This Week in Gastroenterology — Aug 25, 2026

Welcome to This Week in Gastroenterology. This week we're covering 10 notable papers spanning hepatocellular carcinoma and liver disease management, colonoscopy quality and colorectal cancer biology, and new directions in inflammatory bowel disease diagnosis and treatment. Let's dive in.

We start with liver cancer, where two phase 3 trials published in the Lancet family push combination immunotherapy into earlier, locoregional territory. In the Lancet Gastroenterology and Hepatology, Kudo and colleagues reported TALENTACE, which randomised 342 patients across 40 centres in China and Japan with systemically untreated, high tumour burden unresectable hepatocellular carcinoma to on-demand transarterial chemoembolisation alone, or the same chemoembolisation plus atezolizumab and bevacizumab every three weeks. The primary endpoint of chemoembolisation progression-free survival — time to untreatable progression, chemoembolisation failure, or death — favoured the combination, with a median of just over eleven months versus about seven months, a reduction in the risk of that composite event of roughly thirty percent. Importantly, overall survival remains immature and was reported only at the first prespecified interim analysis, so this is a delay-of-progression signal, not yet a survival claim [2]. Alongside it, in the Lancet itself, Sun and colleagues reported TALENTOP, which asks a different and rather bold question: if a patient with macrovascular invasion responds to systemic therapy, should you take them to the operating theatre? Nearly 500 treatment-naive patients with macrovascular invasion but no extrahepatic disease received induction atezolizumab plus bevacizumab, and the 201 who had at least stable disease and were judged resectable were randomised to resection followed by twelve months of the same systemic therapy, or to maintenance therapy alone. Median time to treatment failure was about twenty months with surgery versus roughly twelve months with maintenance, a forty percent relative reduction in the hazard. That benefit came at a cost: grade three or four treatment-related adverse events were nearly twice as common in the surgery arm, close to two in five patients, and there were two treatment-related deaths after surgery, including one from liver failure [3]. Taken together, these trials support a more aggressive multimodal posture in advanced hepatocellular carcinoma, but both use composite progression endpoints rather than mature overall survival, and the surgical question in particular demands careful multidisciplinary patient selection.

Staying with the liver, two papers reframe how we manage risk and bleeding in cirrhosis. In Clinical Gastroenterology and Hepatology, Janko and colleagues randomised 57 procedures in patients with cirrhosis and abnormal clotting parameters to rotational thromboelastometry-guided transfusion or standard of care. Roughly three quarters OF PROCEDURES were high bleeding risk. Prophylactic blood products were given before 71 percent of procedures in the viscoelastic-guided arm versus 97 percent under standard care, with a shift away from fresh frozen plasma toward cryoprecipitate, and — critically — no major procedure-related bleeding in either group, and no difference in thrombosis, length of stay, or mortality. This is a small trial, and it was not powered to exclude a modest bleeding excess, but it adds to a growing consensus that conventional INR and platelet thresholds drive a lot of transfusion that patients with cirrhosis simply do not need [4]. Complementing that, a review in Gut from Jeffrey and colleagues argues we should stop treating a hepatic venous pressure gradient of ten millimetres of mercury as the destination and instead use non-invasive tests to predict the events patients actually care about. They note that clinically significant portal hypertension carries roughly a 29 percent four-year risk of decompensation versus about ten percent without it, and that elastography-based and blood-based models — ANTICIPATE, the non-invasive estimated risk score, the Portal Hypertension Decompensation Score — achieve prognostic accuracy comparable to invasive pressure measurement, with the advantage of repeatability over time. Their call is for standardised, classical decompensation endpoints and outcome-based language rather than surrogate thresholds [5].

Now to colonoscopy quality, where this week gives us a technology trial, a consensus document, and a provocative challenge to the whole adenoma-carcinoma timeline. In the American Journal of Gastroenterology, Nishikawa and colleagues randomised 606 patients to standard colonoscopy or to an additional artificial intelligence-assisted second forward-view examination of the right colon. Right colon adenoma detection rose from about 28 percent to just over 36 percent — an absolute gain of eight and a half points — driven by small, flat lesions in the ascending colon, and the benefit held regardless of endoscopist experience. The authors are appropriately candid that this design cannot separate the contribution of the artificial intelligence from the contribution of simply looking twice [1]. That matters because detection is only half the story; complete resection is the other half. Also in Clinical Gastroenterology and Hepatology, van Bokhorst and colleagues report the SCOPE Delphi consensus, in which 67 endoscopy experts from 18 countries reached agreement on 18 of 20 statements about assessing complete resection of polyps under 20 millimetres. The recommendations are concrete: high-definition white-light inspection of the defect as the primary assessment, image-enhanced inspection and routine post-resection photo documentation for polyps in the ten to nineteen millimetre range, standard histopathological margin assessment for en bloc specimens, extended margin resection as the preferred research method for measuring incomplete resection, and — the one that will land in your unit — endoscopists should know and monitor their own personal incomplete resection rate [9]. Set against both of those, a conceptual piece in Gut from Truninger and colleagues argues that some colorectal cancers will never be prevented by finding and removing precursors, because they may not follow the slow precursor pathway at all. They propose that microsatellite-stable early-onset colorectal cancer and a subset of post-colonoscopy cancers reflect exposure-driven accelerated carcinogenesis, in which subclinical inflammation, microbiome disruption, and epigenetic remodelling compress the timeline from mutation to tumour — potentially arising from very brief precursor phases or directly from dysplastic mucosa. This is a hypothesis, not data, but it's a useful corrective when we discuss interval cancers with patients and when we think about risk stratification beyond polyp counts [7].

