At 24–48 gestational weeks, each participant received an oral glucose tolerance test (OGTT). In brief, after fasting for 10 h or more, the women received the 75-g OGTT. GDM was classified according to the World Health Organization (WHO) [17] endorsed International Association of Diabetes and Pregnancy Study Groups [18] (fasting plasma glucose (FPG) $\ge$5.1 mmol/L or 1-h or 2-h plasma glucose PG $\ge$10.0 or 8.5 mmol/L, respectively). Diagnosis of metabolic abnormalities according to Chinese gestational metabolic syndrome (GMS) diagnostic criteria, includes a confirmed diagnosis of GDM, triglyceride (TG) $\ge$3.2 mmol/L, and a blood pressure higher than 140/90 mmHg [19]. According to the number of metabolic abnormalities, patients were assigned to a normal metabolic group, one metabolic abnormality group, and two or more metabolic abnormalities group.

(a) The inclusion criteria were as follows:

(1) aged 20–40 years;

(2) planning to receive regular prenatal examinations and deliver at Shengjing Hospital;

(3) diagnosed with GDM at the 24–28 gestational weeks.

(b) Participants conforming to any one of the following criteria were excluded:

(1) no available data on the last menstrual period;

(2) with fetal growth restriction, fetal malformation, or stillbirth history, or those with $\ge$3 abortions;

(3) receiving assisted reproductive technology;

(4) with surgical or medical disorders;

(5) hepatitis B virus, human immunodeficiency virus, or syphilis carrier;

(6) diagnosed with gestational complications in addition to GDM;

(7) the use of insulin therapy or hypoglycemic drugs;

(8) with autoimmune disease, thyroid disease, tumors or hematopathy among others.

The NGT gestational women (NGTGW) did not receive topical or systemic medication, and had no underlying systemic disease. The NGTGW were recruited during their second trimester at the Out-patients Clinic, Department of Obstetrics and Gynecology, Shengjing Hospital of China Medical University between January 2019 and July 2021. During this period, we identified 677 gestational women who received periodic systematic examinations at the outpatient department and were undergoing comprehensive routine inpatient prenatal care, all of whom had no abnormality in physical or OGTT examinations, and had similar exclusion criteria (as the above-mentioned in GDM) matched with those in the GDM group. Informed written consent was obtained from all 649 NGTGW before participating in this study. Fig. 1 displays the participant screening and recruitment procedure.

Fig. 1.

Flowchart displaying the patient and blood sample screening procedures of Shengjing Hospital of blood samples for the Gestational Diabetes Mellitus Blood sample from the Shengjing Hospital Biobank (GDMb/SJH) and for the Normal Glucose Tolerance gestational women Blood sample from the Shengjing Hospital Biobank (NGTb/SJH).

This study adopted the propensity score matching (PSM) approach to reduce bias in case control selection and to reduce the candidate clinical confounding factors at the matching ratio of 1:1 [20]. Eventually, 132 eligible NGT cases were collected based on PSM, and additional analyses were performed to match with the GDM group. Each participant and the corresponding clinical measurements were previously presented [21]. In total, 132 pairs of GDM cases and controls in their second trimester were recruited based on age (years), gestational week (weeks), and pregestational body mass index (BMI, kg/m${}^2$). With regard to the GDM cases in their second trimester, no significant difference in age or pregestational BMI was observed in NGT subjects (both p $>$ 0.05).

Finally, the present case-control study recruited 264 pregnant women (132 GDM and 132 normal glucose tolerance (NGT) who were in their second trimester at the out-patients Clinic, Department of Obstetrics and Gynecology, Shengjing Hospital of China Medical University in Shenyang, China, between January 2019 and July 2021. The majority (approximately 80%) of pregnant women living in this region (latitude 123${}^{\circ }$E and 41${}^{\circ }$N) attend the Shengjing Hospital for antenatal care. Pregnant women who received periodic systematic examination at the outpatient department of Shengjing Hospital and GDM patients whose glucose levels were under good control were enrolled.

After excluding the unqualified participants, we eliminated another 25 participants because they were unwilling to provide their inpatient records for the purpose of publication. We included clinical information and blood samples, respectively (Fig. 1). Finally, blood was sampled from 132 eligible GDM patients for further analysis of the Biobank that was registered as the Gestational Diabetes Mellitus Blood sample from the Shengjing Hospital Biobank (GDMb/SJH). At the same time, 132 blood was sampled from 649 NGTGW, and registered as the Normal Glucose Tolerance gestational women Blood sample from the Shengjing Hospital Biobank (NGTb/SJH). After admission, each eligible participant received an intravenous puncture for blood collection into 5.0 mL silica/gel plastic tubes (SST BD Vacutainer® gold) and 5.0 mL anticoagulant EDTA plastic tubes (BD Vacutainer® lavender). Additionally, the plasma was extracted from the sampled blood by centrifugation, which was stored at –80 ${}^{\circ }$C before use (up to 1 year). Furthermore, peripheral blood mononuclear cells were collected from each eligible GDM patient by Ficoll-sodium diatrizoate density gradient centrifugation, as performed previously [22].

