Glycated albumin levels in the third trimester of women with gestational diabetes mellitus are associated with adverse pregnancy-related outcomes Authors: Yueshuai Pan, Ruting Gu, Qianqian Li, Jingyuan Wang, Yan Zhang, Lin Zhao, Yue Wu, Lili Wei Published: 2024 ABSTRACT AbstractBackgroundGlycated albumin (GA) levels have been considered as a promising biomarker for estimating glycemic control during pregnancy, but the relationship between GA levels and the incidence of adverse pregnancy outcomes in women with gestational diabetes mellitus (GDM) remains unclear. Our study aimed to investigate the relationship between GA levels during the third trimester and 13 different adverse pregnancy-related outcomes among women with GDM in China.MethodsWe retrospectively extracted clinical data from the medical records of 819 pregnant women with GDM who underwent prenatal examinations and child delivery at the Affiliated Hospital of Qingdao University between January 2022 and October 2022. The cohort was divided into GA-high (GA-H) and GA-low (GA-L) groups based on the median GA level of 10.6%. Then, the incidence rates of 13 specific adverse pregnancy outcomes were compared between the two groups. Furthermore, we estimated the mean GA levels in pregnant GDM women with or without specific adverse outcomes. Multivariate logistic regression analysis was performed to assess whether the GA levels (high or low) were independent risk factors for specific adverse outcomes in pregnant women with GDM. Receiver operating characteristic (ROC) curve analysis was performed to assess the predictive value of GA levels for the adverse pregnant outcomes in women with GDM. E-value for sensitivity analysis was performed to assess the robustness of the findings to unmeasured confoundings.ResultsWe included 819 pregnant women with GDM, whose average age was 33.09 ± 4.47 years, average pre-pregnancy BMI was 23.51 ± 3.67 kg/m2, and the average gestational week in which GDM diagnosed was 24.80 ± 1.79 weeks. The analysis showed that 80.71% (661/819) pregnant women with GDM were associated with adverse pregnancy-related outcomes. Pregnant women in the GA-L group showed higher incidence of the premature rupture of membranes (PROM), whereas those in the GA-H group showed higher incidence of neonatal hypoglycemia. The GA levels showed acceptable clinical performance for predicting neonatal hypoglycemia with an area under the ROC curve (AUC) value of 0.700 (P = 0.010), sensitivity of 71.4%, and specificity of 70.2%. The optimal cut off value for GA was 11.55%.ConclusionsThis study demonstrated that GA levels were significantly associated with specific adverse pregnancy outcomes, especially PROM and neonatal hypoglycemia. Furthermore, GA levels in the third trimester showed acceptable clinical performance for predicting neonatal hypoglycemia among pregnant women with GDM. In the future, the potential role of GA as a predictor of adverse pregnancy outcomes need to be further confirmed and explored in GDM women.Keywords: Gestational diabetes mellitus, Glycated albumin, Adverse pregnancy outcomes, Premature rupture of membranes, Neonatal hypoglycemia Background Gestational diabetes mellitus (GDM) is one of the most common pregnancy-related complications and is characterized by a varying degree of glucose intolerance during the second or third trimester of pregnancy [1]. According to the IDF Diabetes Atlas (10th edition), hyperglycemia was reported for approximately 21.1 million (16.7%) pregnant women in 2021 worldwide, and 80.3% of these cases were categorized as GDM [2]. The prevalence of GDM is estimated to increase consistently because of a gradually rise in the living standards and the delayed childbearing age of women. Furthermore, GDM is associated with a higher short-term and long-term risk of metabolic disorders for both the mothers and their offspring [3]. GDM is also associated with significantly higher risk of adverse pregnancy outcomes such as premature birth, fetal distress, fetal growth restriction, macrosomia, postpartum hemorrhage, fetal malformations, and even stillbirths [4, 