Top 5 Papers
#1
Source: Annals of Oncology | Authors: Kopetz S, Elez E, et al. (multicenter) | Published: May 31, 2026
Score: 13/20 — Ann Oncol base (7) + Phase 3 RCT (+3) + new standard of care (+3) = 13. Establishes EC + FOLFIRI as the second 1L BRAF-targeted backbone for BRAF V600E mCRC, complementing the EC + mFOLFOX6 regimen from earlier BREAKWATER cohorts. Irinotecan-backbone patients now have a targeted option.
BREAKWATER Cohort 3 randomized 147 previously untreated BRAF V600E mCRC patients to encorafenib + cetuximab + FOLFIRI (n=73) vs FOLFIRI ± bevacizumab (n=74). The targeted arm delivered ORR 64.4% vs 39.2% (OR 2.76, p=0.0011), mPFS 15.2 vs 8.3 months (HR 0.44, p=0.0002), and OS HR 0.56 (not reached vs 20.3 mo). With Cohort 1/2 already establishing EC + mFOLFOX6 as 1L standard, Cohort 3 closes the loop — now any BRAF V600E mCRC patient has a targeted 1L regimen regardless of chemotherapy backbone preference. Irinotecan-tolerant patients no longer have to choose between targeting and backbone fit.
Post angle: BREAKWATER Cohort 3: BRAF V600E mCRC patients now get EC + irinotecan too — mPFS doubles (15.2 vs 8.3 mo, HR 0.44), ORR 64% vs 39%, OS HR 0.56. Whichever 1L backbone you pick, you can layer targeted therapy on top. The BRAF-mut 1L decision is no longer a tradeoff. #GIOnc #CRC #BRAF #PrecisionMedicine
#2
Source: Journal of Clinical Oncology | Authors: Qi C, Shen L, et al. (Peking University Cancer Hospital + Beijing Imunopharm) | Published: June 4, 2026
Score: 10/20 — JCO base (8) + survival/PFS benefit in a graveyard population (+2) = 10. Phase 1 trials get no study-type bonus per CLAUDE.md, but mPFS 7 mo at DL3 in 3L+ mCRC — and 9 mo in GUCY2C-high — is a real signal in a tumor type where every prior CAR-T (CEA, GUCY2C, HER2, EpCAM) has failed.
Single-center Phase 1 of an autologous CAR-T targeting GUCY2C, an intestinal-epithelial antigen stably expressed across all CRC stages. Twenty 3L+ mCRC patients received CAR-T after lymphodepletion across 4 dose levels. DLT was not reached; only 1 patient (5%) had Grade 3 CRS/neurotoxicity; Grade 3 diarrhea in 55% — the on-target/off-tumor signature for a gut-expressed antigen. ORR 26.3% overall, 40% at DL3, and 50% with mPFS 9 mo in GUCY2C-medium/high tumors. Solid-tumor CAR-T has been a graveyard, and CRC the deepest grave. A safety-acceptable, dose-responsive, biomarker-enriched signal opens a credible path to Phase 2 expansion — and validates GUCY2C as the rare antigen broad enough to matter and selective enough to spare normal tissue.
Post angle: GUCY2C CAR-T in mCRC: Phase 1 ORR 40% at DL3, mPFS 7 mo, 9 mo in biomarker-high subset. No Grade 4/5 CRS. Solid-tumor CAR-T finally lands in CRC after a decade of failed attempts. The antigen choice (gut-epithelium-restricted) was the unlock. #GIOnc #CRC #CARt #CellTherapy
#3
Source: JAMA Oncology | Authors: Heslin RT, Kim AC, et al. (UT Southwestern) | Published: June 4, 2026
Score: 8/20 — JAMA Oncol base (7) + retrospective/real-world (+1) = 8. Population-level disparities work in a Tier-1 journal. Identifies a modifiable, system-level lever — interpreter access — that maps directly to a measurable survival outcome in the fastest-growing CRC population.
Population-based cross-sectional analysis of 112,672 Texas Cancer Registry CRC patients (2004–2019) compared early-onset CRC (EOCRC, <50; n=12,079, 11%) vs average-onset CRC. EOCRC patients had better baseline OS than AOCRC (HR 0.56) — younger, fitter, less comorbidity — but treatment delays >6 weeks erased much of that advantage (delay HR 1.35 for OS, p<0.001). The actionable finding: after controlling for clinical and demographic factors, language barriers independently increased the odds of treatment delay in EOCRC (OR 1.45). EOCRC patients were also disproportionately Hispanic (28.1% vs 19.9%). This isn't a paper about biology — it's a paper about access. Interpreter availability and Spanish-language patient navigation are the immediate operational fix.
