Predicting gestational diabetes mellitus risk at 11–13 weeks’ gestation: the role of extrachromosomal circular DNA

Jinfeng Wang, Pengyu Huang, Fei Hou, Dongdong Hao, Wushan Li, Hua JinCardiovascular 2024 · DOI

This study, published in Cardiovascular Diabetology, explores a novel frontier in prenatal diagnostics: using extrachromosomal circular DNA (eccDNA) to predict gestational diabetes mellitus (GDM) months before traditional screening takes place.

TL;DR

Researchers identified a specific circular DNA fragment, $PRDM16^{circle}$, in maternal plasma that can predict the risk of gestational diabetes as early as 11–13 weeks of gestation. With an impressive AUC of 0.941, this biomarker significantly outperforms traditional clinical indicators like HbA1c and BMI, offering a potential path toward much earlier intervention for at-risk pregnancies.

What Problem Does This Paper Address?

Gestational diabetes mellitus (GDM) affects up to 31% of pregnancies globally, leading to complications like pre-eclampsia for mothers and neonatal hypoglycemia or overgrowth for infants. Currently, the "gold standard" for diagnosis is the Oral Glucose Tolerance Test (OGTT), typically administered between 24 and 28 weeks of gestation.

The problem is twofold: first, the OGTT is invasive and often unpleasant for patients; second, by the time a diagnosis is made in the late second trimester, maternal and fetal health may already be compromised. There is a critical need for stable, early-pregnancy biomarkers that can identify high-risk women when lifestyle interventions or treatments are most effective.

Methods

The researchers employed a multi-stage workflow to move from broad discovery to specific validation:

  1. Discovery (Circle-seq): They performed high-throughput eccDNA sequencing on plasma samples from 10 women (5 who later developed GDM and 5 healthy controls) at 11–13 weeks. This involved digesting linear DNA to leave only circular forms, followed by rolling circle amplification (RCA).
  2. Bioinformatics: They mapped these fragments to the human genome to identify "differentially abundant" eccDNAs and used KEGG/GO analysis to see which biological pathways were involved.
  3. Validation (Outward PCR & Sanger Sequencing): To ensure the identified fragments were truly circular, they used "outward-facing" primers that only yield a product if the DNA is a closed loop.
  4. Large-Scale Verification (qPCR): They tested their top candidate, $PRDM16^{circle}$, in a larger cohort of 140 women (70 GDM, 70 controls) to confirm its predictive power.
  5. Modeling: They used LASSO regression and ROC curve analysis to compare the eccDNA biomarker against standard clinical variables.

Datasets

  • Discovery Cohort: 10 pregnant women (5 GDM, 5 NGT) from Jinan, China.
  • Validation Cohort: 140 pregnant women (70 GDM, 70 NGT) from the same region.
  • Reference Genome: Human reference genome (HG19).
  • Public Data: The Circle-seq data is deposited in the Genome Sequence Archive (GSA) under accession number HRA007356.

Key Insights and Ideas

  • The Stability of Circles: Unlike linear cell-free DNA (cfDNA), which degrades quickly, eccDNA’s closed-loop structure makes it highly resistant to exonucleases. This makes it an exceptionally stable candidate for "liquid biopsies."
  • Non-Random Distribution: eccDNAs are not just random genomic "trash." The study found they correlate strongly with coding gene density and often originate from regulatory regions like CpG islands and 5′UTRs, suggesting they play a role in gene regulation.
  • The PRDM16 Connection: The most promising marker, $PRDM16^{circle}$, is derived from a gene known to regulate glucose homeostasis and brown fat thermogenesis. Interestingly, levels of this circular DNA decreased in women who later developed GDM, potentially reflecting early metabolic dysregulation.
  • Pathway Enrichment: Differentially expressed eccDNAs were linked to GDM-relevant pathways, including Insulin Resistance, MAPK, and PI3K-Akt signaling.

Key Results

  • Differential Abundance: The study identified 2,217 eccDNAs that differed between future GDM patients and healthy controls (1,289 increased, 928 decreased).
  • Superior Prediction: $PRDM16^{circle}$ achieved an Area Under the Curve (AUC) of 0.941. For context, traditional markers performed much worse: HbA1c (0.714), Pre-pregnancy BMI (0.700), and Maternal Age (0.673).
  • Early Detection: The biomarker was effective at 11–13 weeks, providing a roughly 12-week head start over the standard OGTT.
  • Model Comparison: Adding clinical variables (age, BMI, etc.) to the $PRDM16^{circle}$ data in a nomogram did not significantly improve the AUC (0.935), suggesting the eccDNA fragment itself carries the bulk of the predictive information.

Limitations and Future Work

  • Sample Size: While the validation cohort (n=140) is respectable, the initial discovery was based on only 10 patients. Larger, multi-center studies are needed to ensure these results generalize across different ethnicities and populations.
  • Mechanistic Gap: The study shows a correlation between decreased $PRDM16^{circle}$ and GDM, but it doesn't explain the why. It remains unclear if the decrease in circular DNA is a cause or a consequence of early metabolic shifts.
  • Technical Complexity: Circle-seq and qPCR for circular junctions are currently more complex and expensive than standard blood tests, which may limit immediate clinical adoption.
  • Future Directions: The authors suggest investigating $PRDM16$ transcript levels in placental tissue to see if the plasma eccDNA levels reflect local changes in the womb.

Summary updated

Abstract

ChiCTR2300075971, http://www.chictr.org.cn . Registered 20 September 2023.
BibTeX
@inproceedings{wang2024predicting,
  author = {Jinfeng Wang and Pengyu Huang and Fei Hou and Dongdong Hao and Wushan Li and Hua Jin},
  title = {{Predicting gestational diabetes mellitus risk at 11–13 weeks’ gestation: the role of extrachromosomal circular DNA}},
  booktitle = {Cardiovascular},
  year = {2024},
  doi = {10.1186/s12933-024-02381-1},
}
Figure