Characteristics of women with NGT and with GDM across all subjects and in the matched case-control study

Among the 1682 women, 358 (21.28%) developed GDM, with the probability of developing GDM greater in those with older age, previous GDM history and higher pre-pregnancy BMI (P < 0.001). Compared to women with NGT, those with GDM exhibited significantly higher FBG, WBC count, neutrophil count, RBC count and Hb concentration (120 ± 11 vs. 125 ± 10 g/L in T1, 112 ± 10 vs. 115 ± 9 g/L in T2, 110 ± 12 vs. 115 ± 12 g/L in T3, all P < 0.001) throughout pregnancy (all P < 0.05), increased first-trimester blood pressure, creatinine, UA, TG, and HDL (all P < 0.05), second-trimester FBG, 1-h blood glucose (BG), 2-h BG, HbA1c, FINS, and HOMA-IR (all P < 0.001), and third-trimester diastolic blood pressure, creatinine, and TG (all P < 0.05). Conversely, first-trimester lymphocyte count, TC, LDL, second-trimester HOMA-β and third-trimester ALT, AST were lower in women with GDM (all P < 0.05) (Supplementary Table S1). Histograms revealed that women with GDM had consistently higher Hb concentrations than those with NGT throughout pregnancy (P < 0.001) (Fig. 2A–C).

Fig. 2: Hemoglobin concentration distribution and correlation with glucose metrics during pregnancy.

Histograms of the distribution of Hb concentration throughout pregnancy for all subjects with NGT and with GDM, in first trimester (A), in second trimester (B) and in third trimester (C). Data are presented as mean ± SD and are analyzed by independent sample t test. Spearman correlation analysis of first-trimester Hb level with FBG (D), 1-h BG (E), 2-h BG (F), HOMA-IR (G) during OGTT. Continuous association of Hb level in the first trimester with the incidence of GDM, adjusted for GDM history, age and pre-pregnancy BMI (H). The OR is expressed per absolute increase in 10 g/L in the Hb value at baseline. The shaded area represents the 95% CI from the restricted cubic spline model. The model is centered at the median (122 g/L) with knots at the 25th, 50th, 75th percentiles.

Compared with mothers with NGT, those with GDM tended to deliver heavier newborns, had higher rates of delivering macrosomic or LGA infants, and were more likely to require cesarean sections. Composite neonatal complications and events of neonatal hypoglycemia were significantly higher in women with GDM than those with NGT (12.26% vs. 6.92%, P < 0.01; 3.77% vs. 0.63%, P < 0.001) (Supplementary Table S2). A 1:1 case-control matching procedure was performed based on age, pre-pregnancy BMI, and parity to minimize potential bias from uneven distribution of covariates. After matching, significant differences remained in RBC count and Hb concentration (NGT vs. GDM, 116 ± 11 vs. 123 ± 10 g/L in T1, 110 ± 11 vs. 114 ± 9 g/L in T2, 109 ± 13 vs. 115 ± 12 g/L in T3, all P < 0.001) and blood glucose metabolic profiles, as well as delivery and neonatal outcomes (all P < 0.05) (Supplementary Tables S1, S2).

Comparison of parameters during each trimester among three groups categorized by first-trimester Hb level in the retrospective cohort study

Subjects were then divided into three groups according to Hb level in the first trimester: anemia (<110 g/L), normal Hb (≥110 and ≤130 g/L) and high Hb (>130 g/L). There was a stepwise increase in the level of pre-pregnancy BMI, diastolic blood pressure, creatinine, RBC count and Hb concentration throughout pregnancy (all P < 0.05), first-trimester WBC count, neutrophil count and UA (all P < 0.01) and second-trimester WBC count, neutrophil count and lymphocyte count (all P < 0.05), as well as FBG, 1-h BG, 2-h BG, HbA1c, HOMA-IR (all P < 0.01) and incidence of GDM (13.4%, 19.1%, and 32.1%; P < 0.001) across the three groups (Table 1). In contrast there was a stepwise decrease with increase Hb level in first-trimester TC and LDL (both P < 0.01), but no significant differences in age, previous GDM history, family history of diabetes, AST, TG, HDL, or HOMA-β among the groups (Table 1). Importantly, Pearson correlation analysis showed that first-trimester Hb level was significantly and positively correlated with FBG, 1-h BG, 2-h BG, and HOMA-IR during OGTT, but had no significant correlation with HOMA-β (Fig. 2D–G).

