Top 5 Papers
#1
Source: FDA — Oncology Center of Excellence | Authors: FDA Center for Drug Evaluation and Research / Revolution Medicines | Published: August 26, 2026
Score: 17/20 — FDA regulatory action (base 8) + Phase III RCT evidence base, RASolute 302 (+3) + new standard of care, first RAS-targeted therapy ever approved in pancreatic cancer (+3) + colleague engagement (+3, capped: six GI oncology KOLs posted the approval within 48h). Note: the RASolute 302 efficacy data led the June 2, 2026 edition from the ASCO plenary and were folded into the August 20 guideline update — the approval itself is the new event here, not the numbers.
On August 26 the FDA approved daraxonrasib for adults with metastatic pancreatic adenocarcinoma who have received at least one prior systemic therapy, or who are not candidates for multiagent chemotherapy. This is the first approval of a RAS(ON) multi-selective inhibitor in any solid tumor, and it lands in the disease where RAS has been the defining, undruggable lesion for four decades. Efficacy came from RASolute 302 (NCT06625320), which randomized 500 patients 1:1 to daraxonrasib or physician's-choice chemotherapy. In the overall population median OS was 13.2 months (95% CI 10.0–NE) versus 6.7 months (95% CI 5.8–8.0), HR 0.40 (95% CI 0.30–0.53), p<0.0001; median PFS was 7.2 versus 3.6 months, HR 0.49 (95% CI 0.38–0.64), p<0.0001; and ORR was 30% versus 11%, p<0.0001. Benefit held in both the RAS G12-mutant population and the overall population. The dose is 300 mg orally once daily. Labeled warnings include dermatologic and soft-tissue toxicity, stomatitis, diarrhea, GI perforation, and ILD/pneumonitis — a real management burden for a drug patients may stay on for many months. Reviewed under Project Orbis with Health Canada, with EMA and PMDA as observers, via the Real-Time Oncology Review pilot.
Post angle: A hazard ratio of 0.40 in second-line metastatic pancreatic cancer is a number this field has never had. Frame it as the end of 'RAS is undruggable' and the beginning of the harder question: what do we combine it with? #PDAC #PancreaticCancer #GIOnc #KRAS #PrecisionMedicine
#2
Source: Clinical Cancer Research | Authors: Shah MA, Hidalgo M, et al. (Weill Cornell / City of Hope, multicenter) | Published: August 26, 2026
Score: 10/20 — Clinical Cancer Research (base 6) + Phase II trial (+2) + biomarker-guided, prospectively stratified by MMR status (+1) + colleague engagement (+1: one GI oncology KOL posted it within 48h). Held below the lead despite an eye-catching response rate because n=24 is small and pathologic response is a surrogate, not an outcome.
Effective immunotherapy for mismatch-repair-proficient colorectal cancer does not exist — pMMR tumors are the immunologically cold majority, and checkpoint blockade has failed in them repeatedly. NEST tested the Fc-enhanced anti-CTLA-4 antibody botensilimab plus the PD-1 inhibitor balstilimab before surgery in 24 patients with resectable non-metastatic colon cancer (26 tumors: 22 pMMR, 4 dMMR). The major pathologic response rate was 41% (95% CI 21–64%) in pMMR tumors and 100% (95% CI 40–100%) in dMMR. Critically for a preoperative regimen, treatment did not delay planned surgery in any patient. Exploratory tumor-microenvironment analysis showed increased CD8+ T-cell density and reduced FOXP3+ regulatory T cells in responders, consistent with the Fc-enhanced CTLA-4 mechanism actually remodeling a cold tumor rather than simply releasing a pre-existing response. The caveats are real and should be stated plainly: 24 patients, a single-arm design, and a pathologic endpoint that has never been formally validated as a surrogate for survival in colon cancer. But a 41% major pathologic response in pMMR disease is well outside what PD-1 blockade alone has ever produced here, and it justifies a randomized trial.
