Gestational diabetes mellitus (GDM) refers to abnormal glucose tolerance first detected during pregnancy and is one of the most common metabolic complications encountered in pregnant women (1). Globally, the prevalence of GDM is approximately 16.7% (2). In recent years, with increasing obesity rates and a rising proportion of advanced maternal age pregnancies, the incidence of GDM has continued to increase, posing significant threats to both maternal and fetal health (3). GDM is closely associated with multiple adverse pregnancy outcomes, including hypertensive disorders of pregnancy, preterm birth, and macrosomia (4). It also significantly increases the long-term risk of type 2 diabetes in mothers, as well as the likelihood of obesity and metabolic syndrome in offspring (5, 6). Existing evidence suggests that women who develop GDM exhibit enhanced insulin resistance (IR) and impaired β-cell compensation early in pregnancy (7), indicating that metabolic disturbances may occur before clinical diagnosis. Therefore, applying effective metabolic evaluation tools for risk stratification of GDM and timely intervention is important for improving pregnancy outcomes and interrupting the intergenerational transmission of metabolic disorders.

IR is one of the key pathophysiological mechanisms underlying the development of GDM and serves as an important marker for predicting and assessing GDM risk (8, 9). However, traditional methods for evaluating IR, such as the euglycemic-hyperinsulinemic clamp (EHC), although considered the gold standard, are technically complex and invasive, which limits their applicability in routine clinical screening (10). Thus, there is a pressing need to develop simple, non-invasive, and accurate surrogate markers for IR to improve early risk stratification of GDM. The metabolic score for insulin resistance (METS-IR), developed by the team of Professor Bello-Chavolla OY, is a novel cardiometabolic risk scoring model. This score incorporates routinely available clinical parameters, including glucose-related indices (fasting plasma glucose, FPG), lipid profiles (such as triglycerides (TG) and high-density lipoprotein cholesterol (HDL-C)), and obesity-related measures (such as body mass index (BMI)) (11). Studies have demonstrated that METS-IR outperforms EHC in detecting impaired insulin sensitivity (11), with good reproducibility and ease of calculation, making it a promising indirect tool for assessing IR. Some cohort studies have consistently shown a positive association between METS-IR and the risk of developing type 2 diabetes (12–15). Recent studies have demonstrated an association between elevated METS-IR and an increased risk of GDM. One study using NHANES data included 5,189 pregnant women (417 with GDM) and found a significant association between higher METS-IR and GDM, particularly among women with high school education or higher (16). Another cohort study in Iran (n = 1,845) reported that first-trimester METS-IR may predict GDM in Iranian women (17). While these studies support the potential of METS-IR as a predictive tool, they have limitations, including single-center design, small sample size, and reliance on self-reported data, which may affect generalizability and accuracy. To address these limitations, our multicenter study with a large sample size rigorously controlled for confounders and applied multiple statistical methods to examine the association between METS-IR and both GDM and related adverse pregnancy outcomes. Our findings aim to provide a simple, practical clinical indicator for early risk stratification and prediction of GDM.

In this study, we conducted a retrospective cohort analysis of 37,770 pregnant women from three hospitals in China, aiming to investigate the association between METS-IR and GDM, as well as its related adverse pregnancy outcomes. After adjusting for potential confounding factors, the results demonstrated a positive correlation between METS-IR and the risk of GDM. Subgroup analyses stratified by maternal age confirmed a consistent positive association between METS-IR and GDM. Moreover, METS-IR exhibited good discriminative ability in predicting GDM and its complications. Notably, it showed strong predictive performance for GDM&PE and GDMA2, with AUC values of 0.741 (95% CI: 0.715–0.767) and 0.712 (95% CI: 0.691–0.733), respectively. These findings indicate that METS-IR, as a simple and non-invasive metabolic assessment tool, holds potential clinical value in risk stratification and severity prediction of GDM.

