GI Oncology Weekly — Issue 12

This week’s calls — 5 readouts · ~9 min read. Tap any row to jump.

VIRAGE Oncolytic adenovirus VCN-01 plus gemcitabine/nab-paclitaxel: durable responses and a landmark-survival gap, but ITT overall survival missed Worth watching ICI downstaging before liver transplant Checkpoint-inhibitor downstaging before transplant for HCC beyond Milan: median OS 4.57 vs 2.53 years, with acute rejection doubled Worth watching RAMIE Robotic vs conventional minimally invasive esophagectomy: phase 3 win on lymph-node yield, no survival difference at 1 year Worth watching INTERCEPT Routine-care ctDNA monitoring: MRD-positive stage II–III CRC recurs more often than MRD-negative stage IV Worth watching RO7589831 (WRN inhibitor) Covalent Werner helicase inhibitor in MSI/dMMR tumours: very well tolerated, 74.2% disease control, 10.6% response rate On the radar

This week · 5 readouts

The Verdicts

Every trial that matters this week, distilled to a single call — act on it, watch it, or keep it on the radar.

Worth watching Pancreas · first-line metastatic

VIRAGE

The most interesting pancreatic cancer mechanism in a while, and a trial whose headline needs reading carefully. VCN-01 did not make tumours shrink more — response rate, disease control and CA19-9 were flat — but responses lasted more than twice as long and 18-month survival more than tripled, which is what you would expect if digesting the stroma keeps chemotherapy reaching the tumour. Overall survival in the intent-to-treat population was not significant; the authors' 'primary endpoint met' rests on a full analysis set five patients smaller in the experimental arm. A mechanism that has earned a phase 3, not yet a regimen.

Go deeper

VCN-01 (zabilugene almadenorepvec) is an oncolytic adenovirus engineered to express soluble hyaluronidase, degrading the hyaluronic-acid-dense stroma that makes pancreatic tumours impenetrable to chemotherapy. This randomized phase 2b gave two intravenous doses of VCN-01 with gemcitabine/nab-paclitaxel versus gemcitabine/nab-paclitaxel alone as first-line therapy in metastatic pancreatic ductal adenocarcinoma. In the intent-to-treat population (VCN-01 + GnP n=53; GnP n=48), median OS was 10.6 versus 8.6 months — HR 0.69 (95% CI 0.42–1.12), P = 0.196, not significant — while PFS was 5.6 versus 4.6 months, HR 0.63 (95% CI 0.4–1.00), P = 0.046. In the full analysis set (n=48 per group) median OS was 10.8 versus 8.6 months, HR 0.57 (95% CI 0.34–0.96), P = 0.055, and PFS 7.0 versus 4.6 months, HR 0.55 (95% CI 0.34–0.88), P = 0.011; the authors report the primary OS endpoint as met in the FAS population. The most striking signal is durability rather than depth: duration of response was 11.2 versus 5.4 months, HR 0.22 (95% CI 0.08–0.63), P = 0.004, and landmark survival was 35.5% versus 12.8% at 15 months and 31.1% versus 8.5% at 18 months — yet there were no significant differences in overall response rate, disease control rate or CA19-9. Patients given two doses 14 weeks apart benefited most, with sustained circulating viral genomes indicating preserved second-dose bioactivity despite persistent neutralizing antibodies. Serious VCN-01-related events occurred in 22.6%, chiefly pyrexia, flu-like symptoms, transaminase elevation and thrombocytopenia, and were milder after the second dose; two fatal events occurred, one per arm, neither treatment-related. This digest covered the AACR 2026 topline on April 18, 2026 — this is the full peer-reviewed dataset behind it. Full daily entry →

Survival over time · schematic — OS*

VCN-01 + gem/nab-PGem/nab-P

· ITT OS 10.6 vs 8.6 mo (HR 0.69, 95% CI 0.42–1.12, P = 0.196, NS)· FAS OS 10.8 vs 8.6 mo (HR 0.57, 0.34–0.96, P = 0.055)· ITT PFS 5.6 vs 4.6 mo (HR 0.63, P = 0.046)· DOR 11.2 vs 5.4 mo (HR 0.22, P = 0.004)· 18-mo survival 31.1% vs 8.5%· n=101 randomised

HR 0.69 No significant difference 95% CI 0.42–1.12 · Δ +2 mo median · p=0.196

Nature Medicine · randomised phase 2b Debate on X →

Worth watching Liver · transplant oncology

ICI downstaging before liver transplant

The efficacy signal is large and consistent across overall survival, recurrence-free survival and competing-risk recurrence, which makes it hard to dismiss as noise. The cost is equally clear and should be presented to patients as part of the decision: acute rejection within 90 days doubled. This is a retrospective, propensity-matched, hepatitis-B-dominant Chinese cohort, so confounding by who was offered immunotherapy cannot be excluded and Western MASH-driven HCC is untested. Transplant programmes downstaging beyond Milan now need an explicit position on checkpoint inhibitors — including washout and rejection surveillance.

