The impact of early pregnancy uric acid on gestational diabetes risk: a prospective cohort study in Northwest China Authors: Doudou Zhao, Nan Li, Yanfang Song, Jie Chen, Xiayang Li, Yishuai Huang, Danmeng Liu, Li Shan, Yang Mi, Pengfei Qu, Lei Shang Published: 2025 ABSTRACT AbstractBackgroundElevated uric acid (UA) has become a global health concern. We aimed to investigate the impact of UA in women during early pregnancy on the subsequent risk of gestational diabetes mellitus (GDM).MethodsPregnant women in their first trimester were recruited from a birth cohort study conducted in Xi’an, China, between July 2018 and December 2023. A generalized linear model (GLM) was used to calculate risk ratios (RRs) and 95% confidence intervals (CIs) for the effects of UA levels.ResultsCompared with the first quartile of UA, the second (RR = 1.15, 95% CI = 1.04–1.28), third (RR = 1.27, 95% CI = 1.15–1.41), and fourth (RR = 1.46, 95% CI = 1.32–1.61) quartiles were associated with an increased risk of GDM. Moreover, UA was the strongest predictor among all variables in the random forest model except for BMI. Importantly, higher UA levels were associated with an increased risk of GDM (OR = 1.39, 95% CI = 1.29–1.50) as long as BMI exceeded 18.5 kg/m². The highest GDM risk was observed in high UA/high BMI combinations.ConclusionsThis study demonstrated that higher UA levels during early pregnancy were associated with an increased risk of GDM. These findings suggest a link that warrants further investigation to determine if UA plays a causal role or is a marker of underlying pathology.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12884-025-08559-z.Keywords: Gestational diabetes mellitus, Uric acid, Early pregnancy, Body mass index, Cohort study Introduction Gestational diabetes mellitus (GDM) is a prevalent pregnancy complication characterized by glucose intolerance that first manifests during pregnancy, and its global incidence is rising [1]. According to studies spanning 1990 to 2020, the worldwide standardized prevalence of GDM was estimated at 14.0%, with the highest rates observed in the Middle East and North Africa (27.6%) and South-East Asia (20.8%) [2]. Hyperglycemia during pregnancy has been shown to adversely affect maternal health, intrauterine growth, and fetal development [3, 4]. Furthermore, it exerts long-term metabolic consequences on offspring, extending from childhood and adolescence into adulthood [5–7]. Identifying modifiable risk factors associated with GDM is therefore critical for developing early prevention strategies. Uric acid (UA), the final product of human purine metabolism, exhibits a dual biological role: it acts as a systemic pro-inflammatory mediator contributing to disease pathogenesis while simultaneously displaying protective antioxidant properties [8]. Elevated serum UA levels are frequently associated with metabolic and hemodynamic abnormalities, including glucose intolerance, insulin resistance, dyslipidemia, and hypertension [9]. Extensive research has established a link between elevated UA levels and type 2 diabetes mellitus (T2DM) [10, 11], as well as increased all-cause and cardiovascular mortality in diabetic populations [12]. However, evidence regarding the relationship between maternal serum UA levels during early pregnancy and GDM remains limited and inconsistent. While some studies suggest that elevated UA levels increase GDM risk [13, 14], others report no significant association with GDM, insulin sensitivity, or β-cell function [15, 16]. Additionally, existing studies predominantly rely on retrospective designs and involve concurrent measurements of UA and blood glucose levels, which may obscure causal interpretations. Moreover, it is well-established that maternal pre-pregnancy BMI is a significant risk factor for GDM [17, 18]. Importantly, both elevated UA and high BMI share common pathophysiological pathways in diabetes, such as promoting insulin resistance and chronic inflammation [19, 20]. This biological synergy