In this phase 2 trial (NCT04744649), 17 patients with locally advanced Epstein-Barr virus-associated gastric or gastroesophageal junction adenocarcinoma (cT2-4aN1-3M0) received four cycles neoadjuvant toripalimab plus capecitabine/oxaliplatin. The primary endpoint was major pathological response; secondary endpoints included pathological complete response, R0 resection, adverse events, event-free survival, overall survival, and tumor microenvironment. Paired pre-/post-treatment tissues were assessed by immunofluorescence. 16 patients underwent curative resection; 1 declined surgery. Major pathological response, pathological complete response, and ypN0 were 37.5% (6/16, 95% CI 0.15-0.65), 25.0% (4/16, 95% CI 0.07-0.52), and 81.3% (13/16, 95% CI 0.54-0.96), respectively. Major pathological response was more frequent in patients with programmed death ligand 1 ≥20 (57.1%, 95%CI 0.18-0.90). Major pathological response was associated with higher pretreatment CD8⁺ T-cell density. 35.3% reported grade 3-4 adverse events. These findings suggest that neoadjuvant immunochemotherapy demonstrated favorable efficacy with a manageable safety profile in Epstein-Barr virus-associated gastric or gastroesophageal junction adenocarcinoma.

BackgroundPostoperative complications (POCs) significantly affect recovery and prognosis in patients with gastric cancer. This retrospective cohort study aimed to develop and validate a nomogram integrating preoperative clinical laboratory indicators and sarcopenia to predict POCs in patients undergoing radical gastrectomy.MethodA retrospective analysis was conducted on 181 patients who underwent radical gastrectomy for gastric cancer between October 2019 and June 2022. Patients were randomly assigned to a training set (n = 126) and a validation set (n = 55). Sarcopenia was defined using the skeletal muscle index (SMI) measured at the third lumbar vertebra on preoperative computed tomography (CT) images. Univariate and multivariate logistic regression analyses were performed to identify risk factors for POCs (Clavien–Dindo grade ≥ II). A predictive nomogram was constructed based on independent risk factors, and model performance was assessed using the area under the receiver operating characteristic curve (AUC), calibration curves, and decision curve analysis (DCA).ResultsInter-observer consistency for SMI measurement was excellent (intraclass correlation coefficient [ICC] = 0.86). Univariate analysis revealed that clinical N stage, carcinoembryonic antigen (CEA) level, sarcopenia, and CT-assessed serosal invasion were significantly associated with POCs (all p < 0.05). The nomogram incorporating these four factors showed good predictive performance, with AUCs of 0.865 (95% confidence interval [CI]: 0.794–0.936) in the training set and 0.826 (95% CI: 0.704–0.948) in the validation set. Calibration curves demonstrated good agreement between predicted and observed probabilities, and DCA confirmed the model’s clinical utility.ConclusionThe nomogram integrating preoperative CEA, sarcopenia, CT-assessed serosal invasion, and clinical N stage effectively predicts the risk of postoperative complications in gastric cancer patients, facilitating preoperative risk stratification and personalized perioperative management.

The information regarding tumor cells, the immune environment and microorganisms can provide insights into the formation and progression of gastric cancer (GC) from various perspectives. However, there remains a need for a diagnostic strategy capable of integrating and analyzing these multiple dimensions to accurately assess the progression of GC. Extracellular vesicles (EVs), which carry multidimensional information, represent a promising non-invasive biomarker for the diagnosis and monitoring of GC. Despite their potential, the development of an efficient method for the enrichment and precise analysis of individual EVs remains a significant challenge. Herein, we present a versatile droplet digital assay, termed the Single-EV MagaDecoder, designed to enable both the enrichment of EVs and the decoding of individual EVs. This assay consisted of two main components: efficient reversible magnetic enrichment of EVs and single EV multidimensional biomarker decoding analysis. To evaluate the clinical utility of the single-EV MagaDecoder for GC diagnosis, screening (n=64) and validation (n=79) GC cohorts were utilized. This assay demonstrated exceptional performance in predicting gastric disease risk, achieving 100% accuracy, and distinguishing between benign and malignant conditions with an area under the curve (AUC) of 0.8875. These results were obtained through the detection of multidimensional signatures of GC-EVs in a validation cohort (n=79). In summary, this study proposes a promising strategy for the precise analysis of multidimensional EV subpopulations at single EV level, thereby advancing the field of precision medicine for GC.

