Association between FPG and T2DM
Over the course of the study period, 730 women developed diabetes (17, 135, 300, 512, 730, cumulative diabetes at 1,2,3,4 and 5 years follow-up). Women with higher FPG developed more T2DM at any year of follow-up and for all study periods. Overall, 11,497 (19%) of women had obesity. Regardless of obesity status, increments in FPG level were related to higher T2DM incidence (Figure S2).
Next, we employed the ROC statistics and the Youden index analysis to determine the best cut-off value of 1st trimester FPG levels for predicting T2DM and compare FPG levels and GDM prediction performance for T2DM. 1st trimester FPG levels as well as the Youden index, which was calculated to be 86.5 mg/dl (sensitivity 53.3% and specificity 72.4%), both had higher prediction performance for T2DM compared to GDM (sensitivity 20% specificity 97.2%) (Figure 1A- ROC-AUC analysis). When the analysis was stratified by obesity status (Figure 1B-1C), 1st trimester FPG levels had superior prediction performance for T2DM, as compared to GDM, for women with and without obesity.
Most significantly, we applied survival analysis to consider the occurrence of diabetes over the study follow-up period while accounting for potential confounding variables. Employing the Cox regression model, we found that maternal age, a diagnosis of GDM, a BMI greater than or equal to 30 kg/m², and FPG levels were all identified as independent predictors for diabetes development over time (Table 2, p<0.001 for all). Notably, FPG levels exceeding 110 mg/dL exhibited a higher aHR in comparison to GDM for predicting the onset of diabetes (aHR 4.92(2.74-8.81), p<0.001, surpassing the aHR of 3.92 (3.08-4.99), p<0.001, with 95%CI) associated with GDM.