Machine learning combining multi-omics data and network algorithms identifies adrenocortical carcinoma prognostic...

Area of Science:

• Endocrinology
• Oncology
• Bioinformatics
• Systems Biology

Background:

• Adrenocortical Carcinoma (ACC) is a rare endocrine cancer with incomplete pathogenesis knowledge and limited treatment options.
• Accurate diagnosis and prognostication are challenging, necessitating the identification of novel molecular drivers and biomarkers.
• Effective biomarkers are crucial for timely diagnosis and patient stratification for optimal treatment.

Purpose of the Study:

• To demonstrate the utility of machine learning and multi-omics data integration for identifying prognostic biomarkers in ACC.
• To discover new molecular drivers and biomarkers for improved ACC diagnosis and patient stratification.
• To develop a robust prognostic signature for clinical application in ACC management.

Main Methods:

• Utilized gene expression and DNA methylation datasets from TCGA Adrenocortical Carcinoma cohort.
• Employed the DIABLO method for multi-omics data integration to identify a discriminating signature.
• Applied Clarivate CBDD systems biology tools for identifying regulators of the signature.
• Validated the prognostic value using random forest classification and Kaplan-Meier analysis on independent datasets.

Main Results:

• Generated a multi-omics signature comprising genes, microRNAs, and methylation sites.
• The signature demonstrated high power in classifying ACC patients by stage (I-II vs. III-IV).
• Identified 8 genes and 4 microRNAs significantly associated with Overall Survival (OS) in ACC patients.
• Developed a 9-feature prognostic signature capable of predicting high-risk ACC patients.

Conclusions:

• Machine learning and integrative multi-omics analysis, coupled with systems biology tools, successfully identified ACC biomarkers with high prognostic value.
• Multi-omics data represent a valuable resource for discovering drivers and prognostic biomarkers in rare cancers.
• The identified biomarkers hold potential for clinical application in improving ACC diagnosis and patient management.