Validating Multicenter Cohort Circular RNA Model for Early Screening and Diagnosis of Gestational Diabetes Mellitus

Abstract

Background Gestational diabetes mellitus (GDM) is a metabolic disorder posing significant risks to maternal and infant health, with a lack of effective early screening markers. Therefore, identifying early screening biomarkers for GDM with higher sensitivity and specificity is urgently needed.

* Methods High-throughput sequencing was employed to screen for key circular RNAs (circRNAs), which were then evaluated using reverse transcription quantitative polymerase chain reaction. Logistic regression analysis was conducted to examine the relationship between clinical characteristics, circRNA expression, and adverse pregnancy outcomes. The diagnostic accuracy of circRNAs for early and mid-pregnancy GDM was assessed using receiver operating characteristic curves. Pearson correlation analysis was utilized to explore the relationship between circRNA levels and oral glucose tolerance test results. A predictive model for early GDM was established using logistic regression.

* Results Significant alterations in circRNA expression profiles were detected in GDM patients, with hsa_circ_0031560 and hsa_ circ_0000793 notably upregulated during the first and second trimesters. These circRNAs were associated with adverse pregnancy outcomes and effectively differentiated GDM patients, with second trimester cohorts achieving an area under the curve (AUC) of 0.836. In first trimester cohorts, these circRNAs identified potential GDM patients with AUCs of 0.832 and 0.765, respectively. The early GDM prediction model achieved an AUC of 0.904, validated in two independent cohorts.

* Conclusion Hsa_circ_0031560, hsa_circ_0000793, and the developed model serve as biomarkers for early prediction or mid-term diagnosis of GDM, offering clinical tools for early GDM screening.

Highlights

• This study identified sensitive biomarkers for early GDM screening using circRNAs.

• We validated hsa_circ_0031560 and hsa_circ_0000793 as reliable biomarkers for GDM.

• We developed an early GDM prediction model (E-GDMM) with high diagnostic accuracy.

• The model was validated in multiple cohorts, showing strong sensitivity and specificity.

INTRODUCTION

Gestational diabetes mellitus (GDM) is a pregnancy complication involving abnormal glucose tolerance, with increasing incidence due to dietary changes [1,2]. It poses serious risks to mothers, such as postpartum hemorrhage, metabolic syndrome, gestational hypertension, and type 2 diabetes mellitus, and to fetuses, increasing the chances of macrosomia, hypoglycemia, and distress [3,4]. The oral glucose tolerance test (OGTT), conducted between the 24th and 28th weeks of pregnancy, is the standard for diagnosing GDM [5]. Early identification of predictive and diagnostic biomarkers is essential to reduce perinatal complications and improve long-term health outcomes for mothers and infants.

Circular RNAs (circRNAs), a distinct class of non-coding RNAs, possess a unique covalently closed loop structure, providing resistance to RNase R enzyme degradation and greater biological stability compared to linear RNAs [6-9]. They act as potent biomarkers and have significant effects in diseases like cancer and GDM [10-12]. Research indicates their diagnostic potential, such as hsa_circRNA_0054633’s correlation with glycated hemoglobin A1 (GHBA1) and GHBA1c levels during pregnancy, and hsa-circRNA_0039480’s association with OGTT results in GDM patients [13,14]. Despite these promising findings regarding the diagnostic potential of circRNAs for GDM, most research has predominantly concentrated on identifying diagnostic markers during the second trimester, serving as an auxiliary tool. These studies have not yet provided a comprehensive framework for the early detection and diagnosis of GDM.

In this study, we aimed to identify more sensitive and specific early diagnostic markers for GDM by comparing the expression profiles of circRNAs in serum samples from GDM patients and matched controls. We identified two key circRNAs, hsa_circ_0031560 and hsa_circ_0000793, which are significantly upregulated in the serum of GDM patients. Our findings indicate that hsa_circ_0031560 and hsa_circ_0000793 can effectively distinguish between GDM patients and controls during the second trimester, and this distinction is also applicable during the early stages of GDM. By constructing an early GDM prediction model (E-GDMM) utilizing these circRNAs, we were able to significantly differentiate potential patients who may develop GDM. Thus, our study presents a novel opportunity for the early diagnosis of GDM.

METHODS

Study design

This study, approved by the Ethics Committee of Nanjing Maternity and Child Health Care Hospital (Ethical Review Report Number: NFKSL-077) and the Clinical Research Ethics Committee of Zhongda Hospital, Southeast University (Ethical Review Report Number: 2024ZDSYLL299-P01), is both a retrospective case-control study and a nested case-control study. All participants provided informed consent. Pregnant women underwent a 75-g OGTT between the 24th and 28th weeks of gestation. According to the criteria set forth by the International Diabetes and Pregnancy Study Groups, GDM was diagnosed if any of the following glucose levels were observed: fasting plasma glucose ≥5.1 mmol/L, 1-hour post-load glucose ≥10.0 mmol/L, and 2-hour post-load glucose ≥8.5 mmol/L. Exclusion criteria encompassed individuals with other pregnancy-related pathologies, chronic hypertension, multiple gestations, gynecological disorders, hepatic or renal diseases, malignancies, pre-existing type 1 or type 2 diabetes mellitus, or obesity. Subjects who were smokers were also excluded. Serum samples were obtained from all pregnant women, and a comprehensive set of clinical parameters was recorded at the time of sample collection. These parameters included maternal age, gestational age, systolic and diastolic blood pressure, body mass index (BMI), levels of alanine aminotransferase, aspartate aminotransferase, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, triglycerides, total cholesterol, serum creatinine, and blood urea nitrogen.