suggests a potential synergistic effect of UA and BMI on GDM risk. Investigating this interaction to identify subgroups of women who may benefit from intensified monitoring could further optimize GDM risk stratification. However, direct evidence regarding this specific interaction remains limited. To address these limitations, this prospective cohort study aimed to investigate the impact of serum UA in women during early pregnancy on the subsequent risk of GDM, utilizing data from 16,064 pregnant women in Northwest China. We further examined the interaction effects of UA with pre-pregnancy BMI and the relationships between UA levels and maternal blood glucose levels at three time points during the second trimester. Methods Study population This prospective birth cohort study [21] was conducted at Northwest Women’s and Children’s Hospital, the largest specialized hospital for women and children in Shaanxi province, with approximately 30,000 annual births. Pregnant women were recruited during their first prenatal visit to the obstetrics outpatient clinic between July 2018 and December 2023. Inclusion criteria required participants to be pregnant women between 1 and 13+ 6 weeks of gestation who received prenatal care and delivered at the hospital. Exclusion criteria were as follows: (1) voluntary withdrawal from the cohort; (2) multiple pregnancies; (3) lack of data on UA detection before 14 weeks or results of the 75 g oral glucose tolerance test (OGTT) at 24–28 weeks of pregnancy; and (4) pre-existing diabetes mellitus, urinary system disease, or heart disease before pregnancy. Figure S1 presents the participant enrollment flowchart. Women included in the final analysis exhibited similar maternal characteristics to those excluded (Table S1). The study protocol was approved by the Human Research Ethics Committee of Beijing Obstetrics and Gynecology Hospital (No. 2018-KY-003-02) and Northwest Women’s and Children’s Hospital (No. 2018018). Written informed consent was obtained from all participants prior to their inclusion in the study. Data collection Following their visit to the obstetrical early pregnancy clinic, general demographic characteristics and lifestyle behaviors of the enrolled pregnant women were systematically collected. These included maternal age, ethnicity, education level, occupation, pre-pregnancy body mass index (BMI) based on Chinese BMI criteria, annual household income (categorized as poverty: < 50,000 Chinese yuan [CNY], medium: 50,000–200,000 CNY, or rich: >200,000 CNY), parity, mode of conception, complications in previous pregnancies, use of folic acid or multivitamin supplements, smoking or passive smoke exposure, alcohol consumption, and occurrences of cold or fever during early pregnancy. Data were collected using the Birth Cohort Enterprise Data Center (EDC) cloud platform online system. Maternal serum UA levels and 75 g OGTT results were obtained from the hospital outpatient laboratory system. All maternal identifying information was anonymized to ensure privacy protection. Detection of serum UA levels Maternal fasting blood samples were collected between 6 and 14 weeks of gestation and transported to the laboratory center within one hour. Following centrifugation at 4500 rpm, serum UA levels were measured using an enzymatic colorimetric assay based on the uricase–peroxidase method, performed on the Beckman Coulter biochemical analysis system. The laboratory center maintained a coefficient of variation (CV) of less than 5.0%. Within the serum UA concentration range of 89 to 1785 µmol/L, the linear correlation coefficient for reagent detection was ≥ 0.99. Diagnosis of GDM At 24–28 weeks of gestation, pregnant women were required to fast overnight and undergo a 75 g OGTT in the morning to measure fasting plasma glucose (FPG), 1-hour plasma glucose (PG), and 2-hour PG at three time points. GDM was diagnosed according to the criteria set by the International Association for Diabetes in Pregnancy Study Group (IADPSG) [22]: FPG ≥ 