5]. Therefore, GDM is associated with significant physical and emotional burden for the affected mothers and their offspring, and is a high economic burden on the family and the public healthcare system [6]. Therefore, strict glycemic control throughout the gestational period is critical in preventing or decreasing the incidence of adverse pregnancy outcomes [7]. Currently, glycated hemoglobin A1c (HbA1c) is the gold standard biomarker for assessing glycemic control because it gives an estimate of the blood glucose levels for the previous 2–3 months [8]. However, HbA1c levels may not be an appropriate biomarker to monitor the glycemic status during pregnancy because of an increased turnover of the red blood cells (RBCs) and the increased incidence of iron deficiency-related anemia during pregnancy, especially during the second and the third trimesters [9]. Glycated albumin (GA) is recommended as an alternate biomarker for estimating glycemic control during pregnancy because it estimates the average glucose level over the previous 2–3 weeks and is not affected by hemoglobin metabolism or iron deficiency-related anemia [10]. Furthermore, previous studies have shown that GA is more sensitive than HbA1c in estimating the glycemic fluctuations during pregnancy [11, 12]. Thus, GA levels during the third trimester are more likely to be associated with adverse pregnancy outcomes in GDM women. Multiple studies have demonstrated that GA is a promising biomarker for the diagnosis and management of diabetes mellitus [13–15], but the relationship between GA levels and the incidence of adverse pregnancy outcomes in women with GDM is not clearly established. Moreover, multiple studies have reported contradictory findings regarding the association between GA levels and adverse pregnancy outcomes in women with GDM. Mendes et al. reported that higher levels of GA in the pregnant mothers with GDM were associated with at least one neonatal complication in their newborns [16]. Sugawara et al. reported that infants of mothers with higher GA levels during pregnancy were large for gestational age (LGA) and diagnosed with hypoglycemia and respiratory disorders [17]. Wu et al. reported that GDM patients with GA ≥ 10.33% were associated with increased incidences of cesarean delivery, hypertension, and fetal distress [18]. However, Dong et al. reported that higher GA levels during the 3rd trimester only in GDM women with postpartum hemorrhage, but GA levels were not associated with adverse outcomes such as LGA, hypertension, preterm delivery and preeclampsia (PE) in GDM women [19]. Furthermore, Zhuang et al. showed that GA levels in the third trimester (34–40 weeks) were not associated with any adverse pregnancy outcomes [20]. Therefore, Our study investigated the relationship between GA levels in the third trimester and adverse pregnancy outcomes in Chinese pregnant women with GDM, and estimated the predictive value of GA for the adverse pregnancy-related outcomes in order to guide better clinical management for GDM and reduce the occurrence of adverse pregnancy outcomes. Methods Study design This retrospective study was approved by the Ethics Committee of the Affiliated Hospital of Qingdao University (No. QYFY WZLL 28129). Participants We reviewed medical records of pregnant women with GDM who underwent obstetric examinations and delivered at the Affiliated Hospital of Qingdao University between January 2022 and October 2022. Pregnant women with singleton pregnancies who were older than 18 years and diagnosed with GDM were included in this study. The exclusion criteria included patients with other pregnancy complications (e.g., pregnancy hypertension) or other diseases (e.g., pulmonary hypertension, congenital heart disease, and others). GDM was diagnosed according to the diagnostic criteria recommended by the International Association of Diabetes and Pregnancy Study Groups (IADPSG) in 2010 [21], and was based on a 75 g oral glucose tolerance test (OGTT) at 24–28 weeks of gestation. GDM was positively diagnosed when the subjects met one of the three following criteria: fasting plasma glucose (FPG) ≥ 5.1 mmol/L; 1-h postprandial blood glucose ≥ 10.0 mmol/L; or 2-h postprandial blood glucose ≥ 8.5 mmol/L. All pregnant women diagnosed with GDM in our study group were given health education on diet, exercise, and blood glucose monitoring, etc. In addition, they were advised to conduct self-monitoring of glycemia daily. The glycemic targets of GDM women during pregnancy are recommended to be as follows: fasting and premeal blood glucose ranges from 3.3 to 5.3 mmol/L, 1-h postprandial blood glucose < 7.8 mmol/L, 2-h postprandial blood glucose < 6.7 mmol/L, and nighttime blood glucose ≥ 3.3 mmol/L.22 Sample size The sample size was calculated using the formula for cohort study, as follows: The GA levels ≥ 10.6% were considered as an exposure factor, and neonatal hypoglycemia with statistically significant and low incidence in univariate analysis was used as the outcome measure. According to the data we collected, was 0.739% and was 2.663%. Considering α of 0.05 (two-sided) and a test power (1-β) of 0.9, was calculated to be 303 and the total sample size for this study was 606. Totally, we included 819 patients with GDM. Collection of clinical data We retrieved the following clinical characteristics of the study participants from the medical records: age, pre-pregnancy height and weight, weight at the last prenatal examination, gravidity and parity, family history of diabetes mellitus, regularity of the menstrual cycle, history of adverse pregnancy, and adverse pregnancy outcomes. Pre-pregnancy body mass index (pre-pregnancy BMI) was calculated by dividing the pre-pregnancy weight (kg) by the pre-pregnancy height (m2). Weight gain during pregnancy was calculated by subtracting the pre-pregnancy weight from the weight at the last prenatal examination. Laboratory parameters included the results of the 75 g OGTT at 16–33 weeks with an average of (24.74 ± 1.80) weeks, triglycerides (TG), total cholesterol (TC) and GA levels at 28–41 weeks with an average of (34.02 ± 2.53) weeks. GA levels were estimated by high-performance liquid chromatography in our study. We assessed 13 specific adverse outcomes in this study, including premature birth, premature rupture of membranes (PROM), placental abruption, caesarean delivery, fetal distress, fetal malformation, fetal growth restriction, shoulder dystocia, stillbirth, macrosomia (birth weights ≥ 4000 g), neonatal asphyxia, neonatal hypoglycemia (plasma glucose<2.6 mmol/L within one hour after birth [23]), and postpartum hemorrhage. Statistical analysis Statistical analysis was performed using the SPSS Statistical software version 26.0 (IBM Corp., Armonk, NY, USA). Probability plot was used to determine whether the data was normally or non-normally distributed. The normally distributed numerical data was represented as mean ± standard deviation. The skewed numerical data was represented as median (interquartile range). The categorical data was expressed as counts and percentages. The median GA value was used as the cutoff value to categorize the study subjects into the GA-high (GA-H) and GA-low (GA-L) groups. The incidences of adverse outcomes between the two groups were analyzed using the chi-square test. Student’s t test was used to analyze whether the differences in the mean GA levels between any two groups with and without specific adverse outcomes were statistically significant or not. Multivariate logistic regression analysis with a backward stepwise approach was used to further determine whether the GA level was an independent risk factor for predicting specific adverse outcomes in pregnant women with GDM. A collinearity test was performed prior to the logistic regression analysis. Receiver operating characteristic (ROC) curve analysis was performed to assess the predictive value of GA for estimating specific adverse pregnant outcomes in pregnant women with GDM. The sensitivity analysis to unmeasured confounding factors using the E-value methodology of VanderWeele and Ding [24]. P < 0.05 was considered statistically significant. Results Baseline characteristics of the study subjects This study included 819 