This study initially included 781 pregnant women at different stages of pregnancy. After excluding six cases with prepregnancy type 2 diabetes mellitus, eight cases of multiple pregnancies, and 25 cases lost to follow-up during the second trimester, a total of 358 first trimester, 344 second trimester, and 40 third trimester samples were collected and divided into five cohorts: (1) second trimester cohort 1: 184 second trimester women from Nanjing Maternal and Child Health Hospital, including 92 cases of GDM and 92 controls; (2) second trimester cohort 2: 160 second trimester women from Zhongda Hospital, Southeast University, including 73 GDM cases and 87 controls; (3) third trimester cohort: 80 third trimester women from Zhongda Hospital, Southeast University, including 40 GDM cases and 40 controls; (4) first trimester validation cohort 1: 228 first trimester women from Nanjing Maternal and Child Health Hospital, with 46 cases developing GDM in the second trimester and 182 cases not developing GDM in the second trimester (48 of these were used for model construction, and the remaining were assigned to the early validation cohort); and (5) first trimester validation cohort 2: 130 first trimester women from Zhongda Hospital, Southeast University, with 22 cases developing GDM in the second trimester and 108 cases not developing GDM in the second trimester. In this study, serum samples were collected from all women during the first trimester (10–14 weeks), second trimester (24–28 weeks) and third trimester (28–35 weeks). First trimester women underwent an OGTT during the second trimester, and based on the results, they were categorized into the group that develop GDM (D-GDM group) and the group that did not develop GDM (UD-GDM group). In addition, we collected serum samples from an additional 40 age-matched healthy non-pregnant women (control group) undergoing routine health check-ups, as well as 40 postpartum women (PW) within 1 week after delivery (PW group) to serve as controls.

Construction and validation of the E-GDMM

The development and validation of the cohort involved the analysis of serum samples from early pregnancy in 20 patients who later D-GDM and 28 who did not. This study focused on evaluating the expression levels of hsa_circ_0031560 and hsa_circ_0000793. Incorporating the expression levels of these circRNAs as independent variables and GDM outcomes as the dependent variable, the E-GDMM was constructed using logistic regression methods. The diagnostic efficacy of the model was validated using receiver operating characteristic (ROC) curve analysis, obtaining key indicators such as sensitivity and specificity to determine the optimal cut-off value.

To verify the stability of the model, early pregnancy cohorts from Nanjing Maternal and Child Health Hospital and Zhongda Hospital, Southeast University, were collected. The expression levels of circRNAs in these samples were similarly measured and input into the model. The sensitivity and specificity in different cohorts were assessed based on the identified cutoffs from the model, thereby validating its robustness and predictive accuracy across multiple populations.

This refined text improves clarity and cohesiveness, making it suitable for academic publication.

Statistical analysis

Statistical analyses were conducted using GraphPad Prism version 9.5 (GraphPad Software Inc., San Diego, CA, USA) and SPSS version 26.0 (IBM Co., Armonk, NY, USA). Variables were expressed as mean±standard deviation, median (interquartile range), or percentages, depending on their distribution. The Kolmogorov-Smirnov test was first applied to determine the normality of data distributions. For normally distributed variables, the homogeneity of variance test was subsequently performed. When both normal distribution and homogeneity of variance were confirmed, an independent samples t-test was utilized. Otherwise, the Mann-Whitney U test was employed. Pearson’s chi-square test was applied to compare frequencies of categorical variables. A P value of less than 0.05 was considered statistically significant. Pearson correlation analysis was used to assess the relationship between two variables. To evaluate the accuracy of circRNAs in distinguishing different groups, ROC curve analysis was performed, and the area under the curve (AUC) was calculated.

RESULTS

Identification of circRNA expression profiles in GDM

To study the expression profile of circRNAs in GDM patients, we reanalyzed next-generation sequencing data from the placental villi tissues of three GDM patients and three normal pregnancies from the study by Yan et al. [15] (Supplementary Methods). A total of 33,574 circRNAs were identified, including 19,359 previously known circRNAs from the circBase database (http://circrna.org/) and 14,215 newly identified circRNAs (Supplementary Fig. 1A). Compared to the 48,270 circRNAs reported by Yan et al. [15], our results differ, possibly due to improved data filtering strategies and updated databases. In addition to features not mentioned in Yan et al.’s study [15], we further discovered that these circRNAs are distributed across all chromosomes, mainly composed of 2–4 exons, with most lengths between 201 and 600 base pairs. Additionally, most genes produce only 1 or 2 circRNAs (Supplementary Fig. 1B-E). This distribution and structural characteristics of circRNAs provide new insights into their functions in GDM. After screening for differentially expressed circRNAs between the two groups, 474 circRNAs exhibited significant changes in expression, with 268 upregulated and 206 downregulated (|log₂ fold change [FC]| ≥1, P<0.05) (Fig. 1A and (B). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis showed that these differentially expressed circRNAs are mainly involved in pathways such as homologous recombination, lysine degradation, and cortisol synthesis and secretion (Fig. 1C). Furthermore, Gene Ontology (GO) analysis revealed that these circRNAs may influence the development of GDM by regulating biological processes like protein degradation, metabolism, and epithelial cell migration (Fig. 1D). Although our pathway analysis results are similar to those of Yan et al. [15], indicating a close relationship between these circRNAs and glucose and lipid metabolism, we further refined the functional speculations of the differentially expressed circRNAs. These circRNAs may play more complex roles in the occurrence and development of GDM, providing new clues for future mechanistic studies.