5.1 mmol/L, 1-hour PG ≥ 10.0 mmol/L, or 2-hour PG ≥ 8.5 mmol/L. Definitions of study variables Maternal characteristics and lifestyle factors associated with GDM from previous studies [23, 24] were included as covariates in this study. “Complications in previous pregnancy” refers to women who developed GDM, gestational hypertension or thyroid disorder during the previous pregnancy. “Smoke or passive smoke” refers to smoking one cigarette or being exposed to smoke for over 15 min at least once a week during early pregnancy. “Drink” refers to drinking alcoholic products such as liquor, beer, and red wine at least once during early pregnancy. “Cold or fever” refers to having a fever or cold at least once during early pregnancy. Early pregnancy was limited to the period from 3 months before conception to the first trimester after pregnancy. Gestational age for laboratory detection was measured from the last menstrual period to the date of serum UA tests and the OGTT. Statistical analysis Categorical variables were presented as frequencies and percentages, with group comparisons performed using the Chi-square test. The distribution of continuous variables was assessed using the Kolmogorov-Smirnov test. Normally distributed continuous variables were expressed as means ± standard deviations (SD), and group comparisons were conducted using Student’s t-test. Non-normally distributed continuous variables were presented as medians and interquartile ranges (IQR), with group comparisons performed using the Mann-Whitney U test. For missing values in maternal age (0.12%) and drinking status (0.23%), we used the multiple imputation (MI) method to address the missing data. We generated 10 imputed datasets under the Missing at Random (MAR) assumption. The imputation model included the outcome (GDM), primary exposure (UA) levels, and all other covariates in the final analysis. Results from the 10 imputed datasets were pooled into a single set of estimates using Rubin’s rules. To estimate risk ratios (RRs) and 95% confidence intervals (CIs) for the binary outcome of GDM, we used generalized linear models (GLM) with a log link function. A Poisson distribution was specified in the model, and robust standard errors were applied to obtain consistent estimates of the RRs. For continuous outcomes such as FPG, GLMs with an identity link and Gaussian distribution were used to estimate mean differences (β) and 95% CIs. The crude model examined the unadjusted association. The adjusted model controlled for the following potential confounders: maternal age, educational levels, occupation, household income, pre-pregnancy BMI, primiparity, mode of conception, history of pregnancy complications, drinking, and fever during early pregnancy. A Random Forest (RF) model was implemented to evaluate the predictive utility of UA and other covariates for GDM. The dataset was randomly split into a training set (70%) for model development and an independent validation set (30%) for performance assessment, with a fixed random seed set to ensure reproducibility. The RF model was constructed with 500 decision trees. The number of variables available for splitting at each node was set to the square root of the total number of predictors. Model performance was evaluated on the validation set by generating the Receiver Operating Characteristic (ROC) curve and calculating the Area Under the Curve (AUC). Sensitivity and specificity were derived at the optimal cutoff point determined by Youden’s index. The importance of each variable was quantified based on its mean decrease in the Gini index. Given that higher levels of UA (4th quartile) appear to be more detrimental, we performed a two-way interaction analysis of UA (higher vs. lower) × BMI (continuous) in the adjusted logistic model to evaluate whether there is a differential effect of higher UA on the risk of GDM across maternal BMI. We created a heatmap to visualize the contribution of UA and BMI to GDM by presenting UA and BMI levels as ordered categorical