Table 1 Comparison of parameters during each trimester and at delivery among three groups categorized by first-trimester Hb level in the retrospective cohort study.

Regarding GDM-related maternal and delivery complications, the incidence of hypertensive disorders of pregnancy and the need for cesarean delivery increased as Hb level increased (P < 0.01), whereas EGWG, absolute GWG, antenatal BMI, fetus sex, neonatal weight, macrosomia, LGA, and SGA were not significantly affected (Table 1). Overall neonatal complications were significantly higher in the high Hb group than in the normal Hb group (10.96% vs. 6.78%, P < 0.01) (Table 1).

Continuous Hb level in the first trimester was closely associated with the incidence of GDM

To graphically visualize the association of Hb with GDM development, a restricted cubic spline model with three knots was applied with adjustment for GDM history, maternal age and pre-pregnancy BMI. We found a significant relationship between continuous Hb level during the first trimester and GDM incidence. The risk of developing GDM increased when Hb level in the first trimester exceeded 122 g/L (Fig. 2H).

Increased first-trimester Hb level was an independent risk factor for development of GDM

To identify independent risk factors for the development of GDM, age, pre-pregnancy BMI, increased Hb level (divided by 122 g/L), neutrophil count, platelet count, TG and creatinine in the first trimester were entered into logistic regression analysis with enter selection in all subjects without GDM history. After adjusting for potential confounding factors, higher first-trimester Hb concentration remained an independent risk factor for development of GDM (OR = 2.214, 95% CI: 1.042–4.331, P = 0.038) (Table 2). This relationship remained significant in the matched case-control cohort, independent of age, pre-pregnancy BMI, neutrophil count or TG level in the first trimester (OR = 4.968, 95% CI: 2.480–9.954, P < 0.001) (Table 2). Additionally, combining Hb concentration with basal factors (age, pre-pregnancy BMI, TG, and FBG in the first trimester) yielded the highest area under the receiver operating characteristic curve (AUC) for predicting GDM. This combination achieved an AUC of 0.795, outperforming basal factors alone (0.786), as well as combinations with other first-trimester blood cell indicators such as neutrophil count (0.787) and platelet count (0.786). Corresponding sensitivities and specificities for these combinations were 0.783 and 0.746 for Hb with basal factors, 0.696 and 0.815 for basal factors alone, 0.710 and 0.821 for neutrophil count with basal factors, and 0.710 and 0.785 for platelet count with basal factors, respectively (Supplementary Fig. S2).

Table 2 Logistic regression analysis to determine the risk factors for development of GDM in all subjects without GDM history and in the retrospective case-control study.Women with Hp2-2 genotype combined with elevated first-trimester Hb concentration were at higher risk of GDM development