Screening and identification of hsa_circ_0031560 and hsa_circ_0000793

To identify circRNAs with biomarker potential, we employed a progressively stringent filtering approach. First, we conducted an initial screening of the raw data using |log₂FC| >1 and P< 0.05 as criteria, identifying 474 differentially expressed circRNAs, with 268 upregulated and 206 downregulated (Supplementary Table 1). Next, we tightened the filtering criteria to |log₂ FC| >2 and P<0.05, selecting 87 differentially expressed circRNAs, which included 43 upregulated and 44 downregulated circRNAs. Given that highly expressed circRNAs exhibit higher sensitivity and better linear relationships, we focused on the 43 upregulated circRNAs (Supplementary Table 2). Further analysis of the expression stability of these circRNAs in samples led us to select those detected in over 90% of the sequencing samples. Ultimately, five circRNAs met this standard and were identified as potential biomarkers (Supplementary Table 3). Subsequently, we assessed the expression levels of these circRNAs in serum samples from 20 GDM patients in the second trimester and normal control (NC) pregnant women. The results indicated that only hsa_circ_0031560 and hsa_circ_0000793 were significantly upregulated in the serum of GDM patients, consistent with the sequencing data (Fig. 2A). The sequences of relevant primers are listed in Supplementary Table 4. Consequently, these two circRNAs were selected for further investigation. Using the circPrimer software, we identified that hsa_circ_0031560 and hsa_circ_0000793 are derived from the HEAT repeat containing 5A (HEATR5A) and ubiquitin specific peptidase 32 (USP32) genes, respectively, comprising 8 and 4 exons with lengths of 1,160 and 419 bp (Supplementary Fig. 2A). Additionally, agarose gel electrophoresis and Sanger sequencing confirmed the primer specificity for these circRNAs (Fig. 2B, Supplementary Fig. 2B). Furthermore, following RNase R and actinomycin D treatment, both circRNAs demonstrated greater stability compared to their corresponding linear parental genes (Fig. 2C and D). To explore the secretion dynamics of these circRNAs, we cultured the human trophoblast cell line human trophoblast 8 (HTR-8)/SV40 large T-antigen-transfected cells (SVNEO) for 48 hours and measured the expression levels of hsa_circ_0031560 and hsa_ circ_0000793 in the culture supernatants at various time points. We observed that the expression of both circRNAs progressively increased over time (Fig. 2E). Collectively, these findings suggest that hsa_circ_0031560 and hsa_circ_0000793 are upregulated in the serum of GDM patients and hold potential as novel biomarkers.

Expression characteristics and clinical significance of hsa_circ_0031560 and hsa_circ_0000793 in GDM

To investigate the expression patterns and clinical significance of hsa_circ_0031560 and hsa_circ_0000793 in GDM, we performed reverse transcription quantitative polymerase chain reaction analysis on serum samples from four groups of women: 140 women who remained free from GDM throughout pregnancy (NC group), 132 women with GDM, 40 PW group, and age-matched non-pregnant women (control group). The analysis revealed that the expression levels of both circRNAs were significantly upregulated during pregnancy compared to the non-pregnant group, with an even more pronounced upregulation observed in the GDM patients. Notably, after delivery and placental removal, the expression levels of both circRNAs rapidly declined, returning to levels comparable to those seen in non-pregnant women (Fig. 3A and B). This suggests a potential link between circRNA expression and the presence of the placenta. Further examination of circRNA expression throughout different pregnancy stages showed that hsa_circ_0031560 expression progressively increased with gestation, with a more marked increase in the GDM group (Fig. 3C). In contrast, hsa_circ_0000793 reached peak expression in the second trimester in GDM patients’ serum but exhibited similar levels between the GDM and control groups during the third trimester (Fig. 3D). Clinical data from the second and third trimester groups showed significant differences in OGTT results (Supplementary Table 5). Correlation analysis revealed a strong association between OGTT results and hsa_circ_0031560 expression (Fig. 3E), while a moderate correlation between hsa_circ_0031560 and hsa_circ_0000793 expression was also observed (Fig. 3F). In conclusion, these circRNAs may play critical roles in the development and progression of GDM, particularly through their involvement in placental function and glucose metabolism regulation.

Association of hsa_circ_0031560 and hsa_circ_0000793 with adverse pregnancy outcomes

We systematically monitored and documented the incidence of adverse pregnancy outcomes among women in both the GDM and NC groups across different stages of pregnancy. Our findings revealed a notably higher incidence of adverse pregnancy outcomes in the GDM group compared to the NC group at all stages (Supplementary Table 6). To investigate the potential association between key clinical characteristics, the expression levels of the two circRNAs, and adverse pregnancy outcomes, we performed univariate and multivariate logistic regression analyses on several critical factors. Both analyses identified maternal age, BMI, GDM status, and elevated expression levels of hsa_circ_0031560 and hsa_circ_0000793 as independent risk factors for adverse pregnancy outcomes (P<0.05) (Fig. 4A and B). These findings suggest that increased expression levels of hsa_circ_0031560 and hsa_circ_0000793 are significantly associated with a heightened risk of adverse pregnancy outcomes. Furthermore, advanced maternal age, higher BMI, and GDM status contribute to this elevated risk. This underscores the critical role these circRNAs may play in the occurrence of adverse pregnancy outcomes.