Background The optimal surgical strategy for adenocarcinoma of oesophagogastric junction (AEG) remains debated, particularly regarding lymphadenectomy extent, gastrectomy type and surgical approach, with real-world prospective evidence being scarce. Objective To map lymph node metastasis (LNM) patterns and assess surgical outcomes in a large multicentre cohort of patients with AEG undergoing radical resection. Design The Chinese League of Adenocarcinoma of Esophagogastric Junction (CLAEG) registry, initiated in 2022 across 44 high-volume Chinese centres, prospectively enrolled AEG patients. This analysis included 2044 radical resections, with LNM assessed by station, stratified by Siewert type and neoadjuvant therapy. Surgical outcomes were compared between total versus proximal gastrectomy and laparoscopic versus open resection. Results Most tumours were Siewert type II (64.6%) or III (33.4%). LNM was substantially higher in abdominal than mediastinal stations; category-1 nodes (metastasis, >10%) comprised stations 1, 2, 3, 4, 7, 8a, 9 and 11p. The LNM rates for mediastinal stations were 2.77% (No. 110), 0.71% (No. 111) and 0.68% (No. 112). Patients who received neoadjuvant therapy had lower LNM rates, indicating nodal downstaging. Among those undergoing gastrectomy, patients who underwent total gastrectomy had a lower postoperative complication rate than those who underwent proximal gastrectomy (14.8% vs 21.0%; p=0.001) and achieved more extensive lymphadenectomy. Compared with open surgery, patients who underwent laparoscopic resection experienced faster postoperative recovery without higher complication rates (16.5% vs 17.3%). No perioperative mortality occurred. Conclusion The CLAEG study shows that abdominal lymphadenectomy should be prioritised in AEG, with neoadjuvant therapy, total gastrectomy and laparoscopy associated with favourable short-term outcomes.

BACKGROUND Primary gastric adenosquamous carcinoma (PGASC) is a rare malignancy with a poor prognosis. The current clinical understanding of PGASC has been derived from Western populations and may not accurately reflect the disease characteristics of Eastern populations. Furthermore, the efficacy of standard adjuvant chemotherapy for the treatment of PGASC has not been determined. AIM To explore the clinicopathological features, prognosis, and chemotherapy efficacy of PGASC in a Chinese cohort. METHODS This retrospective analysis of 108 patients with PGASC was conducted from March 2007 to December 2021 at six medical centers in China. The data from these patients was compared with the data from 468 randomly selected patients with pathologically diagnosed gastric adenocarcinoma (GAC) from the same time period at Renji Hospital Affiliated to Shanghai Jiao Tong University School of Medicine. The characteristics, prognosis, and chemotherapy efficacy were investigated by propensity score matching. RESULTS Patients with PGASC had higher rates of tumors located in the cardia (44.4% vs 22.9%), lymphovascular invasion (55.6% vs 35.5%), and perineural invasion (50.0% vs 32.7%) than patients with GAC. Immunohistochemical analysis revealed that the PGASC group had a significantly higher proportion of cells, with at least 10% expressing programmed cell death-ligand than the GAC group (13.2% vs 2.8%; P < 0.001). Patients with PGASC had significantly poorer 5-year overall survival rates compared with patients with GAC (16.6% vs 49.3%, P < 0.001). In the PGASC group overall survival was similar in patients who received chemotherapy and those who did not. CONCLUSION PGASC has different characteristics than GAC, including a more aggressive presentation, a poorer prognosis, and higher programmed cell death-ligand 1 expression.

