Top 5 Papers
#1
Source: British Journal of Cancer | Authors: Shi Y, et al. (Wang ZJ, Han JG, corresponding authors) | Published: September 29, 2026
Score: 9/20 — Base 6 (British Journal of Cancer) + 1 (prospective non-randomised cohort — scored as observational, not as a trial) + 2 (overall- and disease-free-survival benefit in the propensity-matched cohort) = 9. The survival bonus is granted on the matched and landmark analyses, but treatment was not randomised and the multivariable OS hazard ratio is internally inconsistent as reported (see summary).
Self-expanding metallic stents (SEMS) as a bridge to surgery have become a common way to convert an obstructing left-sided colon cancer from an emergency operation into an elective one, and the stent interval creates a window for neoadjuvant chemotherapy. This multicentre prospective cohort from five Chinese surgical centres enrolled 246 patients with completely obstructing left-sided colon cancer, comparing SEMS followed directly by surgery (n = 105) with SEMS followed by neoadjuvant chemotherapy and then surgery (n = 141); 1:1 propensity matching and an 8-week landmark analysis were used to mitigate selection and immortal-time bias. In the matched cohort (n = 206), SEMS-related complications were similar (15.5% vs 14.6%, P = 0.845), while the neoadjuvant pathway was associated with superior 5-year OS (66.2% vs 51.7%, P < 0.001) and 5-year DFS (65.3% vs 40.0%, P < 0.001), consistent in the landmark analysis. On multivariable Cox regression neoadjuvant chemotherapy was an independent protective factor for DFS (HR 0.661, 95% CI 0.551–0.830, P = 0.002) and for OS (reported as HR 0.436, 95% CI 0.188–0.996, P = 0.015). That OS line is worth flagging: a confidence interval whose upper bound sits at 0.996 is compatible with a P value near 0.05, not 0.015, so one of the three figures is likely misreported. The direction matches FOxTROT-style neoadjuvant thinking in colon cancer, but patients fit and willing to receive chemotherapy after stenting are not a random subset, and no propensity model fully removes that.
Post angle: The stent interval is dead time for most patients — this cohort argues for using it. Five-year OS 66% vs 52% after matching is a big number from a non-randomised design, and the OS hazard ratio does not add up as published. Hypothesis-strengthening, and a trial worth running. #GIOnc #CRC #ColonCancer #SurgicalOncology
#2
Source: BJS Open | Authors: Lal N, et al. (Jenkins JT, senior author) | Published: September 4, 2026 (indexed in PubMed October 2026)
Score: 8/20 — Base 5 (BJS Open — not in the scoring table, so Other = 5; no prior edition has scored it) + 2 (meta-analysis) + 1 (biomarker-guided organ preservation) = 8. No survival-benefit bonus: the pooled outcome is response prediction, not a treatment effect, and the sustained-cCR association was not significant.
Watch-and-wait after total neoadjuvant therapy (TNT) for locally advanced rectal cancer depends on correctly identifying who has truly responded, and ctDNA has been proposed as a systemic biomarker to sharpen that call. This systematic review and meta-analysis pooled six studies comprising 597 patients and found that ctDNA negativity after TNT was associated with a higher likelihood of clinical complete response (OR 8.01, 95% CI 1.43–44.90; I² = 27.0%). The association with sustained cCR, however, was not statistically significant (OR 3.46, 95% CI 0.03–394.64; I² = 59.9%), and across studies post-TNT ctDNA positivity tracked with inferior disease-free, distant recurrence-free and progression-free survival. The decisive caveat is sensitivity: ctDNA showed high specificity and positive predictive value for residual disease but low sensitivity for residual local disease and local regrowth, with a substantial proportion of patients with residual tumour remaining ctDNA-negative at restaging. The authors conclude that ctDNA should complement anatomical assessment for systemic risk stratification, not independently determine organ-preservation eligibility. This is the natural follow-on to the October 3 edition's NEO/CO.28 ctDNA analysis, which reached the same 'hypothesis-generating, not a decision rule' verdict from a single trial.