Serum LPO and GPx4 levels were determined using an enzyme-linked immunosorbent assay kit (Enzyme-linked Biotechnology Co., Ltd. Shanghai, China) according to the manufacturer’s instructions. This assay was highly sensitive to LPO (ml856631), and the range detectable concentration was 0.1–32.0 ng/mL for human LPO. In addition, this assay was also highly sensitive for GPx4 (ml060706), and the detectable concentration range was 1.0–200.0 ng/mL for human GPx4. The coefficients of variation (CV) in the inter-assay and intra-assay were 10% and 8%, respectively.

All participates (132 GDMs and 132 NGTs) were in their third trimester and planned to attend regular visits for prenatal examinations and delivery at the In-patients Clinic, Department of Obstetrics and Gynecology of Shengjing Hospital. Their clinical measurements (In-patient Clinic) in the third trimester were collected, including blood routine test, hepatic panel, lipid panel, blood coagulation function, renal function, and serum electrolyte, and the selective solubilization approach are summarized in Supplementary File 1. Blood tests were completed by the Department of Laboratory Medicine of Shengjing Hospital of China Medical University. After extracting the related clinical data, we constructed the related blood database, followed by implementation of laboratory tests. We also collected maternal clinical information, including delivery mode and gestational weeks at birth. In addition, we extracted neonatal outcomes from birth records, which included body length (cm) and birth weight (g).

We applied the PSM approach in this study to reduce the selection bias of case-controls and to reduce the possible clinical confounding factors at the matching ratio of 1:1. In addition, the model was evaluated by the Hosmer-Lemeshow degree of fitting for the C-statistic test and logistic regression. Prior to PSM, we compared continuous variables in patient clinical features with controls by the Student’s t-test, and compared categorical variables using the chi-squared test. Following PSM, we inspected the variable normality visually. The Mann-Whitney U-test was conducted for comparing continuous variables, whereas differences among groups were compared by the Kruskal-Wallis H test. The chi-squared test was conducted to compare categorical variables. In addition, the associations of GPx4 and LPO with clinical parameters were evaluated by multiple linear regression and Spearman’s correlation analyses. This study adjusted confounders to conduct multiple regression analysis, including age, smoking status, blood pressure (BP), and BMI before pregnancy. Related thresholds were determined by logistic models, receiver operating characteristic (ROC) curves, and area under the ROC curve. Subsequently, the lower and upper strata of the LPO/GPx4 concentrations were determined accordingly. This study compared the LPO/GPx4 concentrations and the pregnancy outcomes of both groups (e.g., lower strata of LPO as the reference group and upper strata of LPO as the case group). Thereafter, this study divided the subjects based on metabolic abnormality numbers (which included blood lipids, PG, and BP) as the one metabolic abnormality, $\ge$2 metabolic abnormalities, and normal metabolic groups. Finally, the odds ratios (ORs) together with 95% confidence intervals (CIs) were estimated by multivariate logistic regression to predict the relationship of LPO with metabolic abnormalities among the gestational women. p 0.05 was considered to indicate statistical significance. SPSS version 25.0 (IBM Inc., Chicago, IL, USA) and GraphPad Prism version 8.0 (GraphPad Software Inc., La Jolla, CA, USA).

Table 1 displays the clinical features of the 649 NGT subjects and 132 GDM cases. Among them, BMI, age, ethnicity, smokers, and hypertension showed statistical significance between the two groups prior to PSM to reduce the selection bias and possible clinical confounding factors. We subsequently compared the 132 pairs of GDM and controls using PSM by logistic regression. p = 0.524 and p = 0.781 were obtained upon Hosmer-Lemeshow goodness of fit test and C-statistic test, respectively. Following PSM, we excluded those heterogeneities in clinical features between the two groups. Clinical features of the GDM cases and NGT participants is elaborated in Table 1.

According to the WHO guidelines released in 2013, all subjects were assigned to the NGT or the GDM group. Fig. 2A exhibits the serum LPO concentrations of both groups. The GDM group displayed a markedly increased LPO concentration relative to the NGT group (174.58 $\pm$ 22.01 ng/mL vs. 119.54 $\pm$ 8.93 ng/mL, p 0.001). By contrast, the GDM group had a markedly decreased GPx4 concentration relative to the NGT group (27.31 $\pm$ 16.88 vs. 33.84 $\pm$ 19.55 ng/mL; p 0.001, Fig. 2B).

Fig. 2.

The LPO (A) and GPx4 (B) concentrations of the GDM and NGT groups at 24–28 gestational weeks. Subgroup analysis between family history of hypertension and normotension (C,D), between the primipara and pluripara (E,F), and between those aged 35 years and those $\ge$35 years (G,H) of serum LPO and GPx4 concentrations, respectively.

