Introduction
The prevalence of pregestational diabetes among women of reproductive age has increased steadily over time. In South Korea, the prevalence of type 2 diabetes among women aged 19-39 years increased from 0.81% in 2010 to 1.71% in 2020 [1]. Similar trends have been observed worldwide. A recent systematic review and meta-analysis reported that the prevalence of pre-existing diabetes during pregnancy approximately doubled globally between 1990 and 2020 [2]. In line with this global trend, the rate of pre-pregnancy diabetes among women giving birth in the USA has increased from 8.6 to 10.9 per 1,000 births between 2016 and 2021 [3].
Hyperglycemia during pregnancy is associated with adverse obstetric and neonatal outcomes [4]. Obstetric complications include a higher rate of primary cesarean sections, spontaneous preterm labor, and preeclampsia [4]. Neonatal consequences include higher rates of respiratory distress syndrome (RDS), neonatal hypoglycemia, congenital malformations, macrosomia, and neonatal mortality [4]. Maternal hyperglycemia leads to fetal hyperinsulinemia, which promotes disproportionate fetal growth and is characterized by an increased abdominal circumference relative to head size, ultimately contributing to large for gestational age (LGA) births and metabolic complications in neonates [5]. Therefore, maintaining blood glucose levels within the recommended targets during pregnancy is crucial for achieving favorable maternal and neonatal outcomes in women with diabetes.
Several indices are used to assess glycemic control during pregnancy, including glycated hemoglobin (HbA1c) levels, self-monitoring of blood glucose (SMBG), and continuous glucose monitoring (CGM) metrics [6]. Since HbA1c levels are affected by pregnancy-related hemodilution and increased red blood cell turnover, they have been primarily used to assess glycemic status before pregnancy rather than during pregnancy [6,7]. However, several studies have shown that HbA1c levels during pregnancy can offer meaningful prognostic information regarding adverse obstetric and neonatal outcomes [8,9]. In a recent multicenter retrospective cohort study conducted by our group, poor glycemic control in the first and second trimesters (HbA1c level >6.5%) was associated with an increased risk of LGA births and pregnancy-associated hypertension with pregestational diabetes [9].
In addition to high blood glucose levels, maternal obesity is also strongly associated with neonatal birth weight. Higher maternal body mass index (BMI) is an established independent risk factor for excessive fetal growth [10,11]. Several studies have examined the relationship between maternal glycemic markers and the risk of delivering an LGA infant in analyses stratified by BMI. A study investigating adverse pregnancy outcomes in women with GDM found that the relationship between HbA1c levels and outcomes was influenced by maternal pre-pregnancy BMI [12]. Similarly, a cohort study conducted in Denmark revealed that among women with type 1 diabetes, a positive association between HbA1c levels and the risk of delivering an LGA infant was observed only among women with BMI >23 kg/m2, while no significant association was found in women with BMI ≤23 kg/m2 [13]. However, evidence regarding the role of BMI stratification in predicting the risk of an LGA birth in women with pregestational diabetes, particularly type 2 diabetes, is limited.
Therefore, this study aimed to evaluate whether maternal BMI modifies the predictive value of trimester-specific HbA1c levels for the risk of LGA births in South Korean women with pregestational diabetes.
Materials and methods
1. Study design and study population
This multicenter retrospective cohort study included pregnant women with pregestational diabetes who delivered at 17 university hospitals in South Korea. Pregestational diabetes included an early diagnosis of gestational diabetes mellitus (GDM; early GDM, diagnosed in the first trimester), type 1 diabetes, or type 2 diabetes. Early GDM was considered a surrogate marker of previously unrecognized pregestational dysglycemia because hyperglycemia detected early in pregnancy is thought to represent pre-existing abnormalities in glucose metabolism rather than true gestational diabetes [14,15].
Women were eligible if they 1) had completed serial prenatal examinations from the first trimester to delivery, 2) had documented HbA1c values, and 3) had available obstetric and neonatal outcome data. The exclusion criteria included multiple gestations, major fetal anomalies, chromosomal abnormalities, nonviable preterm delivery (<24 weeks), or missing records.
This study aimed to evaluate the trimester-specific predictive value of maternal HbA1c levels for LGA births and neonatal adverse outcomes in women with pregestational diabetes, stratified according to pre-pregnancy BMI (<25 kg/m2 vs. ≥25 kg/m2).
