Kinetic Profiling of RAS Mutations With Circulating Tumor DNA in the Canadian Cancer Trials Group CO.26 Trial Suggests the Loss of RAS Mutations in Neo-RAS-Wildtype Metastatic Colorectal Cancer Is Transient

PURPOSE In metastatic colorectal cancer (mCRC), RAS mutations drive resistance to anti-epidermal growth factor receptor antibodies. It is unclear whether RAS mutations ever become clonally undetectable. METHODS CO.26 was a phase II clinical trial that assessed durvalumab + tremelimumab in heavily pretreated mCRC. RAS mutation status was tracked over time using circulating tumor DNA (ctDNA) sequencing at baseline, week 8, and on progression. RESULTS Among the 95 patients with KRAS/NRAS mutations in their archival tumor tissue, 6.3% (6/95) had undetectable RAS mutations in ctDNA collected at baseline or week 8 of the CO.26 study. Of these, 67% (4/6) of disappearances were transient, with the same mutation reappearing with progressive disease. In three cases, the simultaneous persistence of other preexisting CRC-associated truncal mutations could not be demonstrated, suggestive of low tumor shedding of ctDNA, leaving the incidence of true clonal reversion to RAS-wildtype (WT) possibly as low as 3.2% (3/95). Fewer patients in the neo-RAS-WT group (33%) had greater than four lesions at trial baseline compared with patients with persistent RAS mutations (75%), P = .046. The likelihood of synchronous metastases at cancer diagnosis (33% v 63%; P = .15) or liver metastases at trial baseline (50% v 68.5%; P = .17) was not significantly different between patients with disappearing versus persistent RAS mutations. Overall survival from stage IV diagnosis (hazard ratio, 0.77 [95% CI, 0.35 to 1.72]; P = .52) was not significantly different between those with disappearing versus persistent RAS mutations. The disappearance of RAS mutations was not associated with primary tumor sidedness (P = .41), archival BRAF/MEK/ERK-mutant status (P = .16/1.00/.09), nor baseline ctDNA HER2 amplifications (P = 1.00). CONCLUSION We identified a 3.2%-6.3% prevalence of the neo-RAS-WT phenomenon in the CO.26 trial. However, 67% of apparent cases were transient with subsequent re-emergence.