4089 Background: Perioperative PD-L1 inhibitor plus chemotherapy has shown survival benefit in PD-L1-positive patients with locally advanced GC/EGJC. However, evidence for neoadjuvant PD-1/CTLA-4 bispecific blockade remains limited. This study evaluates neoadjuvant cadonilimab plus CapeOX in patients with clinical stage III GC/EGJC. Methods: This prospective, single-arm phase II study (NCT06310473) enrolled patients (pts) with cT3-4aN1-3M0 disease (AJCC 8th), staged by contrast-enhanced CT and staging laparoscopy with peritoneal cytology to exclude peritoneal metastasis and positive peritoneal cytology (CY1). Pts received cadonilimab (10 mg/kg) plus standard-dose CapeOX every 3 weeks for 3 cycles. Pts without progressive disease on repeat CT and laparoscopy underwent curative-intent surgery, followed by 3-5 cycles of adjuvant CapeOX. The primary endpoint was the pathologic complete response rate (pCR, defined as no residual tumor in stomach and lymph node). The data cutoff was January 15, 2026. Results: From May 2024 to August 2025, 30 pts with GC/EGJC were enrolled (GC, n = 21; EGJC, n = 9); 83.3% (25/30) were cT4a and 73.3% (22/30) were cN2-3. PD-L1 CPS < 1 was observed in 14.8% (4/27) of evaluable pts. All pts completed 3 neoadjuvant cycles. Three pts did not undergo surgery: one refused resection and subsequent antitumor therapy after achieving a cCR to neoadjuvant therapy; one had a PR and continued systemic therapy; and one had PD with liver invasion on repeat laparoscopy and initiated second-line therapy. The other 27 pts underwent resection; one had CY1 on postoperative assessment and was considered PD, with an R0 resection rate of 96.3% (26/27). The pCR and major pathologic response (MPR) rates were 22.2% (6/27) and 37.0% (10/27), respectively. The tumor downstaging rate was 74.1% (20/27), including a ypN0 rate of 55.6% (15/27). Adjuvant therapy was administered in 23/27 (85.2%), with four pts still on therapy. Treatment-related adverse events occurred in 100% (30/30), including grade≥3 events in 20.0% (6/30). Notable events included one perioperative cardiovascular death (G5); one pulmonary embolism (G3) diagnosed 5 weeks postoperatively that resolved with anticoagulation; one patient with concomitant anastomotic leakage (G4), intra-abdominal infection (G4), and immune-related adrenal insufficiency (G2) who recovered after endoscopic covered-stent placement and conservative management; additional G3 events included ascites, neutropenia, and peripheral neuropathy (n = 1 each). No postoperative recurrence was observed among resected pts at data cutoff. Conclusions: Neoadjuvant cadonilimab plus CapeOX appears safe and effective in patients with clinical stage III GC/EGJC, achieving a moderate pCR rate and warranting further exploration of neoadjuvant immunochemotherapy combination strategies. Clinical trial information: NCT06310473 .

Lymph node metastasis (LNM) is the most frequent metastatic pathway in gastric cancer (GC) and a major determinant of poor prognosis, but the underlying molecular mechanisms remain poorly defined. In this study, we investigated the role of extracellular leucine-rich repeat and fibronectin type III domain-containing protein 2 (ELFN2) in LNM and its therapeutic potential in GC. ELFN2 expression was examined in clinical GC tissues and found to be significantly downregulated in cases with LNM, with low ELFN2 levels correlating with unfavorable patient survival. Mechanistic studies demonstrated that ELFN2 interacts directly with the catalytic subunit alpha of protein phosphatase-1 (PP1A), thereby inhibiting YAP dephosphorylation at Ser127. This regulation promotes YAP nuclear export and functional inactivation. Pharmacological inhibition of PP1A abrogated ELFN2-induced YAP inactivation, confirming the requirement of PP1A in this process. To further validate the biological significance of ELFN2, we established in vivo models of LNM and peritoneal carcinomatosis, which consistently showed that ELFN2 suppresses metastatic dissemination. Together, our results reveal a novel ELFN2/PP1A/YAP signaling axis that plays a critical role in controlling LNM in GC. Importantly, these findings not only provide new insights into the molecular mechanisms governing GC metastasis but also highlight ELFN2 as a promising biomarker and potential therapeutic target for the management of GC patients at high risk of LNM.

Background and Aim: Lymphovascular invasion (LVI) is a negative prognostic factor for gastric cancer, but detection limitations hinder its clinical utility and subtype analysis. This study aimed to explore the predictive value of LVI and its subtypes in the prognosis and recurrence patterns of gastric cancer using our enhanced detection method. Methods: We reviewed 2057 patients who underwent gastrectomy in 2018, of whom 1073 met the inclusion criteria. Propensity score matching (PSM) was performed to balance baseline clinicopathological characteristics. Results: After PSM, 311 patients were assigned to the LVI+ group and 311 to the LVI- group. The LVI+ group demonstrated a poorer prognosis. Subtype analysis revealed that lymphatic invasion (LI), but not venous invasion (VI), was associated with poor prognosis in the matched cohort. Stratified by pathological tumor-node-metastasis (TNM) stage, LVI+ and LI+ patients had worse prognosis in Stages I and III, while VI+ patients had worse prognosis in Stage III. Stratified by lymph node status, LVI+ predicted poorer prognosis in both node-negative (N0) and node-positive (N+) patients, and LI+ was also associated with worse prognosis among N+ patients, whereas VI+ was not significantly associated with prognosis in either subgroup. Recurrence analysis indicated that LVI+ was associated with distant and peritoneal metastases, whereas LI+ was associated with local recurrence, distant and peritoneal metastases. Conclusions: Lymphovascular invasion was associated with adverse prognosis in resectable gastric cancer, with lymphatic invasion showing a stronger prognostic impact than venous invasion. These findings indicate that refined assessment of lymphovascular invasion may complement conventional TNM staging in postoperative risk stratification.