variables (50 µmol/L for UA, 3 kg/m² for BMI). Analyses were performed using R 4.2.3 software (R Foundation for Statistical Computing), and statistical significance was determined at P < 0.05. Results Characteristics of the participants We recruited 26,944 women, of whom 16,064 were included in this study, consisting of 3,477 with GDM and 12,587 without GDM. Table 1 summarizes the baseline maternal characteristics and lifestyles of women with and without GDM. Women with GDM were more likely to be older, have lower educational and household income levels, be unemployed, be overweight or obese, not be in their first pregnancy, have had in vitro fertilization conception, have alcohol consumption, not have had a fever during early pregnancy, and have a history of complications in previous pregnancies and higher serum UA levels and blood glucose levels during pregnancy (P < 0.05). Table 1General characteristics of study subjectsCharacteristicsTotal (n = 16064)Non-GDM (n = 12587)GDM (n = 3477)t/χ2/ZP valueAge, Mean ± SD, n (%)29.47 ± 3.4929.28 ± 3.4030.17 ± 3.6913.474< 0.001< 308688 (54.08)7081 (56.26)1607 (46.22)160.749< 0.00130 ~ 346023 (37.49)4593 (36.49)1430 (41.13)≥ 351353 (8.42)913 (7.25)440 (12.65)Ethnicity, n (%)0.5730.449 Han15,814 (98.44)12,396 (98.48)3418 (98.30) Other250 (1.56)191 (1.52)59 (1.70)Education, n (%)24.823< 0.001 Senior high school or lower1955 (12.17)1458 (11.58)497 (14.29) Junior or regular college11,638 (72.45)9134 (72.57)2504 (72.02) Graduate or above2471 (15.38)1995 (15.85)476 (13.69)Occupation, n (%)14.365< 0.001 Working regularly12,382 (77.08)9785 (77.74)2597 (74.69) Workers/peasants/migrant workers217 (1.35)164 (1.30)53 (1.52) Unemployed3465 (21.57)2638 (20.96)827 (23.78)BMI before pregnancy (kg/m2), n (%)450.228< 0.001 Underweight (< 18.5)2218 (13.81)1947 (15.47)271 (7.79) Normal (18.5 ~ 23.9)11,018 (68.59)8812 (70.01)2206 (63.45) Overweight (24.0 ~ 27.9)2289 (14.25)1495 (11.88)794 (22.84) Obese (≥ 28.0)539 (3.36)333 (2.65)206 (5.92)Household income, n (%)12.9910.002 Poverty1776 (11.06)1335 (10.61)441 (12.68) Medium10,347 (64.41)8126 (64.56)2221 (63.88) Rich3941 (24.53)3126 (24.84)815 (23.44)Primiparity, n (%)35.109< 0.001 No6943 (43.22)5287 (42.00)1656 (47.63) Yes9121 (56.78)7300 (58.00)1821 (52.37)Mode of conception, n (%)45.108< 0.001 Spontaneous conception15,026 (93.54)11,858 (94.21)3168 (91.11) Artificial fertilization135 (0.84)101 (0.80)34 (0.98) In vitro fertilization903 (5.62)628 (4.99)275 (7.91)Complications in previous pregnancy, n (%)62.287< 0.001 No15,252 (94.95)12,041 (95.66)3211 (92.35) Yes812 (5.05)546 (4.34)266 (7.65)Smoke or passive smoke, n (%)0.0100.919 No13,986 (87.06)10,957 (87.05)3029 (87.12) Yes2078 (12.94)1630 (12.95)448 (12.88)Drink, n (%)4.0520.044 No15,476 (96.34)12,146 (96.50)3330 (95.77) Yes588 (3.66)441 (3.50)147 (4.23)Folic acid supplement, n (%)0.3540.552 No515 (3.21)409 (3.25)106 (3.05) Yes15,549 (96.79)12,178 (96.75)3371 (96.95)Multidimensional nutrients supplement, n (%)2.0940.148 No8278 (51.53)6524 (51.83)1754 (50.45) Yes7786 (48.47)6063 (48.17)1723 (49.55)Cold during early pregnancy, n (%)3.5320.060 No13,022 (81.06)10,165 (80.76)2857 (82.17) Yes3042 (18.94)2422 (19.24)620 (17.83)Fever during early pregnancy, n (%)16.016< 0.001 No15,636 (97.34)12,218 (97.07)3418 (98.30) Yes428 (2.66)369 (2.93)59 (1.70)UA levels, median (Q25, Q75)212.27 (184.80, 243.54)209.50 (183.00, 240.38)222.33 (193.35, 253.40)13.672< 0.001Gestational week for UA detection, median (Q25, Q75)12.57 (12.14, 13.00)12.57 (12.14, 13.00)12.57 (12.14, 13.00)0.7330.463FPG, median (Q25, Q75)4.66 (4.41, 4.93)4.57 (4.36, 4.78)5.19 (4.97, 5.42)71.935< 0.0011-h PG, median (Q25, Q75)7.39 (6.31, 8.56)7.07 (6.09, 8.03)9.22 (7.88, 10.42)63.094< 0.0012-h PG, median (Q25, Q75)6.47 (5.74, 7.31)6.27 (5.60, 6.95)7.67 (6.69, 8.79)55.040< 0.001Gestational week for OGTT detection, median (Q25, Q75)26.00 (25.43, 26.71)26.00 (25.43, 26.71)26.14 (25.43, 26.71)0.1470.883 General characteristics of study subjects Distribution of women with GDM and