Association between umbilical cord oxidative stress and early neonatal outcomes in gestational diabetes mellitus

Abstract

Background

Gestational diabetes (GDM) is frequently associated with adverse neonatal outcomes, but the underlying mechanisms linking maternal hyperglycemia to fetal harm are not fully understood. Oxidative stress has been suggested as a key pathway; however, studies simultaneously evaluating multiple oxidative and antioxidant biomarkers in umbilical cord blood are limited. This study aimed to investigate the impact of GDM on the fetus by comprehensively evaluating oxidative stress and antioxidant defense mechanisms in umbilical cord blood and to examine their potential associations with neonatal clinical outcomes.

Methods

Umbilical cord blood samples were analyzed for malondialdehyde (MDA), membrane attack complex (MAC), total antioxidant status (TAS), total oxidant status (TOS), oxidative stress index (OSI), ischemia-related modified albumin (IMA), 8-hydroxy-2’-deoxyguanosine (8-OHdG), advanced glycation end products (AGEs), reduced glutathione (GSH), and glutathione peroxidase (GPx). Neonatal clinical outcomes were recorded and compared between pregnancies with and without GDM.

Results

After correction for multiple biomarker comparisons, MAC, 8-OHdG, and TOS differed significantly between the GDM and non-GDM groups. MAC and 8-OHdG levels were higher in the GDM group, whereas TOS was lower. OSI was higher in the Mann–Whitney analysis but did not remain significant after BH-FDR correction. Exploratory multivariate analyses supported a group-level difference in the combined oxidative-stress biomarker profile, as confirmed by MANOVA and permutation testing. The Spearman heatmap showed that maternal HbA1c was most prominently associated with OSI and also correlated positively with 8-OHdG and GSH. Although neonatal complications were more frequent in the GDM group, oxidative-stress markers were not consistently associated with specific neonatal complications.

Conclusion

These findings suggest that GDM is associated with a selective fetal redox-related biomarker pattern rather than a uniform global increase in oxidative stress. The combined biomarker profile and HbA1c-related heatmap correlations indicate that long-term maternal glycemic exposure may influence fetal oxidant–antioxidant balance and oxidative DNA-damage signals, while antioxidant-related responses may remain partially preserved. The comprehensive evaluation of multiple oxidative and antioxidant biomarkers, particularly the inclusion of rarely studied AGEs and IMA, provides a broader perspective on fetal exposure to GDM and its pathophysiological mechanisms.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12884-026-09914-4.

Background

Gestational diabetes mellitus (GDM) is an endocrine disorder that develops during pregnancy, affecting approximately 14% of pregnancies worldwide, and is not present prior to conception [1]. In recent years, the prevalence of GDM has increased, bringing with it a range of maternal and neonatal complications [2]. Infants born to mothers with GDM have been reported to experience a higher incidence of adverse neonatal outcomes [3]. These newborns are more likely to be macrosomic and to sustain birth-related injuries during delivery [4]. In addition, increased rates of neonatal jaundice, hypoglycemia, and respiratory distress syndrome (RDS) have been documented [5].

The interaction between maternal and fetal development during the antenatal period is primarily mediated by the placenta. As a key determinant of the intrauterine environment, the placenta controls oxygen delivery to the fetus, facilitates the transport of nutrients, and regulates the synthesis and transfer of hormones essential for fetal growth [6]. Metabolic alterations that develop in the mother during pregnancy are reflected directly or indirectly in placental function and fetal development [7]. In GDM, maternal hyperglycemia leads to increased transplacental transfer of glucose and lipids to the fetus. This metabolic milieu is associated with structural and functional abnormalities in placental trophoblast cells, impaired angiogenic processes, enhanced cellular stress responses, and overall placental dysfunction [8, 9].

Oxidative stress refers to an imbalance between oxidant production and antioxidant defense systems within the organism. Oxidation–reduction (redox) reactions are integral to fundamental biological processes, particularly cellular energy production, which is largely dependent on oxidative respiration occurring in the mitochondria. During this process, numerous reactive oxygen species (ROS) with oxidizing properties are inevitably generated [10]. Under normal physiological conditions, a tightly regulated balance exists between ROS production and elimination, maintaining ROS at low and controlled levels [11]. At these physiological concentrations, ROS function as signaling molecules involved in key cellular processes such as proliferation, differentiation, and apoptosis. However, when ROS generation exceeds antioxidant capacity, oxidative stress ensues, leading to damage of cellular macromolecules and contributing to the development of pathological processes [12].

The placenta plays a central role in regulating the intrauterine environment by facilitating oxygen and nutrient transfer, waste removal, and the neutralization of ROS. Therefore, placental oxidative adaptation may modulate the effects of oxidative stress on the fetus [7]. Mitochondria, abundant in placental tissue, play a critical role in maintaining a healthy pregnancy [13]. Cellular energy production is largely mediated through the tricarboxylic acid cycle and oxidative phosphorylation within mitochondria, processes that inherently generate ROS as by-products [14]. In GDM, the placenta is exposed to a hyperglycemic intrauterine environment accompanied by increased inflammation, oxidative stress, disturbances in lipid metabolism, and alterations in hormonal balance. These metabolic and biochemical changes have the potential to impair normal placental development and function, thereby negatively impacting fetal well-being [15]. Current evidence suggests that these unfavorable intrauterine conditions associated with GDM disrupt fetal metabolic functions. Excessive oxidative stress can induce oxidative damage at the cellular and biochemical levels, affecting nucleic acids, proteins, lipids, and carbohydrates; however, the precise mechanisms underlying these processes remain incompletely understood and warrant further investigation [16].

In a hyperglycemic environment, alternative glucose metabolic pathways are activated. Elevated glucose levels can enhance glycolytic flux and increase the production of advanced glycation end products (AGEs) [17]. Activation of the polyol pathway leads to reduced intracellular glucose levels via sorbitol dehydrogenase–mediated conversion, resulting in the accumulation of sorbitol and fructose-3-phosphate. This, in turn, contributes to sorbitol-induced osmotic stress, reduced Na⁺/K⁺-ATPase activity, a decreased NADPH/NADP ratio, and ultimately depletion of the glutathione system, reflecting a reduction in antioxidant capacity [18]. High levels of intrauterine oxidative stress may lead to acute organ injury and complications in the neonatal period. Organs with high oxygen consumption, such as the lungs and brain, are particularly vulnerable to increased free radical exposure in the newborn [19].