We followed the ethical standards for human experimentation established by the Declaration of Helsinki. The study protocol was approved by the Institutional Review Board of Seoul St. Mary’s Hospital (No. KC23RIDI0434) and by the Institutional Review Boards of all participating centers in accordance with local regulatory requirements.
2. Study assessments
Maternal demographic and clinical characteristics, including maternal age, type of diabetes, parity, mode of conception, use of antidiabetic medication, duration since diabetes diagnosis, use of antihypertensive agents, use of aspirin, pre-pregnancy BMI, and HbA1c levels measured during each trimester, were collected from the medical records. The study period preceded the widespread clinical implementation of CGM. Consequently, glycemic control during pregnancy was monitored primarily through self-monitored blood glucose levels, with HbA1c levels used as an adjunctive indicator.
Obstetric variables included hypertensive disorders of pregnancy, maternal peripartum complications (chorioamnionitis, postpartum endometritis, postpartum hemorrhage, and wound complications), gestational age at delivery, and mode of delivery. Preterm birth was defined as delivery before 37 completed weeks of gestation.
Neonatal clinical data were collected from both delivery and neonatal records.
3. Study outcomes
The primary outcome of the study was the risk of delivering an LGA infant, which was defined as birth weight greater than the 90th percentile of gestational age. Secondary outcomes included a composite of adverse neonatal outcomes, defined as the occurrence of one or more of the following: admission to the neonatal intensive care unit (NICU), neonatal hypoglycemia (blood glucose <40 mg/dL), RDS, neonatal sepsis, seizure, fetal death in utero, or neonatal death within 72 hours after birth.
4. Statistical analyses
Statistical analyses were performed using SPSS software version 25.0 (IBM Corporation, Armonk, NY, USA). Descriptive statistics were used to compare the baseline characteristics and obstetric and neonatal outcomes between the two BMI groups. Categorical variables were analyzed using the chi-square test or Fisher’s exact test, as appropriate, and are presented as numbers and percentages. Continuous variables were analyzed using Student’s t-test and are reported as mean±standard deviation.
Univariable logistic regression analysis was performed to assess the association between maternal HbA1c levels and the risk of delivering an LGA infant within each BMI group. Clinically relevant variables previously associated with fetal overgrowth were included in the multivariable logistic regression model. Adjusted odds ratios (ORs) and 95% confidence intervals (CIs) were determined. Receiver operating characteristic (ROC) curve analysis was conducted to evaluate the discriminatory ability of HbA1c levels in predicting the risk of LGA infants in each BMI group. The optimal HbA1c cutoff value was determined using the Youden index. Sensitivity, specificity, and area under the ROC curve (AUC) were calculated for each cutoff value. Statistical significance was defined as a two-sided P-value of <0.05.
Results
Among the 402 women with pregestational diabetes in the study, 183 (45.5%) had a pre-pregnancy BMI <25 kg/m2 and 219 (54.5%) had BMI ≥25 kg/m2. Baseline characteristics stratified by BMI are presented in Table 1.
Maternal baseline characteristics, including parity, mode of conception, pre-pregnancy or pregnancy diabetes medication use, and antihypertensive medication use, did not differ significantly between the BMI groups. Peripartum maternal complications, gestational age at delivery, preterm birth before 37 weeks, and postpartum bleeding occurred at comparable rates in the BMI groups.
Women in the BMI ≥25 kg/m2 group were older than those in the BMI <25 kg/m2 group (34.6±4.1 vs. 33.7±4.1 years; P=0.044). The distribution of diabetes type differed significantly between the two groups (P<0.001), with type 2 diabetes being more common in women with BMI ≥25 kg/m2 (79.1% vs. 55.0%). Conversely, type 1 diabetes was more common in the BMI <25 kg/m2 group (34.5% vs. 11.2%). The duration of diabetes was significantly longer in the BMI <25 kg/m2 group (6.7±7.0 vs. 4.3±5.5 years; P<0.001).
Mean BMI differed significantly between the groups (21.8±2.0 vs. 30.0±4.4 kg/m2; P<0.001). Among trimester-specific HbA1c values, only second-trimester HbA1c levels were significantly higher in the BMI ≥25 kg/m2 group than in the lower BMI group (6.0±1.5 vs. 5.7±0.9; P=0.043). A higher proportion of women with BMI ≥25 kg/m2 had gestational hypertension than the BMI <25 kg/m2 group (21.5 vs. 10.9; P=0.005). Aspirin use was also more frequent in the higher BMI group (14.3 vs. 6.6; P=0.013).
