Radiation Response Prediction Model Based on Integrated Clinical and Genomic Data Analysis.

Area of Science:

• Oncology
• Genomics
• Radiotherapy

Background:

• Predicting patient response to radiotherapy is crucial for optimizing cancer treatment.
• Genomic profiling offers a potential avenue for enhancing treatment prediction accuracy.

Purpose of the Study:

• To evaluate the utility of targeted gene-sequencing for predicting radiation therapy response.
• To develop a predictive model for radiotherapy response integrating genomic and clinical data.

Main Methods:

• Analysis of 88 patients undergoing radiotherapy with tumor sequencing.
• Classification into durable local control (DLC) and non-durable local control (NDLC) groups.
• Development of a Bayesian Network model using genomic factors and clinical information.

Main Results:

• NOTCH2 and BCL mutations were linked to non-durable control, while CHEK2, MSH2, and NOTCH1 mutations were associated with durable control.
• Altered DNA repair pathways correlated with improved local failure-free survival (HR 0.40, P=0.014).
• A Bayesian network model achieved an AUC of 0.83 for predicting 6-month local control.

Conclusions:

• Durable radiation response is associated with DNA repair pathway alterations and smoking somatic signatures.
• The Bayesian network model shows promise for personalized radiotherapy.
• Further validation in prospective cohorts is recommended for individualizing treatment.