Background: RAS mutations in metastatic colorectal cancer (mCRC) drive resistance to anti-EGFR antibodies. It is unclear if RAS mutations are ever clonally lost, potentially uncovering a new therapeutic option for patients over time. Methods: We examined the temporal evolution of RAS mutation status among patients enrolled in CO.26 [NCT02870920] – a phase II clinical trial that assessed durvalumab + tremelimumab in patients with heavily pretreated mCRC – using whole exome sequencing (WES) of archival primary tumor tissue and circulating tumor DNA (ctDNA) sequencing of serial plasma samples that were taken at baseline, week 8 and on progression. Results: Six out of 95 (6.3%) patients diagnosed with KRAS/NRAS-mutated mCRC showed ‘neo-RAS-wildtype’ reversions at the time of baseline or week-8 ctDNA collection for the CO.26 trial. The majority (4/6) had falling tumor mutation burden (TMB) but stable or rising maximum mutation allele frequency (MAF) when reversions occurred. These were unlikely false negatives from non-secreting cancers as there were continued strong presence of other somatic clonal mutations (e.g., TP53, ATM). The majority (4/6) of reversions were transient, with mutations reappearing with progressive disease. ‘Neo-RAS-wildtype’ revertors had numerically longer median overall survival (OS) from date of cancer diagnosis to death compared those with persistent RAS mutations (7.7 vs. 4.2 yrs, HR = 0.65, 95% CI 0.31 to 1.36, log-rank P= 0.26). However, ‘neo-RAS-wildtype’ revertors were earlier-stage at diagnosis compared to those with persistent RAS mutations (33% vs. 63% had synchronous metastases, χ2 P= 0.15), and had lower disease burden upon enrollment (50% vs. 68.5% had liver metastases, P= 0.17; 33% vs. 75% had > 4 lesions, P= 0.03*). Survival from stage IV diagnosis to death did not differ between those with reverted vs. persistent RAS mutations (median 3.8 vs. 3.2 yrs, HR = 0.77, 95% CI 0.35 to 1.72, P= 0.52). ‘Neo-RAS-wildtype’ reversions were not associated with side of primary tumors (P= 0.41), archival BRAF/MEK/ERK-mutant status (P= 0.16, 1.00, 0.09), baseline HER2 amplifications (P= 1.00), or baseline TMB (P= 0.21). Conclusions: We identified a 6.3% prevalence of ‘neo-RAS-wildtype’ reversions in the CO.26 trial, however only 2.1% of patients had persistent loss when serial time points were considered. Further research is needed to understand if ‘neo-RAS-wildtype’ revertors may benefit from anti-EGFR therapy. Clinical trial information: NCT02870920.