Multi-modal data integration reveals functionally credible predictive biomarkers in ovarian cancer

Area of Science:

• Genomics
• Cancer Biology
• Bioinformatics

Background:

• Precision oncology seeks to personalize cancer treatment based on molecular alterations.
• Current aberration-guided therapies show limited success in clinical trials.
• Distinguishing functional driver aberrations from non-functional alterations is crucial.

Purpose of the Study:

• To develop and validate an integrated whole-genome and transcriptome workflow.
• To systematically identify functionally credible, predictive driver aberrations.
• To improve the reliability of biomarker detection in precision oncology.

Main Methods:

• Applied an integrated omics workflow to 335 ovarian high-grade serous carcinoma (HGSC) patients.
• Utilized whole-genome and transcriptome sequencing with advanced bioinformatics pipelines (Anduril 2, HMW toolkit, nf-core/rnafusion).
• Tested aberration-specific drug sensitivity in patient-derived organoids.

Main Results:

• Identified clinically relevant ESCAT Tier II-III alterations in over 40% of patients.
• Found that 58% of nominally pathogenic variants were false positives.
• NF1 deficiency was the most recurrent actionable event, showing sensitivity to KRAS- and MEK-inhibition.
• Integrated DNA-RNA analysis effectively distinguished true aberrations from false positives.

Conclusions:

• The study provides a reliable strategy for biomarker detection in precision oncology.
• Findings inform biomarker-guided clinical trial design.
• Revealed unexploited therapeutic vulnerabilities in HGSC.