Dose-response association between OGTT and adverse perinatal outcomes after IVF treatment: A cohort study based on a twin population.

Abstract

Background: Investigate the association between Oral Glucose Tolerance Test (OGTT) after in vitro fertilization (IVF) treatment and adverse maternal and neonatal outcomes in twin pregnancies.

Methods: This retrospective study encompassed 2,541 twin pregnancies conceived through IVF treatment. Adverse maternal and neonatal outcomes were compared across different subgroups based on individual and combined OGTT classifications. A Spearman correlation regression model examined associations between OGTT levels at different time points and parameters such as gestational age, birth weight, and length. Subsequently, a Logistic regression model with restricted cubic splines (RCS) explored the relationships between OGTT levels at different time points and adverse pregnancy outcomes. Ultimately, nine types of machine learning models were developed using OGTT glucose values at different times to predict the risk of adverse pregnancy outcomes.

Results: In subgroup analysis based on individual OGTT diagnosis, three time points were examined: fasting glucose (OGTT0), 1-hour post-glucose (OGTT1), and 2-hour post-glucose (OGTT2). OGTT0 ≥ 5.1 mmol/L was significantly associated with increased risks of ICP and neonatal hypoglycemia (p = 0.031; p = 0.022). OGTT1 ≥ 10 mmol/L correlated with higher risks of ICP and neonatal hyperbilirubinemia (p = 0.001; p = 0.002). OGTT2 ≥ 8.5 mmol/L was also linked to neonatal hyperbilirubinemia (p < 0.001). In combined impaired OGTT subgroups, the impaired fasting glucose (IFG) group had a higher incidence of neonatal hypoglycemia than the impaired glucose tolerance (IGT) group and IFG & IGT group, but a lower risk of neonatal hyperbilirubinemia. OGTT2 was negatively correlated with gestational age at delivery (β = - 0.08, p = 0.018), and both OGTT1 and OGTT2 were negatively correlated with neonatal birth weight (β = - 10.54, p = 0.008; β = - 15.04, p < 0.001), as well as OGTT2 with birth length (β = - 0.16, p = 0.009). The RCS logistic regression model indicated that the increase OGTT values was associated with the ICP risk, and the relationship between OGTT2 and neonatal hyperbilirubinemia was U-shaped. Among the various machine learning models predicting adverse outcomes, RandomForest exhibited superior performance.

Conclusion: OGTT values in twin pregnancies under IVF treatment are closely linked to adverse maternal and neonatal outcomes, with post-load glucose levels potentially serving as an early biomarker for identifying poorer outcomes. The inflection points in the RCS suggest a new indication point for the association between OGTT and adverse pregnancy outcomes in twin pregnancies conceived through IVF.