Journal of Gastroenterology最新文献

Background: Mismatch repair (MMR) testing is recommended for all colorectal cancer (CRC) patients, but this assay necessitates the involvement of specialized institutions and is time-consuming. This study aims to develop a deep learning model for MMR prediction using macroscopic images to provide a rapid and cost-free screening tool.

Methods: This diagnostic study enrolled 809 CRC patients who underwent surgical resection without neoadjuvant therapy at Peking University Third Hospital (from January 2020 to July 2025). Macroscopic images of surgical specimens were captured immediately after resection. MMR status was confirmed by postoperative immunohistochemical staining for MMR proteins (MLH1, MSH2, MSH6, and PMS2). Deep learning models were developed by a two-step approach: automated lesion segmentation using DeepLabV3 + , followed by MMR classification using vision transformer (ViT). MMR prediction performance was mainly evaluated utilizing area under the curve (AUC). Gradient-weighted Class Activation Mapping (Grad-CAM) appraisal and principal component analysis (PCA) were performed to assess the explainability of the model.

Results: The proposed model achieved an average AUC of 0.896 (95% CI, 0.763-0.959) on internal test and 0.860 (95% CI, 0.644-0.921) on independent test for MMR prediction. High NPVs of 0.987 (95% CI, 0.928-0.999) and 0.978 (95% CI, 0.925-0.994) were observed in internal and independent testing, respectively, using a threshold of 0.323. Grad-CAM analysis and PCA demonstrated that the deep-learning model was of explainability.

Conclusions: The new deep-learning model accurately identified MMR status using macroscopic specimen images and showed potential for MMR screening among CRC patients, particularly in a rapid-response scenario.

Background: Tyrosine kinase inhibitors (TKIs) targeting KIT and PDGFRA mutations are the mainstay of treatment for gastrointestinal stromal tumors (GISTs). Comprehensive genomic profiling (CGP) has recently been incorporated into routine practice for various solid tumors, facilitating personalized treatment strategies. We elucidated the current status and clinical significance of CGP in patients with advanced GISTs.

Methods: In this multicenter retrospective cohort study, we analyzed 36 patients with unresectable or recurrent GISTs who underwent CGP testing between November 2019 and February 2025. Specimen details, genomic alterations, expert panel assessments, and treatment implementations were collected. Turnaround time and overall survival (OS) were evaluated.

Results: Among the patients, 75% underwent CGP during or post-third-line treatment. Actionable and druggable alterations were identified in 97.2% and 52.8%, respectively. Expert panel-recommended treatment was proposed for 12 patients (33.3%); however, only one received CGP-guided therapy. The median turnaround time and OS from CGP were 63 days (range, 41-100) and 19.9 months (range, 0.8-68.0), respectively. Most patients were microsatellite-stable with low tumor mutational burden. Among non-KIT genes, CDKN2A and CDKN2B alterations were the most frequently identified. Furthermore, patients harboring these alterations showed poorer OS post-recurrence (CDKN2A: HR 3.09, 95% CI 1.17-8.18, p = 0.017; CDKN2B: HR 2.55, 95% CI 0.86-8.18, p = 0.081).

Conclusions: CGP revealed actionable alterations in most patients with advanced GISTs; however, its impact on treatment decisions remains limited under current practice. Early CGP implementation may facilitate access to matched therapies or clinical trials, enabling prognostic stratification in patients with advanced GISTs.

Objectives: The oncologic outcomes of pedunculated-type T1 colorectal cancer (CRC) remain unknown. We determined the risk factors for lymph node metastasis (LNM) and recurrence and evaluated the survival according to the treatment method.

Methods: In this multicenter retrospective study involving 4673 patients with T1 CRC, we analyzed 444 patients with pedunculated-type T1 CRC treated between 2009 and 2016. Treatment included local resection (LR) alone (n = 169), surgery with lymph node (LN) dissection alone (n = 83), and LR followed by additional surgery with LN dissection (n = 192). Factors associated with LNM and recurrence, relapse-free survival (RFS) and overall survival (OS) by treatment were analyzed. The median follow-up period was 64 months.

