Clinical and functional characterization of atypical KRAS/NRAS mutations in metastatic colorectal cancer


Jonathan M Loree 1, Yucai Wang 2, Muddassir A Syed 3, Alexey V Sorokin 4, Oluwadara Coker 4, Joanne Xiu 5, Benjamin A Weinberg 6, Ari M VanderWalde 7, Anteneh Tesfaye 8, Victoria M Raymond 9, Benjamin Miron 10, Gabi Tarcic 11, Ori Zelichov 12, Russell R Broaddus 13, Patrick Kwok Shing Ng 14, Kang Jin Jeong 15, Yiu Huen Tsang 16, Gordon B Mills 17, Michael J Overman 4, Axel Grothey 18, John L Marshall 19, Scott Kopetz 20


Purpose: Mutations in KRAS/NRAS (RAS) predict lack of anti-EGFR efficacy in metastatic colorectal cancer (mCRC). However, it is unclear if all RAS mutations have similar impact, and atypical mutations beyond those in standard guidelines exist.

Experimental Design: We reviewed 7 tissue and 1 cell-free DNA cohorts of 9,485 patients to characterize atypical RAS variants. Using an in vitro cell-based assay (functional annotation for cancer treatment), Ba/F3 transformation, and in vivo xenograft models of transduced isogenic clones, we assessed signaling changes across mutations.

Results: KRAS exon 2, extended RAS, and atypical RAS mutations were noted in 37.8%, 9.5%, and 1.2% of patients, respectively. Among atypical variants, KRAS L19F, Q22K, and D33E occurred at prevalence ≥0.1%, whereas no NRAS codon 117/146 and only one NRAS codon 59 mutation was noted. Atypical RAS mutations had worse overall survival than RAS/BRAF wild-type mCRC (HR, 2.90; 95% confidence interval, 1.24–6.80; P = 0.014). We functionally characterized 114 variants with the FACT assay. All KRAS exon 2 and extended RAS mutations appeared activating. Of 57 atypical RAS variants characterized, 18 (31.6%) had signaling below wild-type, 23 (40.4%) had signaling between wild-type and activating control, and 16 (28.1%) were hyperactive beyond the activating control. Ba/F3 transformation (17/18 variants) and xenograft model (7/8 variants) validation was highly concordant with FACT results, and activating atypical variants were those that occurred at highest prevalence in clinical cohorts.

Conclusions: We provide best available evidence to guide treatment when atypical RAS variants are identified. KRAS L19F, Q22K, D33E, and T50I are more prevalent than many guideline-included RAS variants and functionally relevant.

External Link