Identification of Urinary MMP9 as a Potential Non-invasive Biomarker for Endometrial Cancer through Multi-omics Analysis with Machine Learning–assisted Validation

Objectives: Endometrial cancer (EC) is a prevalent gynecological malignancy with rising incidence.Given the limitations of invasive diagnostics, there are urgent needs for non-invasive biomarkers.Urine represents an ideal biofluid due to its ease of collection.This study aimed to identify and validate urinary biomarkers for EC using a multi-omics approach.Methods: An in-silico framework integrating tissue proteomic, tissue transcriptomic, and urine proteomic datasets was applied to screen for candidates upregulated in EC.Differential expression was validated, and clinical significance was analyzed across independent multiple cohorts EC datasets, together with a machine learning-based cross-cohort validation to assess discriminative potential.Results: Integrated analysis revealed enhanced proinflammatory and profibrotic signaling and identified matrix metalloproteinase 9 (MMP9) as a consistently upregulated across tissue and urine.Validation with independent cohorts confirmed elevated MMP9 levels, significantly associated with advanced stage, progesterone resistance, and higher mortality.Machine learning analysis demonstrated that MMP9 discriminated EC from controls.Moreover, high MMP9 expression correlated with poorer overall survival.Conclusion: This comprehensive multi-omics analysis demonstrates that MMP9 is elevated in cancer and detectable in urine in EC.These findings support the potential of MMP9 as a biomarker for disease detection and prognostic assessment, with potential relevance for non-invasive evaluation.