Background

Uric acid (UA) is one of the major factors that were associated with an increased risk of cardiovascular disease [1, 2]. Evidence suggests that higher UA levels also contribute to other chronic diseases such as type 2 diabetes mellitus [3], metabolic syndrome [4], and non-alcoholic fatty liver disease [5]. During pregnancy, higher UA levels are known to be associated with an increased risk of gestational hypertension, pre-eclampsia/eclampsia [6,7,8], and subsequent adverse birth outcomes [9]. However, the associations between UA and other complications, such as gestational diabetes mellitus (GDM) and adverse birth outcomes, have been relatively unclear in normotensive subjects. In addition, previous studies in this area have produced inconsistent results. Several studies have suggested that hyperuricemia is associated with an increased risk of GDM [10,11,12,13] or adverse birth outcomes [14,15,16]. One study showed that both high and low UA levels contribute to adverse fetal growth [17]. Marginal or null associations with these outcomes have been reported for UA in several other studies [18,19,20,21,22]. The small sample size of most studies, racial heterogeneity, and different trimesters studied may partially explain the heterogeneity of these findings. However, it should be recognized that gestational cardiovascular complications (hypertension, pre-eclampsia, eclampsia) share many common mechanisms with GDM, and independently lead to adverse birth outcomes. Therefore, without eliminating the influence of these patients, the associations of UA with other outcomes in the general population may be overestimated. In addition, with the potential function of antioxidant, UA has also been suggested to be beneficial for degenerative diseases such as osteoporosis [23], and showed U-shaped associations for several diseases [24,25,26]. A U-shaped rather than linear association of UA with fetal growth has been reported [17]. The functional forms of UA with pregnancy outcomes have been inconsistent or poorly investigated in other studies and warrant further investigation.

Therefore, we aimed to perform a retrospective analysis based on medical record data from a large obstetric center in southern China to better illustrate these problems.

Methods

Subjects

The study was based on retrospective data from the Southern Medical University Affiliated Maternal & Child Health Hospital of Foshan from January 1, 2012 to July 31, 2018. The hospital is the largest obstetric center in Foshan City, Guangdong Province, China, covering a large population of more than 7.7 million people. Singleton pregnant women were included in the analyses if they had serum UA measured at antenatal care visits during the first 20 weeks of pregnancy and delivered a live birth at the hospital. A total of 27,582 subjects provided matching information on exposures and outcomes during the same pregnancy. Subjects were further excluded if they met the following criteria: (a) with UA concentrations detected after 20 weeks’ gestation (7,107 subjects); (b) a history of serious medical conditions: including gestational hypertension/pre-eclampsia/eclampsia (608 subjects), type 1 or type 2 diabetes (47 subjects), malignancy (15 subjects), and thyroid dysfunction (836 subjects); (c) missing core data (407 subjects) or suspicious outliers (> 3SD or <-3 SD, 312 subjects). Finally, a total of 18,250 mother-infant pairs were included in the analyses (Fig. 1).

Fig. 1
figure 1

Flow characteristic of the study

Full size image

Data collection

Serum UA concentration data were obtained from clinical laboratory records or medical records. Blood samples were collected during routine obstetric examinations and measured immediately by the clinical laboratory at the hospital without freezing. UA was detected by the uric acid enzyme colorimetric method using an automated biochemical analyzer (AU5800, Beckman Coulter, Inc, USA). Commercial kits were purchased from Ningbo Ruiyuan Biotechnology Co., Ltd, China. The intra- and inter-assay coefficients of variation were less than 4% and 15%, respectively. To reduce the possibility of influence from pregnancy complications, we restricted the data to UA measured before the first 20 weeks of gestation. Hyperuricemia in early pregnancy is defined as serum UA concentrations ≥ 1 SD of reference values [27] at the gestational ages (6 ~ 12, > 12 ~ 13, > 13 ~ 14, > 14 ~ 15, > 15 ~ 16, > 16 ~ 17, > 17 ~ 18, > 18 ~ 19, and > 19 ~ 20 gestational weeks) in our study, as shown in Supplemental Table 1.