Fig. 2C–F present subgroup analyses of the GDM patients. The serum LPO concentrations were not statistically significantly different between a family history of hypertension and normotension (Fig. 2C). The GPx4 level in the GDM without hypertension group was significantly elevated compared with the GDM with a family history of hypertension group (49.50 $\pm$52.05 vs. 25.12 $\pm$ 3.96 ng/mL, p = 0.041; Fig. 2D). The serum LPO concentrations were not statistically significantly different between the primipara and pluripara within the GDM patients (140.95 $\pm$ 62.06 vs. 129.19 $\pm$ 33.31 ng/mL, p = 0.451; Fig. 2E), as was the case for the GPx4 level (26.32 $\pm$ 13.47 vs. 29.63 $\pm$ 22.67 ng/mL, p = 0.097; Fig. 2F). Additionally, no significant difference was observed in the serum LPO level between those aged 35 years and those $\ge$35 years (141.02 $\pm$ 62.36 vs. 129.60 $\pm$ 32.59 ng/mL; p = 0.364; Fig. 2G) nor for the GPx4 concentration (26.46 $\pm$ 13.55 vs. 29.36 $\pm$ 23.02 ng/mL, p = 0.272; Fig. 2H). To determine the predictive value of LPO and PGx4 for GDM, we performed ROC curve analysis. The area under the ROC curve was 0.623 (95% CI: 0.54–0.71, p = 0.008) for LPO and 0.607 (95% CI: 0.52–0.69, p = 0.044) for GPx4 (Table 2). Therefore, LPO and GPx4 exhibited an acceptable capacity to distinguish the GDM patients from the general population (Fig. 3).

Fig. 3.

ROC analysis of the LPO and GPx4 levels for the evaluation of the GDM population. The vertical axis represents the sensitivity and the horizontal axis represents the 1-specificity.

As presented in Table 3, patient features and anthropometric variables were analyzed according to their serum GPx4 and LPO concentrations (classified as four quartiles). As revealed by the Kruskal-Wallis H test, as the GPx4 concentration was elevated, differences in the percentage glycated albumin (GA%), FPG, and 2h-PG were significant between the different groups, and vice visa (p = 0.004, p = 0.028, and p = 0.010, respectively) (Fig. 4). By contrast, the Kruskal-Wallis H test analysis showed that with the rise in LPO, there were no significant differences in glucose metabolism-related indicators within the GDM group (Table 2). Using bivariate correlation, this study analyzed the association of the GPx4 concentration with glucose metabolic parameters of the GDM group. The serum GPx4 level was significantly negatively correlated with the FPG (r = –0.237, p = 0.004), 2h-PG (r = –0.258, p = 0.040), and GA% (r = –0.395, p 0.001). In addition, as revealed by partial correlation after adjusting for gestational week and age, GPx4 displayed a positive correlation with the FPG (r = –0.282, p = 0.020), 2h-PG (r = –0.302, p = 0.006), and GA% (r = –0.343, p 0.001).

Fig. 4.

The relationship between GPx4 and fasting plasma glucose (A), 1h-PG (B), 2h-PG (C), fasting plasma insulin (D), glycated hemoglobin (E), and glycated albumin (F).

Table 3 presents the clinical outcome features for each subject. Using LPO = 160.54 ng/mL in the second trimester as the threshold, this study classified GDM cases into LPO upper and lower strata groups. We found that GDM patients in the second trimester with a high LPO level were more likely in the third trimester to have higher levels of total bilirubin, direct bilirubin, unconjugated bilirubin, and serum creatinine (all p 0.05). When using GPx4 = 24.81 ng/mL as the cut-off point in the second trimester, the GDM patients were assigned to the GPx4 lower strata and GPx4 upper strata groups. We also found that GDM patients in the second trimester with a high LPO level were more likely in the third trimester to have lower TG levels (p = 0.036). Differences in pregnancy and neonatal outcomes were not significant between the two groups (p $>$ 0.05; Table 3). Furthermore, by bivariate correlation, this study analyzed the association of the LPO/GPx4 concentrations with clinical features among GDM cases, which are summarized in Fig. 4.

In addition, multivariate logistic regression was conducted. Supplementary Table 1 presents the associations of metabolic syndrome components with the upper stratum of the LPO and GPx4 levels when compared with the lower stratum, according to Chinese GMS diagnostic criteria. After the GDM subjects were grouped based on the metabolic abnormality component, the metabolic abnormality risk was found to be elevated with increased LPO concentration (elevated DBP, OR = 1.04, p = 0.048; and high TGs, OR = 2.19, p 0.001, respectively), together with a greater incidence of multiple metabolic abnormalities. According to Fig. 5, the serum LPO concentrations were greatly elevated among subjects displaying metabolic abnormalities. Subjects displaying multiple metabolic abnormalities had the highest LPO concentrations. When ‘no metabolic abnormality’ was utilized as the control, the presence of metabolic abnormalities in the one and multiple metabolic abnormalities groups were associated with LPO (Fig. 6).

Fig. 5.

The correlations between the LPO and GPx4 levels and the clinical features of the blood routine (A), serum lipid profile (B), blood coagulation function (C), liver function (D), thyroid function (E), and serum electrolyte (F).

Fig. 6.

The serum LPO level (A) and GPx4 level (B) in the no metabolic abnormality, one metabolic abnormality, and two or more metabolic abnormalities groups.