In this study, to evaluate the potential effects of GDM-associated oxidative stress and antioxidant defense capacity on neonatal outcomes, levels of total antioxidant status (TAS), total oxidant status (TOS), malondialdehyde (MDA) [20], membrane attack complex (MAC), 8-hydroxy-2′-deoxyguanosine (8-OHdG) [21], reduced glutathione (GSH), glutathione peroxidase (GPx), ischemia-modified albumin (IMA), and advanced glycation end products (AGEs) were measured in umbilical cord blood. MDA and MAC levels were evaluated as indicators of lipid peroxidation and oxidative damage to cellular macromolecules, while 8-OHdG was considered a marker of oxidative DNA damage. TAS [20], GSH [22], and GPx were used to reflect fetal antioxidant defense capacity, whereas increased IMA levels indicated protein modifications associated with oxidative stress and intrauterine hypoxia. AGE levels were assessed as a biochemical reflection of intrauterine hyperglycemia and glycation burden [23]. Collectively, these biomarkers were selected to provide complementary information on distinct but interconnected aspects of fetal redox homeostasis, including lipid, protein, and DNA oxidation, overall oxidant and antioxidant status, ischemia-related protein modification, and glycation-related oxidative processes. This comprehensive biomarker panel was designed to enable a more integrated assessment of oxidative stress and antioxidant defense in cord blood from pregnancies complicated by gestational diabetes mellitus.

By analyzing these biomarkers in conjunction with clinical neonatal outcomes—such as birth weight, gestational age, Apgar scores, neonatal hypoglycemia, respiratory morbidity, and the need for neonatal intensive care— this study aims to elucidate comprehensively the impact of fetal oxidant–antioxidant balance on neonatal adaptation within the context of gestational diabetes.

Materials and methods

Study population and data collection

Umbilical cord blood samples were collected from all newborns immediately after delivery under sterile conditions. Eligible term newborns delivered by planned cesarean section were screened during the study period and enrolled consecutively until 40 newborns of mothers with diet-controlled GDM and 40 newborns of mothers without GDM were included. The final study population therefore consisted of 80 newborns.

The study group comprised newborns delivered at 37–41 weeks of gestation to mothers aged 18–45 years who had been diagnosed with GDM. The control group consisted of healthy, term newborns delivered during the same study period to mothers of a similar age range without gestational diabetes or other metabolic disorders who were consecutively recruited.

To ensure standardized timing and conditions for sample collection, all women in both groups underwent planned cesarean delivery at or after 37 weeks of gestation. GDM was defined according to standard oral glucose tolerance test (OGTT) criteria: fasting plasma glucose ≥ 92 mg/dL, plasma glucose ≥ 180 mg/dL at 1 h, or ≥ 153 mg/dL at 2 h following a 75-g OGTT. Women with at least one abnormal value were classified as having GDM [24].

Prenatal data were obtained through medical record review and maternal interviews and included maternal age, gravidity, gestational age at delivery, OGTT results, and HbA1c levels. For mothers with GDM, additional information regarding treatment modality (diet and lifestyle modification or insulin therapy) and glycemic control during pregnancy was recorded. Maternal conditions that could influence oxidative stress parameters—such as pregestational diabetes, insulin use, infections during pregnancy, medication use, premature rupture of membranes, or other obstetric complications—were screened, and such cases were excluded from the study. To maintain group homogeneity, only newborns delivered by planned cesarean section after 37 weeks of gestation to mothers with diet-controlled GDM were included.

Sample size estimation was performed using G*Power software (version 3.1.9.2). Based on an assumed medium effect size (d = 0.5) for oxidative stress biomarkers between the gestational diabetes and control groups, with a significance level of 0.05 and 95% statistical power, the minimum required sample size was calculated as 40 participants.

This prospective cross-sectional study included 80 newborns: 40 newborns delivered to healthy pregnant women without GDM and 40 newborns delivered to women diagnosed with diet-controlled GDM. The study was conducted between November 2025 and February 2026 at Balıkesir University Faculty of Medicine Hospital. Participants were consecutively selected from eligible postpartum women who provided written informed consent.

Umbilical cord blood sampling protocol

Umbilical cord blood samples were obtained from the umbilical vein under sterile conditions immediately after placental separation. Samples were transferred into vacuum tubes containing anticoagulant and promptly transported to the laboratory, where they were centrifuged at 3000 rpm for 10 min at 4 °C. The separated plasma was aliquoted into single-use tubes and stored at − 80 °C until analysis. All procedures were performed in accordance with standardized protocols designed to preserve the stability of oxidative stress biomarkers.

Neonatal data and complications

Collected neonatal data included birth weight, sex, and 1- and 5-minute Apgar scores. In addition, neonatal blood glucose levels measured 30 min after birth and umbilical cord hemoglobin (Hb) values were recorded. Postnatal complications—including hypoglycemia, respiratory distress, need for neonatal intensive care, feeding difficulties, and other clinical conditions—were evaluated and documented based on clinical records.

Enzyme-linked Immunosorbent Assay (ELISA) analysis

The concentrations of serum oxidative-stress parameters AGEs, MDA, GSH, GPx, 8-OHdG, IMA, and MAC were determined using commercially available ELISA kits (Sunred Biological Technology Co., Shanghai, China) according to the manufacturer’s protocols. The intra- and inter-assay coefficients of variation for all kits were < 10% and < 12%, respectively.

The sensitivities and assay ranges of the kits were 22.25 ng/mL and 30-9000 ng/mL for AGEs (Cat. No: 201-12-0004), 0.515 nmol/mL and 0.75–100 nmol/mL for MDA (Cat. No: 201-12-1372), 0.125 Umol/L and 0.2–30 Umol/L for GSH (Cat. No: 201-12-1463), 0.714 ng/mL and 0.8–200 ng/mL for GSH-Px (Cat. No: 201-12-0726), 0.558 ng/mL and 1-100 ng/mL for 8-OH-dG (Cat. No: 201-12-1437), 2.226 ng/mL and 3-600 ng/mL for IMA (Cat. No: 201-12-1173) and 9.882 ng/mL and 10-3000 ng/mL for MAC (Cat. No: 201-12-0360).