Diagnostic potential of hsa_circ_0031560 and hsa_circ_0000793 for GDM

To assess the viability of hsa_circ_0031560 and hsa_circ_0000793 as diagnostic biomarkers for GDM, we evaluated their expression levels in two independent cohorts and examined their diagnostic performance using ROC curves. In the first cohort during the second trimester, both hsa_circ_0031560 and hsa_circ_0000793 showed significant upregulation in the serum of GDM patients (Supplementary Fig. 3A). The AUC values for diagnosing GDM were 0.802 for hsa_circ_0031560 and 0.789 for hsa_circ_0000793, with a combined AUC of 0.809 (Supplementary Fig. 3B and C). In the second cohort of the second trimester, the upregulated hsa_circ_0031560 and hsa_circ_0000793 demonstrated AUC values of 0.850 and 0.788, respectively, for the diagnosis of GDM. When combined, the AUC increased to 0.865 (Supplementary Fig. 3D-F). Additionally, pooled data from both cohorts facilitated further ROC curve analysis across the entire sample set, resulting in consistent findings. The AUC values for diagnosing GDM using hsa_circ_0031560 and hsa_circ_0000793 were 0.829 and 0.789, respectively, with the combined AUC reaching 0.836 (Fig. 5). The combined diagnostic model exhibited a sensitivity of 78.79% and a specificity of 79.21% (Supplementary Table 7). These results suggest that the combined use of hsa_circ_0031560 and hsa_circ_0000793 holds substantial promise as diagnostic biomarkers for GDM, demonstrating strong sensitivity and specificity across independent cohorts.

Development of an early prediction model for GDM (E-GDMM) using circRNAs and validation in independent cohorts

Recognizing the critical importance of early detection and intervention in GDM to reduce adverse maternal and neonatal outcomes [16], we investigated the roles of hsa_circ_0031560 and hsa_circ_0000793 in the early diagnosis of GDM. Our findings revealed that the expression levels of these two circRNAs were significantly elevated in patients who D-GDM compared to those who did not (UD-GDM) (Supplementary Fig. 4A). ROC curve analysis demonstrated that hsa_circ_0031560 and hsa_circ_0000793 could effectively distinguish D-GDM, with AUC values of 0.894 and 0.835 (Supplementary Fig. 4B). To enhance early screening accuracy, we developed an early prediction model (E-GDMM) using logistic regression, with the formula: E-GDMM=–3.663+1.063×Exp (hsa_ circ_0031560)+0.973×Exp (hsa_circ_0000793). This model yielded an AUC of 0.904 and a sensitivity of 85% (Fig. 6A), with an optimal cut-off value of –0.9695 (Fig. 6B). To validate the model’s accuracy, we collected early pregnancy serum samples from two independent centers. The clinical characteristics of these validation cohorts showed significant differences in hypertension, OGTT at 0 and 1 hour, and triglyceride levels (Supplementary Table 8). Consistent with our earlier findings, expression levels of hsa_circ_0031560 and hsa_circ_0000793 were elevated in the D-GDM groups across both validation cohorts (Supplementary Fig. 5A, D, and G). The AUC values for hsa_circ_0031560 were 0.860 and 0.774 in the two first trimester validation cohorts, and 0.832 in the combined first trimester validation cohort, while the AUC values for hsa_circ_0000793 were 0.807 and 0.763 in the two first trimester validation cohorts, and 0.765 in the combined first trimester validation cohort (Supplementary Fig. 5A, D, and G). The prediction model showed AUC values of 0.895 in first trimester validation cohort 1, with a sensitivity of 88.46% and specificity of 72.08%, and 0.798 in first trimester validation cohort 2, with a sensitivity of 81.82% and specificity of 66.67% (Supplementary Fig. 5B, C, E, and F). The AUC for the overall cohort was 0.852, with a sensitivity of 89.13% and specificity of 73.63% (Fig. 6C and D, Supplementary Table 9). The model’s cut-off value of –0.9695 demonstrated robust performance in distinguishing D-GDM patients, suggesting that it could serve as a valuable tool for the early prediction of GDM.

DISCUSSION

GDM is one of the common metabolic disorders during pregnancy, characterized by elevated blood glucose levels that typically resolve after delivery [17]. GDM poses significant impacts on pregnant women and fetuses worldwide [17,18]. Thus, early prediction of GDM is crucial for preventing pregnancy complications and long-term metabolic diseases. Currently, besides the widely accepted OGTT screening test conducted between 24 and 28 weeks of pregnancy, numerous studies have reported various biochemical markers as independent risk factors for GDM [19]. These include glycosylated hemoglobin [20], peripheral blood leukocyte count [21], C-reactive protein [22], interleukin-6 [23], tumor necrosis factor-alpha [24], and adipocytokines [25]. However, current clinical markers are somewhat limited in sensitivity and finding accurate cut-off points. Therefore, there is a need to identify more effective and accurate early biomarkers to address these challenges.

