Abstract SY31-03: Predictive biomarkers in circulating tumor cells to direct management metastatic castration-resistant prostate cancer (mCRPC)

Abstract

Abstract Tissue and cfDNA NGS profiling is improving our ability to detect and clinically validate the significance of molecular alterations that drive tumor growth resulting in actionable therapeutic strategies that can improve outcomes for subsets of patients and accelerate the regulatory approval of targeted agents. Confounding our knowledge base are three critical aspects: 1) intra patient and intra-tumor heterogeneity for which the clustering of alterations and signaling pathways cannot be easily disassociated without cellular context, 2) epigenetic and functional cellular biology alterations which perturb genomic pathways presumed to drive tumor growth, and 3) understanding the clinical significance of “variants of uncertain significance” (with respect to changes in protein function) (1) (https://www.ncbi.nlm.nih.gov/pubmed/26582918). Our objective is to improve the outcomes for men with mCRPC starting by identifying unmet needs at decision points in management, credentialing biomarkers to better inform the decision, followed by the analytical and clinical validation of the assay and the biomarker in a sequence of trials in which samples are obtained at the decision point and associated with outcomes, and showing that use of the test result to direct management improves patient outcomes compared to non-use of the test. Our focus is on predictive biomarkers of sensitivity and resistance assessed at the level of the single circulating tumor cell (CTC) through Functional Cell Profiling (FCP) analyzed by 1) protein characterization & localization, 2) cellular and sub-cellular morphology aggregated with machine learned CTC profiles, and 3) low pass whole genome copy number changes. Establishing a phenotype genotype relationship enables a rapid turnaround of test results to inform decision making for an individual patient. Building on the initial results reported by Investigators at JHU (2) showing the association of the presence of the ARV7 (androgen receptor splice variant 7) in CTC using a PCR based assay predicted for resistance to androgen receptor signaling inhibitors (ARSI) but not taxane based chemotherapy, we applied a protein based assay that enabled distinguishing the localization of the splice variant, nuclear vs. cytoplasmic that showed increased specificity, resistance to ARSI, and clinical utility based on a treatment specific interaction that showed in a survival benefit for treatment with a taxane for patients with AR-V7+ CTC (3, 4). The assay, now marketed, has also received a favorable coverage determination from CMS. Later, building on the regulatory precedent for image based phenotyping that enabled rapid cervical cancer screening, we analyzed CTCs based on protein expression and localization and digital pathology features similar to what is applied in facial recognition analyses, to define phenotypically unique cell type (5), studied independently for the relationship to sensitivity to standard of care (SOC) drugs. The aim was to enable rapid test turnaround to enable real time treatment selection decisions. Such profiling has identified unique cell types associated with resistance to ARSi’s and taxanes, and single cell genomic analyses confirmed differences between cell types, providing insight into the underlying mechanisms. More recently we have focused on the development of a phenotypic biomarker of genomic instability, an important driver of tumor progression, that utilizes morphology to identify CTCs with an abnormal number of single cell Large Scale Transitions (LSTs). The resultant biomarker, phenotypic LST (pLST), was then examined in patient CTCs prior to and shortly after treatment with an ARSI or taxane. Those determined to be pLST biomarker positive have worse OS than those who don’t. An important aspect of liquid biopsy analysis is the biologic characterization of the cells that persist, i.e. that are resistant de novo, and which emerging during therapy that have acquired resistance to treatment. In the case of CTCs, we have validated through 5 prospective phase III trials that the elimination of all CTCs on therapy portends a better clinical response than those who have residual CTCs (6). In this context, pLST CTC biomarker positivity ~1 mo after the start of therapy portends a poor outcome whereby fewer than 15% of patients survive beyond 1 yr. Importantly however, is that these pLST+ cells are/were eliminated in patients treated with a PARP inhibitor and had improved PSA response rates7, suggesting that specifically targeting genomic instability is a valid approach where other standard of care drugs fail. Functional Cell Profiling or FCP has also provided an improved understanding of the tumor heterogeneity common in many patients, occurrence of discordance between phenotype and genotype, and the greater biologic learnings of single cell analysis. The framework has allowed us to postulate a novel strategy for analyzing CTCs through FCP for prediction of therapeutic benefit and subclonal disease resistance. Characterizing CTCs in individual patients may provide insight into disease evolution, inform treatment decisions for individual patients and guide new drug development. Longitudinal monitoring of phenotypic CTC subtypes in multiple clinical trials is currently underway.