Top 5 Papers
#1
Source: Nature Medicine | Authors: Xu C, Mannucci A, Han H, et al.; Goel A (senior). City of Hope, TGen, and centres across the USA, Japan, South Korea, China and Italy | Published: September 16, 2026
Score: 12/20 — Base 9 (Nature Medicine) + prospective international multicentre biomarker validation, scored at the Phase II tier (+2) rather than the retrospective tier because the testing cohort was prospectively enrolled and prespecified + biomarker-guided/precision (+1) = 12. The study-type table has no diagnostic-accuracy row; this is the judgment applied, and it is stated here so future editions can follow or revise it.
An international, multicentre, observational, prospective biomarker study across four countries enrolled 1,785 individuals with and without PDAC to develop and validate a 10-microRNA blood signature, then combined it with CA19-9 into a composite score called PANXEON. The miRNA signature alone reached an AUC of 88.6% in the testing cohort with 83.8% sensitivity for early-stage PDAC, and showed minimal cross-reactivity with other GI cancers. Adding CA19-9 lifted sensitivity to 86.8% for stage I–II disease, at a false-positive rate of 3.2% in low-risk controls but 15.6% in high-risk controls. PANXEON also flagged high-grade dysplasia in 64.3% of individuals with high-risk pancreatic cysts, and in a small serial cohort of 19 people the signature fell during neoadjuvant chemotherapy and after surgery and rose again before recurrence. The authors pair the assay with a Markov cost-effectiveness model against annual MRCP/EUS surveillance, and are explicit that larger prospective studies are still required.
Post angle: The headline is 86.8% sensitivity for stage I–II PDAC — but the number to argue about is the 15.6% false-positive rate in high-risk controls, because that is precisely the population you would deploy this in. #PDAC #LiquidBiopsy #GIOnc
#2
Source: ESMO Gastrointestinal Oncology | Authors: Tong Y, Nakayama I, Sakamoto N, Miyashita H, Shiba S, et al.; Shitara K (senior). National Cancer Center Hospital East, Japan | Published: September 2026 (online first; issue dated December 2026)
Score: 9/20 — Base 5 (ESMO Gastrointestinal Oncology — deliberately pinned at the Other tier on 2026-09-06 and not revised here) + retrospective cohort (+1) + biomarker-guided/precision (+1) + colleague engagement (+2: two GI oncology KOLs posted or reposted it within 48 hours) = 9.
A single-institution retrospective cohort of 272 patients with gastric or GEJ adenocarcinoma, 169 of them metastatic, characterised TROP2 expression by immunohistochemistry against the other biomarkers that currently drive treatment selection. TROP2 was expressed in 87%, 79% and 62% of tumours at cut-offs of ≥25%, ≥50% and ≥75% respectively. TROP2-high status was independent of CLDN18.2, PD-L1 and MMR status, and correlated with HER2 positivity (17.8% vs 5.8%, P=0.005). Most notably, 31.3% of TROP2-high metastatic patients were negative for all four currently actionable biomarkers. TROP2 status did not predict chemotherapy outcome, so this is a prevalence and co-expression map rather than a treatment-effect study — which is exactly what is needed before TROP2-directed ADCs are tested seriously in upper GI disease.
Post angle: Forget the prevalence number for a second: 31.3% of TROP2-high metastatic patients had nothing else actionable. That is a biomarker-orphan population with a druggable surface target sitting on it. #GastricCancer #ADC #GIOnc
#3
Source: JHEP Reports | Authors: Neely J, Yau T, Kudo M, Finn RS, Sangro B, Melero I, El-Khoueiry A, Harding JJ, et al.; Kelley RK (senior) | Published: September 19, 2026
Score: 9/20 — Base 6 (JHEP Reports, matching the 2026-08-29 precedent for this journal) + ancillary biomarker analysis of a randomised Phase III (+2, not the full +3 — this is a correlative study, not the primary trial report) + biomarker-guided/precision (+1) = 9.
Whole-exome sequencing, RNA-seq, immunohistochemistry and serology were applied to roughly 200–343 patients per arm from CheckMate 459, the randomised Phase III of first-line nivolumab versus sorafenib in advanced HCC, at a minimum follow-up of 33.6 months. Baseline tumour mutational burden was not predictive of survival with nivolumab. Three signals separated: tumours without Wnt/β-catenin alteration had an OS HR of 0.66 (95% CI 0.47–0.91), PD-L1 CPS≥10 gave an OS HR of 0.55 (0.33–0.93), and a high T-cell-inflammation signature gave 0.68 (0.47–0.98). Circulating correlates pointed the same way — high IFN-γ, low pro-collagen 11 and low granulocytic MDSC frequency all tracked with better nivolumab efficacy. The honest framing matters here: the parent trial was negative for OS, and the abstract states explicitly that outcomes with sorafenib were improved in Wnt/β-catenin-mutated tumours, so part of that hazard ratio reflects the comparator performing better rather than nivolumab performing worse.