The samples were tested without dilution. Absorbance was measured at 450 nm using a microplate reader (Varioskan Flash Multimode Reader, Thermo Scientific, Waltham, MA, USA). Quantification of parameters in serum samples was performed from a calibration curve generated through a four-parameter logistic (4PL) regression model.

Measurements of TAS and TOS levels

TAS and TOS levels were measured using colorimetric assay kits (Rel Assay Diagnostics, Mega Tıp, Gaziantep, Türkiye). TAS levels were expressed as mmol Trolox equivalent/L, whereas TOS levels were expressed as µmol H₂O₂ equivalent/L. The oxidative stress index (OSI) was calculated using the formula (TOS/TAS) × 100 after unit conversion.

Statistical analysis

Statistical analyses were performed using SPSS version 25 and R software version 4.5.3 (R Foundation for Statistical Computing, Vienna, Austria). Continuous variables were assessed for distributional characteristics, and nonparametric methods were used when normality assumptions were not met. Continuous variables are presented as mean ± standard deviation and median values with minimum–maximum or interquartile range, as appropriate. Between-group comparisons of continuous variables were performed using the Mann–Whitney U test. Categorical variables were compared using Pearson’s chi-square test or Fisher’s exact test, as appropriate, and results are reported as column percentages. A two-sided p value < 0.05 was considered statistically significant. For oxidative-stress biomarker comparisons, Benjamini–Hochberg false discovery rate (BH-FDR) correction was applied to reduce the risk of type I error due to multiple testing, and BH-FDR-adjusted p values were used for interpretation of biomarker significance.

Spearman rank correlation analysis was used to evaluate monotonic relationships between clinical/neonatal variables and oxidative-stress biomarkers. Correlation coefficients (Spearman r) were visualized using a heatmap to summarize the direction and relative strength of associations. Because multiple correlations were evaluated, the heatmap findings were interpreted descriptively and were not considered confirmatory outcome analyses. Factors associated with GDM status were first evaluated using conventional multivariate logistic regression based on clinically relevant variables and variables identified in univariate analyses. Regression results are presented as regression coefficients (B), odds ratios (OR = Exp[B]), 95% confidence intervals, and p values.

Exploratory multivariate analyses were conducted to evaluate the combined oxidative-stress biomarker profile. Principal component analysis (PCA) was used as an unsupervised method to explore overall biomarker patterns and was performed using the FactoMineR package. Partial least squares discriminant analysis (PLS-DA) was applied as an exploratory supervised method to assess group discrimination, with model performance evaluated using 5-fold cross-validation; the mixOmics package was used for this analysis. MANOVA was used to test whether the overall oxidative-stress biomarker profile differed between groups, and the robustness of the group effect was assessed using a permutation-based validation test with 5000 permutations; MANOVA and permutation testing were performed using the vegan package. Finally, Elastic Net logistic regression with 5-fold stratified cross-validation was used as an exploratory penalized model to evaluate the combined discriminative contribution of clinical and biomarker variables while accounting for correlated predictors; this analysis was performed using the glmnet package. Spearman heatmap visualization was generated using the heatmaply package, and additional visualizations were prepared using the matplotlib library in Python. All multivariate prediction models were interpreted as exploratory and requiring validation in independent cohorts.

Results

Continuous variables were compared according to GDM status using the Mann–Whitney U test. Age distribution was similar between groups, with a mean age of 28.77 ± 5.62 years (median 28; range 19–39) in the non-GDM group and 29.62 ± 5.06 years (median 30; range 21–40) in the GDM group; this difference was not statistically significant (U = 738.5; p = 0.275). Similarly, HbA1c levels did not differ significantly, with values of 5.02 ± 0.37% (median 5; range 4–6) in the non-GDM group and 5.33 ± 0.52% (median 5; range 5–6) in the GDM group (U = 94.0; p = 0.393). No significant differences were observed in gestational age between groups: 37.99 ± 1.36 weeks (median 38.29; range 32.00–40.57) in the non-GDM group versus 37.88 ± 1.18 weeks (median 38.14; range 35.29–40.14) in the GDM group (U = 768.0; p = 0.411).

When metabolic parameters were examined, postnatal 30-minute blood glucose levels were significantly higher in the non-GDM group. The mean glucose was 82.58 ± 12.75 mg/dL (median 81; range 61–118) in the GDM group compared to 98.26 ± 18.25 mg/dL (median 98; range 63–155) in the non-GDM group, a difference that was statistically significant (U = 381.5; p < 0.001). In contrast, birth weights were similar between groups, with 3169.66 ± 327.04 g (median 3150; range 2500–3650) in the non-GDM group and 3180.51 ± 465.68 g (median 3150; range 2330–4300) in the GDM group (U = 816.5; p = 0.705). Apgar scores also did not differ significantly: 1-minute Apgar scores were 7.80 ± 0.76 (median 8; range 4–9) versus 7.72 ± 0.65 (median 8; range 6–9) (U = 782.0; p = 0.390), and 5-minute Apgar scores were 8.93 ± 0.55 (median 9; range 7–10) versus 8.87 ± 0.70 (median 9; range 7–10) (U = 813.0; p = 0.616) (Table 1).

Table 1.

Comparison of Age and Biochemical Continuous Variables According to GDM Status

Variables	Gestational Diabetes		MWU	P*
	No	Yes
	Median (Min–Max)	Median (Min–Max)
Age (years)	28 (19–39)	30 (21–40)	738,5	0,275
HbA1c(%)	5 (4–6)	5 (5–6)	94,0	0,393
Gestational age (weeks)	38,29 (32,00–40,57)	38,14 (35,29–40,14)	768,0	0,411
30-min blood glucose (mg/dL)	98 (63–155)	81 (61–118)	366.0	< 0,001
Birth weight (gr)	3150 (2500–3650)	3150 (2330–4300)	816,5	0,705
Apgar 1 min	8 (4–9)	8 (6–9)	782,0	0,390
Apgar 5 min	9 (7–10)	9 (7–10)	813,0	0,616
Hemoglobin (g/dL)	18 (10–21)	19 (12–22)	392,0	0,009

* Mann–Whitney U test

Among laboratory parameters, neonatal Hb levels were higher in the GDM group. Mean Hb was 17.86 ± 2.66 g/dL (median 18; range 10–21) in the non-GDM group and 19.27 ± 2.31 g/dL (median 19; range 12–22) in the GDM group, and this difference was statistically significant (U = 392.0; p = 0.009) (Table 1).