To investigate the moderating effect of Hp genotype on the association between Hb levels and GDM, Hp genotypes were determined in 180 women with NGT and 180 women with GDM, who were 1:1 matched based on pre-pregnancy BMI, age, and parity from a pool of 400 subjects. The proportion of genotype Hp1-1, Hp1-2, and Hp2-2 was 15.6%, 50.0%, and 34.4%, respectively, in healthy controls, close to the reported frequency in China. Nonetheless, the frequency of the Hp1-1 genotype in women with GDM was much lower than in healthy controls while the frequency of Hp2-2 was much higher (P < 0.001) (Fig. 3A). Compared to those with Hp1 carrier genotype (both Hp1-1 and Hp1-2 genotypes), women with the Hp2-2 genotype had much higher RBC count and Hb concentration throughout pregnancy (all P < 0.001). They also had higher first-trimester blood pressure and creatinine, second-trimester FBG, 1-h BG, 2-h BG, HbA1c, FINS, HOMA-IR, and GDM incidence, as well as third-trimester FBG (all P < 0.05) (Supplementary Table S3). Additionally, overall neonatal complications were more frequent in women with the Hp2-2 genotype (P < 0.05) (Supplementary Table S4). Meanwhile, the frequency of Hp1-1 and Hp1-2 genotypes was significantly lower in women with Hb > 122 g/L than in those with Hb ≤ 122 g/L (4.69% vs. 10.94%, 30.21% vs. 46.09%, P < 0.001), whereas the frequency of Hp2-2 was much higher (65.10% vs. 42.97%, P < 0.001) (Fig. 3B). After adjusting for potential confounding factors, logistic regression analysis showed that both higher first-trimester Hb concentration and the HP2-2 genotype remained independent risk factors for the development of GDM (Supplementary Table S5).

Fig. 3: Distribution, joint and mediation analysis of Hp phenotypes and Hb levels in relation to GDM risk.

Frequency distribution of Hp phenotypes in women with NGT and with GDM (A) and in Hb >122 g/L group versus Hb ≤122 g/L group (B) in the validation cohort study. Multivariate OR of GDM risk according to Hb status and Hp genotype (C). Mediation analysis of first-trimester Hb level and Hp genotype on GDM risk without (D) or with (E) adjusting for confounding factors, age and pre-pregnancy BMI.

To identify independent risk factors for the development of GDM, age, pre-pregnancy BMI, increased Hb level (divided by 122 g/L), neutrophil count, platelet count, TG and creatinine in the first trimester were entered into logistic regression analysis with enter selection in all subjects without GDM history. After adjusting for potential confounding factors, higher first-trimester Hb concentration remained an independent risk factor for the development of GDM (OR = 2.214, 95% CI: 1.042–4.331, P = 0.038) (Table 2).

Interaction moderate effect, joint analysis, and mediation analysis of first-trimester Hb level and Hp genotype with GDM risk

After adjusting for age and pre-pregnancy BMI, a significant multiplicative interaction effect was observed between Hp genotype and first-trimester Hb level on GDM development (OR = 4.801, 95% CI:1.705–13.517, P for interaction = 0.003) (Table 3). Further analysis revealed a statistically significant synergistic stimulation additive interaction effect of first-trimester Hb level and Hp genotype on the occurrence of GDM (RERI = 3.472, 95% CI: 1.629–6.697) (Table 3).

Table 3 Interaction analysis of Hp genotype and Hb level for GDM risk.

To explore the joint associations of first-trimester Hb level and Hp genotype with GDM risk, pregnant women were stratified according to their Hb level and Hp genotype. As shown in Fig. 3C, after adjusting for age and pre-pregnancy BMI, the OR of risk for GDM development increased in a stepwise manner from women with Hb ≤ 122 g/L and Hp1 allele to women with Hb > 122 g/L and Hp2-2 genotype (OR = 1.034, 95% CI: 0.274–1.469, P = 0.288; OR = 1.472, 95% CI: 0.795–2.727, P = 0.219; OR = 4.355, 95% CI: 2.437–7.784, P < 0.001; respectively).

Mediation analysis was performed to evaluate the influence of first-trimester Hb level and Hp genotype on GDM risk (Fig. 3D, E). Hp genotype was found to be an independent risk factor for GDM without adjusting for confounding factors (OR = 2.66, 95% CI: 1.74–4.10), and it remained an independent risk factor for GDM after adjusting for age and pre-pregnancy BMI (OR = 2.39, 95% CI: 1.54–3.75). Additionally, first-trimester Hb level partially mediated the association of Hp genotype with GDM risk, explaining 29.16% of the association without adjusting for other factors (Fig. 3D). After adjusting for age and pre-pregnancy BMI, first-trimester Hb level explained 34.89% of the association (Fig. 3E).