Post angle: A positive ctDNA after TNT is a red flag worth acting on; a negative one is not a licence to skip surgery. ctDNA reads the systemic picture and misses what is still sitting in the rectal wall — MRI and endoscopy stay in charge of organ preservation. #GIOnc #RectalCancer #ctDNA #WatchAndWait
#3
Source: ESMO Real World Data and Digital Oncology | Authors: Elbasan S, et al. (Virchow I, senior author) | Published: September 21, 2026
Score: 8/20 — Base 5 (ESMO Real World Data and Digital Oncology — not in the scoring table, so Other = 5; no prior edition has scored it) + 1 (retrospective / real-world) + 2 (overall-survival benefit in the propensity-matched cohort) = 8. The survival bonus is granted because OS is the primary endpoint and is significant after matching, but the comparator is a historical treatment era rather than a randomised arm.
TOPAZ-1 and KEYNOTE-966 made checkpoint-inhibitor chemoimmunotherapy the first-line standard in advanced biliary tract cancer (BTC), but real-world data with a chemotherapy-only comparator remain limited. This single-centre retrospective analysis from the West German Cancer Centre included 172 patients with advanced BTC treated between 2018 and 2024: 64 who received durvalumab with gemcitabine/platinum (treated 2022–2024) and a historical comparator of 108 treated with chemotherapy alone. Because cohort assignment followed treatment era rather than randomisation, the authors propensity-matched on age, sex, metastatic burden, CA19-9, tumour location and chemotherapy regimen. In the matched cohort (n = 77), durvalumab was associated with longer median OS (14.6 vs 11.1 months; HR 0.54, 95% CI 0.307–0.956, p = 0.032) and a higher ORR (45% vs 20%, p = 0.015); DCR was numerically higher (65% vs 47.5%, p = 0.083). The unmatched comparison was larger still (mOS 18.7 vs 12.8 months, p = 0.003), and no new safety signals were observed. The authors themselves describe the findings as supportive of, rather than exceeding, the pivotal trial data.
Post angle: Reassuring that durvalumab holds up outside the trial — but an HR of 0.54 against a 0.75 in TOPAZ-1 is a confounding warning, not a better drug. Era comparators flatter the newer arm. #GIOnc #BiliaryCancer #Cholangiocarcinoma #Immunotherapy
#4
Source: The Lancet Regional Health – Europe | Authors: Risumlund SL, et al. (Larsen PN, senior author) | Published: September 25, 2026
Score: 7/20 — Base 5 (The Lancet Regional Health – Europe — not in the scoring table, so Other = 5; no prior edition has scored it, and a regional Lancet title does not inherit the flagship's tier for one paper) + 2 (randomised phase II) = 7. No impact bonus: the efficacy comparison is null and the trial was not powered for equivalence, so the case for SBRT rests on safety alone.
Thermal ablation is widely used for oligometastatic colorectal liver metastases, while stereotactic body radiotherapy (SBRT) is a non-invasive alternative with little prospective comparative evidence. LAVA-CRLM randomised adults with one to three colorectal liver metastases, each ≤4.0 cm and judged suitable for both modalities with curative intent, 1:1 to microwave ablation (MWA) or SBRT at two Copenhagen centres. Between January 2019 and December 2024, 100 patients were randomised and 92 formed the modified intention-to-treat population. At a median follow-up of 34.9 months, 1-year freedom from local progression was 77.7% (95% CI 66.5–90.9) with MWA and 83.0% (72.3–95.4) with SBRT (HR 0.87, 95% CI 0.43–1.74; p = 0.70). Grade ≥2 treatment-related toxicity occurred in 13 of 52 patients (25%) treated with MWA versus 1 of 40 (3%) treated with SBRT, and grade ≥3 toxicity occurred only with MWA (4 patients, 8%), including one treatment-related death. A null efficacy result in a 92-patient phase II cannot establish equivalence, and the safety populations are unbalanced (52 vs 40), but a randomised head-to-head comparison is what this choice has largely lacked.