abnormal PG values As shown in Table 2, with the quartile of serum UA levels increasing, the incidence of GDM (16.41% vs. 19.48% vs. 22.67% vs. 28.01%), abnormal FPG (10.96% vs. 12.93% vs. 16.38% vs. 20.07%), 1-hour PG (5.71% vs. 6.34% vs. 7.99% vs. 11.44%), and 2-hour PG (5.46% vs. 6.11% vs. 7.54% vs. 10.39%) also increased with statistical significance. Table 2Distribution of women with GDM and abnormal blood glucoses (N = 16046)Study variablesUric acid quartile (µmol/L)χ2P value1 st (≤ 184.79)(n = 4015)2nd (184.80 ~ 212.26)(n = 4015)3rd (212.27 ~ 243.54)(n = 4018)4th (≥ 243.55)(n = 4016)GDM, n (%)659 (16.41)782 (19.48)911 (22.67)1125 (28.01)174.447< 0.001Abnormal FPG, n (%)440 (10.96)519 (12.93)658 (16.38)806 (20.07)151.080< 0.001Abnormal 1-h PG, n (%)228 (5.71)253 (6.34)319 (7.99)456 (11.44)108.769< 0.001Abnormal 2-h PG, n (%)218 (5.46)244 (6.11)301 (7.54)414 (10.39)83.875< 0.001 Distribution of women with GDM and abnormal blood glucoses (N = 16046) Association of UA levels and GDM Table 3 presents the results of univariate and multivariate GLM regression analyses for serum UA levels with GDM. Compared to the first quartile, serum UA levels in the second (RR = 1.15, 95%CI = 1.04–1.28), third (RR = 1.27, 95%CI = 1.15–1.41), and fourth (RR = 1.46, 95%CI = 1.32–1.61) quartiles were associated with an increased risk of GDM (P for trend < 0.001). Serum UA levels in the third and fourth quartiles were associated with an increased risk of abnormal FPG, 1-hour PG, and 2-hour PG (P for trend < 0.001). Similar relationships were observed between PG levels from OGTT as continuous variables and serum UA levels, as detailed in Table S2. Table 3Results of GLM of associations between serum UA levels with GDM and abnormal blood glucoses [RR (95%CI), P value]Study variablesUric acid quartile (µmol/L) P trend 1 st (≤ 184.79)2nd (184.80 ~ 212.26)3rd (212.27 ~ 243.54)4th (≥ 243.55)RefRR (95%CI) P RR (95%CI) P RR (95%CI) P GDM Crude model1 (Reference)1.19 (1.07, 1.32)0.0011.38 (1.25, 1.53)< 0.0011.71 (1.57, 1.88)< 0.001< 0.001 Adjusted model a1 (Reference)1.15 (1.04, 1.28)0.0081.27 (1.15, 1.41)< 0.0011.46 (1.32, 1.61)< 0.001< 0.001Abnormal FPG Crude model1 (Reference)1.18 (1.04, 1.34)0.0111.49 (1.32, 1.69)< 0.0011.83 (1.64, 2.06)< 0.001< 0.001 Adjusted model a1 (Reference)1.12 (0.99, 1.28)0.0701.33 (1.18, 1.50)< 0.0011.46 (1.30, 1.65)< 0.001< 0.001Abnormal 1-h PG Crude model1 (Reference)1.11 (0.93, 1.33)0.2581.40 (1.18, 1.66)< 0.0012.00 (1.71, 2.35)< 0.001< 0.001 Adjusted model a1 (Reference)1.08 (0.90, 1.29)0.4251.27 (1.07, 1.50)0.0071.65 (1.40, 1.95)< 0.001< 0.001Abnormal 2-h PG Crude model1 (Reference)1.12 (0.93, 1.34)0.2291.38 (1.16, 1.64)< 0.0011.90 (1.61, 2.24)< 0.001< 0.001 Adjusted model a1 (Reference)1.11 (0.92, 1.32)0.2991.30 (1.09, 1.55)0.0031.68 (1.42, 1.99)< 0.001< 0.001a Adjusted for maternal age, educational levels, occupation, household income, pre-pregnancy BMI, primiparity, mode of conception, history of pregnancy complications, drinking, and fever during early pregnancy Results of GLM of associations between serum UA levels with GDM and abnormal blood glucoses [RR (95%CI), P value] a Adjusted for maternal age, educational levels, occupation, household income, pre-pregnancy BMI, primiparity, mode of conception, history of pregnancy complications, drinking, and fever during early pregnancy RF prediction model for GDM The RF model included eleven statistically significant variables identified through univariate analysis. The efficacy of the RF model in predicting GDM was assessed using the ROC curve and the AUC values (Table S3). As shown in Fig. 1, the AUCs were 0.673 and 0.611 for the training set and validation set, respectively. The relatively lower AUC in the validation set indicates that the model may lack robustness for accurate prediction. The top three variables contributing to GDM, based on their importance scores, were maternal BMI before pregnancy (0.33), serum UA level (0.25), and age (0.11). The importance score of UA (0.25) among the eleven