Information including maternal age, body mass index (BMI), gestational week of UA measurement, parity, time of last menstrual period and delivery, history of related diseases were obtained from medical records. The outcomes of GDM, PB, LBW, and macrosomia were diagnosed by professional obstetricians with the same criteria, and extracted from medical records. GDM was assessed by the oral glucose tolerance test at 24 to 28 weeks’ gestation, and diagnosed if subjects met any items of the following criteria: fasting blood glucose ≥ 5.1 mmol/L; one-hour blood glucose after oral glucose ingestion ≥ 10.0 mmol/L; two-hour blood glucose after oral glucose ingestion ≥ 8.5 mmol/L. PB was defined as delivery at ≥ 28 and < 37 weeks; gestation. Neonatal birth weight < 2500 g was defined as LBW, and ≥ 4000 g was defined as macrosomia. Based on the latest criteria (WS/T 800—2022) promulgate by the National Health Commission of the People’s Republic of China [28], appropriate for gestational age (AGA) was defined as neonates whose birth weights were between 10th to 90th percentile by gestational age. Small for gestational age (SGA) and (LGA) was defined as neonates whose birth weights were < 10th percentile and > 90th percentile by gestational age, respectively.

Statistical analyses

Continuous variables were presented as mean ± standard deviation (SD) or median (interquartile range) and tested by Student’s t-test or non-parametric test. Categorical variables were presented as frequencies (percentages) and tested by chi-squared test. Logistic regression analyses were performed to examine the associations between UA and outcomes, including GDM, PB, LBW, macrosomia, SGA, and LGA. UA was analyzed as both a categorical (hyperuricemia versus normal) and continuous variable (per one SD increase). Two different analysis models were performed, model 1 as a univariate model without any adjustment and model 2 adjusted for maternal age, BMI, parity, and gestational week of UA measurement. Stratified analyses were performed for different groups of maternal age (< 35 or ≥ 35 years), BMI (< 24, overweight: ≥24 and < 28, obesity: ≥28 kg/m2) [29], parity (0 or ≥ 1), gestational week of UA measurement (< 17 or ≥ 17 weeks). Analyses were performed with SPSS software (version 21.0, Chicago, IL, USA). Generalized additive regression models (GAMs) were used to explore the functional forms of the association between UA and related outcomes and were performed using the R software (version 4.3.1, Vienna, Australia). A two-sided P < 0.05 was considered as statistically significant.

Results

A total of 18,250 pregnant women aged 29.2 ± 4.53 years were included in the analyses, of whom 2,896 (15.9%) had hyperuricemia. As shown in Table 1, maternal serum UA concentrations were measured at a median of 17.6 (IQR: 16.3, 18.6) weeks, with a mean value of 0.22 ± 0.05 mmol/L. The incidence was highest for LGA (13.2%), followed by SGA (9.2%), GDM (7.6%), PB (5.0%), LBW (4.4%), and lowest for macrosomia (2.4%). Subjects with hyperuricemia tended to have higher BMI (26.6 vs. 25.8 kg/m2, P < 0.001) and serum UA concentration (0.30 vs. 0.21 mmol/L, P < 0.001); higher incidence of LGA (15.0% vs. 12.9%, P = 0.007), GDM (9.8% vs. 7.2%, P < 0.001), CS (45.8% vs. 42.7%, P = 0.002), and preterm birth (6.1% vs. 4.8%, P = 0.004); and have a lower delivery gestational age (38.7 vs. 38.9 weeks’ gestation, P < 0.001).

