545P Epidermal growth factor receptor (EGFR) copy number (CN) as a biomarker of prognosis and panitumumab (Pan) benefit in RAS-wt advanced colorectal cancer (aCRC)

Background: Whilst RAS mutations predict which aCRC patients (pts) will not benefit from anti-EGFR agents, RAS-wt status does not reliably predict who will. Several studies report that high EGFR ligand expression (EREG/AREG) is predictive of anti-EGFR agent benefit but progression towards clinical utility is limited by lack of consensus on a clinical dichotomisation point and additional tumor material required. Here we explore EGFR copy number as a biomarker of prognosis and predictor of Pan benefit in a randomised trial in aCRC.

Methods: EGFR CN, EREG/AREG RNA expression, RAS/RAF mutations were assessed in tumor from 275 pts randomised to 2nd-line irinotecan (Ir) or IrPan (PICCOLO, Lancet Onc 14:749-59). EGFR CN status was measured by the Affymetrix OncoScan array, analysed using Biodiscovery Nexus software and defined as normal (2 copies) or gain (>2 copies). Prognostic analysis was in Ir alone pts. Predictive analysis, in the 234 RAS-wt pts, compared baseline values with outcomes using Cox proportional hazards models.

Results: 196 (71.3%) pts were classified as EGFR gain and 79 (28.7%) as normal. EGFR gain was significantly associated with high EREG and AREG RNA expression (both p < 0.001). EGFR gain was not prognostic for OS (p = 0.97) or PFS (p = 0.98). However, it was predictive of Pan benefit: in RAS-wt pts with EGFR gain, median PFS was 5.7 mo (IrPan) vs 3.7 mo (Ir) (HR = 0.60[0.43-0.83], p = 0.002), but pts with normal EGFR had no benefit: 3.4 mo (IrPan) vs 2.9 mo (Ir) (HR = 1.23[0.72-2.08], p = 0.45); interaction p = 0.02. Significant PFS biomarker/treatment interaction was also seen in all 275 pts, including RAS or RAF mutants (p = 0.01). In RAS-wt pts EGFR gain was associated with higher response rates than normal with IrPan (45.3% vs 18.7%, p = 0.01) but not with Ir (13.3% vs 12.9%, p = 1.0). Interaction was not significant (p = 0.22) or for OS (p = 0.20).

Conclusions: EGFR CN status may allow for further stratification for Pan benefit in RAS-wt patients. Normal EGFR CN status identified nearly 1/3 of pts without Pan benefit. This biomarker is consistent with EGFR ligand data, and is a DNA-based assay with a clearly definable dichotomised cut-point and therefore is of potential clinical utility.