CircRNAs, a newly identified class of non-coding RNAs characterized by their unique circular structure, have increasingly been recognized due to advancements in high-throughput sequencing technologies [26]. Notably, several circRNAs have shown potential as biomarkers for GDM. For example, hsa_circRNA_0039480, which is highly expressed in GDM patients, has been reported to distinguish GDM with an AUC of 0.898 [14]. Similarly, hsa_circRNA_0054633, also upregulated in GDM patients, exhibits a diagnostic AUC of 0.793 for second trimester GDM [13]. Consistent with these previous findings, our study identified key circRNAs, hsa_circ_0031560 and hsa_circ_0000793, in the serum of GDM patients and validated their expression in two independent cohorts. Both circRNAs were upregulated in the serum of GDM patients across different cohorts, effectively distinguishing second trimester GDM. The optimal AUC values for diagnosing mid-pregnancy GDM were 0.850 for hsa_circ_0031560 and 0.789 for hsa_ circ_0000793. Among these, hsa_circ_0031560 consistently demonstrated higher AUC values, sensitivity, and specificity across second trimester cohort 1, second trimester cohort 2, and all second trimester cohort compared to hsa_circ_0000793, suggesting a stronger potential for differentiating GDM. To enhance diagnostic accuracy, we performed a combined ROC analysis of both circRNAs, which yielded a maximum AUC of 0.865 in different cohorts, with improved sensitivity and specificity compared to the individual markers. This combined analysis highlights the potential of these circRNAs as robust biomarkers for GDM diagnosis.

While combined diagnostics can effectively differentiate second trimester GDM, diagnosing GDM at this stage may not be optimal, as earlier prediction could substantially reduce the risk of severe outcomes. Indeed, recent studies have increasingly focused on early GDM prediction. For example, circLRP6 was found to be upregulated in GDM patients, distinguishing potential GDM cases with an AUC of 0.8809 [27]. Another study identified hsa_circ_0003218, which was significantly downregulated in GDM patients, achieving an AUC of 0.743, with a sensitivity of 54.35% and specificity of 87.10% [28]. Additionally, hsa_circRNA_0039480, highly expressed in the plasma exosomes of GDM patients, demonstrated diagnostic value for early pregnancy GDM with an AUC of 0.704. Moreover, combining hsa_circRNA_0039480 and hsa_circRNA_0026497 significantly improved the diagnostic capability for early pregnancy GDM, reaching an AUC of 0.754 [14]. However, previous early diagnostic biomarkers have faced challenges, such as low sensitivity, making it difficult to screen potential GDM populations effectively, and many studies were based on single-center cohorts, limiting representativeness. To address these limitations, we investigated the potential of hsa_circ_0031560 and hsa_circ_0000793 in identifying patients at risk for GDM. Our study demonstrated that both circRNAs were upregulated in the serum of early pregnancy patients in the D-GDM group, serving as independent diagnostic markers across different cohorts. To further enhance screening accuracy, we developed an early prediction model for GDM using logistic regression, which demonstrated high sensitivity and specificity in distinguishing potential GDM patients. This model was validated across multiple cohorts, showing a robust ability to differentiate D-GDM patients. Unlike previous studies, our approach improved diagnostic accuracy through combined analysis and model construction, thereby increasing screening efficiency and addressing the limitations of single-molecule detection. Although specificity slightly decreased with the combined diagnosis and model, the sensitivity was significantly enhanced. Therefore, our model offers a more effective tool for identifying patients at high risk for GDM, facilitating timely intervention and improving outcomes.

This study found that the expression of hsa_circ_0031560 and hsa_circ_0000793 was significantly upregulated during pregnancy, particularly in GDM patients, and rapidly declined to non-pregnant levels after delivery. Therefore, this finding suggests that the expression of these two circRNAs may be closely related to the presence of the placenta and plays an important role in the pathogenesis of GDM. Further analysis showed that high expression levels of these two circRNAs, along with maternal age, BMI, and GDM status, were independent risk factors for adverse pregnancy outcomes. This finding suggests that, in addition to traditional risk factors, abnormal circRNA expression may be a key mechanism affecting pregnancy outcomes, thus supporting their potential as biomarkers for GDM. Moreover, our study observed significant differences in various metabolic parameters between the GDM and control groups at different stages of pregnancy, with the most notable differences in glucose metabolism and BMI occurring in late pregnancy. This phenomenon further indicates that GDM begins to affect the maternal metabolic environment early in pregnancy and progressively worsens as the pregnancy advances. Previous studies have also supported the regulatory role of circRNAs in GDM [29,30]. For example, circular RNA produced by mitogen activated protein kinase saucer 4 (circMAP3K4) has been positively correlated with pregnancy weight gain and the OGTT AUC, regulating insulin resistance via the miR-6795-5p/protein tyrosine phosphatase, non-receptor type 1 (PTPN1) axis [31]. Additionally, hsa_circ_0005243 was significantly downregulated in the placenta and plasma of GDM patients, potentially contributing to the pathogenesis of GDM by regulating the β-catenin and nuclear factor-κB signaling pathways [32]. These studies suggest that circRNAs play multiple regulatory roles in the pathological processes of GDM. Our research further revealed the specific expression patterns and potential mechanisms of hsa_circ_0031560 and hsa_circ_0000793 in GDM. In line with previous research on the critical regulatory role of circRNAs in GDM, our study further confirmed the specific expression patterns and potential mechanisms of hsa_circ_0031560 and hsa_circ_0000793 in GDM. Notably, the rapid decline in the expression of these two circRNAs postpartum strongly suggests that their high expression may primarily originate from placental tissue. Therefore, future research should focus on further validating their specific functions and regulatory mechanisms in the placenta, with the goal of providing new insights and potential targets for the diagnosis and treatment of GDM.

In summary, hsa_circ_0031560 and hsa_circ_0000793 are significantly upregulated during the different stages of GDM, demonstrating potential as independent biomarkers for early prediction and mid-term diagnosis of the condition. Notably, we have developed a robust early prediction model for GDM, which may serve as a valuable clinical tool to assist healthcare professionals in screening and identifying GDM at an earlier stage.

