Ultra-selection of metastatic colorectal cancer patients using next-generation sequencing to improve clinical efficacy of anti-EGFR therapy.

ABSTRACT

BACKGROUND: Extended RAS analysis is mandatory in metastatic colorectal cancer (mCRC) patients. The optimal threshold of RAS mutated subclones to identify patients most likely to benefit from antiepidermal growth factor receptor (EGFR) therapy is controversial. Our aim was to assess the clinical impact of detecting mutations in RAS, BRAF, PIK3CA and EGFRS492R in basal tissue tumour samples by using a highly sensitive next-generation sequencing (NGS) technology in mCRC patients treated with chemotherapy plus anti-EGFR or anti-vascular endothelial growth factor. PATIENTS AND METHODS: Five hundred and eighty-one tumour samples from untreated mCRC patients from 7 clinical studies were collected. Mutational analysis was carried out by standard-of-care (therascreen pyro) with a sensitivity detection of 5% mutant allele fraction (MAF), and compared with NGS technology using 454GS Junior platform (Roche Applied Science, Germany) with a sensitivity of 1%. Molecular results were correlated with clinical outcomes. RESULTS: After quality assessment, 380 samples were evaluable for molecular analysis. Standard-of-care mutational analysis detected RAS, BRAFV600E or PIK3CA mutations in 56.05% of samples compared with 69.21% by NGS (P = 0.00018). NGS identified coexistence of multiple low-frequency mutant alleles in 96 of the 263 mutated cases (36.5%; range 2-7). Response rate (RR), progression-free survival (PFS) and overall survival (OS) were increasingly improved in patients with RAS wild-type, RAS/BRAF wild-type or quadruple (KRAS/NRAS/BRAF/PIK3CA) wild-type tumours treated with anti-EGFR, assessed by standard-of-care. No additional benefit in RR, PFS or OS was observed by increasing the detection threshold to 1% by NGS. An inverse correlation between the MAF of the most prevalent mutation detected by NGS and anti-EGFR response was observed (P = 0.039). EGFRS492Rmutation was not detected in untreated samples. CONCLUSIONS: No improvement in the selection of patients for anti-EGFR therapy was obtained by adjusting the mutation detection threshold in tissue samples from 5% to 1% MAF. Response to anti-EGFR was significantly better in patients with quadruple wild-type tumours.