GI Oncology Daily Digest

Revolution Medicines released topline results from RASolute 302, the pivotal Phase 3 of daraxonrasib (RMC-6236, a pan-RAS(ON) multiselective inhibitor) versus chemotherapy in previously treated metastatic PDAC: median OS 13.2 versus 6.7 months, HR 0.40, p<0.0001. PFS also significantly improved. The full plenary dataset will be presented at ASCO 2026 (LBA5, Sunday May 31). After two decades of failed direct-RAS targeting, this is the first Phase 3 OS win in mPDAC for a RAS-directed agent and reframes 2L treatment for the ~92% of pancreatic cancers driven by KRAS mutations. (The May 7 and May 13 digests covered the NEJM Phase 1–2 publication and the topline Ph3 announcement; this entry captures the formal pre-plenary topline confirmation ahead of the ASCO 2026 plenary debut.)

Jazz Pharmaceuticals has confirmed that the PD-L1 subgroup analysis of HERIZON-GEA-01 will be presented at ASCO 2026 (Abstract 4010, Rapid Oral). The pivotal Phase 3 enrolled 914 patients with 1L HER2+ metastatic gastroesophageal adenocarcinoma and randomized between zanidatamab + chemotherapy ± tislelizumab and trastuzumab + chemotherapy. The primary analysis (ASCO GI January 2026, JCO LBA285) showed mPFS 12.4 vs 8.1 months and an OS signal across the HER2+ population. The clinical question motivating Abstract 4010 is the post-KEYNOTE-811 one: whether IO benefit in HER2+ GEA is gated by PD-L1 CPS or extends across all PD-L1 subgroups. Full PD-L1 subgroup data remains embargoed until the ASCO 2026 presentation — but if the subgroup readout confirms benefit irrespective of PD-L1, zanidatamab + IO + chemo could replace trastuzumab + chemo as 1L SOC across the entire HER2+ GEA population, not just CPS ≥1.

CIRCULATE (AIO KRK-0217 / ABCSG) is the long-awaited Phase 3 randomized trial of ctDNA-guided adjuvant decisions in stage II microsatellite-stable colon cancer. The trial design (already published in 2021) randomizes ctDNA-positive patients 2:1 to adjuvant chemotherapy (capecitabine, with oxaliplatin at investigator's discretion) versus follow-up, and ctDNA-negative patients 1:4 to follow-up within or outside the trial. The full disease-free survival and time-to-recurrence readout is scheduled for the Rapid Oral Abstract Session at ASCO 2026 on Saturday May 31, 8:00 AM CDT (LBA3500). Specific results remain embargoed. The clinical stakes are high: stage II colon cancer is precisely the setting where the benefit-versus-harm calculus for adjuvant chemotherapy is most closely balanced, and a positive trial would give clinicians a Phase 3 evidence base to escalate in ctDNA-positive patients and de-escalate in ctDNA-negative ones — replacing the ~25-year-old T-stage and pathologic-risk heuristics that currently drive these decisions.

This Bayesian network meta-analysis pooled 11 RCTs (8,999 patients) comparing six immune checkpoint inhibitor (ICI) strategies in 1L advanced gastric/GEJ cancer: PD-1 monotherapy, PD-1 + chemo, PD-L1 monotherapy, PD-L1 + chemo, dual ICI, and dual-target ICI-chemo (e.g., HER2 + PD-(L)1). Dual-target ICI-chemo combinations ranked highest for OS and PFS, with cadonilimab-chemo emerging as the most effective regimen overall and SHR1701-chemo best in CPS ≥5. No ICI strategy demonstrated benefit over chemo alone in CPS <1, reinforcing current biomarker-based deployment. The indirect-comparison framework remains hypothesis-generating until head-to-head trials confirm, but the dataset is the largest synthesis to date and aligns with the trajectory seen in KEYNOTE-859, CheckMate-649, and the HER2+ HERIZON-GEA-01 results above.

This translational study identifies ZNF274 (a KRAB-zinc-finger transcription factor) as a constraint on lineage plasticity in pancreatic ductal adenocarcinoma and as a determinant of CDK7 inhibitor sensitivity. ZNF274 loss derepresses ZEB1, driving a classical-to-basal PDAC subtype switch and conferring intrinsic resistance to CDK7 inhibition. The mechanism also engages an endogenous retrovirus (HERV) dsRNA response that reinforces the basal transition. Clinically, this provides a candidate biomarker framework for matching CDK7 inhibitors to ZNF274-intact (classical-leaning) tumors, while flagging ZNF274-deficient PDACs as poor candidates for CDK7-directed regimens. The work strengthens the case for PDAC subtype-aware drug development — long discussed but rarely actionable.