Biomarkers of Primary Resistance to Trastuzumab in HER2-Positive Metastatic Gastric Cancer Patients: the AMNESIA Case-Control Study.

Abstract

<b>Purpose:</b> Refining the selection of HER2-positive metastatic gastric cancer patient candidates for trastuzumab is a challenge of precision oncology. Preclinical studies have suggested several genomic mechanisms of primary resistance, leading to activation of tyrosine kinase receptors other than HER2 or downstream signaling pathways.<b>Experimental Design:</b> We carried out this multicenter, prospective, case-control study to demonstrate the negative predictive impact of a panel of candidate genomic alterations (AMNESIA panel), including <i>EGFR/MET/KRAS/PI3K/PTEN</i> mutations and <i>EGFR/MET/KRAS</i> amplifications. Hypothesizing a prevalence of candidate alterations of 30% and 0% in resistant and sensitive HER2-positive patients, respectively, 20 patients per group were needed.<b>Results:</b> AMNESIA panel alterations were significantly more frequent in resistant (11 of 20, 55%) as compared with sensitive (0% of 17) patients (<i>P</i> < 0.001), and in HER2 IHC 2⁺ (7 of 13, 53.8%) than 3⁺ (4 of 24, 16.7%) tumors (<i>P</i> = 0.028). Patients with tumors bearing no candidate alterations had a significantly longer median progression-free [5.2 vs. 2.6 months; HR, 0.34; 95% confidence interval (CI), 0.07-0.48; <i>P</i> = 0.001] and overall survival (16.1 vs. 7.6 months; HR, 0.38; 95% CI, 0.09-0.75; <i>P</i> = 0.015). The predictive accuracy of the AMNESIA panel and HER2 IHC was 76% and 65%, respectively. The predictive accuracy of the combined evaluation of the AMNESIA panel and HER2 IHC was 84%.<b>Conclusions:</b> Our panel of candidate genomic alterations may be clinically useful to predict primary resistance to trastuzumab in patients with HER2-positive metastatic gastric cancer and should be further validated with the aim of molecularly stratifying HER2-addicted cancers for the development of novel treatment strategies. <i>Clin Cancer Res; 24(5); 1082-9. ©2017 AACR</i>.