Table 1 Characteristic of subjects
Full size table

After adjustment for several potential covariates, higher UA levels in early pregnancy were associated with several adverse outcomes (Table 2). The ORs (95%CI) per one SD increase in UA levels were 1.205 (1.136, 1.277) for GDM, 1.137 (1.060, 1.221) for PB, 1.134 (1.051, 1.223) for LBW, and 1.077 (1.020, 1.137) for SGA, respectively. Similarly, hyperuricemia (vs. normal) was associated with a 39.4% (OR: 1.394, 95%CI: 1.211–1.606) higher risk of GDM, and a 38.5% (OR:1.385, 95%CI: 1.168–1.643) higher risk of PB, but not with LBW, macrosomia, SGA, and LGA. In the analysis of GAMs, when the equivalent degrees of freedom (edf) value was closer to 1.0, the relationship between uric acid and outcome tended to be linear. When the edf value was further from 1.0, the relationship between uric acid and outcome tended to be non-linear. A p-value of less than 0.05 indicates a statistical association between uric acid and outcome in the GAMs model. Serum UA levels showed linear relationships with GDM (edf = 1.11, P < 0.001), PB (edf = 1.00, P < 0.001), LBW (edf = 1.00, P = 0.002), and SGA (edf = 1.00, P = 0.007) based on GAMs (Fig. 2). The association between UA and LGA tended to be non-linear (edf = 2.62, P = 0.008). No significant associations were found between UA and macrosomia (edf = 1.00, P = 0.115).

Table 2 Associations between early pregnancy serum uric acid levels and gestational diabetes & adverse infant birth outcomes (N = 18,250)
Full size table
Fig. 2
figure 2

The relationships between early pregnant serum uric acid levels and outcomes of gestational diabetes mellitus (A), preterm birth (B), low birth weight (C), macrosomia (D), small for gestational age (E), and large for gestational age (F) based on the generalised additive regression models (n = 18,250). The covariates were maternal age, BMI, parity, and gestational weeks of uric acid measurement. Dotted lines represented the 95% confidence intervals. When the equivalent degrees of freedom (edf) value was close to 1.0, the relationship between uric acid and outcome tended to be linear. When the edf value was further from 1.0, the relationship between uric acid and outcome tended to be non-linear. A-p value of less than 0.05 indicates a statistical association between uric acid and outcome in the GAMs model

Full size image

No significant interactions were observed between higher serum UA levels (both hyperuricemia or per one SD increase in UA concentration) and the gestational age of UA measurement on the risk of adverse outcome, as shown in Table 3. As shown in Table 4, significant interactions were observed between higher serum UA levels (per one SD increase in UA concentration) and the BMI groups (< 24, ≥ 24 and < 28, ≥28 kg/m2) on the risk of PB (P interaction = 0.001), and LBW (P interaction = 0.008). Significant interactions were observed between hyperuricemia and the BMI groups (< 24, ≥ 24 and < 28, ≥28 kg/m2) on the risk of SGA (P interaction = 0.028). No significant interactions were found between higher serum UA levels and age (< 35 and ≥ 35 years) or parity (0 and ≥ 1) on the risk of all outcomes, and the p-values of the interaction ranged from 0.097 to 0.826 and from 0.133 to 0.986, respectively (data not shown).

Table 3 Associations between early pregnancy serum uric acid levels and gestational diabetes & adverse infant birth outcomes stratified by gestational weeks of measurement
Full size table
Table 4 Associations between early pregnancy serum uric acid levels and gestational diabetes & adverse infant birth outcomes stratified by BMI
Full size table

Discussion

In this study based on retrospective data from an 18,250 mother-infants cohort in southern China, we observed that higher serum UA (both as a categorical and continuous variable) in early pregnancy was associated with higher risk of GDM, PB, LBW, and SGA. The dose-response models tended to be linear for most outcomes. Results tended to be more pronounced in subjects with higher BMI for the outcomes of PB, LBW, and SGA, with significant interactions being found for these associations.

The prevalence of hyperuricemia in our study was 15.9% (subjects), which was relatively higher than that of nonpregnant Chinese adults in a study of 5,939 subjects (14.1%) [30] and a study of 11,601 subjects (11.15%) in Shanghai City [31]. Serum UA concentrations increased during pregnancy and are affected by gestational age, which may partly explain why the prevalence of hyperuricemia was higher than that in nonpregnant subjects, although we tried to adjust hyperuricemia for gestational age. However, the prevalence of hyperuricemia in our study was relatively lower when compared with another study of 404 normotensive singleton women (103 subjects, 25.4%) [15]. The prevalence of hyperuricemia was barely mentioned in the previous study, moreover, hyperuricemia was common defined as ≥ 1 SD values at corresponding age, was sample-based, and UA was measured at different trimesters. These increased the difficulty in comparing the prevalence of hyperuricemia among different populations, and these issues should be further considered in the future.