Oxidative stress markers

Evaluation of oxidative stress markers showed that MAC, 8-OHdG, and TOS differed significantly between the GDM and non-GDM groups after BH-FDR correction for multiple biomarker comparisons (Table 2; Fig. 1). MAC and 8-OHdG levels were higher in the GDM group, whereas TOS levels were lower in the GDM group. No statistically significant between-group differences were observed for MDA, AGEs, GPx, GSH, IMA, or TAS after correction (Table 2).

Table 2.

Comparison of Oxidative Stress Markers According to GDM Status

Variable	GDM Mean ± SD	GDM Median (IQR)	No-GDM Mean ± SD	No-GDM Median (IQR)	MW p	BH-FDR p
MAC	507.13 ± 563.41	248.71 (181.41–447.07)	212.08 ± 77.43	186.21 (153.18–260.66)	0,0072	0,0382
MDA	30.29 ± 30.49	14.79 (12.05–28.29)	18.61 ± 9.75	15.06 (12.51–18.93)	0,6179	0,6866
8-OHdG	29.48 ± 30.88	13.96 (11.21–29.26)	14.55 ± 8.34	11.50 (9.95–15.12)	0,0115	0,0382
AGEs	2288.78 ± 2260.71	1178.06 (1007.35–2328.61)	1404.54 ± 766.04	1138.09 (925.10–1472.08)	0,1076	0,2045
GPx	57.07 ± 62.55	24.78 (20.50–56.56)	29.55 ± 14.71	24.91 (20.33–28.92)	0,2844	0,4063
GSH	9.37 ± 8.74	4.87 (4.15–8.67)	6.03 ± 3.06	4.67 (4.20–6.53)	0,4676	0,5845
IMA	133.43 ± 148.48	61.88 (49.75–124.62)	72.57 ± 46.01	55.64 (48.25–71.22)	0,1227	0,2045
TAS	8.20 ± 8.16	4.49 (3.29–10.19)	4.99 ± 1.71	4.54 (4.13–5.71)	0,7950	0,7950
TOS	1.24 ± 0.43	1.26 (1.06–1.45)	1.46 ± 0.32	1.40 (1.23–1.63)	0,0102	0,0382
OSI	0.70 ± 0.59	0.43 (0.25–0.92)	0.36 ± 0.15	0.32 (0.26–0.42)	0,0318	0,0794

Abbreviations: GDM Gestational diabetes mellitus, SD Standard deviation, IQR Interquartile range, MW Mann–Whitney U test, BH-FDR Benjamini–Hochberg false discovery rate

Only biomarker variables were included in the BH-FDR adjustment

Statistical significance was interpreted using a two-sided p value threshold of 0.05; for biomarker comparisons, BH-FDR-adjusted p values were used to account for multiple testing

Fig. 1.

Biomarker significance after multiple-testing correction. Bars show −log10 (BH-FDR-adjusted p). Red bars indicate biomarkers remaining significant after FDR correction (MAC, 8-OHdG, TOS)

OSI values were higher in the GDM group in the Mann–Whitney analysis; however, this association did not remain statistically significant after BH-FDR correction (Fig. 1). Together, these results suggest that GDM is associated with an altered fetal oxidative-stress biomarker profile, characterized by higher MAC and 8-OHdG levels, lower TOS levels, and a trend toward higher OSI, rather than a uniform increase across all oxidative-stress markers.

Principal Component Analysis (PCA) of oxidative stress markers

Principal component analysis (PCA) was performed as an exploratory multivariate analysis to evaluate the overall oxidative-stress biomarker profile (Fig. 2). The first two principal components explained 65.7% of the total variance, with PC1 accounting for 51.5% and PC2 for 14.1%. PC1 was mainly characterized by positive loadings from 8-OHdG, GPx, MDA, GSH, MAC, OSI, and IMA, suggesting a shared oxidative-stress pattern. PC2 was primarily influenced by TAS, TOS, and OSI, reflecting variation in oxidant–antioxidant balance. These findings support the presence of a broader biomarker profile difference between the GDM and non-GDM groups (Fig. 2).

Fig. 2.

PLS-DA variable weights. Positive weights favor classification as GDM; negative weights favor non-GDM. PLS-DA showed moderate discrimination (5-fold CV AUC ≈ 0.795), but should be considered exploratory

Partial Least Squares Discriminant Analysis (PLS-DA) of oxidative stress markers

PLS-DA was performed as an exploratory supervised multivariate analysis to assess whether the combined oxidative-stress biomarker profile could discriminate between the GDM and non-GDM groups (Fig. 3). A one-component model provided the best cross-validated performance, with a 5-fold cross-validation AUC of 0.795, indicating moderate discriminatory ability. The variables contributing most strongly to group discrimination were OSI, MAC, 8-OHdG, GPx, and TOS. Positive variable weights indicated contribution toward GDM classification, whereas negative weights indicated contribution toward non-GDM classification. These findings suggest that the combined biomarker pattern may help distinguish GDM status; however, because of the modest sample size and supervised nature of the method, the PLS-DA results should be interpreted as exploratory and require validation in an independent cohort (Fig. 3).

Fig. 3.

PCA loading plot. PC1 captures the shared oxidative-stress biomarker pattern, while PC2 separates TAS/TOS-related variation. PCA is unsupervised and should be interpreted as pattern exploration

MANOVA was performed as an exploratory multivariate test to determine whether the overall oxidative-stress biomarker profile differed between the GDM and non-GDM groups (Table 3). The model included 10 oxidative-stress and antioxidant biomarkers and demonstrated a significant group effect using both Wilks’ lambda and Pillai’s trace criteria (Wilks’ λ = 0.601; Pillai’s trace = 0.399; F = 4.116; p = 0.0002). A permutation-based validation test with 5000 permutations confirmed the robustness of this group difference (p = 0.0002). These findings indicate that, although not all individual biomarkers differed after correction, the combined fetal oxidative-stress biomarker profile was significantly different between groups (Table 3).

Table 3.

