Top 5 Papers
#1
Source: Clinical and Translational Oncology | Authors: Yajima S, Yoshida S, Hirose K, … Masuda H, Fujii Y (National Cancer Center Hospital East; Institute of Science Tokyo) | Published: October 17, 2026
Score: 8/20 — Base 5 (Clinical and Translational Oncology — not in the scoring table, so Other = 5; no prior edition scored it) + meta-analysis (+2) + biomarker-guided (+1) = 8. No survival-benefit bonus: the only randomised contrast in the pool (adding an investigational agent to standard chemotherapy) was null at HR 0.95, and under scoring rule 4 the randomised subgroup governs, not the exploratory pooled estimate.
This PROSPERO-registered systematic review reconstructed individual patient data from published Kaplan–Meier curves to ask whether a negative postoperative ctDNA result identifies colorectal cancer patients who can safely receive less adjuvant treatment. Thirteen studies (9,552 patients) were included, with reconstructed data for 5,323 ctDNA-negative patients, 4,236 of them with CRC from five studies. The answer depended on the contrast: chemotherapy versus observation favoured treatment in the non-randomised GALAXY registry (HR 0.72, 95% CI 0.52–0.99), 6 versus 3 months favoured the longer course in IDEA-France (HR 0.75, 95% CI 0.58–0.97), while adding an investigational agent to standard chemotherapy in three randomised trials showed no clear benefit (HR 0.95, 95% CI 0.75–1.20). The exploratory pooled HR was 0.84 (95% CI 0.73–0.96; I² 0%), a 36-month DFS difference of only 2.2 percentage points (95% CI −0.3 to 4.6), and the CI crossed the null once GALAXY was analysed as a single study with small-sample methods. The authors conclude that current evidence does not support omitting or broadly de-escalating adjuvant therapy on a negative ctDNA result alone; complete omission has only been studied in non-randomised data.
Post angle: A negative ctDNA after CRC surgery is great prognostic news, but it does not yet tell you to stop treating. Across 4,236 ctDNA-negative patients, more treatment still edged out less, the randomised data were null, and the absolute gap was small. On this evidence, ctDNA-negative means a better prognosis, not permission to skip adjuvant therapy. #CRC #ctDNA #MRD #GIOnc
#2
Source: The Innovation | Authors: Liu ZC, Wang YX, Xu HM, … You WC, Pan KF, Li WQ (Peking University Cancer Hospital & Institute) | Published: Online April 16, 2026; October 2026 issue, newly indexed in PubMed
Score: 8/20 — Base 5 (The Innovation — not in the scoring table, so Other = 5; no prior edition scored it) + ancillary biomarker analysis of randomised trials (+2, the correlative-study value rather than the +3 for a primary trial report) + biomarker-guided (+1) = 8. Published online in April; included now because it entered PubMed in this window and has not appeared in a prior edition.
Using two randomised gastric cancer prevention trials from Shandong, China — the Mass Intervention Trial in Linqu (MITS; 1,740 participants analysed) and the Shandong Intervention Trial (SIT; 1,811) — the authors looked for host genetic variants that change how much H. pylori eradication prevents gastric cancer. They identified 11 variants that modified the preventive effect and combined their genotype-specific treatment effects into a polygenic treatment benefit score (PTBS). Treating the high-benefit group gave stronger protection than treating everyone infected or treating those at high polygenic risk: HR 0.44 (95% CI 0.35–0.54; NNT 56) in MITS and HR 0.20 (95% CI 0.10–0.41; NNT 26) in SIT. After eradication, PTBS also separated residual gastric cancer risk (MITS HR 0.65, 95% CI 0.52–0.81; SIT HR 0.27, 95% CI 0.13–0.56, for high-benefit individuals) and tracked with slower progression of precancerous lesions. Both cohorts come from one high-incidence region, so the score needs external validation before it informs programmes elsewhere.
Post angle: Screen-and-treat for H. pylori works, but not equally for everyone. An 11-variant genetic benefit score from two randomised trials picked out people whose NNT fell to 26–56. I would not use it to withhold eradication from anyone infected. Its practical value is in deciding who gets surveillance endoscopy after eradication. #GastricCancer #Hpylori #CancerPrevention #GIOnc
#3
Source: Critical Reviews in Oncology/Hematology | Authors: Ding S, Ni M, Tang Z, … Ye X, Chen B (Ningbo No. 2 Hospital; Zhejiang Chinese Medical University) | Published: October 17, 2026
Score: 8/20 — Base 5 (Critical Reviews in Oncology/Hematology — not in the scoring table, so Other = 5; no prior edition scored it) + meta-analysis (+2) + biomarker-guided (+1) = 8. No survival-benefit bonus: these are prognostic associations, not a treatment effect.
This meta-analysis pooled 44 studies of circulating tumour DNA in oesophageal and gastric cancer, including GEJ (28 in resectable and 17 in unresectable disease), and stratified the results by treatment time point. In resectable cancer, ctDNA positivity predicted worse DFS at baseline (HR 2.45, 95% CI 1.18–5.07), after neoadjuvant therapy (HR 3.89, 2.50–6.06), after radical surgery (HR 6.76, 4.72–9.67) and during follow-up (HR 19.17, 12.69–28.96), with a matching OS gradient (baseline HR 1.89; post-neoadjuvant 3.40; post-surgery 4.52). The association differed significantly by time point (DFS Q-test P<0.001; OS P=0.05). In unresectable disease, baseline positivity predicted worse PFS (HR 2.17) and OS (HR 2.04), as did on-treatment positivity (PFS HR 2.72; OS HR 2.63). These are prognostic estimates from heterogeneous assays and cohorts; the very large follow-up HR partly reflects ctDNA detecting recurrence that is already under way.