Results: LNM and recurrence were observed in 25 (5.6%) and 13 (2.9%) cases, respectively. Submucosal invasion depth ≥ 1000 μm (p = 0.0036), positive lymphovascular invasion (p = 0.0007), and budding grade 2/3 (p = 0.0171) were risk factors for LNM. The risk factor for recurrence was tumor size ≥ 20 mm (HR 5.488; 95% CI 1.199-25.12; p = 0.028) in a multivariate analysis. The 5-year RFS rates were 92.5% for LR alone, 94.3% for LR+ surgery, and 90.5% for surgery alone; the 5-year OS rates were 93.1%, 97.1%, and 94.0%, respectively, with no significant difference.

Conclusion: Even in the specific subset of pedunculated-type T1 CRC, submucosal invasion depth ≥ 1000 μm and budding grade 2/3 are risk factors for LNM. Tumors ≥ 20 mm require careful surveillance for recurrence risk. High RFS and OS rates in LR-alone and LR+ surgery groups suggest LR is appropriate for pedunculated lesions.

Background: Owing to the rarity of metastatic tumors in the small bowel, their clinicopathological features, and prognostic factors remain poorly understood. This study aimed to clarify the clinicopathological features and factors associated with the prognosis of patients with small bowel metastasis from other organs in Japan.

Methods: We retrospectively examined 253 patients who were histopathologically diagnosed with small bowel metastases between January 2008 and December 2017 at multiple institutions in Japan. We identified the clinicopathological features of the condition and determined the factors associated with the prognosis of these patients.

Results: Obstructive symptoms were the most frequent clinical presentations (39% abdominal pain and 18% vomiting), while gastrointestinal bleeding was observed in 27% of patients. The diagnostic modalities included enteroscopy (33%), balloon-assisted enteroscopy (30%), and capsule endoscopy (13%). The most common primary tumor was lung cancer (38%), followed by colorectal cancer (18%), gastric cancer (9%), and malignant melanoma (6%). Surgical intervention, including tumor resection or bypass surgery, was performed in 79% of patients. The cumulative survival rates of patients at 12, 24, and 60 months were 49%, 36%, and 22%, respectively. Multivariate analysis identified surgery as a significant factor for improving overall survival (HR = 0.56, 95% CI 0.35-0.89, p = 0.01).

Conclusions: The lung cancer is the most frequent primary tumor of metastatic tumors in the small bowel. Surgical intervention was associated with improved survival outcomes.

Objectives: Sedation is increasingly essential for gastrointestinal endoscopy. Remimazolam, an ultra-short-acting benzodiazepine, has a shorter pharmacokinetic half-life than midazolam. The aim of this study was to determine whether remimazolam provides superior procedural sedation in Japanese patients.

Methods: The cohort of this prospective, multicenter, randomized, single-blind controlled trial comprised adults (18-80 years) scheduled for sedated upper gastrointestinal endoscopy. Participants were randomized to remimazolam and midazolam groups. The primary outcome was the proportion of ambulatory patients 5 min after endoscopy. Secondary outcomes were successful pre-procedure sedation (Modified Observer's Assessment of Alertness/Sedation ≤ 4), dose of sedative to achieve sedation, time to ambulation, and adverse events.

Results: From October 2024 to January 2025, 40 patients were enrolled. After excluding two outliers 38 were analyzed (remimazolam, n = 20; midazolam, n = 18). Ambulation at 5 min occurred in 85.0% (17/20) of the remimazolam versus 0.0% (0/18) of the midazolam group (p < 0.0001). Mean time from procedure end to walking was 4.25 min (range 0.0-10.0) for remimazolam and 35.56 min (10.0-60.0) for midazolam (p < 0.0001). Pre-procedure sedation was successful (MOAA/S ≤ 4) in 100% of both groups. Mean doses to achieve sedation were 4.30 mg (3.0-7.0) for remimazolam and 3.11 mg (2.0-5.0) for midazolam (p = 0.003). Hypoxemia occurred in 5.0% of the remimazolam and 33.3% of the midazolam group (p = 0.038).

Conclusions: In upper gastrointestinal endoscopy, remimazolam achieved markedly faster recovery and a lower incidence of hypoxemia than midazolam. Rates of achieving target sedation were equivalent. These findings indicate remimazolam is an effective and potentially safer sedative option for Japanese patients undergoing endoscopy.

Trial registration: This research was registered with the Japan Registry of Clinical Trials (trial number jRCTs071240062) on September 26, 2024.

