Abstract
Spontaneous preterm birth (sPTB) is the leading cause of infant morbidity and mortality worldwide. Deficiency of effective predict methods is an urgent problem that needs to be solved. To evaluate the value of differential indicators, a prospective nested case-control study was carried out. Among an overall cohort of 1,050 pregnancies, 20 sPTB pregnancies and 20 full-term pregnancies were enrolled. Participants were followed up until labor. The psychological profile was evaluated utilizing the Zung Self-Rating Depression Scale (ZSDS) at 11–14 weeks. Stress-related biomarker cortisol and metabolites were detected by Electrochemiluminescence Immunoassay (ECLIA) and Gas Chromatography-Mass Spectrometry (GC-MS) in serum samples. The expression level of cortisol and the ZSDS score were significantly higher in the sPTB group. Additionally, 29 metabolites were differentially expressed. A prediction model established with five variables (ZSDS, cortisol, Eicosane, methyltetradecanoate, and stearic acid) yielded an AUC of 89.5% with 100% sensitivity. This study establishes a prediction model that may predict sPTB more accurately in the second trimester.
Introduction
Spontaneous preterm birth (sPTB), defined as delivery before 37 weeks of gestation, remains a critical challenge in global maternal and neonatal health. It is the primary driver of infant morbidity and mortality. Despite its prevalence, current clinical tools—often relying on single indicators like cervical length or specific protein markers—frequently lack the sensitivity and specificity required for early and accurate identification of at-risk pregnancies.
This study explores the "biopsychosocial" nature of pregnancy, investigating whether the physiological manifestations of psychological stress—specifically hormonal changes and metabolic shifts—can serve as a robust "early warning system" during the second trimester.
Methods
The researchers conducted a prospective nested case-control study at The First Affiliated Hospital of Chongqing Medical University. From an initial cohort of 1,050 pregnant women, 20 cases of sPTB and 20 full-term controls were selected for detailed analysis.
Data collection involved three primary streams:
- Psychological Profiling: The Zung Self-Rating Depression Scale (ZSDS) was administered during the first visit (11–14 weeks).
- Hormonal Analysis: Serum cortisol levels were measured across all three trimesters using ECLIA.
- Metabolomics: GC-MS was employed to profile 119 compounds in maternal serum, identified using the MCF and NIST spectral libraries.
Feature selection was performed using LASSO (Least Absolute Shrinkage and Selection Operator) logistic regression to identify the most predictive variables from the high-dimensional metabolomic data. The final model's performance was validated using Support Vector Machines (SVM).
Results
The study identified a clear link between psychological stress and biological markers. Women in the sPTB group exhibited significantly higher ZSDS scores and elevated serum cortisol levels compared to the control group.
Key findings include:
- Temporal Significance: While cortisol levels naturally rise throughout pregnancy, the disparity between sPTB and control groups reached peak statistical significance during the second trimester ($P = 0.005$).
- Metabolic Profiling: 29 metabolites were differentially expressed. Notably, long-chain saturated fatty acids like stearic acid were downregulated in sPTB cases, while certain alkanes (e.g., Eicosane) and amino acid derivatives were upregulated.
- The Five-Variable Model: The LASSO regression narrowed the optimal predictors to: ZSDS score, cortisol level, Eicosane, methyltetradecanoate, and stearic acid.
The resulting model demonstrated exceptional diagnostic performance, achieving an Area Under the Curve (AUC) of 89.5% (95% CI: 0.8006–0.9889). Most notably, the model achieved 100.0% sensitivity and 78.9% specificity. Validation via SVM with a Sigmoid kernel further confirmed the model's robustness with an accuracy of 97.4%.
Discussion and Conclusion
This study demonstrates that combining psychological assessments with metabolic profiling in the second trimester can accurately flag high-risk pregnancies well before the onset of labor. The high sensitivity of the five-variable model suggests it could be a powerful tool for early clinical intervention.
However, the study is limited by its small sample size (n=40), necessitating further validation in larger, more diverse cohorts. Future research should also explore less invasive sampling methods (e.g., saliva or urine) and investigate the underlying biological mechanisms of the identified metabolites. In conclusion, the integration of mental health screening and metabolic biomarkers represents a promising frontier for the early prediction and management of spontaneous preterm birth.
Summary updated