The mode of delivery differed between groups (P=0.021). Women with BMI ≥25 kg/m2 were more likely to undergo cesarean delivery (82.2 vs. 71.0), whereas vaginal delivery was more common among women with BMI <25 kg/m2 (26.3 vs. 15.1).
Neonatal outcomes stratified by BMI category are summarized in Table 2. The distribution of neonatal sex did not differ significantly between the groups (P=0.078). Overall, 22.4% of the neonates were LGA, and the rate of LGA births was significantly higher in the ≥25 kg/m2 group than in the <25 kg/m2 group (26.5 vs. 17.5; P=0.031).
The composite neonatal outcomes showed no significant differences between the groups (P=0.304). The composite neonatal outcomes included NICU admission, neonatal hypoglycemia, RDS, sepsis, seizures, and perinatal death. Among the individual components, only neonatal seizures occurred more frequently in the <25 kg/m2 group than in the ≥25 kg/m2 group (2.2 vs. 0.0; P=0.047).
Table 3 presents the results of the multivariable regression analysis evaluating the association between trimester-specific HbA1c levels and the risk of LGA births and composite neonatal outcomes stratified by BMI groups. The results of the univariable logistic regression analyses are shown in Supplementary Table 1. Among women with BMI <25 kg/m2, higher HbA1c levels across all three trimesters were consistently associated with an increased risk of LGA births. First-, second-, and third-trimester HbA1c levels showed ORs of 1.89 (95% CI, 1.15-3.10; P=0.012), 2.21 (95% CI, 1.15-4.27; P=0.018), and 2.28 (95% CI, 1.46-3.58; P<0.001), respectively. Among women with BMI ≥25 kg/m2, only third-trimester HbA1c levels were significantly associated with the risk of LGA births (OR, 1.46; 95% CI, 1.07-1.98; P=0.017), whereas first- and second-trimester HbA1c levels did not show statistically significant associations (P=0.192 and P=0.061, respectively). When stratified by BMI, HbA1c levels in any trimester were not significantly associated with the risk of composite neonatal outcomes (all P>0.05).
Supplementary Table 2 presents additional subgroup analyses stratified by BMI category and type of diabetes using univariable logistic regression models, which were performed because of the limited sample size within the individual subgroups. HbA1c levels showed relatively consistent associations with the risk of LGA births across both type 1 and type 2 diabetes among women with BMI <25 kg/m2, whereas these associations appeared to be attenuated among women with BMI ≥25 kg/m2, particularly among women with type 1 diabetes.
Table 4 presents the optimal HbA1c cutoff values for each trimester and their predictive performance for the risk of LGA births in relation to BMI groups. Among women with BMI <25 kg/m2, trimester-specific HbA1c levels demonstrated acceptable to moderate discrimination, with AUCs ranging from 0.70 to 0.82. The HbA1c cutoff values decreased progressively throughout gestation. Second-trimester HbA1c levels showed high specificity (90.5) and a high positive likelihood ratio (6.56), indicating a good rule-in performance for identifying the risk of an LGA birth. Third-trimester HbA1c levels demonstrated high sensitivity (84.4) and a low negative likelihood ratio (0.22), suggesting that a negative test result significantly reduced the likelihood of an LGA birth.
In contrast, among women with BMI ≥25 kg/m2, HbA1c levels showed poor to modest discriminatory performance across all trimesters, with AUCs consistently ≤0.65. The cutoff values and likelihood ratios varied substantially by trimester, and neither the positive nor negative likelihood ratios reached clinically meaningful thresholds. These findings are visually illustrated in Fig. 1, which depicts the ROC curves for trimester-specific HbA1c levels stratified by BMI.
Discussion
This study demonstrated that in a cohort of women with pregestational diabetes, the predictive value of trimester-specific HbA1c levels for the risk of LGA births differed significantly in relation to the pre-pregnancy BMI. Among women with BMI <25 kg/m2, higher HbA1c levels in all three trimesters were consistently and independently associated with an increased risk of LGA births, with stronger associations observed in the later trimesters. In contrast, among women with BMI ≥25 kg/m2, first- and second-trimester HbA1c levels were not significantly associated with the risk of LGA births, whereas third-trimester HbA1c levels showed a significant association, although the magnitude of this association was smaller than that observed among women with BMI <25 kg/m2. The discriminatory performance further reflected these BMI-dependent differences in the predictive strength of HbA1c levels for the risk of LGA births. Among women with BMI <25 kg/m2, HbA1c levels demonstrated acceptable-to-good discrimination for LGA births, with the highest AUC observed in the third trimester. In addition, a second-trimester threshold of >6.4% yielded high specificity and a positive likelihood ratio. However, overall discrimination was poor to modest in women with BMI ≥25 kg/m2 across all trimesters. We found no significant differences in the relationship between HbA1c levels and composite neonatal outcomes between BMI groups.
