Construction of a pathway-level model for preeclampsia based on gene expression data

Area of Science:

• Biomedical Science
• Genomics
• Computational Biology

Background:

• Preeclampsia (PE) is a complex pregnancy disorder with significant maternal and fetal health risks.
• Delayed diagnosis due to atypical symptoms and lack of sensitive assays hinders timely intervention.
• Current diagnostic methods often lack the precision needed for early and accurate detection.

Purpose of the Study:

• To develop a robust diagnostic model for preeclampsia using pathway-based analysis and machine learning.
• To identify key biological pathways and molecular markers associated with preeclampsia.
• To evaluate the diagnostic performance of the model in independent cohorts using various sample types.

Main Methods:

• Personalized pathway-based analysis integrated with machine learning algorithms.
• Utilized multiple cohorts from the Gene Expression Omnibus (GEO) database for model development and validation.
• Identified hub genes within the core pathways using network analysis.

Main Results:

• A nine-core pathway diagnostic model for preeclampsia achieved high AUROC values (0.959 in development, 0.898 and 0.876 in validation cohorts).
• The model demonstrated good performance with maternal plasma samples (AUROC: 0.815).
• Tyrosine-protein kinase Lck (LCK) was identified as a crucial hub gene, with its protein levels (LCK and pLCK) found to be downregulated in PE placentas.

Conclusions:

• The developed pathway-level model offers a promising novel approach for preeclampsia diagnosis.
• This model can guide the development of molecular diagnostic assays for earlier and more accurate PE detection.
• Further investigation into the identified pathways and the role of LCK may elucidate potential mechanisms underlying preeclampsia.