DISCUSSION
Our study which utilizes the CGM sensor data to analyze GV and GC parameters, is the first to be conducted in Singaporean pregnant women who are relatively at higher risk for GDM, compared to women of Western ethnicity 23. In the second trimester of pregnancy, we find that MAGE was significantly higher in the group of women who subsequently developed GDM, compared to those who did not. Our findings contribute to the much-needed evidence examining the associations between GV and GC parameters early in pregnancy, and the subsequent development of GDM in pregnant women.
Only three studies 11, 13, 16 thus far have included MAGE from CGM data as a GV parameter when examining GDM as an outcome. Our findings concur with the cross-sectional study by Su et al.13, which reported significantly higher levels of MAGE in pregnant women who were diagnosed with GDM at an average of 25 weeks of gestational age, compared to pregnant women who did not develop GDM, and non-pregnant healthy women with normal glucose regulation. Similarly, Dalfra et al . 16 reported an overall trend of slightly higher MAGE levels across the first, second and third trimester of pregnancy in women who developed GDM, compared to healthy control, although the significance was merely borderline. However, when MAGE levels were assessed independently in different trimesters of pregnancy in another study by Dalfa et al., the levels of MAGE were reported to be lower, but higher in the second and third trimester respectively, in participants with GDM outcomes compared to the healthy controls 11. Still, they only used CGM data after a short wear-time of 2 days which might not be sufficient to optimally asses glycaemic control 20, and a GDM diagnosis across a wide timepoint with an average of 21 weeks of gestational age with an SD of 6.3 weeks which might result in misclassifications1, possibly explaining the discrepancies in study findings.
MAGE was the first diabetes-specific GV metric to be developed primarily to capture mealtime-related and intra-day glucose excursions, and has been considered a gold standard for assessing GV. It had been associated with increased insulin resistance, and early-phase insulin secretion deterioration 13, which is characteristic of the pathophysiology of GDM and Type 2 Diabetes development24. Furthermore, higher MAGE levels which represents greater glucose fluctuations had been associated with adverse maternal and neonatal outcomes attributed by GDM, including large for gestational age, small for gestational age, higher birth weight, neonatal hypoglycaemia 25.
The differences in MAGE levels in the group who developed GDM compared to the group who did not was only significantly different in the second trimester, but not the first. This observation can be explained by the physiological changes in the action and secretion of insulin, where serum insulin levels progressively increase from the first to the third trimester of pregnancy, signifying an increase in insulin resistance as the pregnancy advances 26. This can then be correlated to higher glucose variability exacerbated by glycaemic instability in the later trimesters of pregnancy as seen in a study by Dalfra et al. The authors of this study reported an overall increasing trend of GV parameters (MAGE, SDBG, CONGA continuous overlapping net glycaemic action, interquartile range (IQR)) from the first to the third trimester of pregnancy in both healthy pregnant women, and in pregnant women with GDM 16. Our findings coupled with the available evidence from literature allude to inherent differences in MAGE levels in pregnant women diagnosed with GDM, compared to those without GDM which is likely to be more pronounced starting from the second trimester of pregnancy.
In our study, MBG, SDBG, CV, J-Index, %TIR, %TAR and %TBR were not significantly associated with GDM outcomes. However, across existing literature these CGM-derived parameters had not been shown to be consistently associated with GDM outcomes. Three studies reported MBG to be significantly higher in the group of women who developed GDM13, 14, 16, one study reported significant associations only the second trimester but not the third trimester of pregnancy, 11 while another three reported null associations 12, 15, 17. Amongst these studies, only four analyzed SDBG 11, 13, 16, 17, and only two found significant associations with GDM outcomes 13, 16, one study reported higher SDBG in the GDM group only in the third trimester, but lower SDBG in the second trimester compared to the controls11. So far, only one other study has analyzed CV, J-Index and time in glucose ranges (%TIR, %TBAR and %TBR) and reported null associations with GDM outcomes 17.
The strengths in our study lie in its’ prospective design, which enables us to assess the temporal sequence of the GV and GC parameters measured by the CGM in the first and second trimester of pregnancy on subsequent GDM development. Unlike other studies which have explored similar associations, our study has analyzed CGM data from an average of 14 days wear-time to acquire a more accurate and meaningful interpretation of the glucose data. Furthermore, our GDM diagnosis uses the IADPSG criteria adopted by the World Health Organization, while other studies in literature have used different diagnostic criteria for GDM12, 14, 16, 17, and thus different glucose thresholds in a two- or three-time-point antenatal OGTT for diagnosis. All in all, our findings corroborate with other studies in terms of the applicability of CGM use in detecting GV and GC parameters during pregnancy.
A few limitations were noted in this study. Firstly, our small sample size limits the generalizability of its findings, and secondly, this observational study cannot establish a causal relationship between MAGE and the increased risk for the development of GDM.