Post angle: pMMR colon cancer is where immunotherapy goes to die. A 41% major pathologic response with Fc-enhanced CTLA-4 blockade is the first signal worth randomizing. Small n, surrogate endpoint — but the right kind of surprising. #CRC #ColorectalCancer #GIOnc #Immunotherapy
#3
Source: Nature Communications | Authors: Multicenter preclinical study (MD Anderson-led) | Published: August 26, 2026
Score: 7/20 — Nature Communications (base 6) + biomarker-guided / mechanism-directed combination selection (+1). No study-type bonus — this is preclinical, not a clinical trial. Included in the top tier on editorial grounds: it lands in the same week as the daraxonrasib approval and answers the immediate next question that approval raises.
Daraxonrasib was approved this week; the obvious next question is what to combine it with, and this paper makes an unusually specific argument. Using MRTX1133 or daraxonrasib in mouse PDAC models, the authors show that KRAS inhibition recruits a broad T-cell infiltrate — effector, exhausted, and regulatory — into a microenvironment that is normally immune-excluded. That influx opens a therapeutic window, but not a general one: the benefit was specific to anti-CTLA-4 and was not reproduced with anti-PD-1, anti-TIM-3, anti-LAG-3, anti-VISTA, or 4-1BB agonism. Mechanistically, anti-CTLA-4 transcriptionally reprogrammed effector Tregs toward a naive phenotype, reversed CD8+ T-cell exhaustion, and recruited functional tertiary lymphoid structures. Single-cell ATAC sequencing traced the Treg reprogramming to epigenetic downregulation of AP-1 family transcription factors at the IL-35 promoter. This is mouse data and should be read as such. But it is a falsifiable, mechanism-anchored prediction about which checkpoint to pair with a newly approved drug — and it converges with the NEST result above, where Fc-enhanced CTLA-4 blockade is the agent moving the needle in another GI tumor that PD-1 blockade cannot reach.
Post angle: Two independent papers this week point at the same axis: CTLA-4, not PD-1, is the checkpoint that matters in cold GI tumors. One is mouse PDAC after KRAS inhibition; the other is a colon cancer trial. Worth noticing. #PDAC #KRAS #GIOnc #Immunotherapy
Additional Papers of Interest
-
Lancet Oncology — the peer-reviewed publication of a trial whose PFS result (13.0 vs 9.8 mo, HR 0.70, p=0.0007) led our June 2, 2026 edition from ASCO LBA4000. New here: at the second data cutoff, overall survival was 39.5 vs 34.7 months, HR 0.84 (95% CI 0.65-1.09), p=0.18 — not statistically significant. The PFS win did not convert to an OS win on this analysis.
-
Annals of Surgical Oncology — prospective cohort (NCT04616131), 113 patients and 219 samples in localized PDAC. Co-occurrence of KRAS and TP53 ctDNA at diagnosis independently predicted shorter OS, adjusted HR 5.66 (95% CI 2.43-13.18), p<0.001; detectable ctDNA at diagnosis predicted lower odds of reaching resection, aOR 0.38, p=0.049. KRAS ctDNA detection fell from 19% at diagnosis to 4.6% after neoadjuvant chemotherapy, p=0.010.
-
Cancer Medicine — retrospective series of 31 patients with unresectable gallbladder cancer, a setting with almost no prospective data. ORR 48.4%, DCR 90.3%, median PFS 8.4 months (95% CI 5.0-11.8), median OS 20.9 months (95% CI 16.8-24.9); 7 of 11 locally advanced patients converted to resectable disease and underwent surgery. Hypothesis-generating given the sample size, but conversion rates like this deserve a prospective test.
-
Nature Communications — Norwegian population-based study using leftover material from 1,034 FIT-positive screening participants for shotgun metagenomic profiling. Adding microbial profiles to quantitative FIT values improved detection of premalignant lesions beyond optimizing the FIT threshold alone, even after accounting for established risk factors — but FIT alone remained the better discriminator for CRC itself. Fusobacterium nucleatum and Peptostreptococcus stomatis were confirmed as enriched in CRC.
-
Gastroenterology — addresses the diagnostic gap in patients with adenomatous polyposis and no identified germline cause, examining both APC somatic mosaicism and colibactin-associated mutational damage from genotoxic E. coli as explanatory mechanisms. Abstract not yet posted to PubMed at time of writing.
-
Gut — prospective multi-kingdom microbiome profiling in Lynch syndrome carriers, proposed as a model for tracking colorectal cancer evolution in a genetically defined high-risk population under surveillance. Abstract not yet posted to PubMed at time of writing.
Back to all digests