Association between artificial intelligence (AI) -assisted tumor AREG and EREG immunohistochemistry (IHC) and outcomes from anti-EGFR therapy during the routine management of metastatic colorectal cancer (mCRC): An observational cohort study.

Abstract

203 Background: AREG and EREG are ligands of EGFR. AI assisted IHC evaluation of tumor AREG/EREG expression predicted benefit from anti-EGFR therapy in a retrospective analysis of the PICCOLO trial of second-line irinotecan ± panitumumab. Here, we sought to validate those findings through an analysis of patients who received anti-EGFR therapy during routine care for mCRC. Methods: Patients (pts) with available archival FFPE tumor tissue who received panitumumab or cetuximab ± chemotherapy at any time for treatment of mCRC at 8 UK cancer centers were eligible. Central RAS testing by next generation sequencing (NGS) was performed for pts where extended RAS testing had not been previously undertaken. RAS-mutant (mut) and RAS-unknown pts were excluded from the primary analysis population. AREG and EREG positive tumor cells were identified by IHC, performed locally at 6 of 8 sites. Pathologists annotated tumor areas on digital images of glass slides. AI algorithms calculated the percentage of tumor cells staining positive for AREG and EREG within tumor areas. More than 20% AREG and/or EREG tumor cell positivity was regarded as high biomarker expression, the optimal cut-point identified in PICCOLO. Study endpoints were progression-free survival (PFS), overall survival (OS), and locally reported response rate (RR) and disease control rate (DCR). Results: Of 541 pts recruited, 494 (91.3%) had adequate archival tissue for analysis. Central RAS testing was successfully performed in 255 of 393 (64.9%) pts without existing extended RAS results, leading to 45 exclusions, leaving 449 pts in the primary analysis population. 26 (5.8%) pts were BRAF-mut. 110 (24.5%) received concomitant FOLFOX and 304 (67.7%) FOLFIRI. After adjustment for additional prognostic factors, high AREG/EREG expression (n = 360; 80.2%) was associated with significantly prolonged PFS (HR 0.73; 95% CI, 0.56-0.95; p = 0.02), OS (HR 0.66 [0.50-0.86]; p = 0.002), and DCR (OR 1.92 [1.05-3.54]; p = 0.04), but not RR (unadjusted OR 1.39 [0.83-2.33]; p = 0.21). Median PFS in the high vs low biomarker groups was 8.5 vs 4.4 months; median OS 16.4 vs 8.9 months. The significant difference in OS (high vs low) was maintained in the subgroup with right-sided primary tumor location (n = 107; 23.8%) (HR 0.56 [0.37-0.86]; p = 0.007). Conclusions: High tumor AREG/EREG expression was associated with significantly prolonged PFS, OS and DCR among a cohort of pts treated with anti-EGFR therapy during routine care of mCRC. The prognostic effect observed validates the predictive effect of AREG/EREG seen in the PICCOLO trial, where AREG/EREG had no prognostic effect in patients receiving chemotherapy alone. A prospective biomarker-led trial is in set-up to support the use of AREG/EREG IHC in clinical practice.