variables suggests its potential significance in the development of GDM. ROC curve (a) and feature importance of variables (b) in the RF model Associations of higher UA and BMI with GDM We further explored how higher versus lower UA levels affect GDM risk across maternal BMI before pregnancy. As shown in Fig. 2 (a), the test for interaction between UA and BMI was not statistically significant (P = 0.364). The stratified analysis presented in Table 4 showed that the association between higher UA and GDM risk was significant in women with BMI ≥ 18.5 kg/m² (OR = 1.39, 95%CI = 1.29–1.50) but not in those with BMI < 18.5 kg/m² (OR = 1.16, 95% CI = 0.84–1.60). In Fig. 2 (b), the highest absolute risks for GDM were concentrated in the upper right corner of the heatmaps, where UA and BMI levels were highest. Conversely, the lowest risks were observed in the lower left corner, where UA and BMI levels were lowest. The interdependent impact of UA and BMI on GDM risk. Panel (a) shows the effect of higher vs. lower UA on GDM risk across the BMI spectrum. Blue lines and blue shadow areas represent odds ratios with 95% confidence intervals. Panels (b) shows heatmaps of UA and BMI combinations, respectively, displaying absolute GDM risk at 50 µmol/L and 3 kg/m2 intervals. Numbers in each square represent absolute risk, derived from GLM models. Adjusted for maternal age, educational levels, occupation, household income, pre-pregnancy BMI, primiparity, mode of conception, history of pregnancy complications, drinking, and fever during early pregnancy Table 4Subgroup analysis of the higher UA on GDM riskSubgroupLower UA, n (%)Higher UA, n (%)OR a95% CIP valueOverall2352 (19.52)1125 (28.01)1.421.32–1.53< 0.001BMI< 18.5226 (11.93)45 (13.94)1.160.84–1.600.364≥ 18.52126 (20.94)1080 (29.24)1.391.29–1.50< 0.001a Adjusted for maternal age, educational levels, occupation, household income, primiparity, mode of conception, history of pregnancy complications, drinking, and fever during early pregnancy Subgroup analysis of the higher UA on GDM risk a Adjusted for maternal age, educational levels, occupation, household income, primiparity, mode of conception, history of pregnancy complications, drinking, and fever during early pregnancy Discussion In this study, we investigated the relationship between serum UA levels and the risk of GDM in a cohort of 16,046 pregnant women from Northwest China. Our findings indicated that compared to women in the lowest quartile of serum UA levels, those in the second, third, and fourth quartiles were associated with higher risks of developing GDM. Moreover, UA levels in the third and fourth quartiles were also associated with increased risks of abnormal FPG, 1-hour PG, and 2-hour PG. The two most important variables in the GDM prediction model were maternal BMI before pregnancy and serum UA level. Importantly, higher UA was associated with an increased risk of GDM as long as BMI exceeded 18.5 kg/m2. The highest GDM risk was observed with high UA/high BMI combinations, and the lowest risk with low UA/low BMI. These findings underscored the importance of monitoring serum UA levels in women during early pregnancy, as this may help identify individuals at higher risk of GDM who could benefit from further clinical attention. This birth cohort study found that pregnant women with GDM had significantly higher serum UA levels than those without, consistent with earlier studies showing higher serum UA levels in women who develop GDM during 7–24 weeks of gestation [25]. Additionally, our findings suggest that higher serum UA levels during early pregnancy may be significantly associated with a higher risk of GDM, which consolidates existing evidence from large-scale meta-analyses that have consistently reported positive associations between elevated UA and GDM risk [26, 27]. Notably, this pattern mirrors the established link between UA and type 2 diabetes in the general population [28, 29]. Although evidence for a causal relationship between serum UA and diabetes