Post angle: No difference in local control, a 25% vs 3% gap in grade 2+ toxicity, and a treatment-related death on the ablation arm. Not an equivalence trial — but it moves SBRT from 'alternative' to 'reasonable default to discuss' for small CRC liver metastases. #GIOnc #CRC #SBRT #LiverMetastases
Additional Papers of Interest
-
The Oncologist — single-centre retrospective cohort of 339 resected gastric cancers after neoadjuvant chemotherapy, 78% of whom received adjuvant chemotherapy. No overall EFS or OS difference (EFS 7.0 vs 7.5 years, p = 0.089), and none in ypN0; in ypN+ disease EFS was longer (4.8 vs 2.7 years, p = 0.031) but OS was not (p = 0.071). High lymph-node ratio and Becker TRG 2–3 identified subgroups where adjuvant receipt tracked with longer EFS and OS. Retrospective, and confounded by fitness to continue therapy — but it points the same way as the 6,600-patient response-stratified meta-analysis this digest covered on September 3.
-
The Oncologist — randomised phase II closed prematurely for slow enrolment and lack of funding after 22 patients (12 regorafenib-first, 10 anti-EGFR-first). Median OS 21.9 vs 13.3 months, one-sided p = 0.114 — not significant — and sequential treatment was actually started in only 33% vs 80% of patients. The sequencing question remains open; this trial cannot answer it.
-
Int J Radiat Oncol Biol Phys — prospective PRO registry, 206 patients with baseline EORTC QLQ-CR29 (116 long-course chemoradiation, 90 short-course RT). Long-course caused greater declines in bowel and urinary symptoms during the preoperative period, but there were no sustained differences at 3 months, 6 months, 1 year or 2 years; a transient 1-year detriment after low anterior resection had resolved by 2 years. Non-randomised, and long-course patients had more high-risk features.
-
ESMO Gastrointestinal Oncology — 23-centre retrospective study of 171 patients with refractory colorectal cancer. Median rwPFS 2.5 months and OS 5.8 months, DCR 38.1% — numerically below FRESCO-2. Grade ≥3 TEAEs in 49.1%; proteinuria was the most frequent adverse event and the leading cause of dose modification, and initial dose reduction was associated with shorter rwPFS but not OS.
-
Hepatology — computational-pathology framework trained on TCGA-LIHC derives a metabolic (CTNNB1-enriched) versus proliferative (TP53-enriched) subtype split (OS HR 4.20) and a morphologic index that predicted OS (HR 3.17) with a C-index of 0.73, remained independent on multivariable adjustment (HR 2.97), and stratified OS in an in-house validation cohort (HR 2.16). Released as a public pipeline. Retrospective and TCGA-derived.
-
Nature Communications — in 71 gastric cancers resected after neoadjuvant anti-PD-1 plus chemotherapy, treatment-resistant tumours had greater TLS density and area in the muscularis propria, and those TLS were dysfunctional (more naïve B cells, fewer germinal-centre B cells). Proposed mechanism: PLA2G2A-positive tumour cells drive inflammatory CAFs that secrete CXCL14 and recruit naïve B cells. Translational, with no outcome hazard ratio — but a useful reminder that TLS presence alone is not a favourable biomarker.
-
Nature Cancer — published online October 6, 2026, ahead of PubMed indexing. In mouse models with multi-omics profiling, epithelial GSK3β drives a lactate–fumarate axis that polarises tumour-infiltrating neutrophils into an immunosuppressive, ferroptosis-resistant PD-L1-positive 'T3' state; pharmacological GSK3β inhibition with tideglusib attenuated that polarisation and potentiated checkpoint blockade in preclinical CRC models. Preclinical only, but a druggable route at the neutrophil arm of CRC immune resistance.
Back to all digests