0.60), but well in children under 10 (AUC 0.83), where more than half of tests could have been avoided without missing MODY.

A negative MODY panel often leaves the clinician with no diagnosis. A genetic score that points towards type 1 gives that result meaning and can change monitoring and family counselling. This is a diagnostic-accuracy study from one specialist laboratory; the thresholds need validation in other populations, including South Asian ones.

Before sending a MODY panel, check islet autoantibodies and C-peptide; positive antibodies make MODY unlikely.
When a MODY panel is negative in an insulin-treated patient, ask the laboratory whether a type 1 genetic risk score is available.
In young children with suspected MODY, a high type 1 score may justify deferring gene testing.
Score thresholds were derived mainly in European-ancestry populations; interpret cautiously in other ancestries.

Why it matters

Unsolved MODY referrals are often type 1 diabetes hiding in plain sight.

The statistics, in plain English

An AUC of 0.60 is barely better than a coin toss (0.50); 0.83 in under-10s is useful discrimination. The 20% estimate is inferred from score distributions, not from confirmed diagnoses in each person.