remains limited [30, 31], current research has consistently demonstrated associations between serum UA levels and diabetes in the general population regardless of pregnancy status. Incorporating serum UA testing into routine early pregnancy check-ups may help identify high-risk groups for GDM several months prior to traditional OGTT screening, thereby providing a time window for early intervention. This study highlights a significant association between elevated serum UA levels and dysglycemia across three time points during pregnancy. This aligns with a growing body of evidence from diverse populations, including large-scale adult cohorts [32] and studies of youth [33], which positions hyperuricemia as a correlate of insulin resistance and impaired glucose metabolism. Notably, the association often appears stronger in women [32, 34], a finding that lends epidemiological context to our results in a pregnant cohort. While previous research has primarily focused on FPG, our study provides crucial new insights by demonstrating that high UA is equally relevant to postprandial glycemic control in pregnancy, suggesting that comprehensive glycemic assessment including postprandial metrics may be particularly important for women with elevated early-pregnancy UA levels. Although our RF model identified pre-pregnancy BMI and serum UA as the most important predictors, we acknowledge its limited predictive performance. The validation set AUC of 0.611 is modest for clinical use, limiting direct utility for individual prediction. UA-incorporating prediction models for GDM show significant variability in predictive performance across studies [14, 35, 36]. These discrepancies likely reflect the complex multifactorial nature of GDM involving genetic, metabolic, and lifestyle factors not fully captured in typical datasets. Thus, our model serves mainly as proof of concept underscoring the associative role of UA alongside established risk factors like BMI. The high importance score of UA despite modest overall AUC reinforces its potential biological significance and warrants further study in larger cohorts with richer feature sets. Our analysis showed that although the UA and BMI interaction was not statistically significant, the stratified analysis revealed that the association between high UA and GDM was strong in women with BMI ≥ 18.5 kg/m² but absent in underweight women. This finding suggests that higher pre-pregnancy BMI may create a pro-inflammatory and insulin-resistant metabolic environment [37] that amplifies the detrimental effects of UA on glucose regulation. The risk heatmap visually corroborates this relationship, showing the highest absolute GDM incidence in the high UA/high BMI subgroup. Therefore, the combination of these two readily available clinical markers effectively identifies a high-risk phenotype that warrants prioritization for early screening, close monitoring, and targeted preventive strategies in routine clinical practice. Elevated serum UA levels may be associated with increased GDM risk, and this association may be linked to mitochondrial dysfunction, oxidative stress, the generation of inflammatory mediators such as TNF-α, and decreased lipocalin levels [38–40]. UA can induce insulin resistance by interfering with the IRS2/AKT pathway and impair pancreatic β-cell function [41]. It also activates inflammatory responses, including the NLRP3 inflammasome and HMGB1-TLR4-NF-κB signaling pathway [42]. Additionally, UA promotes hepatic steatosis by inducing endoplasmic reticulum stress and activating SREBP-1c, further exacerbating metabolic dysfunction [43, 44]. Collectively, these processes disrupt glucose metabolism and insulin secretion, thereby increasing susceptibility to GDM. Future research is needed to determine whether interventions for lowering uric acid can effectively prevent GDM in high-risk populations. This study had both strengths and limitations. Firstly, it was one of the largest prospective birth cohort studies investigating the