Researchers studied 1,129 insulin-treated people referred from routine clinical practice for genetic testing for MODY (maturity-onset diabetes of the young). Every referral had a gene panel that also generated a type 1 diabetes genetic risk score built from 10 variants. Score distributions were compared with reference populations with and without type 1 diabetes, to estimate how much undiagnosed type 1 diabetes was sitting inside the referred group.
About 20% (95% CI 14.9 to 25.2) of people referred for MODY testing in real-world practice in fact had type 1 diabetes. Among those whose panel returned no MODY variant, the enrichment was higher still, consistent with 26.0% type 1 diabetes overall, and 76.2% in children against 16.8% in adults. Age-specific thresholds flagged 16.1% (13.8 to 18.5) of the genetically unsolved group as probable type 1 diabetes. Used the other way round, as a triage test before sending the panel at all, the score discriminated poorly overall (area under the curve 0.60) but well in children under 10 (0.83), where more than half of panels could have been avoided without missing a case of MODY.
Two things follow for clinic. First, a negative MODY panel is not a diagnosis. In a child it most often means type 1 diabetes, and an antibody-negative, insulin-treated child should not be left indefinitely with an unexplained label. Second, the triage use is narrow and age-dependent: under 10 the score can spare the panel, in an adult it cannot and you should simply send the panel. Where genetic testing is paid out of pocket, as it usually is in India, that distinction is the difference between a test worth ordering and an expensive one.
- Before referring for a MODY panel, recheck islet autoantibodies and C-peptide - one in five referrals in this cohort turned out to be type 1 diabetes.
- In a child under 10, ask whether a type 1 genetic risk score is available first; it could avoid more than half the panels.
- In an adult, do not use the score to decide whether to test - its discrimination was barely better than chance.
- Treat a 'no variant found' result as an unfinished diagnosis and go back to the clinical phenotype.
- Take a three-generation family history before testing; the score does not replace it.
The statistics, in plain English
Area under the curve is the single number that says how well a test separates two groups: 0.5 is a coin toss and 1.0 is perfect. The score's 0.60 overall is close enough to a coin toss to be clinically useless, while 0.83 in children under 10 is genuinely usable - the same test, two different jobs, decided by age. The thresholds were set to give a positive predictive value of at least 80%, which means that even among those flagged as probable type 1 diabetes, up to one in five will not have it. And the prevalence figures such as 20% and 26.0% are modelled by comparing score distributions against reference populations, not counted from individual confirmed diagnoses, so they describe the group rather than any one patient in front of you.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for diabetes & endocrinology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free