Negative hyper-selection of RAS wild-type (wt) metastatic colorectal cancer (mCRC) patients randomized to first-line FOLFOX plus panitumumab (Pan) followed by maintenance therapy with either 5FU/LV plus pan or single-agent pan: Translational analyses of the VALENTINO study

Background: RAS wt unresectable mCRC pts were randomized to FOLFOX þ Pan (8
cycles) followed by maintenance with Pan (arm B) or Pan þ 5FU/LV (arm A). A pre specified translational endpoint was the evaluation of PRESSING panel, that groups
rare genomic markers beyond RAS/BRAFto predict anti-EGFR resistance in addition
to primary tumor location (PTL) (Cremolini, Ann Oncol ’17).
Methods: Primary endpoint was PFS. A sample size of 224 pts had 90% power to detect
50% 10-month PFS in arm A, max 15% less in arm B, significance level 0.1 (non-inferi ority margin of arm B vs A: 1.515). PRESSING panel analyses: ISH for HER2/
METamplification, IHC þ/- RNA-seq for ALK/ROS/TRKs/RETfusions, NGS (Hotspot
Cancer Panel, Ion TorrentVR ) for HER2/PI3K/PTEN/low % RAS mutations, PCR for
MSI.
Results: 229 pts randomized (117 arm A/112 arm B). At updated median follow-up of
18 mos, the upper boundary of 1-sided 90% CI of HR was 1.857. 10-m PFS was 49% in
arm B vs 59.9% in arm A (HR ¼ 1.51 [1.11-2.07]; p ¼ 0.009). A subgroup of 189 RAS/
BRAF wt evaluable pts had available tumor tissue for PRESSING analyses, with 46
(24%) PRESSING-pos tumors. Table 1 shows PFS according to PTL and PRESSING
panel, overall and by treatment arm. In post-hoc combined analyses of PTL and
PRESSING, right-sided and/or PRESSING-pos tumors were “predicted resistant (R)”
(arm B/A: 31/32); left-sided þ PRESSING-neg “predicted sensitive (S)” (arm B/A: 58/
68). mPFS: 8.1 vs 13.2 mos for predicted R vs S (HR ¼ 2.08 [1.47-2.93]; p < 0.0001); 7.7
vs 9.9 mos for arm B vs A in predicted R (HR ¼ 2.12 [1.16-3.89]), 12.4 vs 14.2 mos for
arm B vs A in predicted S (HR ¼ 1.54 [0.98-2.40]) (interaction p ¼ 0.126).
Conclusions: RAS/BRAFwt, right-sided and/or PRESSING-pos pts receiving mainte nance with Pan alone had extremely poor PFS. The PFS benefit of 5FU/LV added to
Pan was consistent in all subgroups