SUPPLEMENTARY MATERIALS

Supplementary materials related to this article can be found online at https://doi.org/10.4093/dmj.2024.0205.

Supplementary Methods.

[See PDF]

Supplementary Table 1.

Differentially expressed circRNAs identified with |log₂FC|>1 and P<0.05

[See PDF]

Supplementary Table 2.

Screen the specific information of 43 circRNAs that meet the criteria of log₂FC >2 and P<0.05

[See PDF]

Supplementary Table 3.

Detailed information on the five selected circRNAs

[See PDF]

Supplementary Table 4.

Primer sequences used in this study

[See PDF]

Supplementary Table 5.

Clinical characteristics of independent cohorts in the first, second, and third trimesters

[See PDF]

Supplementary Table 6.

Specific details of adverse pregnancy outcomes across different groups in cohorts at various stages of pregnancy

[See PDF]

Supplementary Table 7.

The diagnostic value for gestational diabetes mellitus in different cohorts

[See PDF]

Supplementary Table 8.

Clinical characteristics of an independent cohort in early pregnancy

[See PDF]

Supplementary Table 9.

The diagnostic value of hsa_circ_0031560 and hsa_circ_0000793 for D-GDM in different cohorts

[See PDF]

Supplementary Fig. 1.

Sequencing information of circular RNAs (circRNAs). (A) Types, (B) chromosomal locations, (C) lengths, (D) exon compositions, and (E) number of circRNAs generated from individual genes after sequencing.

[See PDF]

Supplementary Fig. 2.

Basic information of hsa_circ_0031560 and hsa_circ_0000793. (A) Basic information of hsa_circ_0031560 and hsa_circ_0000793. (B) Sanger sequencing results of hsa_circ_0031560 and hsa_circ_0000793. HEATR5A, HEAT repeat containing 5A; USP32, ubiquitin specific peptidase 32.

[See PDF]

Supplementary Fig. 3.

Expression and receiver operating characteristic (ROC) curve analysis of hsa_circ_0031560 and hsa_ circ_0000793 in different mid-pregnancy cohorts. (A) Expression levels of hsa_circ_0031560 and hsa_circ_0000793 in the serum of gestational diabetes mellitus (GDM) and normal control (NC) groups of pregnant women in second trimesters cohort 1. (B) ROC curve analysis using hsa_circ_0031560 and hsa_circ_0000793 for GDM patients in second trimesters cohort 1. (C) Combined ROC curve analysis of hsa_circ_0031560 and hsa_circ_0000793 for GDM patients in second trimesters cohort 1. (D) Expression levels of hsa_circ_0031560 and hsa_circ_0000793 in the serum of GDM and NC groups of pregnant women in second trimesters cohort 2. (E) ROC curve analysis using hsa_circ_0031560 and hsa_circ_0000793 for GDM patients in second trimesters cohort 2. (F) Combined ROC curve analysis of hsa_circ_0031560 and hsa_circ_0000793 for GDM patients in second trimesters cohort 2. AUC, area under the curve; CI, confidence interval. ^aP<0.0001.

[See PDF]

Supplementary Fig. 4.

The expression levels of hsa_circ_0031560 and hsa_circ_0000793 in the serum of pregnant women with gestational diabetes mellitus (GDM) and normal control (NC) groups in the modeling cohort (A) and receiver operating characteristic (ROC) curve analysis (B). UD-GDM, did not develop GDM; D-GDM, develop GDM; AUC, area under the curve; CI, confidence interval. ^aP<0.01, ^bP<0.0001.

[See PDF]

Supplementary Fig. 5.

Expression and diagnosis of hsa_circ_0031560, hsa_circ_0000793, and early gestational diabetes mellitus (GDM) prediction model (E-GDMM) in the independent validation cohort. (A) Expression levels and receiver operating characteristic (ROC) curve analysis of hsa_circ_0031560 and hsa_circ_0000793 in the serum of pregnant women from the GDM and normal control (NC) groups in the first trimester validation cohort 1. (B, C) ROC curve analysis and cut-off values of E-GDMM in GDM patients from the first trimester validation cohort 1. (D) Expression levels and ROC curve analysis of hsa_circ_0031560 and hsa_circ_0000793 in the serum of pregnant women from the GDM and NC groups in the first trimester validation cohort 2. (E, F) ROC curve analysis and cut-off values of E-GDMM in GDM patients from the first trimester validation cohort 2. (G) Expression levels and ROC curve analysis of hsa_circ_0031560 and hsa_circ_0000793 in the serum of pregnant women from the GDM and NC groups in the all first trimester validation cohorts. UD-GDM, did not develop GDM; D-GDM, develop GDM; AUC, area under the curve; CI, confidence interval. ^aP<0.001, ^bP<0.0001.

[See PDF]

NOTES

CONFLICTS OF INTEREST

No potential conflict of interest relevant to this article was reported.

AUTHOR CONTRIBUTIONS

Conception or design: S.M., Y.C., G.W.

Acquisition, analysis, or interpretation of data: all authors.

Drafting the work or revising: S.M., Y.C., Y.Y., M.M.

Final approval of the manuscript: all authors.

FUNDING

This work was Supported by the National Natural Science Foundation of China (82373781 and 82302609), Natural Science Foundation of Jiangsu Province (BK20230840), Jiangsu Provincial Medical Key Discipline (Laboratory) Cultivation Unit (JSDW202240), the Jiangsu Provincial Key Laboratory of Critical Care Medicine (JSKLCCM202202015), Southeast University Doctoral Students Innovation Ability Enhancement Program (CXJH_SEU_24219), Zhongda Hospital Affiliated to Southeast University, Jiangsu Province High-Level Hospital Pairing Assistance Construction Funds (zdyyxy10) and Zhongda Hospital Affiliated to Southeast University, Jiangsu Province High-Level Hospital Pairing Assistance Construction Funds (zdlyg09).