Background: For right hepatocellular carcinoma (HCC) involving multiple segments, right hemihepatectomy (RH) is standard but often precluded by insufficient future liver remnant (FLR). This study aimed to evaluate the feasibility and safety of augmented reality navigation and indocyanine green fluorescence imaging (ARN-FI)-guided tailored subsegmental resection (TSR) to optimize parenchymal preservation without compromising oncologic outcomes. Methods: A total of 89 consecutive patients with right HCC ineligible for RH due to insufficient FLR from January 2020 to October 2024 were retrospectively analyzed and divided into the TSR group ( n = 24) and the NAR (non-anatomical resection) group ( n = 65). Both groups utilized multimodal guidance but with different navigational strategies: TSR employed ARN-FI for anatomical subsegmental guidance, while NAR used it for tumor localization. Propensity score matching was used to balance baseline characteristics. Perioperative and survival outcomes were compared. Results: After matching, there were 24 patients in the TSR group and 48 patients in the NAR group. Preoperative three-dimensional planning showed that TSR significantly increased FLR% from 34.5% (simulated RH) to 57.6% ( P < 0.001). Compared with the NAR group, the TSR group exhibited longer operation time and postoperative hospital stays, higher alanine aminotransferase levels and total bilirubin on postoperative days 1 and 3, but lower incidences of ascites and pleural effusion. The 1- and 3-year recurrence-free survival (RFS) rates in the TSR and NAR groups were 90.9% versus 75.6%, and 74.7% versus 46.2%, respectively ( P = 0.028). Multivariable analysis identified TSR as a protective factor for recurrence (hazard ratio = 0.30, P = 0.015). Conclusions: For patients with right HCC ineligible for RH because of insufficient FLR, ARN-FI-guided TSR is associated with superior parenchymal preservation and RFS compared to NAR in this propensity-matched analysis, meriting consideration as a viable alternative surgical strategy.

Gastric cancer remains a highly prevalent malignancy worldwide, with surgical resection currently constituting the cornerstone of treatment aimed at improving long-term patient survival. Owing to their notable advantages, including reduced surgical trauma and accelerated postoperative recovery, minimally invasive procedures are progressively supplanting conventional open surgery and have become the mainstream approach in gastric cancer management. Concurrently, the rapid advancement of artificial intelligence (AI) technologies has enabled real-time intraoperative monitoring of surgical scenes, thereby furnishing novel technical support for adjunctive decision-making, surgical navigation, and skill assessment during gastrectomy. This article provides a systematic review of the current status of AI applications in minimally invasive gastric cancer surgery, with a particular focus on research progress pertaining to instrument recognition, surgical phase identification, delineation of normal anatomical structures, detection of metastatic foci, and early warning of intraoperative adverse events. Furthermore, we discuss the potential value of AI in enhancing surgical efficiency, ensuring patient safety, and optimizing surgical education. On this basis, we further analyze the principal challenges and inherent risks confronting current AI systems, with the aim of informing future technological innovation and facilitating clinical translation.