Background: Intrahepatic cholangiocarcinoma (ICC) is an aggressive malignancy with limited treatment options and a poor prognosis. Although immune checkpoint inhibitor (ICI)-based chemotherapy has recently been introduced as a first-line treatment, clinical responses remain highly variable. Currently, no reliable molecular biomarkers are available to predict the therapeutic efficacy of ICC.

Methods: We conducted a retrospective, multicenter cohort study of 25 patients with unresectable ICC treated with gemcitabine, cisplatin, and durvalumab (GCD therapy) between September 2022 and August 2024. Comprehensive genomic profiling was performed in 19 patients (76%) to identify genomic alterations. In selected cases, spatial transcriptomic analysis using the Xenium platform was employed to characterize tumor-immune spatial dynamics.

Results: The median progression-free survival (PFS) was 7.8 months (95% CI: 3.6-10.6), and the median overall survival (OS) was 11.1 months (95% CI: 6.9-18.6). Among genomic alterations, TP53 mutation was the only independent predictor of shorter PFS in multivariate analysis (HR: 4.20; p = 0.036). TP53-mutant tumors were associated with significantly shorter PFS (p = 0.011) and a trend toward worse OS (p = 0.051). Spatial transcriptomic profiling revealed a distinct immunosuppressive microenvironment in TP53-mutated tumors marked by poor CD8⁺ T-cell infiltration, enrichment of CD109⁺ tumor-associated macrophages, and downregulation of antigen-presentation genes TAP1 and TAP2.

Conclusions: This study is the first to identify TP53 mutation as a biomarker of resistance to ICI-based therapy in ICC, and to uncover its immune-evasive phenotype using spatial profiling. These findings provide mechanistic insight into immune evasion and support the development of TP53-guided immunotherapeutic strategies in ICC.

Background: The transient receptor potential cation channel subfamily V member 6 (TRPV6) gene, encoding a calcium-selective ion channel, was recently identified as a susceptibility gene for pancreatitis. This study aimed to clarify the natural history of TRPV6-related pancreatitis and the impact of pancreas-specific deletion of Trpv6 on pancreatitis in mice.

Methods: Clinical information of the patients carrying functionally impaired TRPV6 variants, defined by Ca²⁺ imaging and minigene assays, was collected from six international centers. Cumulative rates were assessed using Kaplan-Meier analysis. As controls, Japanese patients with alcohol-unrelated pancreatitis carrying pathogenic variants in PRSS1 or SPINK1, as well as those without pathogenic variants in pancreatitis susceptibility genes, were enrolled. A pancreas-specific Trpv6 conditional knockout mouse was established by crossing the Trpv6 floxed mouse and the Pdx-1-Cre mouse. Pancreatitis was induced by repeated intraperitoneal injections of caerulein.

Results: Ninety-four patients with functionally impaired TRPV6 variants, including six splice-site variants, were enrolled. The median age at symptom onset was 16 years. The cumulative rates of pancreatic calcification, pancreatic exocrine insufficiency, diabetes mellitus, and interventions for pancreatitis were 55.5%, 20.1%, 10.8%, and 41.6% at 30 years, and 81.5%, 49.6%, 45.4%, and 69.9% at 50 years, respectively. Pancreas-specific Trpv6 knockout mice developed more severe acute and chronic pancreatitis than the control mice. Caerulein treatment increased the TRPV6 expression in pancreatic acinar cells.

Conclusions: Functionally impaired TRPV6 variants significantly influenced the clinical outcomes of chronic pancreatitis. TRPV6 in pancreatic acinar cells might play a protective role against pancreatitis in mice.

Background: The long-term local control and effect of photodynamic therapy (PDT) using talaporfin sodium and a diode laser (talaporfin PDT) on overall survival (OS) for local failure after chemoradiotherapy (CRT) for esophageal cancer are unknown. Here, we present a 5-year survival analysis for talaporfin PDT.

Methods: This was a prospective follow-up analysis of an open-label, multicenter, phase 2 study of local failure after CRT or radiotherapy in patients who received talaporfin PDT. The primary endpoint was the overall OS. The secondary endpoints were progression-free survival (PFS), local progression-free survival (L-PFS) with local progression or recurrence and death as events, and local time to progression (L-TTP) with only local progression or recurrence as events.