ACKNOWLEDGMENTS

We appreciate all patients implicated in this study and thank all individuals contributed to this study.

Fig. 1. Expression profile of circular RNAs (circRNAs) in gestational diabetes mellitus (GDM) and screening of key circRNAs. (A) Heatmap and (B) volcano plot representing differential circRNAs post-sequencing. (C) Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis of differential circRNAs. (D) Gene Ontology (GO) functional enrichment analysis of differential circRNAs. BP, biological process; CC, cellular component; MF, molecular function; GTPase, guanosine triphosphatase; ncRNA, non-coding RNA; RING, really interesting new gene.

Fig. 2. Screening and characterization of hsa_circ_0031560 and hsa_circ_0000793. (A) The expression levels of the five selected circular RNAs (circRNAs) in the serum of gestational diabetes mellitus (GDM) patients (n=20) compared to the matched normal control (NC) group (n=20). (B) Agarose gel electrophoresis results of hsa_circ_0031560 and hsa_circ_0000793. Expression of circRNAs and their host genes after treatment with (C) RNase R enzyme or (D) actinomycin D. (E) Expression of circRNAs in cell culture supernatant after continuous cultivation for 48 hours. NS, not significant; HEATR5A, HEAT repeat containing 5A; USP32, ubiquitin specific peptidase 32. aP[ less than ]0.01, bP[ less than ]0.001, cP[ less than ]0.0001.

Fig. 3. Expression characteristics of hsa_circ_0031560 and hsa_ circ_0000793. (A, B) The expression of hsa_circ_0031560 and hsa_circ_0000793 in the serum of non-pregnant women (control, n=40), pregnant women without gestational diabetes mellitus (GDM) (normal control [NC], n=140), pregnant women with GDM (GDM, n=132), and postpartum women (PW, n=40). (C, D) The expression levels of hsa_circ_0031560 and hsa_ circ_0000793 in the serum of pregnant women in early, mid, and late pregnancy in the NC and GDM groups. (E) Correlation analysis of hsa_circ_0031560 and hsa_circ_0000793 with oral glucose tolerance test (OGTT). (F) Correlation analysis of hsa_circ_0031560 and hsa_circ_0000793. NS, not significant. aP[ less than ]0.05, bP[ less than ]0.01, cP[ less than ]0.0001.

Fig. 4. (A) Univariate and (B) multivariate logistic regression analysis of hsa_circ_0031560 and hsa_circ_0000793 with key clinical characteristics and adverse pregnancy outcomes. OR, odds ratio; CI, confidence interval; BMI, body mass index.

Fig. 5. hsa_circ_0031560 and hsa_circ_0000793 can distinguish gestational diabetes mellitus (GDM) patients in the mid-trimester of pregnancy. (A) Expression levels of hsa_circ_0031560 and hsa_circ_0000793 in the GDM and normal control (NC) groups in all second trimester cohort. (B) Receiver operating characteristic (ROC) curve analysis of hsa_circ_0031560 and hsa_ circ_0000793 for GDM patients in all second trimester cohort. (C) Combined ROC curve analysis of hsa_circ_0031560 and hsa_ circ_0000793 for GDM patients in all second trimester cohort. AUC, area under the curve; CI, confidence interval. aP[ less than ]0.0001.

Fig. 6. Construction and validation of the early warning model for gestational diabetes mellitus (GDM). (A) The receiver operating characteristic (ROC) curve for distinguishing GDM patients in the modeling cohort using the model. (B) The optimal cut-off value set by the model. (C, D) The ROC curve and cut-off value for distinguishing GDM patients in all first trimester validation cohort using the model. AUC, area under the curve; CI, confidence interval; UD-GDM, did not develop GDM; D-GDM, develop GDM.

References

1. Metzger BE, Coustan DR. Summary and recommendations of the Fourth International Workshop-Conference on Gestational Diabetes Mellitus: the organizing committee. Diabetes Care 1998;21 Suppl 2:B161-7.

2. Chiefari E, Arcidiacono B, Foti D, Brunetti A. Gestational diabetes mellitus: an updated overview. J Endocrinol Invest 2017;40:899-909.

3. Zhu Y, Zhang C. Prevalence of gestational diabetes and risk of progression to type 2 diabetes: a global perspective. Curr Diab Rep 2016;16:7.

4. ACOG practice bulletin no. 190 summary: gestational diabetes mellitus. Obstet Gynecol 2018;131:406-8.

5. Agarwal MM. Gestational diabetes mellitus: an update on the current international diagnostic criteria. World J Diabetes 2015;6:782-91.

6. Xu T, Wang M, Jiang L, Ma L, Wan L, Chen Q, et al. CircRNAs in anticancer drug resistance: recent advances and future potential. Mol Cancer 2020;19:127.

7. Ma S, Kong S, Wang F, Ju S. CircRNAs: biogenesis, functions, and role in drug-resistant tumours. Mol Cancer 2020;19:119.

8. Lei M, Zheng G, Ning Q, Zheng J, Dong D. Translation and functional roles of circular RNAs in human cancer. Mol Cancer 2020;19:30.

