FOxTROT: Predictive effects of <i>ERBB2</i> and <i>ERBB3</i> on neoadjuvant panitumumab benefit in <i>RAS/BRAF</i> wild-type (-wt) locally advanced colon cancer (LACC).

Abstract

17 Background: The FOxTROT trial reported advantages of neoadjuvant chemotherapy in LACC. An embedded phase II trial tested neoadjuvant FOLFOX+panitumumab vs FOLFOX and reported a trend towards a TTR benefit and a significant DFS and OS benefit from panitumumab (pan) in RAS/BRAF -wt patients (pts). ERBB2 and ERBB3 overexpression have putative negative and positive predictive effects on pan efficacy, respectively. Here we examined whether ERBB2/3 could improve patient selection for neoadjuvant pan. The additional impact of AREG/EREG expression (EGFR ligands) on pan efficacy was tested. Methods: In FOxTROT, 279 pts with cT3-4 N0-2 M0, KRAS -wt LACC were randomized 1:1 to neoadjuvant FOLFOX±pan (6 weeks). All pts received FOLFOX post-op. 169 (61%) were confirmed RAS/BRAF -wt by NGS. ERBB2/3 and AREG/EREG were tested by RNAseq. Endpoints in this exploratory analysis were TTR, DFS, CCSS and OS. Results: 157 (93%) pts had available ERBB2/3 data. Median follow-up was 49 months. There were no associations between ERBB2/3 and stage, primary tumor location or MMR status. ERBB2 and ERBB3 were dichotomized using internally derived cut-points at the 50 th and 40 th percentiles respectively. Trends were consistent across all endpoints but strongest for DFS and OS, presented here. ERBB2 expression had no association with pan efficacy: trends towards benefit from FOLFOX+pan vs FOLFOX were similar in ERBB2 -high (DFS: HR 0.45 [95% CI 0.17-1.19], P = 0.11; OS: 0.40 [0.12-1.38], P = 0.15) and -low pts (DFS: 0.60 [0.23-1.59], P = 0.30; OS: 0.24 [0.05-1.18], P = 0.08). ERBB3 enabled improved patient selection for neoadjuvant pan: ERBB3 -high pts had significant benefit from FOLFOX+pan vs FOLFOX (DFS: 0.37 [0.14-0.97], P =0.04; OS: 0.16 [0.04-0.75], P =0.02), whereas ERBB3 -low pts did not (DFS: 0.80 [0.29-2.22], P =0.67; OS: 0.71 [0.18-2.84], P =0.63). Tests for interaction did not reach significance in this subgroup analysis (DFS: P interaction =0.28; OS: P interaction =0.14). We have previously reported the ability of AREG/EREG to predict pan benefit in this dataset. ERBB3 strengthened the predictive effect of AREG/EREG : AREG , EREG and/or ERBB3 -high pts had significant benefit from FOLFOX+pan vs FOLFOX (DFS: 0.38 [0.17-0.82], P =0.01; OS: 0.22 [0.07-0.67], P =0.008), while there was a trend towards disbenefit from pan in AREG , EREG and ERBB3 -low pts (DFS: 2.58 [0.50-13.31], P =0.26; OS: 3.05 [0.32-29.35], P =0.33). Biomarker-treatment interactions were significant (DFS: P interaction =0.03; OS: P interaction =0.045). AREG/EREG and ERBB3 were not correlated, suggesting differing underlying biology. Conclusions: High ERBB3 expression predicted benefit from neoadjuvant FOLFOX+pan. ERBB3 augmented the known predictive effect of the EGFR ligands, AREG/EREG . Validation in a biomarker-enriched trial is warranted. There was no association between ERBB2 expression and pan efficacy. Clinical trial information: NCT00647530 .