association between serum UA levels and GDM in women during early pregnancy, providing robust data on maternal baseline characteristics. Secondly, by detecting serum UA levels before 14 weeks of gestation, we highlighted the critical role of early serum UA accumulation in its association with a higher risk of GDM compared to later accumulation [45]. Thirdly, we comprehensively explored the correlations between serum UA levels and maternal blood glucose levels at three distinct time points, enhancing our understanding of the association between UA and the dynamics of blood glucose. However, our study also had certain limitations. The generalizability of our findings may be limited by the single-center design in Northwest China and the exclusion of a substantial proportion of the initial cohort due to missing data, which could introduce selection bias, although baseline characteristics were similar between included and excluded participants. Additionally, it is important to note that due to the observational nature of our study design, we cannot establish a causal relationship between elevated UA levels and the development of GDM. While our findings show a significant association, residual confounding or reverse causality cannot be entirely ruled out. Moreover, the predictive performance of our RF model was modest, indicating limited clinical utility for individual-level prediction at this stage. This limitation likely reflects the complex multifactorial etiology of GDM that may not be fully captured by the available variables. Further multi-center intervention studies involving diverse populations of pregnant women are needed to validate the causal relationships and elucidate the underlying mechanisms linking UA levels to GDM. Conclusion In this prospective cohort study of pregnant women in Northwest China, we observed that higher UA levels during early pregnancy were associated with an increased risk of GDM across normal weight, overweight, and obese populations based on BMI classifications. Our findings highlight the potential value of monitoring early pregnancy UA levels as a marker linked to GDM risk. Further research using robust methodologies and larger sample sizes is warranted to confirm and extend these results. Supplementary Information Supplementary Material 1. Acknowledgements The authors would like to express gratitude to the participants of all mothers. We also thank staff members from Northwest Women’s and Children’s Hospital and Xi’an Jiaotong University for their collaborating with data collection in this study. Authors' contributions The authors’ contributions are as follows: D.Z. and L.S. analyzed the data and contributed to the original draft; D.Z., N.L., Y.S., J.C., X.L., Y.S., D.L. and L.S. collected and cleared the data. L.S., Y.M. and P.Q. contributed to the review and revision of the manuscript. Y.M., P.Q. and L.S. conceived and designed the study. All the authors have read and approved the manuscript. Funding This study was funded by the National Natural Science Foundation of China (No. 82574206, 82103924), the Shaanxi Health Science and Technology Innovation Capacity Enhancement Program (No. 2025YF-09), and the Key Research and Development Program of Shaanxi Province (No. 2025SF-YBXM-227, 2024SF-YBXM-238, and 2022ZDLSF02-11). Data availability The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate The study was conducted in accordance with the ethical standards of the Declaration of Helsinki and its later amendments. The study protocol was approved by the Human Research Ethics Committee of Beijing Obstetrics and Gynecology Hospital (No. 2018-KY-003-02) and Northwest Women’s and Children’s Hospital (No. 2018018). Written informed consent was obtained from all participants prior to their inclusion in the study. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Yang Mi, Email: miyangmm@163.com. Pengfei Qu, Email: xinxi3057@163.com. Lei Shang, Email: shanglei@fmmu.edu.cn. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.