Top 5 Papers
#1
Source: Journal of Clinical Oncology (Phase II) | Authors: Gelli, Ewald, Tanguy, et al. — Gustave Roussy & 14-center French collaboration | Published: April 22, 2026 (JCO Vol 44 Issue 15)
Score: 12/20 — JCO base (8) + Phase II RCT (+2) + survival benefit on the primary endpoint h-RFS and trend toward OS (+2) = 12. Practice-relevant for high-risk CRLM patients with ≥4 metastases — addresses an area where adjuvant chemo benefit has been ambiguous since JCOG0603.
99 patients (50 HAI vs 49 IV) with ECOG 0-1 and curative resection/ablation of ≥4 CRLM after preoperative IV chemotherapy were randomized to adjuvant oxaliplatin via HAI plus IV LV5FU2 vs all-IV oxaliplatin + LV5FU2 for ≥3 months. At median follow-up 59 months, median hepatic RFS (primary endpoint) was 25 mo with HAI vs 12 mo with IV (HR 0.63, 95% CI 0.40–0.99, p=0.047), and overall RFS was 14 vs 9 mo (HR 0.63, p=0.03). Median OS was 74 vs 57 mo (HR 0.61, p=0.11) with 5-year OS of 62% vs 47% — a clinically meaningful 15-point absolute gap that didn't quite reach significance in this Phase II powered for h-RFS. Grade 3-4 AEs were higher with HAI (58% vs 32%, p=0.02), but feasibility was similar — 81% of HAI patients completed ≥4 cycles vs 78% IV. The signal supports moving HAI oxaliplatin to a definitive Phase III in selected high-risk CRLM.
Post angle: 🔥 #1 — Adjuvant HAI oxaliplatin doubles hepatic RFS after resection of ≥4 CRLM (Gustave Roussy Phase II): mh-RFS 25 vs 12 mo (HR 0.63, p=0.047), 5-yr OS 62% vs 47%. Trade-off is G3-4 AE 58% vs 32%, but completion rates ~80% in both arms. This is the right population for HAI revival — bring on the Phase III. #GIOnc #CRC #CRLM #PrecisionMedicine
#2
Source: Cancer Discovery (Mechanism) | Authors: Chen, Ren, Zhang, et al. — First Affiliated Hospital of Sun Yat-sen University | Published: May 19, 2026
Score: 7/20 — Cancer Discovery (treated as Tier-2 translational, base 6) + biomarker-guided/precision target identification with in vivo therapeutic validation (+1) = 7. Defines BHLHE40 as a druggable neutrophil-intrinsic vulnerability in CRLM — a novel mechanism in a tumor type where the neutrophil compartment has been a stubborn black box.
Single-cell RNA-seq plus spatial-enhanced-resolution-omics-sequencing (Stereo-seq) of tumor-associated neutrophils (TANs) in CRLM identifies a terminally differentiated pro-tumor subset (TAN1) enriched in liver metastases and tied to poor prognosis. TAN1 displays a glycolysis signature, a senescent phenotype, and arises from other TAN subsets through BHLHE40 upregulation driven by glucose deprivation in the metastatic microenvironment. Functionally, TAN1 promotes angiogenesis via VEGFA and recruits immunosuppressive macrophages through CCL3L1-CCR1 signaling, fostering an angiogenic immunosuppressive niche. Neutrophil-specific Bhlhe40 knockout in mice significantly promotes anti-tumor immunity and suppresses tumor growth — turning this into a tractable target for combination with anti-VEGF or checkpoint blockade in MSS CRLM, the population that has failed every IO strategy to date.