pregnant women with GDM. All GDM participants were treated with diet, and those with poor dietary control were treated with insulin. In our study group, only 7 (0.85%) participants were treated with insulin. The baseline clinical characteristics of the study subjects are shown in Table 1. Adverse pregnancy outcomes were reported for 80.71% (661/819) of the pregnant women with GDM included in this study. The cases and incidences of specific adverse pregnancy outcomes are shown in Table 2. Table 1Baseline clinical characteristicsVariablesMean ± SD/ MedianRangeAge, years33.09 ± 4.4719 ~ 53Pre-pregnancy BMI, kg/m223.51 ± 3.6713.39 ~ 37.11Weight gain during pregnancy, kg13.21 ± 4.800 ~ 31Fasting plasma glucose (FPG), mmol/L5.04 ± 0.573.47 ~ 8.341 h 75 g glucose level, mmol/L9.63 ± 1.654.20 ~ 17.032 h 75 g glucose level, mmol/L8.10 ± 1.563.10 ~ 19.70GA levels (%)10.73 ± 1.922.00 ~ 17.20Triglycerides (TG), mmol/L3.40 ± 1.560.87 ~ 15.08Total cholesterol (TC), mmol/L6.13 ± 1.211.86 ~ 13.50Gravidity (times)2 (1,3)1 ~ 8Parity (times)1 (0,1)0 ~ 3Insulin treatmentYes7(0.85%)-No812(99.15%)Family history of DM-Yes88 (10.74%)-No731 (89.26%)-Regular menstrual cycle-Yes727 (88.77%)-No92 (11.23%)-history of adverse pregnancy-Yes401 (48.96%)-No418 (51.04%)- Baseline clinical characteristics Table 2The cases and incidences of specific adverse pregnancy outcomes among women with GDMAdverse OutcomesCasesIncidence (%)Caesarean47858.36%PROM15018.32%Macrosomia10112.33%Postpartum hemorrhage819.89%Fetal distress708.55%Premature birth506.11%Neonatal hypoglycemia141.71%Placental abruption131.59%Fetal malformation80.98%Fetal growth restriction60.73%Shoulder dystocia30.37%Stillbirth30.37%Neonatal asphyxia20.24%Total adverse outcomes66180.71% The cases and incidences of specific adverse pregnancy outcomes among women with GDM Pregnant women with higher GA levels show lower incidence of PROM and higher incidence of neonatal hypoglycemia The cohort was divided into the following two groups based on the median level of GA (10.6%): GA-L group (GA level<10.6%, n = 406) and GA-H group (GA ≥ 10.6%, n = 413). The GA-H group showed significantly lower incidence of PROM (14.77% vs. 21.92%, P = 0.008) and significantly higher incidence of neonatal hypoglycemia (2.66% vs. 0.74%, P = 0.034) compared to the GA-L group (Table 3). Table 3Comparison of the incidence for specific adverse pregnancy outcomesAdverse OutcomesGA-H (n = 413)GA-L (n = 406) P Premature birth21 (5.08%)29 (7.14%)0.219PROM61 (14.77%)89 (21.92%)0.008**Gestational age for PROM, weeks38.29(36.93, 39.64)38.57(36.93, 39.43)0.832Placental abruption7 (1.69%)6 (1.48%)0.804Caesarean234 (56.66%)244 (60.10%)0.318Fetal distress30 (7.26%)40 (9.85%)0.185Fetal malformation3 (0.73%)5 (1.23%)0.704Fetal growth restriction2 (0.48%)4 (0.99%)0.667Shoulder dystocia2 (0.48%)1 (0.25%)1.000Stillbirth2 (0.48%)1 (0.25%)1.000Macrosomia44 (10.65%)57 (14.04%)0.141Weight of macrosomia, g4240 (4085, 4490)4180 (4080, 4300)0.132Neonatal asphyxia1 (0.24%)1 (0.25%)1.000Neonatal hypoglycemia11 (2.66%)3 (0.74%)0.034*Glucose values in neonates with hypoglycaemia, mmol/L2.05 ± 0.521.90 ± 0.500.669Postpartum hemorrhage44 (10.65%)37 (9.11%)0.460Volume of postpartum hemorrhage, ml575 (500, 867.5)500 (500,740)0.231Total adverse pregnancy outcomes330 (79.90%)331 (81.53%)0.556Note: *P<0.05; **P<0.01 Comparison of the incidence for specific adverse pregnancy outcomes Note: *P<0.05; **P<0.01 We then investigated the potential risk factors for PROM and neonatal hypoglycemia. In our cohort, women with PROM showed significant differences in the pre-pregnancy BMI (23.63 ± 3.73 vs. 22.94 ± 3.33, P = 0.035), gravidity (2(1,3) vs. 2(1,3), P = 0.024), and parity (0(0,1) vs. 1(0,1), P = 0.007) compared with the women without PROM. Furthermore, the TC levels were significantly lower in women with neonatal hypoglycemia compared with the women without neonatal hypoglycemia (5.52 ± 0.94 vs. 6.14 ± 1.21, P = 0.030) (Table 4). Table 4Baseline comparison for premature rupture of membranes and neonatal hypoglycemiaVariablesWomen with PROM (n = 150)Women without PROM (n = 669) P Women with NH (n = 14)Women without NH (n = 805) P Age, years32.51 ± 4.2133.22 ± 4.520.08233.93 ± 5.2033.07 ± 4.460.478Pre-pregnancy BMI, kg/m223.63 ± 3.7322.94 ± 3.330.035*24.05 ± 3.8223.50 ± 3.670.576Weight gain during pregnancy, kg12.84 ± 5.1313.29 ± 4.730.29414.14 ± 2.9213.19 ± 4.830.464FPG, mmol/L5.10 ± 0.575.03 ± 0.570.1735.63 ± 1.135.03 ± 0.550.0691 h 75 g glucose level, mmol/L9.81 ± 1.839.59 ± 1.610.13910.79 ± 3.079.61 ± 1.610.1742 h 75 g glucose level, mmol/L8.19 ± 1.518.08 ± 1.570.4639.83 ± 3.648.07 ± 1.490.095TG, mmol/L3.37 ± 1.553.41 ± 1.560.8293.91 ± 1.533.39 ± 1.560.216TC, mmol/L6.25 ± 1.286.10 ± 1.190.1855.52 ± 0.946.14 ± 1.210.030*Gravidity (times)2 (1, 3)2 (1, 3)0.024*2.5 (1.75, 3)2 (1, 3)0.304Parity (times)0 (0, 1)1 (0, 1)0.007**1 (0, 1)1 (0, 1)0.356Insulin treatmentYes1 (0.12%)6 (0.73%)1.0000(0.00%)7 (0.85%)0.997No149 (18.19%)663 (80.95%)14(1.71%)798 (97.44%)Family history of DM0.9730.997Yes16 (1.95%)72 (8.79%)1 (0.12%)87 (10.62%)No134 (16.36%)597 (72.90%)13 (1.59%)718 (87.67%)Regular menstrual cycle0.8080.360Yes134 (16.36%)593 (72.41%)14(1.71%)713(87.06%)No16 (1.95%)76 (9.28%)0 (0.00%)92 (11.23%)history of adverse pregnancy0.6590.090Yes71 (8.67%)330 (40.29%)10 (1.22%)391 (47.74%)No79 (9.65%)339 (41.39%)4 (0.49%)414 (50.55%)Note: *P<0.05; **P<0.01 Baseline comparison for premature rupture of membranes and neonatal hypoglycemia Note: *P<0.05; **P<0.01 Subsequently, we performed a dichotomous logistic regression analysis with PROM as the dependent variable and categorical data for GA as the independent variable after adjusting for factors such as age, pre-pregnancy BMI, FPG, 1 h glucose level in the 75 g OGTT, 2 h glucose level in the 75 g OGTT, and TC (all P<0.20). The regression analysis results showed that GA level<10.6%, pre-pregnancy BMI, and 1 h glucose level in the 75 g OGTT were independent risk factors for PROM (Table 5). The VIF values for all the variables were less than 5 in the collinearity test, which suggested absence of collinearity. Table 5Multivariate logistic regression analysis of premature rupture of membranesVariablesBSE P OR (95%CI)GA level (<10.6%)0.5610.1970.004**1.752 (1.191–2.576)Pre-pregnant BMI-0.0600.0280.034*0.942(0.892–0.995)1 h 75 g glucose level0.1200.0590.041*1.127(1.005–1.265)Note: *P<0.05; **P<0.01 Multivariate logistic regression analysis of premature rupture of membranes Note: *P<0.05; **P<0.01 Next, we performed a dichotomous logistic regression analysis with neonatal hypoglycemia as the dependent variable after adjusting for parameters such as FPG, 1 h glucose level in the 75 g OGTT, 2 h glucose level in the 75 g OGTT, TC, and the history of adverse pregnancy (all P<0.20). The logistic regression analysis results showed that GA level ≥ 10.6% in the third trimester was not an independent risk factor for neonatal hypoglycemia. However, the incidence of neonatal hypoglycemia was higher in the GA-H group compared with the GA-L group (2.66% vs. 0.74%, P = 0.034). The mean GA levels are higher for pregnant women in the neonatal hypoglycemia group Then, we analyzed whether the mean GA levels were significantly different between GDM women with or without different pregnancy-related adverse outcomes. As shown in Table 6, there were no significant differences in the mean GA levels between pregnant women with and without different pregnancy-related adverse outcomes except for women with and without neonatal hypoglycemia. The mean GA level of pregnant women in the neonatal hypoglycemia group was significantly higher than those in the group without neonatal hypoglycemia (12.01 ± 2.06% vs. 10.70 ± 1.91%, P = 0.012). Table 6Comparison of the mean GA level for specific adverse outcomeAdverse OutcomesGA level P ControlCasePremature birth10.75 ± 1.9210.41 ± 1.850.235Premature rupture of membranes10.78 ± 1.8910.50 ± 2.010.106Gestational age for PROM, weeks──0.549Placental abruption10.74 ± 1.9010.10 ± 2.820.235Caesarean10.87 ± 1.8810.63 ± 1.940.079Fetal distress10.75 ± 1.8910.48 ± 2.210.268Fetal malformation10.73 ± 1.9210.04 ± 1.780.307Fetal growth restriction10.74 ± 1.929.27 ± 1.230.061Shoulder dystocia10.73 ± 1.9210.73 ± 1.100.095Stillbirth10.73 ± 1.9211.00 ± 0.870.805Macrosomia10.71 ± 1.8810.84 ± 2.140.542Weight of macrosomia, g──0.759Neonatal asphyxia10.73 ± 1.9211.10 ± 1.270.783Neonatal hypoglycemia10.70 ± 1.9112.01 ± 2.060.012*Glucose values in neonates with hypoglycaemia, mmol/L──0.279Postpartum hemorrhage10.74 ± 1.8510.60 ± 2.470.626Volume of postpartum hemorrhage, ml──0.108Total adverse outcomes10.65 ± 1.7610.75 ± 1.950.580Note: *P<0.05 Comparison of the mean GA level for specific adverse outcome Note: *P<0.05 GA level in the third trimester is a predictive factor for neonatal hyperglycemia among pregnant women with GDM The logistic regression analysis results for neonatal hypoglycemia are shown in Table 7. In the logistic regression model for neonatal hypoglycemia, we screened for parameters such as GA level in the third trimester, FPG, and 2 h glucose level in the 75 g OGTT. Our data showed that the GA level in the third trimester was the only significant independent risk factor for neonatal hypoglycemia (P = 0.024). Table 7Multivariate logistic regression analysis of neonatal hypoglycemiaVariablesBSE P OR (95%CI)GA level0.3390.1510.024*1.404 (1.045–1.887)FPG0.7490.4200.0742.116 (0.929–4.820)2 h 75 g glucose level0.2450.1440.0881.278 (0.964–1.693) Multivariate logistic regression analysis of neonatal hypoglycemia Furthermore, ROC curve analyses showed that the clinical performance of the GA level in the prediction of neonatal hypoglycemia among pregnant women with GDM was acceptable with an area under curve (AUC) value of 0.700 (P = 0.010), sensitivity of 71.4%, and specificity of 70.2% (Fig. 1). The optimal GA cut off value was 11.55%. Receiver operating characteristic (ROC) curve analysis of GA for predicting neonatal hypoglycemia in women with GDM E-Value analysis for PROM and neonatal hypoglycemia We assessed the robustness of the findings to unmeasured confoundings using the E-value for sensitivity analysis. E-values for the odds ratios were 3.27 and 2.16 for PROM and neonatal hypoglycemia, respectively. Discussion In this study, the incidences of one or more adverse outcomes in the cohort of 819 pregnant women with GDM was 80.71%. Caesarean delivery, PROM, and macrosomia were the top 3 pregnancy-related adverse outcomes with the incidences of 58.36%, 18.32%, and 12.33% among women with GDM, respectively. Compared with our results, Tian et al. study showed the lower incidences of cesarean (40.5%), and macrosomia (9.2%) in a cohort of 25,995 pregnant women with GDM from Hebei Province, China [25]. GA is a biomarker of glucose homeostasis and is significantly associated with the incidence rates of specific adverse pregnancy outcomes among women with GDM. We used 10.6% as the cut-off value for GA and compared the incidences of specific adverse outcomes between subjects with GA ≥ 10.6% (GA-H group) versus subjects with GA<10.6% (GA-L group) and did not find any significant differences in the incidence rates of 11 out of 13 adverse outcomes that were assessed. However, GA-H group showed lower incidence of PROM and higher incidence of neonatal hypoglycemia than the GA-L group. After adjusting for the confounding factors, GA<10.6% was associated with a 1.752-fold higher risk of PROM compared to those with GA ≥ 10.6% (Table 5). Furthermore, our data showed that GA ≥ 10.6% was not an independent risk factor for neonatal hypoglycemia. Premature rupture of membranes (PROM) is defined as the rupture of the amniotic membranes and release of the amniotic fluid before the onset of labor [26]. The etiology of PROM is unknown. Previous studies have shown that maternal hyperglycemia increased the risk of infection at the choriodecidual interface, which weakened the fetal membranes and increased the incidence of PROM [27, 28]. Furthermore, nutrient deficiency causes premature weakening and rupture of the fetal membranes. Pregnant women with GDM are deficient in the energy and nutrient levels because their food intake is restricted to control the blood glucose levels [29]. Therefore, women in the GA-L group are expected to be more deficient in the energy and nutrient levels despite better