9. Ren GL, Zhu J, Li J, Meng XM. Noncoding RNAs in acute kidney injury. J Cell Physiol 2019;234:2266-76.

10. Chen LL. The expanding regulatory mechanisms and cellular functions of circular RNAs. Nat Rev Mol Cell Biol 2020;21:475-90.

11. Ma S, Zhou M, Xu Y, Gu X, Zou M, Abudushalamu G, et al. Clinical application and detection techniques of liquid biopsy in gastric cancer. Mol Cancer 2023;22:7.

12. Yang H, Ye W, Chen R, Zeng F, Long Y, Zhang X, et al. Circulating expression of Hsa_circRNA_102893 contributes to early gestational diabetes mellitus detection. Sci Rep 2020;10:19046.

13. Wu H, Wu S, Zhu Y, Ye M, Shen J, Liu Y, et al. Hsa_circRNA_0054633 is highly expressed in gestational diabetes mellitus and closely related to glycosylation index. Clin Epigenetics 2019;11:22.

14. Jiang B, Zhang J, Sun X, Yang C, Cheng G, Xu M, et al. Circulating exosomal hsa_circRNA_0039480 is highly expressed in gestational diabetes mellitus and may be served as a biomarker for early diagnosis of GDM. J Transl Med 2022;20:5.

15. Yan L, Feng J, Cheng F, Cui X, Gao L, Chen Y, et al. Circular RNA expression profiles in placental villi from women with gestational diabetes mellitus. Biochem Biophys Res Commun 2018;498:743-50.

16. Wang Y, Sun P, Zhao Z, Yan Y, Yue W, Yang K, et al. Identify gestational diabetes mellitus by deep learning model from cell-free DNA at the early gestation stage. Brief Bioinform 2023;25:bbad492.

17. Farrar D, Simmonds M, Bryant M, Sheldon TA, Tuffnell D, Golder S, et al. Hyperglycaemia and risk of adverse perinatal outcomes: systematic review and meta-analysis. BMJ 2016;354:i4694.

18. Lapolla A, Metzger BE. The post-HAPO situation with gestational diabetes: the bright and dark sides. Acta Diabetol 2018;55:885-92.

19. Su Y, Wang HK, Gan XP, Chen L, Cao YN, Cheng DC, et al. Alterations of gut microbiota in gestational diabetes patients during the second trimester of pregnancy in the Shanghai Han population. J Transl Med 2021;19:366.

20. Bender W, Mccarthy C, Elovitz M, Parry S, Durnwald C. Universal HbA1c screening and gestational diabetes: a comparison with clinical risk factors. J Matern Fetal Neonatal Med 2022;35:6430-6.

21. Wolf M, Sauk J, Shah A, Vossen Smirnakis K, Jimenez-Kimble R, Ecker JL, et al. Inflammation and glucose intolerance: a prospective study of gestational diabetes mellitus. Diabetes Care 2004;27:21-7.

22. Wolf M, Sandler L, Hsu K, Vossen-Smirnakis K, Ecker JL, Thadhani R. First-trimester C-reactive protein and subsequent gestational diabetes. Diabetes Care 2003;26:819-24.

23. Hassiakos D, Eleftheriades M, Papastefanou I, Lambrinoudaki I, Kappou D, Lavranos D, et al. Increased maternal serum interleukin-6 concentrations at 11 to 14 weeks of gestation in low risk pregnancies complicated with gestational diabetes mellitus: development of a prediction model. Horm Metab Res 2016;48:35-41.

24. Guillemette L, Lacroix M, Battista MC, Doyon M, Moreau J, Menard J, et al. TNFα dynamics during the oral glucose tolerance test vary according to the level of insulin resistance in pregnant women. J Clin Endocrinol Metab 2014;99:1862-9.

25. Lain KY, Daftary AR, Ness RB, Roberts JM. First trimester adipocytokine concentrations and risk of developing gestational diabetes later in pregnancy. Clin Endocrinol (Oxf) 2008;69:407-11.

26. Holdt LM, Kohlmaier A, Teupser D. Circular RNAs as therapeutic agents and targets. Front Physiol 2018;9:1262.

27. Hao H, Liu Q, Chen J, Zhou H. Overexpression of circRNAs LRP6 in gestational diabetes mellitus predicts foetal malformation and intrauterine death. Endocrinol Diabetes Nutr (Engl Ed) 2023;70:124-9.

28. Zou J, Liu Y, Shen J, Xue A, Yan L, Zhang Y. The role of 25(OH) D3 and circRNAs in early diagnosis of gestational diabetes mellitus. J Clin Lab Anal 2023;37:e24826.

29. Yu T, Wang Y, Fan Y, Fang N, Wang T, Xu T, et al. CircRNAs in cancer metabolism: a review. J Hematol Oncol 2019;12:90.

30. Bao Y, Zhang J, Liu Y, Wu L, Yang J. Identification of human placenta-derived circular RNAs and autophagy related circRNA-miRNA-mRNA regulatory network in gestational diabetes mellitus. Front Genet 2022;13:1050906.

31. Du R, Wu N, Bai Y, Tang L, Li L. circMAP3K4 regulates insulin resistance in trophoblast cells during gestational diabetes mellitus by modulating the miR-6795-5p/PTPN1 axis. J Transl Med 2022;20:180.

32. Wang H, Zhou W, She G, Yu B, Sun L. Downregulation of hsa_ circ_0005243 induces trophoblast cell dysfunction and inflammation via the β-catenin and NF-κB pathways. Reprod Biol Endocrinol 2020;18:51.