Classifying circulating, mutation bearing tumor cells from breast cancer patients.

Abstract

580 Background: The treatment of advanced breast cancer demands systemic therapies that can address disease heterogeneity and the development of treatment resistance without a “real-time” molecular window into disease biology. New technologies have focused on increased capture and molecular analysis of circulating tumor cells (CTCs) including cells undergoing epithelial mesenchymal transition (EMT). We conducted a pilot experiment to test the efficiency of capture and cytokeratin (CK) detection and the presence of single point variants (SNV) to determine the best utility of scoring alternatives for CTC. Methods: EpCAM expressing CTC were recovered from breast cancer patients using CellSearch (Veridex) and LiquidBiopsy (Cynvenio Biosystems). EpCAM recovery and CK scoring were indexed in spiked samples and in 12 inflammatory breast cancer (IBC) patient samples using antibodies against CKs 7, 8 or CKs 1-8, 10, 13-16, 18, 19. Additionally, LiquidBiopsy template was analyzed using an Ampliseq 1.0 panel on the IonTorrent PGM. SNV present in the CTC but not white blood cell (WBC) negative controls were identified and where possible, compared to tissue biopsy SNV analyzed using Foundation One (Foundation Medicine). Results: CTCs were detected using CellSearch 10/12 (83%) (range 0-2502 CTC/7.5ml) and LiquidBiopsy 12/12 (100%) (range 6-2800 CTC/7.5mL). More CK positive events were scored using CKs 1-8, 10, 13-16, 18, 19 than CKs 7, 8 in patient samples. Upon sequencing, shared germline polymorphisms were observed in CTC and WBC. Conversely, 1 or 2 SNV were detected in the Epcam selected population but not WBC controls from 6/12 patients (frequency 1.1%-2.1% with 520-5160x coverage) with SNV observed in TP53, MPL, PIK3ca, MET and IDH1. All but one of the PIK3ca mutations were absent in evaluable tissue biopsy. Conclusions: CTC recovery and scoring are two separate events. Altered CK detection emphasized the need to tailor CTC classification to specific disease settings. Sequence analysis showed one correlated SNV among 6 evaluable comparisons to tissue reflecting variable analysis as well as the biologic disparity of metastatic disease. This pilot demonstrates the feasibility of using CTC for molecular analysis.