Multivariate analysis of oxidative-stress biomarker profile

Analysis / test	Statistic	Value	df/validation	p value
MANOVA	Wilks’ lambda	0,601	10, 62	0,0002
MANOVA	Pillai’s trace	0,399	10, 62	0,0002
Permutation test	Permutation p	—	5000 permutations	0,0002
PLS-DA	5-fold CV AUC	0,795	1 component	—

Neonatal complications according to gestational diabetes status

When neonatal complications were evaluated according to maternal GDM status, at least one complication was observed in 35% of newborns delivered to mothers without GDM and in 70% of newborns delivered to mothers with GDM. This difference was statistically significant, indicating an association between the presence of gestational diabetes and the occurrence of neonatal complications (χ² = 9.269; p = 0.002) (Table 4).

Table 4.

Comparison of Neonatal Complications With and Without Gestational Diabetes

Presence of Complications	GDM (-)	GDM (+)	χ²	P*
	n (%)	n (%)
No	26 (65,0)	12 (30,0)	9,269	0,002
Yes	14 (35,0)	28 (70,0)
Total	40 (100,0)	40 (100,0)

*Percentages are column percentages. Pearson chi-square test was used

Frequency of specific neonatal complications according to GDM Status

The incidence of neonatal hypoglycemia was significantly higher in newborns of mothers with GDM compared to those of mothers without GDM (p = 0.043) (Table 5). Similarly, RDS was significantly more frequent in the GDM group (p = 0.003) (Table 5). All cases were managed according to our institutional neonatal hypoglycemia protocol, and no infant required intensive care admission or experienced persistent hypoglycemia requiring prolonged treatment.

Table 5.

Comparison of Neonatal Complications According to Gestational Diabetes Status

Variables	Category	GDM (+)	GDM (–)	P*
		n (%)	n (%)
Hypoglycemia	Yes	8 (20,0)	2 (5,0)	0,043
	No	32 (80,0)	38 (95,0)
HIE	Yes	6 (15,0)	2 (5,0)	0,263
	No	34 (85,0)	38 (95,0)
Hypocalcemia	Yes	2 (5,0)	4 (10,0)	0,675
	No	38 (95,0)	36 (90,0)
Hypomagnesemia	Yes	0 (0,0)	0 (0,0)	—
	No	40 (100,0)	40 (100,0)
Fetal heart defect	Yes	0 (0,0)	0 (0,0)	—
	No	40 (100,0)	40 (100,0)
IUGR	Yes	0 (0,0)	1 (2,5)	1,000
	No	40 (100,0)	39 (97,5)
Transient neonatal tachypnea	Yes	9 (22,5)	6 (15,0)	0,390
	No	31 (77,5)	34 (85,0)
RDS	Yes	10 (25,0)	1 (2,5)	0,003
	No	30 (75,0)	39 (97,5)
NEC	Yes	2 (5,0)	1 (2,5)	1,000
	No	38 (95,0)	39 (97,5)
Jaundice	Yes	7 (17,5)	1 (2,5)	0,057
	No	33 (82,5)	39 (97,5)

Percentages are column percentages. Statistical analysis was not performed for variables with no observed events in either group

* Pearson chi-square or Fisher’s exact test was used

HIE Hypoxic-ischemic encephalopathy, IUGR Intrauterine growth restriction, RDS Respiratory distress syndrome, NEC Necrotizing enterocolitis

No statistically significant differences were observed between groups for hypoxic-ischemic encephalopathy (HIE), hypocalcemia, intrauterine growth restriction (IUGR), transient tachypnea of the newborn, necrotizing enterocolitis (NEC), or jaundice (p = 0.263, 0.675, 1.000, 0.390, 1.000, and 0.057, respectively). Hypomagnesemia and fetal heart defects were not observed in either group, and therefore statistical comparisons were not performed for these variables (Table 5).

Comparison of oxidative stress parameters according to hypoglycemia and RDS in newborns of mothers with GDM

In newborns of mothers with gestational diabetes, oxidative stress parameters were compared according to the presence of hypoglycemia and RDS (Table 6).

Table 6.

Comparison of Oxidative Stress Parameters According to Hypoglycemia and RDS in the GDM group

Variables	Hypoglycemia		MWU	P*
	No (n = 32)	Yes (n = 8)
Median (Min–Max)	Median (Min–Max)
MAC	243,66 (76,61-2149,01)	248,15 (144,79-2112,88)	113,0	0,612
TOS	1,27 (0,31 − 2,26)	1,25 (0,41 − 1,33)	99,0	0,327
OSI	0,43 (0,18 − 2,27)	0,51 (0,21 − 1,57)	120,0	0,787
Variables	RDS		MWU	P*
	No (n = 30)	Yes (n = 10)
	Median (Min–Max)	Median (Min–Max)
MAC	262,52 (76,61-2149,01)	1193,05 (134,75-1572,53)	126,0	0,453
TOS	1,23 (0,31 − 2,26)	1,37 (0,78 − 2,10)	119,0	0,333
OSI	0,55 (0,18 − 2,27)	0,29 (0,22 − 2,15)	118,0	0,318

* Mann–Whitney U test

For hypoglycemia, median values of MAC, TOS, and OSI were 243.66 (76.61–2149.01), 1.27 (0.31–2.26), and 0.43 (0.18–2.27), respectively, in infants without hypoglycemia, while in infants with hypoglycemia, the corresponding values were 248.15 (144.79–2112.88), 1.25 (0.41–1.33), and 0.51 (0.21–1.57). No statistically significant differences were observed between groups for MAC (p = 0.612), TOS (p = 0.327), or OSI (p = 0.787) (Table 6).

Regarding RDS, infants without RDS had median MAC, TOS, and OSI values of 262.52 (76.61–2149.01), 1.23 (0.31–2.26), and 0.55 (0.18–2.27), respectively, whereas infants who developed RDS had values of 1193.05 (134.75–1572.53), 1.37 (0.78–2.10), and 0.29 (0.22–2.15). However, no statistically significant differences were detected for MAC (p = 0.453), TOS (p = 0.333), or OSI (p = 0.318) according to RDS status (Table 6).

