Genomic biomarkers of primary resistance to anti-EGFR monoclonal antibodies (mAbs) by comprehensive genomic profiling (CGP) in <i>EGFR</i> -amplified (ampl) advanced gastroesophageal adenocarcinoma (aGEA) and correlative analysis from a single-arm study with panitumumab.

Abstract

4052 Background: EGFR ampl is found in up to 6% of patients with aGEA and accounts for the efficacy of anti-EGFRs in this patients’ subgroup. The AMNESIA panel with KRAS / PIK3CA / MET mutations and KRAS / EGFR / MET amplifications represented a paradigm of negative selection to trastuzumab in HER2+ aGEA. Here, we aimed to characterize the prevalence and prognostic relevance of genomic drivers of primary resistance to EGFR blockade in EGFR -ampl aGEA. Methods: Patients with EGFR -ampl aGEA were retrieved from 3 CGP screening sources, i.e. the ARMANI phase 3 trial, an observational cohort established at the Fondazione IRCCS Istituto Nazionale dei Tumori (INT) of Milan and 11 patients with EGFR ISH positive aGEA treated with panitumumab as ≥2L from the 117/15 INT screening platform (thereafter Panitumumab study). CGP was performed by means of FoundationOne CDx next-generation sequencing (NGS). AMNESIA-EGFR panel grouped genomic alterations with clinical and biological rationale as driver of primary resistance to anti-EGFR mAbs, i.e. EGFR rearrangements, FGFR2 / HER2 / MET / RAS co-ampl, RAS / MEK1 / PIK3CA ex20 mut, AKT1/2 mut and PTEN loss. We evaluated prevalence of AMNESIA-EGFR alterations in the ARMANI and observational cohort. Then, we explored the prognostic impact of AMNESIA-EGFR panel and lack of EGFR ampl by NGS in the Panitumumab study. Results: EGFR- ampl by NGS was found in 3/130 (2.3%) and 7/128 (5.5%) pts in the ARMANI and observational cohort. EGFR ampl was confirmed by NGS in 8/11 pts (73%) in the panitumumab cohort. Among patients with EGFR -ampl tumors by NGS, AMNESIA-EGFR alterations were found in 9/18 (50%) pts, including: KRAS ampl (16.7%), EGFR rearrangement (11.1%), PTEN loss (11.1%), FGFR2 co-ampl (5.6%), MET co-ampl (5.6%) and MEK1 mut (5.6%). AMNESIA-EGFR panel alterations were mutually exclusive, except for one case with concurrent MET co-ampl and MEK1 mutation. In the Panitumumab study, disease control rate (DCR) was 18.2% including a partial response and a stable disease. Median progression-free survival (PFS) and overall survival (OS) were 2.0 months (95%CI, 1.8-NA) and 5.3 months (95%CI, 3.38-NA). 7/11 pts had tumors with AMNESIA-EGFR alterations or lacked EGFR ampl by NGS. AMNESIA-EGFR+ or lack of EGFR ampl were associated with inferior DCR (0% vs 50%), PFS (median PFS 2.0 vs 3.8 months; HR 3.17 [95% CI, 0.64-NA] and OS (median OS 5.3 vs 6.4 months; HR 1.80 [95%CI, 0.45–7.14]). Conclusions: Genomic drivers of primary resistance to anti-EGFR mAbs are common in EGFR -ampl aGEA. A negative selection paradigm based on CGP may optimize outcomes with anti-EGFRs. Innovative drugs that may bypass resistance mechanisms to mAbs, such as bispecific antibodies, antibody-drug conjugates and bispecific T-cell engagers, are awaited.