Several studies have reported associations between trimester- specific HbA1c levels and fetal overgrowth outcomes, including LGA or macrosomia, in pregnancies complicated by diabetes. Higher HbA1c levels in early pregnancy have been associated with an increased risk of LGA births, as demonstrated in a recent multicenter Korean study proposing trimester-specific cutoff values (5.4% and 6.6% in the first and second trimesters, respectively) [9], consistent with prior studies reporting early pregnancy cutoff values ranging from 5.6% to 5.7% [16,17]. Our optimal HbA1c cutoff values for predicting the risk of an LGA birth in women with BMI <25 kg/m2 were relatively higher than those in prior studies (approximately 7.0% in the first trimester and 6.4% in the second trimester), while still demonstrating good discriminatory performance. Notably, the second-trimester threshold exhibited strong rule-in characteristics in women with BMI <25 kg/m2, with high specificity (90.5%) and a positive likelihood ratio of 6.56. Thus, mid-pregnancy HbA1c levels may be particularly useful in identifying a subgroup of women at significantly increased risk of LGA. These differences may be attributed to BMI stratification, which has not been consistently applied in previous studies. From a clinical perspective, these higher cutoff values may allow for more realistic and individualized glycemic targets for leaner women with pregestational diabetes.
Across both BMI categories, third-trimester HbA1c levels showed the strongest association with the risk of LGA births, exhibiting the highest ORs and AUC values in comparison with the earlier trimesters. In a cohort of women with type 1 diabetes, HbA1c levels measured in late pregnancy were independently associated with the risk of delivering LGA infants, further emphasizing the predictive value of third-trimester glycemia in pregestational diabetes [18]. Bashir et al. [19] reported that HbA1c levels measured during late pregnancy were positively associated with birth weight, whereas HbA1c levels measured during early pregnancy were inversely associated with birth weight. Another longitudinal analysis of pregnancies with pregestational diabetes found that suboptimal glycemic control in early pregnancy did not significantly affect neonatal outcomes if glycemic control was achieved later in pregnancy [20]. In contrast, inadequate glycemic control during late pregnancy was associated with an increased risk of LGA births and neonatal hypoglycemia [20]. These findings are consistent with our results and support the importance of glycemic exposure in late pregnancy in determining fetal overgrowth.
In women with BMI ≥25 kg/m2, the HbA1c levels in the first and second trimesters did not significantly predict the risk of LGA births and the overall discriminatory performance was low. These findings suggest that fetal overgrowth in overweight or obese women with pregestational diabetes may not be adequately explained by HbA1c levels alone, indicating the need to consider additional metabolic factors. Although the diabetes type distribution differed substantially in relation to the BMI category, supplementary subgroup analyses suggested that the BMI-dependent association between HbA1c levels and the risk of LGA births was not solely explained by the diabetes type. Rather, the HbA1c-LGA association appeared to align more closely with the BMI category, suggesting that higher BMI may attenuate the predictive utility of HbA1c levels for fetal overgrowth. However, these exploratory analyses should be interpreted with caution because they were based on univariable models with limited subgroup sample sizes.
One plausible explanation for the limited predictive value of HbA1c levels in women with higher BMI is the presence of glucose-independent growth-promoting factors in an obesogenic environment. Maternal obesity is associated with increased insulin resistance, hypertriglyceridemia, elevated free fatty acid levels, and low-grade inflammation, all of which may directly contribute to excessive fetal growth [21,22]. Maternal lipid alterations contribute substantially to fetal growth and adiposity and recent studies on diabetic pregnancies have highlighted the role of altered free fatty acid metabolism in fetal overgrowth [23,24]. Consistent with this concept, a prospective single-center study of women with type 1 diabetes demonstrated that maternal lipid profiles, particularly lower high-density lipoprotein cholesterol and higher triglyceride levels, were independently associated with the risk of LGA births [25]. These findings support the idea that fetal overgrowth is not solely driven by maternal glycemia.