Post angle: A negative Phase III still has something to teach. Wnt/β-catenin, CPS≥10 and T-cell inflammation each separate an HCC population that does benefit — but read the β-catenin result as sorafenib doing better, not nivolumab doing worse. #HCC #Immunotherapy #GIOnc
#4
Source: American Journal of Gastroenterology | Authors: Multicentre Japanese prospective cohort investigators | Published: September 17, 2026
Score: 9/20 — Base 6 (American Journal of Gastroenterology — society flagship, consistent with the 2026-08-31 and 2026-09-06 editions) + prospective multicentre cohort (+1) + survival outcomes supporting organ preservation as the standard in LVI-negative disease (+2) = 9.
385 patients (315 squamous cell carcinoma, 70 adenocarcinoma) who underwent endoscopic resection for pathological muscularis mucosae or SM1 oesophageal cancer were followed prospectively across multiple centres — a design point that matters, because the evidence base for organ preservation at this depth has been retrospective. Five-year overall survival was 92.3% (88.8–94.8) for SCC and 94.2% (85.4–97.8) for EAC. In LVI-negative patients managed with no additional treatment, five-year OS was 94.0% for SCC and 93.3% for EAC. Metastatic recurrence in LVI-negative SCC was 6.3% for pMM and 10.5% for pSM1, and in LVI-negative EAC just 2.2% and 0.0%. The cleanest subgroup was type 0-II lesions ≤30 mm, with metastatic recurrence of only 2.9% (1.1–7.5). The counterweight is metachronous disease: 25.8% of SCC patients developed a second primary oesophageal cancer by five years.
Post angle: Organ preservation holds up prospectively in LVI-negative early oesophageal cancer — 94% five-year OS with no additional treatment. The price isn't recurrence, it's surveillance: 25.8% developed a second primary by five years. #EsophagealCancer #Endoscopy #GIOnc
#5
Source: American Journal of Gastroenterology | Authors: UK polyposis registry investigators | Published: September 17, 2026
Score: 9/20 — Base 6 (American Journal of Gastroenterology) + registry cohort with life-table analysis benchmarked against national ONS population data (+1) + mortality difference demonstrating the benefit of registry-based screening (+2) = 9.
A registry cohort of 1,917 patients with familial adenomatous polyposis born after January 1935 recorded 364 deaths against 28 expected from national population life tables. The split by ascertainment route is the finding: relatives identified through screening had a median standardised mortality ratio of 0.66 — mortality comparable to the general population — while the unselected cohort had a median SMR of 11.09 (p<0.01). The causes of death have also shifted over time. Colorectal deaths fell from 64% of the total before 2001 to 31% after 2013 (p<0.01), while duodenal deaths rose from 3% to 12% and gastric deaths from 1% to 8% (both p<0.01). Read together, this is evidence that registry-based screening and colectomy have largely solved the problem they were designed for, and moved the residual risk upstream into the foregut.
Post angle: FAP registry screening works — screen-detected relatives now have essentially population-level mortality (SMR 0.66 vs 11.09 unselected). But colorectal deaths fell from 64% to 31% while duodenal and gastric deaths climbed. The weak link has moved upstream. #FAP #CRC #GIOnc
Additional Papers of Interest
-
ESMO Gastrointestinal Oncology — Pooled intratumoral discordance 35.8% across 13 studies and intertumoral discordance 20.3% across 12, arguing that a single specimen can misclassify zolbetuximab eligibility; this synthesis extends the primary-versus-metastatic discordance study covered in the September 3 edition
-
ESMO Gastrointestinal Oncology — In more than 650,000 patients, early-onset disease is enriched for signet-ring histology (OR 2.52), diffuse type (OR 2.26), CDH1 alteration (OR 2.92) and germline pathogenic variants (OR 2.00), and depleted of MSI-H (OR 0.17) and TMB-high (OR 0.37) — the upper-GI counterpart to the early-onset colorectal story
-
European Journal of Cancer — Multicentre single-arm Phase II met its primary endpoint with ORR 63.3%, DCR 86.7%, mPFS 9.0 months and mOS 15.0 months, though n=30, single-arm, and S-1 is not US-registered for PDAC, so the comparison is historical
-
Gut — 210,978 single-cell transcriptomes plus spatial profiling of 20 sections define a baseline responder programme built on CXCL13+ CD8+ exhausted T cells, TNFRSF9+ CD4+ Tregs, atypical memory B cells and APOE+ macrophages, and nominate LAG-3 as a target to potentiate response
-
Journal of Gastrointestinal Cancer — Postoperative ctDNA positivity predicts recurrence with a pooled HR of 7.60 across 7 prospective cohorts, but I²=80.6% and a 1.60–36.25 prediction interval are the honest headline; DYNAMIC supports selective stage II de-escalation while DYNAMIC-III did not establish non-inferiority for routine stage III de-escalation
-
Journal of the National Cancer Institute — Among nine less-studied PFAS in 853 cases and 853 matched controls, only perfluoroheptanoic acid was associated with liver cancer (OR 1.78, 95% CI 1.26–2.52), rising to 4.53 for hepatocellular carcinoma and 3.72 in men
Back to all digests