Recent vision language models (VLMs) have shown strong generalization and multimodal reasoning abilities in natural domains. However, their application to medical diagnosis remains limited by the lack of comprehensive and structured datasets that capture real clinical workflows. To advance the development of VLMs for clinical applications, particularly in gastric cancer, we introduce Gastric-X, a large-scale multimodal benchmark for gastric cancer analysis. Each case in Gastric-X includes paired resting and dynamic CT scans, a set of structured biochemical indicators, expert-authored diagnostic notes, and bounding box annotations of tumor regions, reflecting realistic clinical conditions. We systematically examine the capability of recent VLMs on five core tasks: Visual Question Answering (VQA), report generation, cross-modal retrieval, disease classification, and lesion localization. These tasks simulate critical stages of clinical workflow, from visual understanding and reasoning to multimodal decision support. Through this evaluation, we aim not only to assess model performance but also to probe the nature of VLM understanding: Can current VLMs meaningfully correlate biochemical signals with spatial tumor features and textual reports? We envision Gastric-X as a step toward aligning machine intelligence with the cognitive and evidential reasoning processes of physicians, and as a resource to inspire the development of next-generation medical VLMs.

BACKGROUND:Biliary tract cancer (BTC) has a high postoperative recurrence rate and limiting long-term survival. Accurate prediction of recurrence is vital for clinical management. Carbohydrate antigen 19-9 (CA19-9) and the neutrophil percentage-to-albumin ratio (NPAR) are established prognostic markers in various cancers, but their combined value in BTC remains unclear. This study assessed the predictive role of the combination of CA19-9 and NPAR on time to recurrence (TTR) and introduced the novel NPAR-CA19-9 stratified score (NCSS). METHODS:A total of 145 BTC patients who underwent resection between March 2019 and December 2023 were analyzed. Preoperative blood samples, collected within two weeks of surgery, were used to measure CA19-9 and NPAR. Cutoff values were defined using the Maximum Selected Rank Statistic (MSRS). NCSS (0-2) were assigned according to marker levels and evaluated using Kaplan-Meier survival analysis and Cox regression models. RESULTS:Higher NCSS were associated with shorter TTR and reduced overall survival (P < 0.001). Cox regression confirmed NCSS as an independent predictor of recurrence [NCSS = 1: hazard ratio (HR) = 1.82, 95% confidence interval (CI): 1.05-3.14; NCSS = 2: HR = 2.95, 95% CI: 1.54-5.64] and overall survival (NCSS = 1: HR = 5.21, 95% CI: 1.19-22.82; NCSS = 2: HR = 11.88, 95% CI: 2.44-57.75). Subgroup analysis showed that patients with NCSS = 1 or 2 benefited from adjuvant chemotherapy, whereas those with NCSS = 0 did not. Incorporating NCSS into a nomogram improved prediction of TTR compared with a model excluding NCSS (C-index: 0.72 vs. 0.70; P = 0.027). CONCLUSIONS:NCSS is a novel, simple, and effective preoperative scoring system that predicts TTR and overall survival in BTC patients, supporting more individualized treatment planning.

Minimally invasive surgery has become an important approach for treating complex surgical conditions.As such surgical procedures are often associated with high risks and technical complexities,the surgeon's skills,efficiency of team collaboration,and capacities for emergency response all significantly affect the surgical outcomes.Establishing a robust surgical skill assessment and quality control system is therefore of vital importance to reduce the risks and enhance perioperative safety.In this context,the conventional methods for surgical skill assessment and quality control(e.g.,subjective scoring,postoperative analysis,etc.)appear insufficient due to their inconsistent standards,time and labor intensity,and lack of real-time feedback.In recent years,artificial intelligence(AI),particularly computer vision-based deep learning technology,has demonstrated immense potential in surgical motion recognition,workflow phase analysis,and intraoperative safety monitoring,and has propelled the development of surgical skill assessment and quality control toward automation,objectivity,and precision.Herein the authors review the latest research progress in the application of AI in minimally invasive surgical skill assessment and quality control,focusing on AI-based skill assessment and quality control algorithm models.This review also addresses model generalization,privacy protection,multicenter applicability,and future clinical applications of intelligent precision surgery.