Correlations between clinical/neonatal variables and oxidative stress markers

Spearman correlation analysis was performed to evaluate the relationships between clinical/neonatal variables and oxidative-stress biomarkers, and the results were summarized visually in a heatmap (Fig. 4). Overall, the correlation pattern was limited, with most clinical and neonatal variables showing weak or inconsistent associations with individual oxidative-stress markers. The most prominent positive correlation was observed between maternal HbA1c and OSI, supporting a relationship between long-term maternal glycemic control and fetal oxidant–antioxidant balance.

Fig. 4.

Spearman Correlation Heatmap Between Clinical/Neonatal Variables and Oxidative Stress Markers. Notes: Spearman correlation heatmap showing associations between clinical variables and oxidative-stress biomarkers. Values represent Spearman r coefficients. The strongest positive correlations in this table were observed for HbA1c with OSI, 8-OHdG, and GSH, and for Apgar 1 min with MDA, 8-OHdG, GPx, and GSH

Maternal HbA1c also showed positive correlations with 8-OHdG and GSH in the heatmap, suggesting that greater maternal glycemic exposure may be accompanied by higher fetal DNA oxidative-damage signal and an antioxidant-related response. In addition, the 1-minute Apgar score showed positive correlations with MDA, 8-OHdG, GPx, and GSH; however, these findings should be interpreted cautiously because Apgar scores were generally clustered within a narrow clinical range (Fig. 4).

No consistent correlation pattern was observed for birth weight, gestational age, 5-minute Apgar score, 30-minute postnatal blood glucose, or hemoglobin in relation to the oxidative-stress biomarkers. Thus, the Spearman heatmap mainly supports an association between maternal glycemic control and selected fetal oxidative-stress markers, rather than a broad relationship between early neonatal clinical parameters and oxidative-stress biomarkers.

In conventional multivariate logistic regression, lower 30-minute postnatal blood glucose and higher MAC levels were independently associated with GDM status, whereas hemoglobin, TOS, and OSI were not independently associated with GDM (Supplementary Table 1).

Because the oxidative-stress biomarkers were interrelated and the sample size was modest, we next performed an exploratory Elastic Net logistic regression analysis to assess whether a penalized multivariable approach could capture the combined clinical and biomarker pattern associated with GDM status.

Elastic net logistic regression analysis for discrimination of GDM status

Elastic Net logistic regression was then performed as an exploratory penalized model to evaluate whether the combined clinical and oxidative-stress biomarker profile could discriminate GDM status while accounting for correlated predictors. Using 5-fold stratified cross-validation, the model showed moderate discriminatory performance, with a cross-validated AUC of 0.800, accuracy of 0.734, sensitivity of 61.5%, and specificity of 85.0%. Predictors retained with non-zero coefficients included 30-minute postnatal blood glucose, OSI, MAC, TOS, GPx, AGEs, 8-OHdG, TAS, MDA, and IMA, whereas age, gestational week, HbA1c, birth weight, GSH, and sex were shrunk to zero (Table 7). These findings suggest that a penalized multivariable approach may capture an exploratory biomarker pattern associated with GDM status; however, because of the modest sample size, this model requires validation in an independent cohort before clinical interpretation (Table 7).

Table 7.

Predictors retained in the Elastic Net logistic regression model for GDM status

Retained predictors with non-zero Elastic Net coefficients
Predictor	Coefficient	OR per 1-SD increase	Direction of association
30-minute postnatal blood glucose	−0.288	0.750	Lower values associated with GDM
OSI	0.196	1.217	Higher values associated with GDM
MAC	0.116	1.123	Higher values associated with GDM
TOS	−0.116	0.891	Lower values associated with GDM
GPx	0.109	1.115	Higher values associated with GDM
AGEs	0.087	1.091	Higher values associated with GDM
8-OHdG	0.056	1.057	Higher values associated with GDM
TAS	0.019	1.019	Higher values associated with GDM
MDA	0.008	1.008	Higher values associated with GDM
IMA	0.001	1.001	Higher values associated with GDM

Coefficients and ORs are based on standardized predictors; ORs therefore represent the change in odds per 1-SD increase in each predictor. Model performance metrics, including cross-validated AUC, accuracy, sensitivity, and specificity, are reported in the Results text

The Elastic Net model should be interpreted as exploratory because of the modest sample size

Abbreviations: GDM Gestational diabetes mellitus, OR Odds ratio, SD Standard deviation

Discussion

This study evaluated fetal oxidative-stress and antioxidant-defense biomarkers in umbilical cord blood from pregnancies with and without GDM and examined their relationship with early neonatal outcomes. The main finding was that GDM was associated with a selective fetal redox-related biomarker pattern rather than a uniform increase in oxidative stress. This pattern was characterized by evidence of increased oxidative damage markers together with relatively preserved antioxidant-related markers. Although neonatal complications were more frequent among newborns of mothers with GDM, oxidative-stress markers were not consistently associated with specific early neonatal complications, suggesting that these biomarkers may better reflect intrauterine redox adaptation than immediate clinical morbidity.

The source and clinical significance of oxidative stress in GDM remain heterogeneous in the literature. Sarıkabadayı et al. proposed that increased oxidative activity in diabetic mothers and their newborns arises primarily from impaired glycemic control rather than inadequate antioxidant defense [25]. Similarly, studies evaluating maternal milk, maternal blood, cord blood, and neonatal outcomes suggest that compensatory antioxidant responses and fetal adaptation may modify the impact of maternal oxidative stress on the newborn [26, 27]. In line with these observations, the present findings support the possibility that fetal redox responses in GDM are influenced not only by maternal hyperglycemia but also by placental buffering, fetal antioxidant adaptation, and biomarker-specific dynamics. The HbA1c-related correlation pattern, particularly its association with OSI and additional links with 8-OHdG and GSH, further supports the biological relevance of long-term maternal glycemic exposure for fetal oxidant–antioxidant balance and oxidative DNA-damage signaling [28, 29]. In addition to the observed differences in oxidative stress biomarkers, early neonatal complications, particularly neonatal hypoglycemia, were more frequent in the GDM group and were consistent with the significantly lower 30-minute postnatal blood glucose levels observed in these newborns [30]. This finding is in accordance with current evidence indicating that infants born to mothers with GDM are at increased risk of transient neonatal hypoglycemia because persistent fetal hyperinsulinism may continue after interruption of the maternal glucose supply at birth [31]. Nevertheless, no consistent correlations were observed between postnatal blood glucose levels and oxidative stress biomarkers in the present study. The absence of a clear association may be explained by the relatively mild maternal hyperglycemia in our diet-controlled GDM cohort and the homogeneous characteristics of the study population, which may have attenuated both glycemic variability and oxidative stress responses.

