The central role of recovery of β-cell function in the remission of prediabetes: a prospective cohort study in Canada.

BACKGROUND

The mechanistic basis underlying the remission of prediabetes (ie, the return to normoglycaemia) has been suggested to be amelioration of insulin resistance without improvement of β-cell function. We aimed to characterise the relative contributions of changes in insulin sensitivity and β-cell function to the remission of prediabetes.

METHODS

In this prospective cohort study, conducted at Mount Sinai Hospital (Toronto, ON, Canada), we screened pregnant women for gestational diabetes, aiming to recruit participants with varying degrees of dysglycaemia and conduct serial postpartum metabolic evaluations. The study population comprised participants who were diagnosed with prediabetes following pregnancy (designated visit 1) and had a subsequent visit at which they could be assessed for remission of prediabetes (visit 2); study visits occurred at 3 months, 1 year, 3 years, or 5 years postpartum. Both visits included oral glucose tolerance tests, enabling assessment of glucose tolerance, insulin sensitivity and resistance (with the Matsuda index and Homeostasis Model Assessment for Insulin Resistance [HOMA-IR]), and β-cell function (with the Insulin Secretion-Sensitivity Index-2 [ISSI-2] and insulinogenic index divided by fasting plasma insulin [IGI/FPI]). Using logistic regression analyses, we evaluated changes in the log-transformed Matsuda index, HOMA-IR, ISSI-2, and IGI/FPI occurring between visits as predictors of remission of prediabetes, after adjustment for clinical risk factors for diabetes (age, ethnicity, family history of diabetes, BMI at visit 1, and duration of breastfeeding at visit 2). The contribution of each of the four indicated changes to the adjusted model was assessed and ranked.

FINDINGS

Between Sept 22, 2003, and Nov 30, 2018, 787 pregnant women were screened for gestational diabetes. We identified 182 participants with prediabetes at visit 1 (median 3·5 months [IQR 3·0-12·3] postpartum) and stratified them into those with (n=100) and without (n=82) remission on reassessment at visit 2 (median 13·0 months [11·8-27·9] postpartum). At baseline (visit 1), participants who subsequently attained remission had higher ISSI-2 than those who did not attain remission (median 576 [IQR 493-674] vs 496 [391-600]; p<0·0004). At follow-up (visit 2), participants in remission had higher Matsuda index values (5·8 [3·9-7·9] vs 3·9 [2·6-6·7], p=0·0004) and better β-cell function (ISSI-2 650 [563-820] vs 478 [389-590], p<0·0001; IGI/FPI 1·9 [1·2-2·5] vs 1·3 [0·8-2·0], p=0·0002) than those who were not in remission. On logistic regression analyses, the between-visit changes in log ISSI-2 (adjusted odds ratio 3·80, 95% CI 1·53-9·41) and log Matsuda index (2·28, 1·21-4·31) were associated with remission of prediabetes. The top-ranked predictor was change in log ISSI-2 (change in model deviance 8·49), with change in log Matsuda index ranked second (change in model deviance 6·56).

INTERPRETATION

Recovery of β-cell function is the dominant determinant of the remission of prediabetes, akin to its role in remission of diabetes but at an earlier point in the natural history of dysglycaemia.

FUNDING

Canadian Institutes of Health Research.