glycemic control than the GA-H group in our study. This nutritional and energy deficiency may explain the higher risk of PROM in the pregnant women belonging to the GA-L group. Moreover, our data also showed that pre-pregnancy BMI was an independent risk factor that was negatively associated with PROM. This result was in agreement with the findings of previous studies, which reported that women with higher pre-pregnancy BMI showed lower incidence of PROM [27, 30]. We then compared the mean GA levels in women with GDM that were sub-grouped into those with or without each of the 13 pregnancy-related adverse outcomes. The results showed that the GA levels were higher in the third trimester for GDM women in the group with neonatal hypoglycemia compared to the group without neonatal hypoglycemia. Sugawara et al. retrospectively analyzed 77 Japanese pregnant women with GDM and reported higher incidence of neonatal hypoglycemia in those with higher GA levels compared to those with lower GA levels [31]. Another study by Sugawara et al. showed similar results which the higher GA levels in third trimester were associated with higher incidence of neonatal hypoglycemia for pregnant women that were diagnosed with hyperglycemia because of GDM, type 1 diabetes or type 2 diabetes [17]. These data were similar to our findings and showed that higher GA levels during the 3rd trimester indicated poor glucose control in the pregnant women. The fetus receives higher levels of glucose during maternal hyperglycemia. This is counteracted by a compensatory hyperinsulinemia in the fetus to metabolize the excess glucose. This hyperinsulinemia is the main cause of neonatal hypoglycemia after delivery [32]. Our data also showed reliable clinical value for the GA levels in predicting neonatal hypoglycemia. Logistic regression analysis results showed that the risk of neonatal hypoglycemia increased by 1.404-fold for every 1% increase in the maternal GA levels (Table 7). As shown in Fig. 1, the predictive performance of GA was acceptable with an AUC value of 0.70 (P = 0.010). The sensitivity and specificity values were 71.4% and 70.2%, respectively, when the GA cut off value was 11.55%. Therefore, in a population of 1000 women with an incidence rate of 1.71% for neonatal hypoglycemia, our data predicted that 5 patients with neonatal hypoglycemia would receive false negative results and 310 patients without neonatal hypoglycemia would receive false positive results. Our data also suggested that the GA levels in pregnant women with GDM should be below 11.55% to decrease the risk of neonatal hyperglycemia. To assess the robustness of the findings to unmeasured confoundings, we used the E-value for sensitivity analysis. E-value was calculated to estimate the extent to which an unmeasured confounder would have to be associated with both GA levels and adverse pregnancy outcomes to nullify the observed association between GA levels and outcomes [24]. The results indicated that unmeasured confounders could fully account for the association of GA levels with PROM and neonatal hypoglycemia if it were associated with the exposure and outcomes by ORs of 3.27 and 2.16, which weaker confounding factors could not do so. Therefore, the robustness of the conclusions that GA levels in the third trimester were associated with the occurrence of PROM and neonatal hypoglycemia were validated. More than 80% of women with GDM were associated with adverse pregnancy outcomes in our study. And our sthdy found that both high and low GA levels were associated with different adverse pregnancy outcomes in women with GDM. Therefore, identifying an appropriate GA range is of great significance to prevent adverse pregnancy outcomes. However, there is no explicitly GA recommended reference interval for GDM. Only a few studies focused on establishing the GA reference ranges in pregnancy and there are differences in the reference values for GA in related studies [18, 19]. Moreover, the results of a meta-analysis showed that the average cut-off values reported in these GDM studies were much lower than those for another type of