Among the oxidative damage markers, 8-OHdG appears particularly relevant because it reflects DNA oxidative damage. In the present cohort, the higher 8-OHdG levels in the GDM group suggest that fetal DNA oxidative-damage signaling may be more evident than lipid peroxidation, as MDA did not differ significantly between groups. This interpretation is supported by previous reports describing increased oxidative DNA damage in women with GDM [32], although the literature remains heterogeneous because of differences in biological sample type, assay methods, timing of sampling, and degree of glycemic control [33–35]. Therefore, the absence of a significant MDA difference should not be interpreted as absence of oxidative stress, but rather as evidence that different oxidative-stress pathways may be affected unequally in the fetal circulation.

Protein and macromolecular oxidation products are frequently reported as increased in GDM and may reflect exposure to a hyperglycemia-related oxidative environment. Previous studies have shown higher maternal levels of protein carbonyls and advanced oxidation protein products in GDM, supporting the presence of maternal systemic oxidative stress [36]. In the present study, MAC was significantly higher in the GDM group after BH-FDR correction and was also independently associated with GDM status in the conventional multivariate logistic regression model. These findings suggest that MAC may be a relevant marker of fetal macromolecular oxidative damage in the setting of GDM. However, MAC did not differ according to neonatal hypoglycemia or RDS status within the GDM group, indicating that increased MAC may reflect the intrauterine redox environment more than specific early neonatal complications. This interpretation is consistent with reports that cord blood oxidative-damage markers may not directly correlate with short-term neonatal outcomes, possibly because placental buffering and fetal antioxidant adaptation modify the clinical expression of oxidative stress [37]. Therefore, MAC should be interpreted as a marker contributing to the GDM-related fetal redox profile rather than as a standalone predictor of neonatal morbidity.

AGEs are long-term biochemical products of glycation and oxidative reactions and have been implicated in diabetes-related vascular and inflammatory complications [38, 39]. Maternal AGE levels have been reported to increase in GDM and to correlate with glucose intolerance, inflammation, oxidative stress, and adverse perinatal outcomes [40–42]. In the present study, however, cord blood AGE levels did not differ significantly between the GDM and non-GDM groups after BH-FDR correction. This discrepancy may reflect differences between maternal and fetal compartments, placental regulation of AGE transfer, relatively good glycemic control in this diet-controlled GDM cohort, and fetal antioxidant adaptation. Therefore, the absence of a significant cord blood AGE difference does not contradict the maternal GDM literature but suggests that fetal AGE accumulation may be limited or more variable under the conditions studied.

IMA is generated when the metal-binding region of albumin is modified under oxidative and ischemic conditions, and it is commonly considered a marker of tissue hypoxia, ischemia, and reperfusion-related oxidative damage [43]. In pregnancy-related disorders, IMA may increase particularly when placental hypoperfusion or dominant hypoxic stress is present, as reported in preeclampsia [44]. In the present cohort, IMA levels did not differ significantly between the GDM and non-GDM groups after correction, suggesting that a strong fetal hypoxic-ischemic signal was not a dominant feature of this diet-controlled GDM population. This interpretation is also consistent with the lack of consistent associations between oxidative-stress markers and specific early neonatal complications.

The antioxidant-related findings should also be interpreted in the context of fetal adaptive capacity. GSH, GPx, and TAS did not differ significantly between groups after correction, indicating that fetal antioxidant defense was not uniformly suppressed in GDM. This is compatible with reports showing preserved or variable antioxidant enzyme activity in cord blood, likely reflecting acute and dynamic responses influenced by sampling conditions, fetal metabolic adaptation, and placental transfer mechanisms [37, 45]. In this context, the lower TOS level observed in the GDM group and the loss of statistical significance for OSI after BH-FDR correction should be interpreted cautiously, as global oxidant indices may not fully capture the complexity of fetal redox alterations in GDM. Overall, these findings support a model in which selective oxidative damage biomarkers, such as MAC and 8-OHdG, are increased, whereas antioxidant-related responses remain relatively preserved or compensatory, suggesting selective rather than generalized fetal redox alterations.

Catalase (CAT), superoxide dismutase (SOD), and glutathione peroxidase (GPx) constitute the principal enzymatic antioxidant defense system against reactive oxygen species. Previous studies investigating these enzymes in GDM have reported inconsistent findings, which may reflect differences in glycemic control, disease severity, biological specimens, and analytical methods. A recent systematic review and meta-analysis demonstrated that while maternal SOD activity is reduced in GDM, SOD activity in umbilical cord blood does not differ significantly from controls, suggesting that fetal antioxidant defenses may be relatively preserved despite maternal oxidative stress [37]. Experimental studies have also demonstrated altered CAT and SOD activity in endothelial cells isolated from umbilical cords of women with GDM, indicating that oxidative alterations may occur at the cellular level even when circulating antioxidant capacity appears maintained [46]. Interestingly, earlier physiological studies in healthy newborns demonstrated that erythrocyte SOD activity in umbilical cord blood is maintained at levels comparable to adults, whereas CAT and GPx activities are significantly lower. These findings suggest that preservation of SOD activity may represent an essential adaptive mechanism protecting fetal erythrocytes against superoxide radicals during intrauterine life, while lower CAT and GPx activities may still be sufficient for detoxification of hydrogen peroxide under physiological conditions [47]. This developmental pattern may partly explain why fetal antioxidant defense may remain relatively preserved despite selective alterations in oxidative stress biomarkers. Although CAT and SOD activities were not assessed in the present study, the absence of significant differences in TAS, GSH, and GPx between groups supports the possibility that fetal antioxidant defense is maintained, at least in part, in well-controlled GDM pregnancies.