The positive associations found in our study were consistent with several previous studies. Higher UA levels (or hyperuricemia) in the first 20 weeks of pregnancy were associated with a 2.34- and 1.11- fold increased risk of GDM in 1000 Chinese [11] and 5507 Israeli women [12], respectively. Similar results have been observed in a retrospective cohort study of 23,843 Chinese with UA measured before 24 weeks of gestation [32] and other studies [10, 13]. However, several studies also reported marginal or null associations between UA and GDM [18, 20, 22]. Adverse associations with PB, SGA, or BW were also found in several studies with small sample size of less than 500 [14,15,16], which were consistent with ours. In a retrospective study of 11,580 Chinese, higher UA in late pregnancy (the corresponding gestational weeks at delivery) was associated with a higher risk of LBW, SGA, but also associated with a lower risk of PB [33]. Marginal or null significant results were found in a cross-sectional study of 885 Germans [19] and in a prospective cohort of 1541 Americans [21]. In another prospective cohort of 1291 Americans, both too high and too low UA levels were detrimental to fetal growth [17]. The UA measured in different trimesters or periods of pregnancy may partly explain the heterogeneity of these results, as UA levels tend to be influenced by the course of pregnancy [27]. In our study, linear rather than U-shaped relationships were observed for UA and most gestational or birth outcomes. Highlighting the associations with adverse outcomes of UA in early pregnancy instead of late pregnancy, may provide a better opportunity for pregnant women to improve their UA status and possibly partially prevent the occurrence of adverse outcomes. In addition, the longer time between UA measurement (early versus late pregnancy) and outcomes increases the ability to avoid the possibility of causal inversion. Our study included a large number of subjects, which may reduce the possibility of false-negative results that may have occurred in previous studies with much smaller sample sizes. Our results, along with several, but not all, studies, underscore the importance of UA in early pregnancy as an indicator of GDM and adverse birth outcomes, even in normotensive subjects.

Several mechanisms may contribute to the adverse associations of UA. Higher UA might affect GDM by increasing the risk of insulin resistance during pregnancy [34], through several potential pathways, including endothelial dysfunction [35], decreased production of nitric oxide [36, 37], and induction of metabolic inflammation [38]. Apart from the mechanisms mentioned above, elevated UA might inhibit amino acid uptake in the placenta [39], and attenuate trophoblast invasion and integration into endothelial cell monolayers [40], leading to poor placental development. Most of these mechanisms have also been implicated in the development of adverse gestational cardiovascular complications. Because adverse associations were observed in pregnant women without the occurrence of cardiovascular complications, other potential mechanisms may exist, and need to be well characterized by further studies.

More pronounced adverse associations were found in subjects with higher BMI for outcomes of PB, LBW, and SGA. Indeed, obesity is one of the most important pathways for adverse outcomes [41]. Moreover, obesity also leads to higher UA levels [42, 43], although we tried to attenuate its influence by adjusting for BMI. Our results suggest that clinicians should make more precise considerations for subjects with both high UA levels and obesity in early pregnancy.

Our study has several strengths. The study was based on a large sample of 18,250 subjects, which provided more precise estimates of the associations. Serum UA data were measured and collected in early pregnancy, which established the temporal sequence of the observed associations. We examined the functional forms of the dose-response associations, and potential interactions from multiple variables, including age, BMI, UA measurement times, and parity. Our results provided further evidence for obstetric clinicians to identify higher UA levels and potential high-risk patients.

There are also several limitations to the study that deserve careful consideration. First, serum levels of UA were measured and recorded only once during the pregnancy in subjects without gestational cardiovascular complications, so we were unable to analyze the dynamic course of UA and its influence on outcomes during the early pregnancy. Second, the study was based on data from an obstetric center instead of a community, which might have weakened the representativeness of the sample, but this could partly be counteracted because the obstetric center was the largest one in the areas and received patients from different regions of the city. In addition, it would be of very difficult to collect medical record information from the community for a retrospective study. Third, although we adjusted for several potential covariates, there may still be residual confounding.

Conclusions

In conclusion, this retrospective cohort study found that higher UA levels in early pregnancy were associated with higher risk of GDM, PB, LBW, and SGA in normotensive Chinese women.