GI Oncology Daily Digest

Pancreatic ductal adenocarcinoma develops from precursor lesions over a decade or more, creating a theoretical window to intercept it before invasive cancer forms — but current imaging surveillance catches only a minority of precursors, and no immune-based interception strategy had been tested prospectively in people at hereditary risk. This phase I, first-in-human study (NCT05013216) treated 20 individuals with hereditary PDAC predisposition and a radiographic pancreatic abnormality with mKRAS-VAX, a peptide vaccine targeting the six most common KRAS mutations that drive most pancreatic cancers and their precursor lesions. The vaccine was well tolerated, with only grade 1-2 adverse events, and elicited a mutant-KRAS-specific T-cell response in 18 of 20 participants (90%). Longitudinal T-cell-receptor sequencing showed the vaccine-induced clonotypes persisted for up to two years, and over a median follow-up of 16.5 months none of the participants developed pancreatic cancer. The senior authorship — Elizabeth Jaffee's group at Johns Hopkins with MD Anderson — signals how seriously the interception field is taking mutant KRAS as a shared neoantigen. The honest read: this is a small, single-arm safety and immunogenicity study without a control group, so the absence of cancers cannot yet be attributed to the vaccine. But demonstrating durable, on-target T-cell immunity in high-risk carriers is a genuine milestone that justifies advancing mutant-KRAS-targeted vaccination into larger, controlled interception trials.

CAR-T therapy has transformed hematologic malignancies but has repeatedly stalled in solid gastrointestinal tumors, so a genuine efficacy signal in refractory colorectal cancer is notable — as is the safety cost that came with it. This phase I trial tested a non-armored CAR-T targeting guanylyl cyclase C (GUCY2C/GCC), a receptor highly restricted to intestinal epithelium and colorectal cancer, in 24 patients with heavily pretreated metastatic colorectal cancer across four dose levels. The objective response rate was 33% and the disease control rate 63% — meaningful activity in a population that had exhausted standard options — but median progression-free survival was only 57 days and median overall survival 190 days, indicating responses were generally not durable. The toxicity was substantial: roughly three-quarters of patients had grade 3 or higher cytokine-release-related hematologic toxicity, and there were two treatment-related deaths, one from an immune-effector-cell HLH-like syndrome and one from an intestinal fistula. The result is best read as proof-of-concept that GCC is a targetable antigen capable of driving real responses in colorectal cancer, tempered by a safety profile — including two deaths among 24 patients — that will need to be engineered down before this approach can move beyond early-phase testing. It is an important signal for the solid-tumor CAR-T field, not a ready-for-practice therapy.

Guidelines recommend twice-yearly ultrasound, often paired with alpha-fetoprotein (AFP), to screen patients with cirrhosis for hepatocellular carcinoma — but the real-world sensitivity of that strategy for the early-stage tumors that are actually curable has been contested. This systematic review and network meta-analysis pooled 170 studies and 62,643 participants, comparing 97 index tests for HCC detection in cirrhosis. The headline finding challenges standard practice: for very-early-stage HCC, ultrasound sensitivity was just 0.34 — no better than AFP alone at 0.39 — meaning the workhorse surveillance test misses roughly two-thirds of the earliest tumors. Cross-sectional imaging performed far better and, importantly, was the least degraded by tumor stage: contrast-enhanced MRI ranged from about 0.70 sensitivity for the earliest lesions up to 0.90 for more advanced disease, with contrast-enhanced CT close behind. The analysis also found that none of the emerging genomic or molecular biomarkers convincingly outperformed conventional blood-marker combinations. For practicing hepatologists and oncologists, this is a substantial evidence base arguing that ultrasound-based surveillance underperforms precisely where detection matters most, and it strengthens the rationale for abbreviated MRI or other cross-sectional strategies in high-risk cirrhotic patients — though cost, access, and the absence of hard outcome (mortality) data from these accuracy studies keep it short of an immediate guideline rewrite.

Why some colorectal tumors resist immune-checkpoint blockade remains incompletely understood, and this mechanistic study identifies an unexpected culprit: the cytokine IL-26, produced by tumor-specific type-17 (Th17) T cells, acting inside cancer cells rather than at their surface. Across patient samples and models, the authors found that IL-26-expressing T cells selectively accumulate in colorectal tumors resistant to checkpoint-blockade therapy. Mechanistically, IL-26 behaves as a noncanonical cytokine — it translocates into the nucleus of tumor cells, binds the transcription factor STAT1, and forms transcriptional complexes with NF-kappaB and AP-1. That reprogramming installs an active-chromatin state marked by BRD4 and H3K27ac enrichment, driving upregulation of CXCL chemokines that recruit neutrophils and suppress CD8+ T-cell responses — a self-reinforcing loop of immune evasion. The translational appeal is that it nominates a specific, potentially druggable axis (IL-26 and the BRD4-dependent chromatin program it switches on) that could, in principle, be combined with checkpoint inhibitors to resensitize resistant colorectal cancers. This is preclinical mechanism rather than clinical data, but it is a well-worked-out and testable hypothesis for a recurrent problem in GI immuno-oncology.