Background: Gastric cancer remains a leading cause of cancer-related mortality worldwide, underscoring the critical need for effective prevention through the management of precancerous lesions. Gastric intestinal metaplasia (GIM) is a key precursor, with its incomplete subtype (IIM) conferring a substantially higher risk of malignant progression compared to complete intestinal metaplasia (CIM). However, IIM is not merely a static histological variant but represents a dynamic, transitional epithelial state of mixed gastric and intestinal differentiation, driven by chronic inflammation. Its accurate identification and clinical management pose significant challenges. Summary: This review synthesizes current evidence on IIM as a high-risk transitional phenotype in gastric carcinogenesis. We detail its cellular origins, primarily from spasmolytic polypeptide-expressing metaplasia (SPEM) via chief cell reprogramming (paligenosis), and its characteristic hybrid molecular signature. Diagnostically, definitive subtyping relies on histochemical staining (e.g., AB-PAS), as conventional endoscopy and histology lack sufficient precision, highlighting a barrier to routine risk stratification. Epidemiologically, IIM is associated with a 4- to 11-fold increased risk of gastric cancer compared to CIM. While Helicobacter pylori eradication may induce partial regression, patients with IIM retain a significant residual cancer risk, necessitating ongoing surveillance. Emerging endoscopic techniques (e.g., narrow-band imaging) and molecular biomarkers (including DMBT1, AQP5, TROP2, and OLFM4) show promise for improving in vivo detection and risk assessment, with combinatorial biomarker strategies potentially better capturing IIM's hybrid phenotype. Current classification systems have limitations, including the frequent coexistence of subtypes and a lack of validated biomarkers for precise risk stratification. Key Messages: 1、IIM is a biologically unstable, transitional lesion with a markedly elevated risk of progression to gastric adenocarcinoma, warranting its distinction from CIM in clinical practice. 2、 Reliance on specialized pathology for diagnosis limits widespread risk stratification, creating an urgent need for reliable endoscopic predictors and validated molecular biomarkers. 3、Even after successful H. pylori eradication, patients with IIM require long-term endoscopic surveillance due to persistent cancer risk. Future strategies must integrate IIM subtyping into personalized surveillance protocols to enhance gastric cancer prevention.

Gastric cancer is the fourth most common cause of cancer-related deaths globally. Cancer stem cells (CSCs) play an essential role in tumor initiation, development, and chemoresistance. However, the molecular mechanisms that regulate CSC traits in gastric cancer, particularly the role of CD44v6 as a key CSC marker, remain poorly understood. Here, we demonstrate that CD44v6 is markedly upregulated in gastric cancer tissues and correlates with poor prognosis. Functional assays, including colony formation, wound healing, proliferation, and apoptosis assays, show that CD44v6 enhances CSC characteristics, such as self-renewal, proliferation, migration, and cisplatin chemoresistance. CD44v6 knockdown effectively suppresses these aggressive phenotypes. Mechanistically, CD44v6 regulates the expression of key CSC markers, including CD24, CD133, EpCAM, as well as stemness-related transcription factors Oct-4 and Nanog. Additionally, CD44v6 enhances cell proliferation and drug resistance in both in vitro and in vivo experiments. Collectively, our findings highlight the significant role of CD44v6 in regulating gastric CSC traits, suggesting it’s a potential as a biomarker and therapeutic target for improving gastric cancer treatment outcomes, particularly in overcoming chemoresistance.

Lymphatic metastasis of gastric cancer is a challenging issue in clinical practice. Recently, copper single‐atom nanozymes (SAZ) have gained tremendous attention due to its superior peroxidase (POD) activity that has good nonocatalytic tumor therapy (NCT) capabilities, and photothermal properties. Therefore, using a high‐expressing P‐selectin platelet membrane (PM) to encapsulate SAZ and cisplatin is proposed, forming PSC nanoparticles. Due to their exquisite nanoscale size and the unique structure of lymphatic vessels, PSC can highly target cancer cells in invasive primary tumors and metastatic lymph nodes that both highly express CD44. It is noteworthy that cisplatin can simultaneously perform chemotherapy and generate H₂O₂ under the action of NADPH oxidases (NOXs) that further enhance the catalytic activity of SAZ and increase intracellular reactive oxygen species (ROS) production. Both in vitro and vivo experiments have demonstrated the superior targeting and elimination capability of the PCS system in primary and metastatic tumor cells. In addition, transcriptomic analysis reveals that PSC + NIR induced apoptosis in MFC cells. This marks the first proposal of combining single‐atom nanozymes and chemotherapy drugs for dual‐targeting in gastric cancer and lymphatic metastasis, providing new insights into a challenging clinical issue in the treatment of gastric cancer lymphatic metastasis.

