Novel potential mechanisms of acquired resistance to anti-EGFR monoclonal antibody (mAb) therapy detected in liquid biopsies (LBx) from patients (pts) with advanced colorectal cancer (CRC).

Abstract

199 Background: While several studies have explored the evolving clonal profile of mCRC under stress of anti-EGFR using circulating tumor DNA (ctDNA), heterogeneity and small cohorts limit our understanding of acquired genomic alterations (GAs) associated with EGFR resistance. Methods: CRC LBx were profiled with FoundationOne Liquid CDx, which reports GAs in 324 cancer-related genes. Foundation Medicine’s (FMI) ctDNA tumor fraction (TF) is a composite algorithm prioritizing aneuploidy at higher levels and variant allele frequency (VAF) of canonical alterations at lower levels. Variants were considered clonal at VAF/TF ≥25%, subclonal at <25%. Among 4,546 non-MSI-H LBx, 2,776 (61%) had TF ≥1% and were used in this analysis. A subset of pts had treatment data in the US-based de-identified Flatiron Health-FM real-world clinico-genomic database (CGDB), originating from ~280 US cancer clinics (01/2011–03/2023). Results: 1,349 (49%) LBx had a clonal RAS GA, 107 (3.9%) had a clonal BRAF V600E, and 1,320 (48%) had no clonal RAS pathway driver GA. GAs in EGFR and MAP2K1 were significantly (FDR<0.0001) more common in LBx with no clonal RAS GA (162, 11%; 105, 7.4%) than with one (27, 2.0%; 24, 1.8%), consistent with prior reports of these GAs appearing in the setting of acquired resistance to anti-EGFR mAb. The 362 LBx with GAs in EGFR, MAP2K1, or subclonal RAS GAs (indicators of anti-EGFR mAb exposure based on a validated exposure signature) were compared to the 1349 LBx with a clonal RAS mutation. Post anti-EGFR LBx were significantly enriched in BRAF, TP53, FGFR3, PTPN11, ERRFI1, and MLH1 GAs (FDR<0.05) and all but TP53 had median VAF/TF <25%. The majority of clonal BRAF GAs were V600E (14/19, 74%) while the majority of subclonal BRAF GAs were Class 2 (36/61, 59%). Other subclonal GAs in receptor tyrosine kinases and RAS pathway components detected in these LBx included: NF1 (26), ROS1 (9), ALK (7), MET (7), RAF1 (7), FGFR2 (6), HRAS (6), MAP2K2 (6), and RET (4). 18/362 (4.4%) of the LBx had ≥2 of these resistance GAs (up to 5 per LBx). In CGDB, 14/25 (56%) LBx collected at progression had potential resistance GAs (median 2, maximum 10). LBx collected at progression (N = 25) had higher prevalence of GA in EGFR (28% vs. 2.3%, p=0.0003), KRAS (28% vs. 0%, p<0.0001), NRAS (20% vs. 0%, p=0.0002) and MAP2K1 (20% vs. 2.3%, p=0.008) than LBx collected before therapy start (N = 87). Conclusions: This study reports on the largest set of LBx from pts likely exposed to anti-EGFR published to date. LBx can provide a more comprehensive and quantitative picture of acquired resistance to anti-EGFR in CRC than tissue biopsy, including some potential novel mechanisms identified in this study. Further investigation into whether detection of resistance can inform clinical decisions about timing of anti-EGFR mAb use and rechallenge is warranted.