Post angle: 📊 #2 — Why MSS CRC liver metastases stay immunologically cold: glucose-starved senescent-like neutrophils flip on BHLHE40 → angiogenic + immunosuppressive niche (VEGFA + CCL3L1-CCR1). Neutrophil-specific Bhlhe40 KO restores anti-tumor immunity in vivo. Beautiful single-cell + Stereo-seq + mouse genetics combo. New druggable target. #GIOnc #CRC #LiverMets
#3
Source: Clinical Cancer Research (Biomarker) | Authors: Yeh, Yu, Lee, et al. — National Health Research Institutes Taiwan + National Cheng Kung University Hospital | Published: May 21, 2026
Score: 7/20 — Clinical Cancer Research base (6) + biomarker-guided precision (+1) = 7. Actionable today — every PDAC clinic uses CA19-9, but ~10% of patients are FUT3-null nonproducers whose 'low' values mislead. A dual-threshold model implementable without genotyping.
Germline whole-exome sequencing of 615 PDAC patients defined FUT2/FUT3 genotypes; ~10% were FUT3-null (Lewis-negative) CA19-9 nonproducers with median baseline CA19-9 of 2.4 U/mL versus 496 in FUT3-intact patients. Critically, their median OS (13.5 mo) was equivalent to FUT3-intact patients with CA19-9 >200 (12.9 mo) — meaning these 'low' CA19-9 patients are actually high-risk. ROC analysis identified 7 U/mL as the optimal CA19-9 cutoff to flag FUT3-null status (PPV 95.1%, accuracy 87.9%). The proposed practical dual-threshold model (≤7 / >7-37 / >37-200 / >200 U/mL) — usable without genotyping — separates four prognostic groups (mOS 13.5 / 23.2 / 22.0 / 12.8 mo, p<0.001), validated in n=308. Should reshape how we read 'reassuring' low CA19-9 in newly diagnosed PDAC.
Post angle: 🎯 #3 — Low CA19-9 in PDAC ≠ good prognosis. ~10% of patients are Lewis-negative FUT3-null nonproducers whose mOS (13.5 mo) matches CA19-9 >200 patients. New cutoff: CA19-9 ≤7 U/mL = high-risk surrogate (PPV 95%). Practical 4-group dual-threshold model (≤7 / >7-37 / >37-200 / >200) without needing genotyping. #GIOnc #PDAC #PancreaticCancer
#4
Source: Gut (Mechanism / Prevention) | Authors: Yang, Dungan, Madhu, et al. — Cadwell/Katona labs, Penn King Center for Lynch Syndrome | Published: May 19, 2026
Score: 7/20 — Gut journal (treated as Tier-2, base 6) + biomarker-guided / risk stratification (+1) = 7. Important for understanding the variable penetrance of Lynch syndrome — and a candidate prevention biomarker.
Multimodal single-cell profiling (Expanded CITE-seq) of tumor-free colonic mucosa from Lynch syndrome (LS) carriers with and without a prior CRC history vs general-population controls. LS mucosa showed widespread remodeling — expanded epithelial stem/progenitor cells, loss of fibroblast populations, and altered lymphocyte subsets. Crucially, LS carriers WITHOUT a CRC history were enriched for cytotoxic mucosal-associated invariant T (MAIT) cells, while those WITH a CRC history showed clonally expanded, terminally exhausted CD8 T cells. MAIT enrichment correlated with riboflavin metabolite production by epithelial progenitors, and the protective role was validated in a murine model of CRC. The data argue MAIT cells are an active arm of immune surveillance keeping LS carriers cancer-free — a candidate biomarker for individualized colonoscopy intervals and a target for chemoprevention strategies that boost MAIT activity (e.g., synthetic MR1 ligands).
Post angle: 🌍 #4 — Why some Lynch syndrome carriers stay cancer-free for decades while others get CRC: MAIT cells. Single-cell atlas of LS colonic mucosa shows MAIT enrichment in cancer-free carriers vs exhausted CD8 in those with CRC history. Validated in mouse CRC model. Biomarker for risk stratification + chemoprevention target. #GIOnc #CRC #LynchSyndrome #Prevention
#5
Source: BMC Cancer (Real-world Validation) | Authors: Virdee, Birks, Holt, et al. — Oxford Nuffield Department of Primary Care + University of Birmingham + Exeter Medical School | Published: May 14, 2026
Score: 7/20 — BMC Cancer base (5) + retrospective real-world validation cohort (+1) + biomarker-guided risk stratification (+1) = 7. Externally validates a primary-care-ready CRC risk model on 6.5 million UK patients — directly deployable in GP electronic-records workflows.