· median OS from transplant 4.57 vs 2.53 years, P < .001· OS HR 0.63 (95% CI 0.43–0.92)· RFS HR 0.59 (0.44–0.80)· recurrence subdistribution HR 0.54 (0.40–0.72)· acute rejection ≤90 days 16.3% vs 8.2%, P = .01· n=416 propensity-matched, 8 centres

HR 0.63 ↓ 37% risk of death 95% CI 0.43–0.92 · Δ +24.4 mo median

Survival over time · schematic — OS*

ICI-based downstagingNo ICI

Go deeper

Downstaging HCC that sits outside the Milan criteria into transplantability is one of the most consequential decisions in liver oncology, and whether checkpoint inhibitors belong in that pathway has been unresolved — partly because of a genuine fear of precipitating graft rejection. This multicentre retrospective cohort drew on 8 high-volume Chinese transplant centres (>55 liver transplants annually) between January 2019 and December 2024 and propensity-matched 416 recipients, 208 who received an ICI-based multimodal pretransplant strategy (typically with locoregional and/or targeted therapy) and 208 who did not. Median overall survival from transplant was 4.57 years (95% CI 4.34 to not estimable) with ICIs versus 2.53 years (95% CI 2.17-3.10) without, P < .001, and median recurrence-free survival 1.47 versus 0.89 years, P = .02. Multivariable Cox models gave an OS HR of 0.63 (95% CI 0.43-0.92) and an RFS HR of 0.59 (95% CI 0.44-0.80), with a Fine-Gray subdistribution HR of 0.54 (95% CI 0.40-0.72) for post-transplant recurrence. The safety signal is real and should not be buried: acute rejection within 90 days occurred in 16.3% (34/208) of the ICI group versus 8.2% (17/208), P = .01. The cohort was 90.6% male with a median age of 53 — a hepatitis-B-dominant population whose generalisability to Western MASH-driven HCC is unproven, and the design cannot exclude residual confounding by who was offered ICIs in the first place. Full daily entry →

JAMA Oncology · multicentre retrospective cohort, propensity-matched Debate on X →

Worth watching Esophagus · surgical technique

RAMIE

The randomised answer to a question that had only low-certainty observational data. Robotic esophagectomy delivered a more radical lymphadenectomy at equal surgical safety in four high-volume centres — but nodal yield is a surrogate, survival did not differ at one year, and the morbidity advantage suggested by earlier pooled, mostly non-randomised data did not replicate. Safe and thorough in expert hands; whether it changes outcomes awaits longer follow-up.

· median lymph nodes resected 36 vs 32, P = .005· luminal R0 99.0% vs 97.0%· 90-day mortality 1.0% vs 3.0%, P = .62· complications similar· no 1-year survival difference· n=218 randomised, 202 resected

Go deeper

RAMIE (robot-assisted minimally invasive esophagectomy) was compared head-to-head against conventional minimally invasive esophagectomy in a phase 3 randomized trial across four high-volume tertiary centres in Mainz, Amsterdam, Zurich and Cologne. Of 218 patients randomized between January 2021 and March 2025, 202 underwent transthoracic esophagectomy for cT1-4a, cN0-3, cM0 adenocarcinoma of the esophagus or esophagogastric junction (101 per arm; mean age 64.8 years, 89.6% male). The primary endpoint was met: median resected lymph nodes 36 with RAMIE versus 32 with MIE (P = .005). Luminal R0 resection was 99.0% versus 97.0%, postoperative complication rates were similar, 90-day mortality was 1.0% versus 3.0% (P = .62), and there was no difference in survival at 1 year. This digest covered a RAMIE-versus-MIE meta-analysis in May whose GRADE certainty was rated very low because the pooled data were predominantly non-randomised — and notably, the randomised trial reproduces the nodal yield but not the morbidity advantage that meta-analysis reported. Full daily entry →

JAMA Surgery · randomised phase 3, four centres in Germany, the Netherlands and Switzerland Debate on X →

Worth watching Colorectal · ctDNA surveillance

INTERCEPT

MRD status is now a stronger prognostic signal than stage, and serial testing nearly doubles sensitivity over a single draw — the most actionable number here. The sobering half: adjuvant therapy in stage IV showed limited durable benefit whatever the MRD result, and a quarter of patients already had macroscopic disease when ctDNA turned positive. Detection has outrun intervention; interventional trials are what is missing.

