ESPE Abstracts

Background: It has been shown that a proportion of relatives with multiple islet auto antibodies do not develop diabetes for many years, indicating that a more accurate marker of advanced insulitis is needed.

Objective and Hypotheses: We evaluated whether the level of HbA1c can discriminate children at risk of T1D onset in a cohort of auto antibody positive relatives.

Method: A total of 74 subjects <18 years of age who were participating in the Czech Prediction Programme for T1D, a longitudinal study which prospectively monitors siblings and offspring of diabetic patients for T1D development, were recruited. The median of their follow-up was 3 years (range 1–12 years). They have risk HLA-DQ genotype. All of them developed persistent islet auto antibodies and six of them progressed to diabetes during follow-up. HbA1c was measured at least once (one to ten times, two measurements on average) during follow up until T1D onset. The Markov chain model addressed the predictive strength of HbA1c levels on T1D development.

Results: HbA1c level was effective in discriminating children at risk of T1D development, fitted log-linear effect 0.052 with 95% CI (0.0137, 0.09). The estimated hazard ratio was 1.05 (95% CI (1.014, 1.094) and tells us that a 1 mmol/mol HbA1c increase is associated with 5% higher risk of T1D development. An estimate of the optimal threshold from this data is 37.7 mmol/mol with associated hazard ratio 1.95 (95% CI (1.23, 3.08), which means that a 1 mmol/mol increase over the threshold is associated with a twofold risk increase.

Conclusion: HbA1c monitoring increases the effectiveness of T1D prediction. Children with persistent islet auto antibody and HbA1c over 37.7 mmol/mol may be at greater risk of T1D onset.