External validation of sex-specific dynamic risk prediction models using longitudinal trends in the full blood count (haemoglobin, mean cell volume, platelets) to predict 2-year CRC risk in primary care. Validation cohort included 2,956,977 males and 3,561,349 females ≥40 years old without prior CRC; 12,578 (0.4%) and 11,939 (0.3%) respectively were diagnosed with CRC. C-statistic was 0.73 (95% CI 0.72-0.73) for males and 0.74 (0.74-0.75) for females, with good calibration (slopes 0.92 and 0.95). Performance was stable across the number of repeat tests, with predictive accuracy peaking when the longitudinal trend window was 2.5-3.5 years. Calibration under-predicted risk in patients ≥70 years and in higher-deprivation groups — a fixable miscalibration. The simple inputs (a CBC repeated over years) make this immediately deployable in any EHR-based primary care system, potentially shifting CRC diagnoses earlier without adding any new tests.
Post angle: 🧪 #5 — BLOODTRACC (BMC Cancer, Oxford, n=6.5M UK primary care) — dynamic CBC-trend prediction model for 2-yr CRC risk: c-stat 0.73 (male) / 0.74 (female), calibration good. Inputs are just hemoglobin, MCV, platelets over the prior ~3 years — already in every EHR. Zero new tests required. Under-predicts in >70s + high-deprivation — fixable. Could move CRC diagnoses earlier at scale. #GIOnc #CRC #EarlyDetection
Additional Papers of Interest
-
Gut RCT — Urgent ERCP <24 h was NOT superior to early ERCP within 24-48 h for 30-day mortality (3.95% vs 6.58%, HR 0.70, p=0.47) and caused more procedure-related adverse events (17.1% vs 9.2%, RR 2.03) — relevant for biliary tract cancer patients with cholangitis.
-
Gut — A definitive null. Gut microbiome composition differences between people with and without neoplasia were statistically significant but minimally (R²=0.0008); microbiome + clinical features AUC 0.64 vs 0.58 clinical alone. The 'microbiome-as-CRC-screening-biomarker' story takes another hit.
-
British Journal of Cancer — DL model on H&E slides of primary GC predicts LNM status AND identifies occult tumor cells in pN0 patients; pN0 + OTC+ had significantly worse survival (p=0.003). Pathology-AI as an independent prognostic biomarker that doesn't need extra tissue.
-
Gastric Cancer (journal) — Multidisciplinary modified-Delphi consensus. Convergence on early-GC endoscopic resection and anti-HER2/anti-VEGF/IO in metastatic disease, but persistent disagreement on extended ESD criteria, perioperative chemo vs CRT for locally advanced disease, oligometastatic definitions, and intraperitoneal chemo indications.
-
BMC Cancer — Combined Gd-EOB-DTPA MRI model AUC 0.819 (95% CI 0.719-0.920) for high vs low Ki-67 in HCC. Habitat analysis alone AUC 0.814 — the habitat (intra-tumor heterogeneity) signal does most of the work. Non-invasive proliferation index estimation pre-resection or pre-transplant.
-
ASCO — Regular abstracts published today at 5pm ET (May 21). Headline GI plenary: RASolute 302 detailed prespecified analyses for daraxonrasib in 2L mPDAC (mOS 13.2 vs 6.7 mo, HR 0.40 — topline reported May 6) ahead of the May 31 Plenary. Watch also SWOG S2107 (encorafenib + cetuximab ± nivolumab in MSS BRAF V600E mCRC), SWOG S2001 (olaparib + pembro maintenance in gBRCA PDAC), and QLS31905 1L in PDAC + gastric.
Back to all digests