ESMO Congress 2026

KRAS is one of the central driver events in pancreatic ductal adenocarcinoma (PDAC): integrated genomic analysis of 456 PDAC samples showed that 32 recurrently mutated genes aggregate into 10 core pathways, with the KRAS pathway among them ^[6]; systematic RNA interference work long ago noted that KRAS-mutant tumors are generally aggressive, respond poorly to standard therapies, and have remained refractory to direct targeting ^[4]. A therapeutic vaccine aims to bypass direct drugging by letting the immune system recognize mutant cells. Another thread directly relevant to this readout is circulating tumor DNA (ctDNA): digital PCR studies showed ctDNA was detectable in more than 75% of advanced pancreatic cancers and 48% of localized pancreatic cancers, and in a panel of 206 metastatic colorectal cancers, ctDNA detection of KRAS mutations had a sensitivity of 87.2% and specificity of 99.2% ^[3], establishing the methodological basis for post-surgical stratification.

At this ESMO 26 meeting, Eileen O'Reilly of MSK presented the randomized phase 2 AMPLIFY-7P study, published concurrently in Nature Medicine: 144 patients with resected KRAS-mutant PDAC received adjuvant off-the-shelf KRAS vaccine camupepimut (live posts noted coverage of G12D, G12R, G12V, G12S, G12A, G12C and G13D variants) versus observation. The primary endpoint of disease-free survival (DFS) was not met: 23.8 vs 21.2 months (HR 0.85; p=0.5329). Exploratory findings: baseline ctDNA positivity predicted worse DFS (HR 6.56); in prespecified subgroups, baseline ctDNA-negative HR 0.51 and no prior radiotherapy HR 0.59 (DFS 23.8 vs 10.4 months); 99% of vaccinated patients developed cancer-fighting T cells, and strong KRAS-specific T-cell responses were linked to better DFS (HR 0.22); grade >=3 adverse events were 10.4% vs 14.6%.

Relative to the long-standing difficulty of directly targeting KRAS ^[4], this readout sketches a viable outline for an off-the-shelf immune approach: as O'Reilly noted, an off-the-shelf vaccine could treat more patients more easily, faster and less expensively; the association between immune response magnitude and recurrence risk, along with the ctDNA stratification signals, echoes the earlier evidence of high sensitivity and specificity for ctDNA-based KRAS mutation detection ^[3], and supports the ctDNA and combinatorial development directions emphasized in the live posts. Two other KRAS threads at the same meeting deserve mention: an update of RASolute 302 raised the question of whether G12R (12-15%) constitutes a roadblock and whether protein degraders could be an alternative; and a lecture outlined the evolution of KRAS G12C from undruggable to approved by trapping the inactive GDP-bound state, with the next wave going after the active RAS(ON) state.

The limitations are equally clear: a phase 2 study with a missed primary endpoint, with all subgroup and biomarker signals being exploratory and based on limited numbers; as the live posts stated, these findings warrant further investigation and do not change the negative primary endpoint result; the G12R question likewise remains at the hypothesis stage.