Patient-matched tissue and liquid biopsies identify shared and acquired genomic alterations in breast cancer.

Abstract

1050 Background: Liquid biopsy-based comprehensive genomic profiling (CGP) is a minimally invasive approach to potentially identify truncal driver alterations and mechanisms of acquired resistance. However, there are limited data comparing solid tissue and plasma biopsies from the same patient in breast cancer. Methods: CGP was performed on matched tumor tissue and plasma samples from 444 patients with breast cancer (Foundation Medicine, Inc.) to identify short variant mutations and rearrangements in at least 62 genes. ER/PR/HER2 status was abstracted for a subset of patients (n = 273). Patients were primarily ER+/HER2- (58%) with fewer HER2+ (16%) and TNBC (25%). Results: Samples were temporally heterogeneous with a median time between matched tissue-plasma collection of 276 days (interquartile range 45 - 650). Positive percent agreement (PPA) to tissue biopsy was 81.5% with higher PPA for collection interval ≤1y v > 1y (82.5% v 76.5%). PPA was highest in known truncal driver genes ( AKT1, PTEN, BRCA2, BRCA1, TP53, PIK3CA: 85-89% PPA). PPA of tissue to liquid biopsy was 50.5% with strong time dependence (62.4% ≤1y, 37.9% > 1y), presumably due to intervening therapies. Acquired alterations in liquid biopsy were primarily known resistance mechanisms ( ESR1, NF1, ERBB2, PIK3CA, PTEN, TP53, BRCA1, and BRCA2), were often polyclonal, associated with longer collection intervals, and exhibited higher prevalence in the ER+/HER2- subtype (62% ER+/HER2-, 33% HER2+, 42% TNBC). Acquired alterations significantly enriched in ER+ cases included ESR1, NF1, ERBB2 and TP53. For PIK3CA, a PPA of 84.6% was observed with solid as baseline and 79.1% with liquid as baseline; prevalence was similar (36.5% solid, 35.6% liquid). Alpelisib companion diagnostic alterations (CDx) had a similar PPA (84.0% solid, 83.3% liquid). Overall percent agreement (OPA) for PIK3CA CDx in paired samples was 91.4% (406/444). When acquired PIK3CA alterations are observed they tend to be subclonal, co-occur with shared PIK3CA alterations, and consist of rare alterations (E726K, E453K, M1043I). Conclusions: While high PPA to tissue was observed for alterations in truncal driver genes, acquired alterations observed in plasma were frequently polyclonal, often identify potential mechanisms of therapeutic resistance and are, in part, a function of time and intervening treatments relative to tissue CGP. These results demonstrate high PPA to tissue suggesting that liquid biopsy has clinical validity in identifying truncal alterations and can also identify acquired alterations in breast cancer.