· MRD+ stage II–III vs MRD− stage IV recurrence, p < 0.0001· sensitivity 79.9% multi-timepoint vs 48.5% single-timepoint· macroscopic disease already present at ctDNA detection in 26%

Go deeper

INTERCEPT is a prospective observational cohort of stage II-IV colorectal cancer patients treated with curative intent and monitored longitudinally with tissue-informed ctDNA assays as part of routine care — which makes it a read on what MRD testing actually does to practice, not just on whether it predicts. The headline finding reorders something clinicians treat as fixed: MRD-positive stage II-III patients had a higher recurrence incidence than MRD-negative stage IV patients (p<0.0001), and the prognostic value persisted beyond 12 months from surgery rather than decaying. Multi-timepoint testing raised sensitivity for recurrence to 79.9% from 48.5% with single-timepoint testing, which is the most directly actionable number here. Two findings deserve more attention than they will get. Adjuvant therapy in stage IV disease showed limited durable benefit in MRD-positive and MRD-negative patients alike — detecting MRD is not the same as having something that works against it. And concurrent macroscopic disease was already present in 26% of cases at the moment ctDNA was detected, so the true lead time is smaller than raw numbers imply and varies across surveillance (p=0.007). ctDNA detection also drove more intensive second-line imaging (MRI, PET; p<0.0001). The authors are explicit that interventional trials are still required to establish clinical utility. Full daily entry →

Gut · prospective observational cohort, routine care

On the radar MSI / dMMR · synthetic lethality

RO7589831 (WRN inhibitor)

WRN has been a known synthetic-lethal dependency of MSI cancers for years; this phase 1 puts a drug against it into patients. It reads exactly as a phase 1 should — outstanding tolerability, high disease control, and a modest response rate — and direct target engagement in tumour tissue could not be demonstrated. The mechanism is the news; whether it matters clinically depends on where it lands relative to checkpoint inhibitors, which already work well in MSI disease.

· DCR 74.2% (49/66)· ORR 10.6% (7/66)· median PFS 6.7 mo (95% CI 4.1–8.5)· median OS 17.6 mo· 1 DLT, MTD not reached· AE discontinuation 3.4%· n=88

Go deeper

Werner syndrome helicase (WRN) has been a known synthetic-lethal dependency of microsatellite-unstable cancers for years without a drug to exploit it; this first-in-human phase 1 is the first clinical test of one. Part 1 escalated RO7589831 (VVD-133214), described by the authors as a first-in-class covalent WRN inhibitor, in 88 patients with MSI and/or mismatch-repair-deficient advanced solid tumours, dosed once daily, twice daily (150–1,000 mg) or three times daily (200–600 mg). Tolerability was the headline: a single dose-limiting toxicity (grade 2 nausea at 600 mg TID), the maximum tolerated dose never reached, mostly grade 1–2 toxicity, and only 3.4% (3/88) discontinuing for adverse events, with no grade 5 events. Among MSI efficacy-evaluable patients, disease control was 74.2% (49/66) but confirmed RECIST v1.1 partial responses numbered only seven (ORR 10.6%, 7/66), with median DOR 10.2+ months, median PFS 6.7 months (95% CI 4.1–8.5) and median OS 17.6 months. Direct target engagement in tumour tissue could not be demonstrated, though ctDNA and FDG-PET analyses supported biological activity; 150 mg and 600 mg BID were taken forward as recommended phase 2 doses. Full daily entry →

Nature Medicine · first-in-human phase 1

* Survival curves are illustrative — modeled from each trial’s reported median and hazard ratio (assuming exponential survival), not the published Kaplan–Meier curves.

The take

Every major result this week came with a gap between the number in the headline and the number a patient would care about most.

VIRAGE: 18-month survival more than tripled, durations of response doubled — and intent-to-treat overall survival was not significant.
RAMIE: the primary endpoint was met — on lymph nodes, not on survival.
Pre-transplant checkpoint inhibitors: survival nearly doubled, and so did acute rejection.
The WRN inhibitor: three in four patients had disease control, one in ten responded.
INTERCEPT: ctDNA finds recurrence better than staging does, and we still lack a treatment proven to act on it.

None of that makes these weak papers. Several are among the more important GI oncology results of the autumn. It means the useful skill this week is not spotting the positive trial — it is naming precisely what each one proved, and what it has left for the next trial to prove.

GI Oncology Weekly · curated by Dr. Allan Pereira, Moffitt Cancer Center. Educational summaries for practising oncologists — not medical advice; verify against primary sources before acting.

All weekly issues