GI Oncology Daily Digest

First-line treatment of HER2-positive advanced gastric/GEJ adenocarcinoma now centers on pembrolizumab added to trastuzumab plus chemotherapy (the KEYNOTE-811 regimen), while the bispecific anti-HER2 antibody zanidatamab improved outcomes in HERIZON-GEA-01 but has never been compared head-to-head with that standard. This analysis tries to bridge the gap by reconstructing individual patient data from published Kaplan-Meier curves across nine first-line phase 2/3 trials — trastuzumab + chemotherapy (n=1,466), pembrolizumab + trastuzumab + chemotherapy (n=430), zanidatamab + chemotherapy (n=350), and zanidatamab + tislelizumab + chemotherapy (n=335) — using Cox frailty models with the pembrolizumab regimen as reference. Median PFS was 12.5 months with both zanidatamab regimens versus 10.0 months with the pembrolizumab standard, and the zanidatamab-tislelizumab triplet significantly reduced progression hazard (HR 0.78, 95% CI 0.62-0.99; p=0.042). Median OS was 27.0 and 25.3 months in the zanidatamab arms versus 20.3 months with pembrolizumab, with both zanidatamab arms showing significantly longer restricted mean survival at 36 months. The critical caveat is methodological: this is a reconstructed cross-trial comparison, not a randomized head-to-head, so cross-trial confounding, differing enrollment eras, and population differences could all inflate the apparent advantage — the authors themselves conclude that direct randomized comparisons are needed. The honest read is that this is a provocative, hypothesis-generating signal that zanidatamab-based regimens may raise the bar in 1L HER2+ gastric cancer, and a reason to watch the randomized HERIZON-GEA-01 readout closely rather than a practice-changing result on its own.

In a regulatory move that quietly reshapes upper-GI biomarker testing, the FDA approved Agilent's PD-L1 IHC 28-8 pharmDx assay as a companion diagnostic to identify patients eligible for nivolumab (Opdivo, and the subcutaneous Opdivo Qvantig) across four gastrointestinal tumor types: esophageal squamous cell carcinoma (ESCC), gastric, gastroesophageal junction (GEJ), and esophageal adenocarcinoma. The approval uses a combined positive score (CPS) cutoff of 1 or greater and rests on the pivotal CheckMate-648 trial in ESCC and CheckMate-649 in gastric/GEJ/esophageal adenocarcinoma, both of which established overall-survival benefit for nivolumab-based therapy in PD-L1-enriched populations. Practically, the value here is standardization: it validates a specific, reproducible assay and scoring cutoff for nivolumab eligibility across the upper-GI landscape, harmonizing the biomarker testing that oncologists and pathologists rely on to select immunotherapy candidates. This is a diagnostics approval rather than a new therapeutic, and it does not by itself change which drug patients receive — but for a tumor group where PD-L1 status increasingly gates first-line immunotherapy decisions, a formally approved companion diagnostic tightens the link between testing and treatment and reduces assay-to-assay variability in real-world practice.

Clavien-Dindo grade II-or-higher complications occur in roughly one in five patients after curative gastrectomy for gastric cancer, and they independently shorten survival by compromising delivery of adjuvant therapy — yet current nutritional, inflammatory, and surgical-risk scores flag only a minority of the patients who go on to have them. DeepComp is a multimodal deep-learning framework that integrates clinical variables with foundation-model imaging features from the tumor, a 5-mm peritumoral region, and L3 body-composition compartments in a dual-task architecture that jointly predicts complications and overall survival. Trained and validated on 5,237 patients from 11 Chinese centers and further tested against prospectively collected data from six registered trials spanning chemotherapy, chemoradiation, and immunochemotherapy neoadjuvant settings, the model reached an AUC of 0.888 in internal validation and 0.824-0.869 across nine external cohorts, outperforming the best clinical baseline by 15.3 percentage points and beating all nine established clinical scores. In a reader study, DeepComp assistance raised the mean sensitivity of ten surgeons from 47.1% to 87.9%. Through target-trial emulation, DeepComp-guided prophylactic ICU monitoring, preoperative nutritional support with delayed surgery, and minimally invasive triage produced absolute complication-risk reductions of 5.9%, 20.6%, and 11.7% respectively. The model also independently predicted survival (adjusted HR 3.08 per standard deviation; pooled C-index 0.766), with 5-year survival ranging from 97.5% in the lowest-risk quintile to 2.4% in the highest. As a decision-support tool this is a strong, well-validated signal, though prospective clinical trials will be needed before DeepComp-guided perioperative management changes routine practice.

Whether to routinely dissect regional lymph nodes during resection of primary liver tumors has long been contested, because the answer depends heavily on tumor type and the trade-off between staging value and operative morbidity. This systematic review and meta-analysis pooled 40 studies and 5,872 patients undergoing resection for hepatocellular carcinoma (HCC), intrahepatic cholangiocarcinoma (ICC), or perihilar cholangiocarcinoma (PCC), and the picture that emerges is cleanly stratified by histology. The incidence of lymph node metastasis was low in HCC (5%) but high in both biliary cancers (37% in ICC and 39% in PCC), and the prognostic weight of nodal involvement tracked the same way: lymph node metastasis was associated with substantially worse overall survival in ICC (HR 2.46) and PCC (HR 2.01), but not in HCC. The authors' bottom line is a practical staging rule — routine lymph node dissection should be performed during resection of intrahepatic and perihilar cholangiocarcinoma, where nodal disease is common and prognostically decisive, but is not warranted for HCC, where node positivity is rare and does not independently drive survival. For hepatobiliary surgeons weighing added operative time and morbidity against staging yield, this is a useful, appropriately histology-specific synthesis rather than a one-size-fits-all recommendation.