Another explanation is that HbA1c levels do not adequately reflect postprandial hyperglycemia and glycemic variability, both of which are increasingly being acknowledged as important determinants of fetal overgrowth. Although SMBG was used in routine clinical care, detailed glucose profile data, including information for SMBG trends, postprandial glucose excursions, and glycemic variability indices, were not consistently available in this retrospective multicenter cohort. This limitation is clinically relevant because recent evidence suggests that HbA1c levels may not fully capture clinically meaningful glycemic dynamics in early or heterogeneous metabolic states, whereas CGM-derived metrics can provide additional information on dynamic glucose exposure and related perinatal outcomes [26,27]. Law et al. [28] reported that CGM-derived metrics, particularly those reflecting postprandial excursions, are associated with birthweight and the risk of LGA births. In women with BMI ≥25 kg/m2, greater insulin resistance may result in larger postprandial glucose excursions and increased glycemic variability, which may not be adequately reflected by HbA1c levels, thereby reducing its predictive performance.
Overall, the limited discriminatory performance of HbA1c levels in women with BMI ≥25 kg/m2 suggests that HbA1c levels alone may be insufficient for assessing the risk of LGA births in overweight or obese women with pregestational diabetes. To capture clinically relevant glucose dynamics, complementary glucose metrics derived from SMBG or CGM may be necessary. Supporting this notion, the CONCEPTT trial, a randomized controlled study of CGM use in pregnant women with type 1 diabetes, demonstrated significant reductions in the incidence of LGA births and neonatal hypoglycemia with CGM use, despite relatively modest improvements in HbA1c levels [29].
Finally, while a strong association was observed between HbA1c levels and the risk of LGA births, trimester-specific HbA1c levels were not significantly associated with composite neonatal outcomes in either BMI group. This may indicate fundamental differences in the underlying mechanisms of fetal overgrowth in comparison with short-term neonatal morbidity. An LGA birth reflects the cumulative effects of chronic intrauterine metabolic exposure, whereas neonatal outcomes such as hypoglycemia, RDS, sepsis, and NICU admission are more strongly influenced by acute peripartum factors, including intrapartum glycemia, gestational age, and mode of delivery [30,31]. In addition, the relatively narrow HbA1c range in our cohort may have limited our ability to detect associations with neonatal morbidity. Consequently, these findings suggest that HbA1c levels may be more informative for predicting fetal overgrowth than for capturing short-term neonatal morbidity and adverse outcomes.
Among baseline characteristics, prophylactic aspirin use was more frequent in women with BMI ≥25 kg/m2. This may reflect real-world clinician prescription patterns, since obesity itself is considered an additional risk factor for preeclampsia, and may prompt more frequent prophylactic aspirin administration [32].
The strength of this study is that it is the first large multicenter study in a Korean population to evaluate BMI-dependent differences in the predictive utility of trimester-specific HbA1c levels for the risk of LGA births. We utilized serial HbA1c measurements across all trimesters and included a broad spectrum of pregestational diabetes phenotypes to enhance the clinical relevance of the findings. Importantly, we assessed pre-pregnancy BMI as an effect modifier and provided BMI-stratified discrimination metrics (AUC and likelihood ratios) that facilitated clinically actionable risk stratification.
This study had several limitations. First, the retrospective multicenter design limited causal inference and may be subject to residual confounding factors. In particular, detailed data regarding the potential determinants of fetal growth and glycemic exposure, such as maternal lipid profiles and SMBG/CGM-derived glucose metrics, were not uniformly available. Additionally, the HbA1c level does not account for postprandial hyperglycemia or glycemic variability and can be influenced by pregnancy-related changes in red cell turnover or iron status, which could attenuate these associations [6,7]. Although multivariable analyses were adjusted for diabetes type, some residual heterogeneity may exist. Further stratified analyses according to both BMI and diabetes type were limited by sample size constraints, particularly among women with type 1 diabetes and a higher BMI.
In conclusion, trimester-specific HbA1c levels showed BMI-dependent differences in their predictive utility for the birth of LGA infants to women with pregestational diabetes. HbA1c levels provided clinically meaningful risk discrimination in women with BMI <25 kg/m2, demonstrating particularly strong rule-in performance during mid-pregnancy and the highest overall discrimination in late pregnancy. In contrast, their discriminatory performance was poor to modest in women with BMI ≥25 kg/m2. These findings support a BMI-tailored glycemic monitoring approach for pregnancies complicated by pregestational diabetes. HbA1c levels may serve as a practical risk-stratification tool in lean and normal- weight women, whereas additional metabolic and dynamic glucose measures (e.g., SMBG/CGM-derived metrics) may be necessary in overweight and obese women.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Supplement
Print
