Gene-Mutation-Based Algorithm for Prediction of Treatment Response in Colorectal Cancer Patients

Area of Science:

• Oncology
• Genomics
• Biomarker Discovery

Background:

• Metastatic colorectal cancer (mCRC) has high mortality with limited predictive biomarkers for treatment response.
• Current predictive factors for mCRC treatment response lack precision.

Purpose of the Study:

• To develop and validate a novel gene-mutation-based algorithm for predicting treatment response in mCRC.
• To enhance precision in identifying patients likely to respond to therapies.

Main Methods:

• Random forest machine learning (ML) was used to identify candidate algorithms.
• The algorithm was trained on the MSK Cohort (n=471) and validated on the TCGA Cohort (n=221).
• Performance was assessed using logistic regression, progression-free survival (PFS), and Cox proportional hazard analyses.

Main Results:

• A 7-Gene Algorithm based on KRAS-associated gene mutations was identified.
• The algorithm achieved high accuracy (AUC 0.97 in training, 0.98 in validation) in predicting responders vs. non-responders.
• It demonstrated significant predictive power for PFS in mCRC patients (HR=16.9, p<0.001).

Conclusions:

• The novel 7-Gene Algorithm shows strong potential as a predictive biomarker for mCRC treatment response.
• This algorithm can aid in developing personalized therapeutic strategies for mCRC patients.
• Further development could lead to improved clinical decision-making in mCRC management.