Results: Between November 2012 and December 2013, 26 patients with oesophageal squamous cell carcinoma underwent talaporfin PDT. The baseline T stages were cT1 in 14 patients, cT2 in 6 patients, and cT3 in 6 patients. Pre-PDT T stages were cT1b for 19 and cT2 for 7 patients, and no lymph nodes or distant metastases were detected. At a median 6.8-year follow-up, the median OS and 5-year OS rates were 4.2 years (95% confidence interval [CI]: 1.6-7.3) and 40.6% (95% CI: 21.7-58.7), respectively. The median PFS and L-PFS were 1.1 and 2.1 years, respectively. The 5-year local progression-free rate was 84.9%. No treatment-related deaths occurred.

Conclusion: Talaporfin PDT for patients with oesophageal cancer with local failure after CRT can achieve long-term local complete response and long-term survival as a minimally invasive salvage treatment.

Trial registration number: UMIN000009184.

Background: This study aimed to evaluate whether laparoscopic splenectomy and azygoportal disconnection (LSD) can promote decompensated cirrhotic portal hypertension (CPH) patients to achieve recompensation as defined by Baveno VII.

Methods: This retrospective study reviewed clinical records and follow-up data of decompensated CPH patients diagnosed with gastroesophageal variceal bleeding (GEVB) and hypersplenism at our hepatobiliary center from 2013 to 2023. According to treatment strategy, patients were categorized into the LSD arm or the endoscopic therapy (ET) arm. Post-treatment liver function, incidence of decompensation events, recompensation, and overall survival were analyzed across the two arms. Furthermore, the mediating effect of LSD on survival through recompensation was analyzed.

Results: This study enrolled 568 eligible patients, with 300 undergoing LSD and 268 receiving ET. Most patients in both groups showed varying degrees of liver function improvement post-treatment. Overall, 307 patients (54.05%) met the Baveno VII criteria for recompensation. More patients achieved recompensation among those treated with LSD as opposed to ET (73.0% VS. 32.8%, OR = 4.569; 95% CI 3.088-6.760; P < 0.001). Mediation assessment indicated that recompensation accounted for 43.3%, 32.4%, and 16.4% of the effect of LSD on mortality at 3, 5, and 8 years, respectively. Furthermore, younger patients were more likely to achieve recompensation after LSD.

Conclusions: LSD was found to significantly promote recompensation and extend survival in decompensated cirrhosis patients with CPH bleeding and hypersplenism.

Background: We sought to evaluate the predictive value of protease-3 (PR3)-anti-neutrophil cytoplasmic antibodies (ANCAs) and myeloperoxidase (MPO)-ANCA for primary nonresponse (PNR) to biologic agents in patients with ulcerative colitis (UC).

Methods: A multicenter, prospective cohort study was conducted. Patients with UC who initiated biologic agents (anti-TNF, anti-integrin, or anti-IL12/23 agents) as their first-line therapy were enrolled. PNR was defined as failure to achieve a partial Mayo Score improvement of at least 2 points by weeks 14-16 or necessitating treatment modification. Logistic regression identified PNR predictors.

Results: Among 95 biologic-naïve UC patients, 38 received anti-TNF, 39 received anti-integrin, and 18 received anti-IL12/23 agents. PR3-ANCA positivity was observed in 47.4% of anti-TNF recipients. The PNR rate among PR3-ANCA-positive patients receiving anti-TNF therapy was 66.7%, significantly higher than in PR3-ANCA-negative patients (25.0%, P = 0.02). Multivariate logistic regression confirmed PR3-ANCA as an independent risk factor for PNR to anti-TNF therapy (OR 5.61; 95% CI 1.37-26.82; P = 0.02). By contrast, PR3-ANCA did not predict PNR to anti-integrin therapy, but the platelet-to-lymphocyte ratio (PLR) and the systemic immune-inflammatory index (SII) were identified as significant predictors. Only one of 95 patients tested positive for MPO-ANCA, precluding statistical analysis for its association with PNR.

Conclusion: In patients with UC, PR3-ANCA is a significant predictor of PNR to anti-TNF therapy, but not to anti-integrin therapy. PLR and SII may predict PNR to anti-integrin therapy. PR3-ANCA could aid biologic selection, optimizing personalized UC treatment and healthcare resource utilization.