Previous studies have established that increased IR is a central mechanism in the development of GDM (24). The EHC technique is considered the gold standard for assessing insulin resistance. However, it is complex, costly, and invasive, making it unsuitable for routine clinical use. As a result, several surrogate indices, including the Quantitative Insulin Sensitivity Check Index (QUICKI), the homeostasis model assessment of insulin resistance (HOMA-IR), the triglyceride-glucose (TyG) index, and the TG/HDL-C, have been used to evaluate insulin resistance and its association with GDM (25, 26). However, the clinical validity of these indices remains uncertain. METS-IR is non-invasive and easily calculated from routine clinical data. Evidence has shown that METS-IR demonstrates better diagnostic performance for insulin resistance compared to EHC (11), and it has been identified as an independent predictor of type 2 diabetes mellitus (13). Our findings indicate that, as METS-IR percentile scores increase, so does the risk of GDM in a graded manner. Compared with the lowest quartile, individuals in the highest METS-IR quartile had a 165% higher risk of developing GDM. This observation aligns with previous findings in general populations. For instance, Cheng et al. (14) reported a significant association between elevated METS-IR and T2DM incidence (OR: 1.804; 95% CI: 1.720–1.891). Another cross-sectional study conducted in China provided similar evidence of a positive correlation (27), and a subsequent 6-year longitudinal study found that each one-standard-deviation increase in METS-IR was associated with an 82% higher risk of developing diabetes (28). More recently, a cohort study in a Japanese population showed that participants in the highest METS-IR quartile had a 215% higher risk of developing diabetes compared to those in the lowest quartile (13). Recent studies have reported a positive association between elevated METS-IR and an increased risk of GDM (16, 17). Building on these findings, our multicenter study with a large sample size, rigorous control of confounders, and comprehensive statistical analyses further confirms the association between METS-IR and GDM. Given the practicality of METS-IR measurement and its strong pathophysiological link to IR, its application in identifying high-risk individuals for GDM appears feasible. To further assess the predictive performance of METS-IR for GDM, we conducted ROC curve analysis, which yielded an AUC of 0.623 (95% CI: 0.614–0.631), with a specificity of 64.4% and a sensitivity of 54.0%. Although the AUC is less than 0.7, METS-IR may be considered as a potential indicator for GDM risk stratification, helping to identify individuals at higher risk during early pregnancy.

In this study, we further evaluated the association between METS-IR and multiple adverse pregnancy outcomes related to GDM. The results showed no significant association between METS-IR and preterm birth, whereas significant positive correlations were observed between METS-IR and macrosomia, GDM&PE, and GDMA2. These findings suggest that METS-IR not only reflects overall IR in pregnant women but may also uncover underlying pathophysiological mechanisms of GDM-related complications. Emerging evidence has demonstrated that IR is closely linked to both GDM and its adverse perinatal outcomes. For instance, compared with women with normal glucose tolerance, those with GDM exhibit higher levels of IR accompanied by more severe metabolic disturbances and worse perinatal outcomes (29). Moreover, insulin resistance has been recognized as one of the key metabolic underpinnings of preeclampsia (PE) (30). However, unlike some previous studies that reported a significant association between IR and preterm birth (8), our study did not detect a statistically significant relationship between METS-IR and preterm birth. This inconsistency may stem from differences in calculation methods and clinical applicability among various IR assessment tools, or it could be due to variations in population characteristics and the extent of confounding adjustment across studies. Notably, our findings revealed that METS-IR demonstrated good predictive performance for GDM&PE and GDMA2, with AUC values of 0.741 and 0.712, respectively, exceeding the performance of several traditional risk parameters. IR not only represents metabolic dysregulation but is also closely associated with endothelial dysfunction (31), and these shared mechanisms may form the basis for the co-occurrence of GDM and PE. Furthermore, accumulating evidence suggests that multiple pathological pregnancy conditions, including PE, GDM, and obesity, are characterized by reduced insulin signaling in the fetal-placental vasculature (32, 33). Therefore, as a surrogate marker of insulin resistance, METS-IR holds clinical value not only in identifying high-risk individuals for GDM but also in predicting severe GDM subtypes complicated by preeclampsia or requiring insulin therapy.

