Cell-free DNA fragmentomics as prognostic and treatment resistance biomarkers in metastatic castration-resistant prostate cancer

Project Summary Metastatic castrate resistant prostate cancer (mCRPC) is the lethal form of the disease, and while multiple therapies such as androgen signaling inhibitors (ARSIs) are available, drivers of resistance such as alterations in androgen receptor and lineage plasticity, including shifts to neuroendocrine prostate cancer (NEPC), are common. A major unmet clinical need in mCRPC is longitudinal monitoring to continually assess for molecular transitions that drive treatment resistance. Current clinical monitoring with PSMA imaging or PSA levels can detect progression, but do not provide the underlying molecular mechanism, which is needed to rationally select the next line of therapy. These methods are also less effective as androgen independence and lineage plasticity emerges. Fragmentomics provides a liquid biopsy approach using fragments of circulating tumor DNA (ctDNA) on standard commercial cancer gene panels. Paired with machine learning, it can be used to monitor for the transition between prostate adenocarcinoma and NEPC, complementing standard variant identification. This innovative integrated method provides tremendous performance, logistical, and cost benefits compared to separate assays. This study will investigate ARSI response and resistance in multiple prospective trials using three specific aims. 1) It will determine if baseline ctDNA fragmentomics scores are prognostic biomarkers in mCRPC in comparison to ctDNA fraction and other clinical variables; 2) It will detect the emergence of diverse ARSI resistance mechanisms in longitudinal ctDNA that are independent of androgen receptor or NEPC- associated DNA alterations to assist in treatment decisions; and 3) It will interrogate ctDNA fragmentomics biomarkers in the Alliance A031201 trial to determine their prognostic value and ability to identify emergence of ARSI resistance and biomarkers of dual ARSI response. Utilizing a single unified liquid biopsy assay powered by machine learning provides the maximum data for each of these aims in a cost-effective manner. In addition, a single targeted panel ctDNA sequencing assay allows for maximal use of a plasma sample, as splitting a sample for multiple assays can decrease the sensitivity of each, especially at very low ctDNA quantities. Thus, successful implementation of these aims would revolutionize the ability to monitor and make treatment decisions in mCRPC and could serve as a model for such work for other cancers.