Dynamic thiol-disulfide homeostasis has recently emerged as a sensitive biomarker reflecting oxidative stress and redox balance. Unlike traditional oxidative stress markers, this system provides a dynamic assessment of the reversible oxidation of thiol groups into disulfide bonds. Several studies have evaluated thiol-disulfide homeostasis in gestational diabetes mellitus. Aktun et al. demonstrated significantly increased disulfide levels and disulfide/native thiol ratios, together with decreased native thiol/total thiol ratios, in umbilical cord blood from pregnancies complicated by GDM, indicating increased fetal oxidative stress [48]. Similarly, Özler et al. reported a shift of thiol-disulfide homeostasis toward oxidation in cord blood of neonates born to mothers with GDM or obesity, with these alterations being associated with adverse neonatal outcomes [49]. In maternal serum, Çakina et al. also demonstrated impaired thiol-disulfide homeostasis in women with GDM, further supporting the role of oxidative stress in disease pathophysiology [50]. More recently, Erdoğan et al. evaluated thiol and disulfide levels together with total antioxidant capacity and total oxidant status in maternal milk and umbilical cord blood, emphasizing the usefulness of combined oxidative stress biomarkers in pregnancies complicated by GDM [26]. Although thiol-disulfide homeostasis was not evaluated in the present study, incorporation of this dynamic redox system into future investigations may provide a more comprehensive assessment of fetal oxidative status and antioxidant defense.

Oxidative stress and inflammation are closely interconnected biological processes in the pathophysiology of gestational diabetes mellitus. Excessive production of reactive oxygen species activates inflammatory signaling pathways, particularly nuclear factor-kappa B (NF-κB), leading to increased production of pro-inflammatory cytokines such as tumor necrosis factor-alpha (TNF-α), interleukin-6 (IL-6), and C-reactive protein (CRP). In turn, these inflammatory mediators further enhance reactive oxygen species generation, creating a self-perpetuating cycle that contributes to insulin resistance, endothelial dysfunction, and placental abnormalities [51]. Recent evidence has demonstrated that women with GDM exhibit higher circulating levels of IL-6, TNF-α, and CRP together with increased oxidative stress biomarkers and altered antioxidant defenses, supporting the close interaction between inflammatory and oxidative pathways [52]. Therefore, assessment of inflammatory biomarkers alongside oxidative stress parameters may provide a more comprehensive understanding of the mechanisms underlying fetal oxidative adaptation and neonatal outcomes in GDM. Future prospective studies integrating both inflammatory and oxidative biomarkers are warranted to further clarify their combined clinical significance.

The exploratory multivariate analyses were included to complement the single-marker comparisons and to evaluate the biomarker profile as a whole. PCA, PLS-DA, MANOVA with permutation testing, and Elastic Net regression all supported the concept that GDM is associated with a combined fetal redox-related pattern. However, these analyses should be interpreted as exploratory rather than diagnostic, especially because supervised models may be sensitive to overfitting in modest sample sizes. External validation in larger independent cohorts is required before any clinical application of these multivariate models can be considered.

Conclusions and limitations

In this study, GDM was associated with a selective fetal redox-related biomarker profile in umbilical cord blood rather than a uniform global increase in oxidative stress. After BH-FDR correction, MAC, 8-OHdG, and TOS differed significantly between the GDM and non-GDM groups; MAC and 8-OHdG were higher in the GDM group, whereas TOS was lower. OSI was higher in the GDM group in the Mann–Whitney analysis but did not remain significant after correction. The Spearman correlation heatmap further supported the relevance of maternal glycemic exposure by showing that HbA1c was most prominently associated with OSI and also positively associated with 8-OHdG and GSH. Overall, these findings suggest that cord blood oxidative-stress biomarkers in diet-controlled GDM may be more informative for understanding fetal redox adaptation and potential long-term metabolic programming than for predicting specific early neonatal complications.

Several limitations should be considered when interpreting these findings. First, the sample size was modest, and several neonatal complications occurred infrequently; therefore, the study may have been underpowered to detect associations between individual oxidative-stress markers and specific neonatal outcomes. Second, the single-center and cross-sectional design limits generalizability and precludes causal inference. Third, the study population was intentionally homogeneous, including only term newborns delivered by planned cesarean section. Although this approach reduced potential confounding from labor-related oxidative stress, it may also have selected a clinically more stable neonatal population with fewer severe complications. Fourth, only diet-controlled GDM pregnancies were included, and women requiring insulin or metformin therapy were excluded; thus, the findings may not apply to more severe or poorly controlled GDM populations. Fifth, several potential modifiers of oxidative stress, including maternal nutrition, environmental exposures, inflammatory status, and genetic predisposition, were not available for analysis. Finally, although BH-FDR correction was applied for biomarker comparisons, the PCA, PLS-DA, MANOVA, Spearman heatmap, and Elastic Net analyses were exploratory and require confirmation in larger independent cohorts. Future prospective, multicenter studies with long-term follow-up are needed to determine whether fetal redox-related biomarker patterns in GDM are associated with later metabolic, respiratory, or neurodevelopmental outcomes.

Abbreviations

GDM

Gestational Diabetes Mellitus

MDA

Malondialdehyde

MAC

Membrane Attack Complex

TAS

Total Antioxidant Status

TOS

Total Oxidant Status

OSI

Oxidative Stress Index

IMA

Ischemia-Modified Albumin

8-OHdG

8-Hydroxy-2′-Deoxyguanosine

AGEs

Advanced Glycation End Products

GSH

Reduced Glutathione

GPx

Glutathione Peroxidase

ROS

Reactive Oxygen Species

Authors’ contributions

Conceptualization, OKA, FEC, MAY; methodology, OKA, MAY; software, OKA; validation, OKA, FEC, MAY, AOK, formal analysis, OKA, data curation, OKA, FEC, MIT, MOC. All authors have read and agreed to the published version of the manuscript.

Funding

This study was supported by the Balıkesir University Scientific Research Projects Coordination Unit (BAP) (Project No: 2025/083).

Data availability

The data supporting the findings of this study are not publicly available due to ethical and legal restrictions. However, the data may be available from the corresponding author upon reasonable request and with permission of the Balıkesir University Ethics Committee.

Declarations

Ethics approval and consent to participate

Ethical approval was obtained from the Balıkesir University Non-Interventional Health Sciences Ethics Committee (Approval No: 2025/300; Date: 16 September 2025). All procedures were conducted in accordance with the principles of the Declaration of Helsinki. The authors confirm that all methods were carried out in accordance with relevant guidelines and regulations along with ethical approval statement and informed consent to participation.

Competing interests

The authors declare no competing interests.