Phase II study of high-sensitivity genotyping of KRAS, NRAS, BRAF and PIK3CA to ultra-select metastatic colorectal cancer patients for panitumumab plus FOLFIRI: the ULTRA trial

Abstract

Background: Several studies show the importance of accurately quantifying not only KRAS and other low-abundant mutations because benefits of anti-EGFR therapies may depend on certain sensitivity thresholds. We assessed whether ultra-selection of patients using a high-sensitive digital PCR (dPCR) to determine KRAS, NRAS, BRAF and PIK3CA status can improve clinical outcomes of panitumumab plus FOLFIRI. Patients and methods: This was a single-arm phase II trial that analysed 38 KRAS, NRAS, BRAF and PIK3CA hotspots in tumour tissues of irinotecan-resistant metastatic colorectal cancer patients who received panitumumab plus FOLFIRI until disease progression or early withdrawal. Mutation profiles were identified by nanofluidic dPCR and correlated with clinical outcomes (ORR, overall response rate; PFS, progression-free survival; OS, overall survival) using cut-offs from 0% to 5%. A quantitative PCR (qPCR) analysis was also performed. Results: Seventy-two evaluable patients were enrolled. RAS (KRAS/NRAS) mutations were detected in 23 (32%) patients and RAS/BRAF mutations in 25 (35%) by dPCR, while they were detected in 7 (10%) and 11 (15%) patients, respectively, by qPCR. PIK3CA mutations were not considered in the analyses as they were only detected in 2 (3%) patients by dPCR and in 1 (1%) patient by qPCR. The use of different dPCR cut-offs for RAS (KRAS/NRAS) and RAS/BRAF analyses translated into differential clinical outcomes. The highest ORR, PFS and OS in wild-type patients with their lowest values in patients with mutations were achieved with a 5% cut-off. We observed similar outcomes in RAS/BRAF wild-type and mutant patients defined by qPCR. Conclusions: High-sensitive dPCR accurately identified patients with KRAS, NRAS, BRAF and PIK3CA mutations. The optimal RAS/BRAF mutational cut-off for outcome prediction is 5%, which explains that the predictive performance of qPCR was not improved by dPCR. The biological and clinical implications of low-frequent mutated alleles warrant further investigations.