4069 Background: Surgery remains the cornerstone of curative therapy for locally advanced GC/EGJC. EBV-positive tumor is a distinct molecular subtype that would be potentially sensitive to immunotherapy, but no consistent reports. Given that chemotherapy may enhance antitumor immunity, perioperative immunochemotherapy may be a promising modality for EBV-positive patients. Methods: The NICE trial is a multicenter, multi-cohort phase II study (NCT04744649) evaluating the safety and efficacy of toripalimab plus CapeOX as perioperative treatment in patients with locally advanced GC/EGJC. The Cohort B was first of its kind to assess the efficacy of the immunochemotherapy on the EBV-positive GC/EGJC , in which patients received toripalimab (240 mg) combined with standard-dose CapeOX every 3 weeks for 4 cycles preoperatively and 4 cycles postoperatively. Eligibility criteria included clinical tumor stages of cT3-4aNxM0 or cT2N+M0 disease as determined by both imaging scan and staging laparoscopy with negative peritoneal cytology. The primary endpoint was major pathologic response (MPR, defined as < 10% viable tumor cells). The tumor immune microenvironment (TIME) of tissue samples obtained before and after treatment was analyzed using multiple immunofluorescence assays to assess changes in immune cell infiltration and other biomarkers related to treatment response. Results: From May 2021 to September 2023, 17 patients with EBV-positive GC/EGJC (GC, n = 15; EGJC, n = 2) were enrolled, with cT2N0 (n = 1), cT3N1-3 (n = 5), and cT4aN1-3 (n = 11). All patients completed 4 preoperative cycles of treatment, and none experienced progression before surgery. Only one patient withdrew the inform content after preoperative therapy, the 16 patients underwent radical resection, achieving a 100% R0 resection rate (16/16). The MPR rate was 37.5% (6/16), and pathological complete response rate (pCR) was 25.0% (4/16). Of the 16 participants, 15 received postoperative adjuvant therapy, while 1 declined further treatment. The TIME analysis results showed that tumor-infiltrating CD8+ T cells in post-treatment tumor tissues significantly clonally expanded compared with pre-treatment paired tissues. Treatment-related grade 3/4 adverse events were observed in 6 patients (35.3%, 6/17). Until Dec 31 2024, none of the patients experienced disease recurrence. Conclusions: Neoadjuvant toripalimab combined with CapeOX is a safe and effective treatment option for patients with EBV-positive, locally advanced GC/EGJC, with moderate MPR and pCR, indicating further investigating for this distinct type of cancer. Clinical trial information: NCT04744649 .

Gastric cancer (GC) is the fifth most common malignant tumor worldwide. Surgical resection remains the primary treatment for GC, with laparoscopic surgery recommended by several international guidelines. Due to complex perigastric vessels, standard D2 lymph node dissection (LND) in laparoscopic GC (LapGC) surgery is challenging. Careful dissection is required to expose, dissect, and ligate vessels without injury, ensuring radical LND. Computer vision has the potential to assist in the identification of key vessels during LapGC surgery, thereby reducing the risk of vascular injury. However, existing publicly available surgical anatomy datasets mainly focus on organ segmentation and simple surgeries. To address the clinical challenges and research needs outlined above, we present the LapGC Key Vascular Anatomy Dataset (LapGC-KVAD-30). This dataset was extracted from thirty complete surgical videos and contains annotations for fifteen types of key vessels across eight D2 LND scenes. The LapGC-KVAD-30 uniquely contains 5303 frames that showcase the dynamic process of key vessels from initial appearance to full exposure (or ligation), providing essential information for effective and safe LND.

Endoscopic submucosal dissection (ESD) is an effective treatment with minimal invasiveness for early gastric cancer (EGC). However, cases undergoing non-curative resection (NCR) may still undergo additional surgical procedures. We aimed to analyze the features of NCR under white light imaging (WLI), and develop a prediction model to assess the risk of NCR before ESD. We retrospectively collected and analyzed WLI and clinicopathological features of 568 EGC patients undergoing ESD between March 2016 and March 2024. A nomogram was developed on 455 patients from the training set after subgroup difference analysis. 91 out of 568 (16.0%) cases had NCR. WLI features including remarkable redness, ulceration, fold convergence, marginal elevation, whitish mucosal change, larger lesions, and Helicobacter pylori (Hp) infection were associated with independent risk factors for NCR. The nomogram based on these features showed good predictive value for NCR, with an area under the curve (AUC) of 0.8095 (95% CI: 0.7538-0.8651) in the training set and 0.7567 (95% CI: 0.6427-0.8707) in the validation set. We developed a nomogram incorporating WLI features that exhibits good predictive performance and could potentially assist in selecting optimal treatment strategies for EGC.