Post angle: EOCRC is rising and the biology is no worse than older CRC. The problem is access. JAMA Oncology n=112,672: language barriers nearly halve OS via treatment delays (delay HR 1.35, language OR 1.45). Hispanic patients overrepresented in EOCRC. Interpreter capacity = survival in 2026. #GIOnc #CRC #HealthEquity
#4
Source: Clinical Cancer Research | Authors: Braganca Xavier C, Rodon J, et al. (MD Anderson + BostonGene) | Published: June 4, 2026
Score: 8/20 — Clin Cancer Res base (6) + retrospective genomic dataset (+1) + biomarker-defined subgroup (+1) = 8. With PRMT5/MAT2A-MTAP-synthetic-lethal agents in active Phase 1/2 and KRAS G12C/G12D/multi-RAS inhibitors moving toward 1L PDAC, a genomically defined co-mutated subgroup is exactly the trial-design substrate the field needs.
Combined analysis of 4,009 BostonGene Tumor Portrait PDAC samples plus a 2,181-sample meta-cohort. 24% of PDAC harbors homozygous MTAP deletion. 18.9% of all PDAC carries the combined KRAS-mutant + MTAP-loss genotype. This subgroup is fibrotic, immune-excluded, and has shorter OS than KRAS-mut/MTAP-intact tumors. Within co-mutated tumors, KRAS G12D and G12V dominate; G12R skews slightly more immune-enriched. Two clinical implications: this is the natural target population for MTA-cooperative PRMT5 inhibitors (already in trials), and the immune-excluded phenotype argues against IO monotherapy and toward stroma-/myeloid-modulating combinations. Worth pulling MTAP status on every PDAC NGS report going forward.
Post angle: PDAC NGS reports should now include MTAP. 18.9% of PDAC = KRAS-mut + MTAP-loss = immune-excluded + worse OS. The natural trial population for PRMT5-MTA inhibitors and the wrong population for IO monotherapy. (n=6,190, MD Anderson/BostonGene) #GIOnc #PDAC #PrecisionMedicine #KRAS
#5
Source: Cancer Discovery | Authors: Liu W, Yu J, et al. (CUHK + Sun Yat-sen + Renji + Zhejiang) | Published: June 4, 2026
Score: 7/20 — Cancer Discov base (6) + biomarker/novel target (+1) = 7. Closes a 5-year gap between the clinical observation (gut dysbiosis correlates with HCC) and a mechanistic, druggable axis (B. fragilis Enolase ↔ hepatocyte Vimentin).
Matched gut + intrahepatic microbiome profiling of HCC patients vs healthy controls showed striking gut-liver microbiome convergence in HCC, with a pathobiont-centered network suggesting translocation. HCC stool transplanted into germ-free mice impaired gut-barrier integrity and increased hepatic bacterial load. Intrahepatic pathobionts in tumor regions correlated with host cytokine and oncogenic-pathway activation. Mechanistically, B. fragilis surface Enolase binds hepatocyte Vimentin and triggers downstream oncogenic signaling; oral B. fragilis administration was sufficient to disrupt gut barrier, drive translocation, and accelerate HCC in mice. The translational hook: a tractable, targetable microbe-host interaction in HCC, distinct from the diet/MASLD inflammation paradigm. Expect a wave of B. fragilis-targeted modulators (vaccines, narrow-spectrum antibiotics, Enolase blockers) entering preclinical HCC pipelines.
Post angle: HCC pathogenesis just got a new axis: Bacteroides fragilis translocates across a leaky gut barrier, lands in the liver, and its surface Enolase activates hepatocyte Vimentin to drive oncogenic signaling. Sufficient to cause HCC in mice. A tractable target for prevention in cirrhotic risk pools. #GIOnc #HCC #Microbiome
Additional Papers of Interest
-
NEJM — Full Phase 3 manuscript of RASolute 302 (covered previously at the May 26 topline and June 2 ASCO Plenary). Final dataset: G12 mOS 13.2 vs 6.6 mo (HR 0.40, p<0.001), mPFS 7.3 vs 3.5 mo (HR 0.45). Grade ≥3 AEs 61.8% vs 69.6%; discontinuation 1.2% vs 11.2% — favorable tolerability vs chemotherapy. The peer-reviewed citation of record.
-
Critical Reviews in Oncology/Hematology — Sintilimab + IBI305 tops OS (SUCRA 0.91); finotonlimab + bevacizumab tops PFS (SUCRA 0.86); atezolizumab + bevacizumab retains the strongest integrated benefit-risk profile via Bayesian TOPSIS. A useful side-by-side for community oncologists choosing between expanding 1L HCC options.
-
Clinical and Translational Gastroenterology — Stool DNA in FIT-positive patients reached AUC 0.94–0.98 for CRC; Q-Cancer hit AUC 0.89–0.92 for PDAC; the GALADUS model achieved AUC 0.94 for HCC in cirrhotic patients; an AI model reached AUC 0.87 for esophageal cancer. The systematic backbone behind the diagnostics that have crowded clinic conversations over the past 18 months.
Back to all digests