During the 2026 American Society of Clinical Oncology (ASCO) Annual Meeting, Triangle CERSI Co-PI Susan Halabi, PhD, sat down for an interview with Andrea Miyahira of the Prostate Cancer Foundation. Their conversation focused on a new model for estimating radiographic progression-free survival in patients with metastatic castration-resistant prostate cancer.
The model combines established clinical measures with 74 genomic features detected through circulating tumor DNA, or ctDNA, a minimally invasive biomarker obtained from a small plasma sample. Researchers developed the model using data from 1,311 patients enrolled in the Alliance A031201 phase 3 clinical trial and a machine-learning method called Random Survival Forest.
Adding genomic information improved the model’s accuracy and produced a 30% net improvement in correctly classifying patients by risk. Median radiographic progression-free survival was 40 months for the low-risk group, 25 months for the intermediate-risk group, and 13 months for the high-risk group.
The model could eventually help researchers better balance risk groups in clinical trials and improve study design. The team has developed a tool for entering patient data and is now seeking clinical trial and real-world datasets for external validation before making it broadly available to the scientific community.
Watch the interview.
Valentín López JC, Yu C, Knutson TP, et al. Evolution of AR and Non-AR Alterations in ctDNA of Patients with Metastatic Prostate Cancer Treated in the Phase III Alliance A031201 Trial. Clin Cancer Res. 2026;32(12):2413-2427. doi:10.1158/1078-0432.CCR-25-2779 [https://pubmed.ncbi.nlm.nih.gov/41587107/]