Finally, three papers on inflammatory bowel disease, spanning diagnosis, deprescribing, and an unexpected candidate therapy. In Alimentary Pharmacology and Therapeutics, Truniger and colleagues prospectively measured anti-integrin alpha-v-beta-6 autoantibodies in 224 patients and found an overall accuracy of about 87 percent for distinguishing ulcerative colitis from Crohn's disease, with sensitivity and specificity both in the mid-to-high eighties. Positivity followed a striking anatomical gradient — around eight percent in ileal and ileocolonic Crohn's, but nearly 44 percent in colonic Crohn's and 86 percent in ulcerative colitis — and in the first head-to-head comparison in the same cohort, these autoantibodies clearly outperformed ANCA and ASCA status [10]. On the therapeutic side, in Clinical Gastroenterology and Hepatology, Patel and colleagues used the prospective multicentre BISCUIT inception cohort of 679 children with Crohn's disease and found that 18 percent were started on aminosalicylate monotherapy within 90 days of diagnosis. That choice was associated with more systemic corticosteroid exposure, a delay of roughly eleven months before biologic initiation, and among those who eventually escalated, a more than fourfold increased risk of biologic discontinuation. Aminosalicylates and immunomodulators did not prevent complications, whereas early anti-tumour necrosis factor therapy was associated with substantially lower odds of perianal disease. The message is blunt: aminosalicylate monotherapy in paediatric Crohn's disease is undertreatment [6]. Most speculative is a matched cohort study in Inflammatory Bowel Diseases from Alqinai and colleagues, in which 150 patients with ulcerative colitis started on liraglutide or semaglutide for metabolic indications were matched to 150 who were not. Symptomatic remission at twelve weeks was about 67 percent with a GLP-1 receptor agonist versus roughly 25 percent in controls, an independently associated roughly sixfold increase in the adjusted odds, and notably weight loss itself did not explain remission. This is retrospective, single-centre, and open to substantial confounding by indication and by concurrent IBD therapy, so treat it as hypothesis-generating rather than a reason to prescribe.

If you only have time for one paper this week, make it the paediatric Crohn's disease cohort in Clinical Gastroenterology and Hepatology [6]. It quantifies the real cost of a habit many of us still see in practice, and the change it asks for — stop using aminosalicylates as monotherapy and move to effective therapy early — costs nothing to implement.

Here are the key takeaways from this week in Gastroenterology. First, in advanced hepatocellular carcinoma, adding atezolizumab and bevacizumab to on-demand chemoembolisation, and offering resection to responders with macrovascular invasion, both delayed treatment failure — but overall survival data are immature and surgical toxicity was meaningful. Second, viscoelastic testing safely cut prophylactic transfusion in patients with cirrhosis undergoing procedures, reinforcing that INR and platelet thresholds overtreat these patients. Third, in the right colon, an artificial intelligence-assisted second look raised adenoma detection by more than eight absolute percentage points, and international consensus now says you should be tracking your own incomplete resection rate for polyps ten millimetres and above. Fourth, a serum autoantibody to integrin alpha-v-beta-6 outperformed ANCA and ASCA for separating ulcerative colitis from Crohn's disease, with high positivity even in colonic Crohn's. And fifth, early aminosalicylate monotherapy in paediatric Crohn's disease means more steroids, later biologics, and shorter biologic durability — while the GLP-1 signal in ulcerative colitis is intriguing but far too preliminary to act on.

That's your roundup for This Week in Gastroenterology. The full transcript and references are available on the episode page in your AudioScholar library. This is an AI-curated summary — for clinical decisions, always consult primary sources and current guidelines. See you next week.

If this weekly briefing is useful, follow the show in your podcast app so new episodes arrive automatically. And think of one colleague — in any specialty — who never has time to keep up with the literature. Tell them about AudioScholar: a free ten-minute weekly for every specialty, to listen to in any podcast app, or to read at audioscholar dot C C.

This is an automated summary generated by artificial intelligence, which can make mistakes. Always review the original source materials.