Post angle: In esophagogastric cancer, ctDNA becomes more prognostic the later you test: a DFS HR of 2.45 at baseline rises to 6.76 after surgery and 19 in surveillance. Read it alongside today's CRC paper, though. Being prognostic is not the same as telling you what to do. We still need trials that act on the result. #GastricCancer #EsophagealCancer #ctDNA #GIOnc
#4
Source: The Lancet Gastroenterology & Hepatology | Authors: Coppi MM, Graciotti M, Digklia A, Kandalaft LE (Ludwig Institute for Cancer Research, Lausanne; CHUV) | Published: October 14, 2026
Score: 7/20 — Base 7 (The Lancet Gastroenterology & Hepatology, per the 2026-09-06 and 2026-09-24 precedents) + no study-type bonus (a qualitative review of registered trials, not a meta-analysis) = 7.
The Lausanne group reviewed every therapeutic cancer vaccine trial in pancreatic ductal adenocarcinoma registered on ClinicalTrials.gov from inception to August 31, 2025, screening trial-linked publications up to January 31, 2026. Of 135 records, 120 met eligibility and went into a qualitative synthesis spanning cell-based, nucleic acid-based, protein and peptide, microorganism-based and multiplatform vaccines. Early-phase studies consistently showed feasibility, favourable safety and tumour-specific immune responses, but immune activation rarely translated into durable clinical benefit. The strongest biological signals came from personalised neoantigen vaccines, particularly mRNA-based ones, which induced high-magnitude CD8+ T-cell responses with long-lived clonal persistence after surgery, and shared neoantigens such as mutant KRAS emerged as scalable targets. The authors see the field moving to combinations that address stromal and myeloid immunosuppression, and say progress depends on adequately powered, biomarker-selected trials.
Post angle: Pancreatic cancer vaccine trials have mostly proven one thing: we can induce an immune response. 120 trials show that immunogenicity rarely turns into durable benefit. Personalised mRNA neoantigen and shared KRAS vaccines are the strongest leads, and the question now is whether adequately powered trials can show a survival gain. #PDAC #CancerVaccines #mRNA #GIOnc
#5
Source: BJS Open | Authors: Larsson P, Fernández Moro C, D'Souza M, Tingstedt B, Sparrelid E, Ghorbani P (Karolinska Institutet; Lund University) | Published: October 2026
Score: 6/20 — Base 5 (BJS Open — not in the scoring table, so Other = 5, per the 2026-10-06 precedent) + retrospective/real-world (+1) = 6.
Using the Swedish National Pancreatic Registry, the authors studied 1,043 adults who underwent pancreatoduodenectomy for primary resectable PDAC between 2017 and 2024: 398 (38.2%) had R0 and 645 (61.8%) R1 resections. Adjusted median survival was 8.2 months shorter after R1 than R0 (95% CI −12.0 to −4.3), and the loss grew with the number of involved margins: 4.6 months for one R1 margin (95% CI −10.2 to 1.0, not significant), 6.7 months for two (−11.9 to −1.5) and 12.4 months for three or more (−16.5 to −8.3). The authors argue that binary R0/R1 reporting is insufficient and that guidelines should adopt standardised, margin-by-margin pathology assessment.
Post angle: R1 is not one category. In 1,043 Swedish pancreatoduodenectomies, a single positive margin did not significantly shorten survival, but three or more cost 12.4 months of median survival. Pathology reports should say how many margins are involved and which ones, because a single yes/no hides that difference. #PDAC #SurgicalOncology #Pathology #GIOnc
Additional Papers of Interest
-
Cancers — 24 studies (20 experimental, four early-phase clinical). The clinical pool is three single-arm KRAS G12C trials only: pooled ORR 36.9% (95% CI 9.8–75.7%), DCR 85.9%, grade ≥3 TRAEs 23.8%, all at very low GRADE certainty. It does not cover the multi-selective RAS(ON) inhibitor daraxonrasib, FDA-approved on August 26 on Phase 3 RASolute 302.
-
Japanese Journal of Radiology — 31 studies, pooled bleeding response 75.9% (95% CI 69.7–81.2%), rebleeding 32.4%, median OS 3.8 months. Higher BED was associated with response, but the authors suspect selection bias and favour shorter fractionation for most patients.
-
Cardiovascular and Interventional Radiology — nine studies (one randomised), 1,534 patients. Tumour response and safety did not differ. The OS signal favouring B-TACE (HR 0.41, 95% CI 0.23–0.73) rests on only two studies with extractable hazard ratios.
-
Cancers — 653,825 preventive colonoscopies. Adenoma detection rate rose from 25.3% to 32.6%, and age-standardised CRC incidence and mortality fell 31.7% and 50.0% from 1998–2000 to 2022–2024. Only 58.0% had colonoscopy within six months of a positive FIT.
Back to all digests