Abstract 575: Prediction of genomic instability in triple-negative breast cancer patients utilizing a CTC phenotype classifier

Abstract

Abstract Background: Genomic scars and HRD gene mutations are biomarkers for PARP inhibitor (PARPi) and platinum agent therapy response in breast cancer. Tests for these markers are usually performed by next-generation sequencing (NGS) of tumor tissue or circulating tumor DNA (ctDNA) and have limitations such as sample accessibility/availability and under-sampling due to inter/intratumor heterogeneity. Previously we predicted genomic instability (GI) using CTC phenotypic features without the use of NGS in metastatic castration-resistant prostate cancer (mCRPC) pts with high accuracy (>76%) (ASCO 2016). In addition, patients with phenotypically predicted GI (pGI) CTCs had significantly better PSA responses on abiraterone + veliparib (92%) vs. abiraterone alone (23%) (ESMO 2016), and also responded better to platinum agents vs. taxane agents (ESMO 2017). Here, we sought to develop and analytically validate an algorithm for predicting CTC pGI in TNBC. Methods: Training set: 521 CTCs from 26 mCRPC pts were detected with the Epic Sciences CTC platform and analyzed for 19 phenotypic digital pathology features, including protein expression and cell morphology. The same CTCs were single-cell sequenced for the number of large-scale transitions (LSTs) as an indicator of GI. A linear regression algorithm to predict GI by CTC phenotype was developed, cross validated, and utilized to generate a CTC pGI score. Test set: 114 CTCs from 8 TNBC blood samples, median of 8 CTCs/pt, were sequenced for GI and phenotypically predicted for pGI. Results: Analytical cutoff for GI (8.74) was determined in the mCRPC training set by separation of a binomial distribution low vs. high GI scores. pGI cutoff (8.74) was set the same as GI. The training set performance was: sensitivity = 69%, specificity = 83%, accuracy = 77%. In the TNBC test set, pGI vs. GI demonstrated: sensitivity = 86%, specificity = 92%, accuracy = 88%. Conclusions: Previous studies showed that pGI was an analytically validated biomarker with clinical utility to predict PARPi or platinum therapy response in mCRPC pts. Here we show the same test concept can be applied to TNBC. Further analytical validation in a larger cohort is ongoing. The ability to identify PARPi/platinum sensitivity using an IF CTC staining method for CTC phenotype without the use of NGS will help to stratify patients more rapidly, at reduced cost, and aid in the acceleration of drug development. Citation Format: Adam Jendrisak, Nadia Ebrahim, Angel Rodriguez, Jerry Lee, Ramsay Sutton, Yipeng Wang, Mark Landers, Ryan Dittamore. Prediction of genomic instability in triple-negative breast cancer patients utilizing a CTC phenotype classifier [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 575.