non-GDM [33]. Therefore, it is reasonable to infer that the GA values that predict adverse pregnancy outcomes in GDM may also differ from those in non-GDM. In the future, a reference range for GA is more valuable to establish for women with GDM. Limitations This study has a few limitations. Firstly, the causality and generalizability of the results was limited because this was a single-center retrospective study. In the future, a multi-center prospective cohort study should be designed to confirm our findings and improve external validity. Secondly, due to the retrospective nature of this study, we could not obtain accurate data for their treatment which may have prevented the occurrence of adverse outcomes in GDM patients, such as diet, exercise, medication and parameters of glycaemic control. And our study may be subject to those unmeasured confounders although we used logistic regression analysis that incorporated major clinically relevant factors (e.g., results of the 75 g OGTT, ) to address the risk of confounding factors. Thirdly, we used the median of GA levels as a cut-off value, which is relatively intuitive and stable, but the actual distribution and characteristics of the data are not taken into account. Thus, our results did not definitely prove that GA levels are not associated with other adverse pregnancy outcomes. In the future, it may be necessary to explore more reasonable classification based on the actual distribution and characteristics of the data. Conclusions Our study showed that altered GA levels in pregnant women with GDM were associated with increased prevalence of PROM and neonatal hypoglycemia. And low GA levels were associated with increased risk of PROM, whereas high GA levels were associated with increased risk of neonatal hypoglycemia. Moreover, GA levels showed acceptable prediction performance for neonatal hypoglycemia. Due to the limitations of our study, the potential role of GA as a predictor of adverse pregnancy outcomes need to be further confirmed and explored in GDM women. Acknowledgements We would like to express our gratitude to all the participants in this study and the staff of the Affiliated Hospital of Qingdao University for their support and assistance in this project. Abbreviations Glycated albumin Gestation diabetes mellitus Receiver operating characteristic Premature rupture of membranes Area under the ROC curve Hemoglobin A1c Large for gestational age Preeclampsia International Association of Diabetes and Pregnancy Study Groups Oral glucose tolerance test fasting plasma glucose Body mass index Triglycerides total cholesterol Author contributions The study was designed by LLW, RTG and YSP. YSP was also a major contributor in writing the manuscript. QQL collected and sorted out the data. YZ and JYW performed the statistical analysis. LLW, YW and LZ reviewed and edited manuscripts. All authors read and approved the final manuscript. Funding This research was supported by Qingdao outstanding Health Professional Development Fund, the Chinese Nursing Association Research Program (ZHKY202118) and the Shandong Nursing Association Research Program (SDHLKT202209). The funding agency played no role in research design, data collection, analysis or interpretation, and manuscript writing. Data availability The datasets generated and/or analysed during the current study are not publicly available due to limitations of ethical approval involving the patient data and anonymity but are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate This retrospective study was approved by the Ethics Committee of the Affiliated Hospital of Qingdao University (No. QYFY WZLL 28129). Because of the retrospective nature of the study, informed patient consent was waived by the Ethics Committee of the Affiliated Hospital of Qingdao University. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement The datasets generated and/or analysed during the current study are not publicly available due to limitations of ethical approval involving the patient data and anonymity but are available from the corresponding author on reasonable request.