Given that advanced maternal age (≥35 years) is a well-established independent risk factor for GDM (34), we conducted a subgroup analysis stratified by age. The results were consistent with expectations, showing a significantly higher incidence of GDM in the ≥ 35 years group. The positive association between METS-IR and GDM was largely consistent across age groups, indicating that METS-IR is a feasible and stable tool for risk stratification in pregnant women of different ages. However, within the ≥ 35 years subgroup, a significant association between METS-IR and GDM was observed only in Q3 and Q4, but not in Q2. To explain this discrepancy, we propose the following possibilities: It has been documented that, with increasing age, pancreatic β-cell function declines, insulin sensitivity decreases, and glucose metabolism becomes more impaired, thereby elevating the risk of GDM (35, 36). Nevertheless, some older pregnant women may already have a certain degree of insulin resistance prior to pregnancy, which could potentially reduce the sensitivity of METS-IR in capturing changes in IR levels. Additionally, age is an important determinant of lipid metabolism. Studies have shown that HDL-C levels tend to decline with advancing age (37). Since both FPG and HDL-C are key components in the calculation of METS-IR, their age-related variations may affect the stability of the score and consequently reduce its accuracy in predicting GDM. Finally, as an independent risk factor for GDM, age may contribute to disease development through mechanisms not directly related to IR. Therefore, in clinical practice, it is essential to integrate other risk factors alongside METS-IR when assessing GDM risk in older pregnant women.

The association between METS-IR and GDM may involve multiple interacting pathophysiological mechanisms. IR is a central mechanism in the development of GDM. During normal pregnancy, a progressive increase in IR serves as an adaptive response to meet the growing energy demands of both mother and fetus (38). However, when IR becomes excessively elevated, it can lead to inadequate β-cell compensation, resulting in glucose metabolic imbalance and increased risk of GDM (39, 40). As a composite indicator of insulin resistance, elevated METS-IR reflects increased IR and thus holds potential value in identifying individuals at higher risk. Dyslipidemia also plays a significant role in GDM progression. Studies have shown that, compared with women with normal pregnancies, those with GDM often exhibit lower HDL-C levels and higher levels of TG, TC, and LDL-C (41, 42). Since METS-IR incorporates both TG and HDL-C into its scoring system, it partially reflects the degree of lipid metabolic disturbance. Chronic inflammation is another key link connecting insulin resistance and GDM. Pro-inflammatory cytokines such as IL-6 can suppress lipoprotein lipase activity, promote TG accumulation, and exacerbate IR (43). In addition, inflammatory mediators, such as IL-1β, can activate multiple signaling pathways and directly impair pancreatic β-cell function (44, 45). C-reactive protein (CRP) further contributes to systemic inflammation and worsens insulin action defects (46). Previous studies have demonstrated a positive correlation between METS-IR and inflammatory markers, such as CRP and IL-6 (47). Oxidative stress also contributes to insulin resistance by activating the NF-κB pathway, leading to endothelial dysfunction and impaired IR (48). Moreover, lifestyle factors such as dietary patterns and physical activity levels significantly influence insulin sensitivity and thereby modulate the risk of developing GDM. In summary, the relationship between METS-IR and GDM likely results from the complex interplay of multiple metabolic, inflammatory, and oxidative stress mechanisms.

The main strengths of this study lie in its multicenter design and large sample size, which effectively reduced bias caused by small sample sizes and enhanced statistical power, thereby improving the generalizability and applicability of the findings. Moreover, METS-IR is a simple, non-invasive, and easily obtainable metabolic index with good clinical feasibility, making it suitable for widespread application in routine clinical practice and highlighting its important clinical value. Despite these strengths, several limitations should be acknowledged. First, this study was based only on baseline measurements before 24 weeks of gestation. Since METS-IR may vary dynamically across gestation, future prospective studies with multiple time points are needed to explore longitudinal changes in METS-IR and their association with GDM. Second, all participants were from China. Given potential differences in genetic background, lifestyle, and metabolic profiles across populations, future studies in multiethnic and multicenter settings are necessary to confirm the generalizability of our findings. Finally, as a retrospective cohort study, our analysis excluded some patients due to incomplete clinical data. Additionally, information on exercise, dietary interventions, and insulin use was not collected. Although we adjusted for multiple confounders, potential selection bias may still influence the results.

Our findings demonstrate that the METS-IR score is positively associated with the risk of GDM and has a certain predictive value for GDM occurrence, particularly in identifying severe subtypes such as GDM&PE or GDMA2. As a novel, simple, and easily accessible marker of insulin resistance, METS-IR holds promise for use in risk stratification and early intervention strategies for GDM, offering valuable clinical insights with strong practical implications.