Identification of biomarkers for endometriosis based on summary-data-based Mendelian randomization and machine learning.
Journal:
Medicine
PMID:
40193647
Abstract
Endometriosis (EM) significantly impacts the quality of life, and its diagnosis currently relies on surgery, which carries risks and may miss early lesions. Noninvasive biomarkers are urgently needed for early diagnosis and personalized treatment. This study utilized the genome-wide association study dataset from FinnGen and performed Multi-marker Analysis of GenoMic Annotation (MAGMA) to identify genes significantly associated with EM. Differentially expressed genes (DEGs) were then analyzed, and an intersection selection was conducted to obtain the MAGMA-related DEGs. Gene Ontology and Kyoto Encyclopedia of Genes and Genomes enrichment analyses were performed to explore the biological functions of these genes. Summary-data-based Mendelian randomization was used to identify potential risk and protective genes. Subsequently, a machine learning model was used to further select key biomarkers. Single-cell RNA sequencing and consensus clustering were applied to analyze the expression of biomarkers and classify the EM samples into subgroups. Immune infiltration analysis was conducted to evaluate the molecular characteristics of these subgroups. MAGMA analysis identified 2832 genes significantly associated with EM, while 3055 DEGs were detected. Intersection analysis resulted in 437 MAGMA-related DEGs. Summary-data-based Mendelian randomization analysis identified 10 candidate genes, and after further selection using a machine learning model, three core biomarkers were validated: adenosine kinase, enoyl-CoA hydratase/3-hydroxyacyl CoA dehydrogenase, and CCR4-NOT transcription complex subunit 7. Single-cell RNA sequencing revealed the expression patterns of these biomarkers. Consensus clustering analysis classified 77 EM samples into two subgroups, with immune infiltration analysis showing significant differences in immune cell composition among the subgroups. This study successfully identified three core biomarkers for EM: adenosine kinase, enoyl-CoA hydratase/3-hydroxyacyl CoA dehydrogenase, and CCR4-NOT transcription complex subunit 7, which exhibit protective roles in EM.