Background: There is an unmet need for accurate cost-effective estimation of future T1D risk.

Methods: We derived a combined T1D prediction model using 7883 children followed closely from birth for a median of 9 yr, considering T1D genetic risk score (GRS), family T1D history (FH), standard islet autoantibodies (IA), race, birth circumstances, early growth and nutrient status. T1D developed in 326.

Results: Machine learning and traditional methods (Cox models) performed equivalently as measured by time-dependent AUC ROC. Future T1D was accurately predicted by combining only 3 variables: GRS, FH and IA status. Accuracy of combined scores increased with age at scoring. By age 2, they were highly predictive for T1D in the next 5 years (AUC >0.91, 95% CI 0.88-0.95) (see X on Figure). A 2-yr-old with 2 IA, positive FH and high GRS (>12) would have T1D risk over the next 1, 3 and 5 years of 14% (8-19%), 36% (25-45%) and 51% (39%-61%) respectively. A 2-yr-old with 1 IA, no FH and moderate GRS (10-11) has T1D risk of 0.8% (0.6-1.2%), 2.6% (1.9-3.2%) and 4.3% (3.4%-5.2%) respectively. After newborn genetic screening, only simple venous sampling in routine healthcare settings is required.

Conclusion: This approach allows updated individual risk estimates by age, and in the future may enable release of low risk individuals from surveillance long after initial newborn screening for more cost-efficient population based pediatric T1D prediction.

Disclosure

L.A. Ferrat: None. K. Vehik: None. S.A. Sharp: None. Å. Lernmark: None. A. Ziegler: None. M. Rewers: None. J. She: None. J. Toppari: None. B. Akolkar: None. J. Krischer: None. M.N. Weedon: None. S.S. Rich: None. R.A. Oram: Other Relationship; Self; Randox Laboratories Ltd. W. Hagopian: Research Support; Self; Novo Nordisk A/S.

Funding

National Institute of Diabetes and Digestive and Kidney Diseases; National Institute of Allergy and Infectious Diseases; Eunice Kennedy Shriver National Institute of Child Health and Human Development; National Institute of Environmental Health Sciences; JDRF; Centers for Disease Control and Prevention

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at http://www.diabetesjournals.org/content/license.