GI Oncology Daily Digest

The FDA granted Breakthrough Therapy designation to olomorasib (LY3537982), an oral next-generation KRAS G12C inhibitor, as monotherapy for adults with advanced pancreatic cancer who have had at least one prior systemic therapy and carry a KRAS G12C mutation. The designation rests on preliminary results from the open-label multicenter Phase 1/2 LOXO-RAS-20001 study (NCT04956640) in KRAS G12C-mutant advanced solid tumours; Lilly has not disclosed response rates or survival figures for the pancreatic cohort, and no efficacy numbers should be inferred from the designation itself. The clinical context is what makes this matter: roughly 90% of pancreatic adenocarcinomas are KRAS-mutant, but G12C accounts for only about 1-2% of them, and there is currently no approved therapy specifically targeting KRAS G12C-mutant pancreatic cancer. This is olomorasib's second Breakthrough Therapy designation, after the September 2025 grant with pembrolizumab in first-line KRAS G12C-mutant, PD-L1-high NSCLC.

SPACE-FLOT analysed 2,240 patients with locally advanced gastroesophageal adenocarcinoma treated with perioperative FLOT and surgery across an international registry, using Fine-Gray competing-risk regression to characterise recurrence patterns. Peritoneal relapse accounted for 41% of all recurrences and was the earliest site of failure (median 9.8 vs 11.4 months, p=.024). It was also the worst: median post-recurrence survival 4.4 vs 9.8 months and median overall survival 17.0 vs 24.3 months compared with non-peritoneal recurrence (both p<.001). Six pre-treatment and eight post-treatment clinicopathologic variables independently predicted peritoneal recurrence and were assembled into two prediction models with good discrimination and calibration at 12, 24 and 36 months post-surgery; decision-curve analysis showed both outperformed treat-none and treat-all strategies. The models are available as web-based calculators.

This is the most granular immune map yet of CLDN18.2-expressing gastric cancer, and it complicates the assumption that a CLDN18.2-high tumour is simply a better target. The authors profiled 103 patients with HER2-negative/low advanced gastric cancer on first-line checkpoint inhibition by multiplex immunohistochemistry — 7,423,483 cells across 2,315 regions of interest — then validated in five published transcriptomic cohorts totalling 1,521 samples, including the CheckMate 649 phase III trial. CLDN18.2-high tumours were reproducibly enriched for humoral pathways, with higher CD20+ B-cell density and B-cell/plasma-cell signatures. CLDN18.2 status alone did not predict survival on first-line immunotherapy. The spatial detail is the finding: in CLDN18.2-high tumours the B cells sat inside the tumour compartment rather than in tertiary lymphoid structure-like neighbourhoods, and that intratumoural localisation was associated with diminished checkpoint-inhibitor benefit — whereas in CLDN18.2-low tumours B-cell enrichment tracked with benefit. Digital spatial profiling of 480 regions confirmed compartment-specific activation of extra-follicular humoral programs.

Only about 30% of patients with advanced HCC respond to first-line immune checkpoint inhibition, and because tissue is rarely obtained in routine HCC practice, the field badly needs a blood-based selection tool. This group profiled immune checkpoint proteins on circulating extracellular vesicles across three cohorts: an explorer cohort (n=40), an early-stage cohort with paired blood and tissue (n=37), and a treatment cohort of 202 patients contributing 600 sequential samples, split into checkpoint-inhibitor training (n=79) and validation (n=82) groups plus a TKI-treated comparator (n=41). PD-1, PD-L1 and CTLA-4 were enriched on the vesicle fraction relative to vesicle-depleted serum. Baseline levels and early on-treatment dynamics separated responders from non-responders in both the training and validation checkpoint-inhibitor cohorts and predicted progression-free and overall survival — but not in the TKI cohort, which is the specificity control that makes the result credible. In patients who initially responded, rising vesicle checkpoint levels predicted acquired resistance approximately 36-42 weeks before progression was visible on imaging.