Clinical Gastroenterology and Hepatology最新文献

Background and aims: Prospective evidence for the long-term efficacy of endoscopic submucosal dissection (ESD) for T1 colorectal cancer (CRC) is lacking. We evaluated the long-term outcomes after ESD for T1 CRC, focusing on risk stratification and the benefits of additional surgery.

Methods: This large-scale, multicenter, prospective cohort study, conducted by Hiroshima GI Endoscopy Research Group, included consecutive patients with pathologically confirmed T1 CRC who underwent colorectal ESD between 2014 and 2018. Patients were classified into low- and high-risk groups per JSCCR guidelines. High-risk patients either underwent additional surgery (surgery group) or were followed without surgery (follow-up group). The primary outcome was the 5-year cumulative recurrence rate.

Results: Among 2358 patients who underwent ESD for 2478 early colorectal neoplasms, 383 patients with T1 CRC were enrolled. Ten recurrences (3.6%) and five T1 CRC-associated deaths (1.8%) occurred in the high-risk group; none occurred in the low-risk group. Five-year cumulative local recurrence rate in the high-risk group was significantly higher in the follow-up group than in the surgery group (6.8% vs. 0%; P<0.001), whereas distant recurrence did not differ significantly (2.7% vs. 1.9%; P=0.704). Five-year overall survival in the high-risk group was significantly worse in the follow-up group than in the surgery group (79.7% vs. 95.2%; P<0.001), whereas 5-year disease-specific survival (DSS) was similar (98.4% vs. 98.5%, P=0.420).

Conclusion: ESD alone appears appropriate for low-risk T1 CRC, with no recurrence or disease-specific mortality observed, suggesting that intensive surveillance is unnecessary after curative ESD. In high-risk T1 CRC, additional surgery could prevent local recurrence; however, its impact on distant metastasis and DSS remains unclear.

Background and aims: Capsule endoscopy (CE) is a non-invasive technique for diagnosing Crohn's disease (CD); however, manual interpretation of CE videos is time-consuming and error-prone. We developed an artificial intelligence system, INTELCAPE, to automate CE video analysis for accurate and efficient CD diagnosis.

Methods: This retrospective, multi-center study used data from two Chinese hospitals. A multi-task deep learning framework segmented small-intestine regions, detected lesions, and diagnosed CD using CE videos from 757 (Cohort 1) and 115 (Cohort 2) patients. INTELCAPE integrated the ResNet, Transformer, and EfficientNet architectures for hierarchical processing. Performance was benchmarked against clinicians using three metrics. This study received Ethics Committee approval (2024ZSLYEC-040).

Results: INTELCAPE achieved state-of-the-art performance across all tasks. For small-intestine segmentation, the model showed intersection over union (IoU) scores of 94.82% (Cohort 1, 95% confidence interval [CI] = 93.28%-96.36%) and 96.87% (Cohort 2, 95% CI = 94.63%-99.12%). For lesion detection, it achieved area under the curve (AUC) values of 0.993 (Cohort 1) and 0.980 (Cohort 2), with 99.33% classification accuracy, which was comparable to that of specialists (97.83%) but superior to that of residents (91.05%, p < 0.001). For CD diagnosis, INTELCAPE demonstrated robust generalizability, achieving AUCs of 0.982 (Cohort 1) and 0.984 (Cohort 2) with 90% diagnostic accuracy, comparable to that of specialists (93.33%) but 10-fold faster (p < 0.001). INTELCAPE improved doctors' diagnostic accuracy (76.7% to 94.8%, p < 0.001), while reducing their interpretation time (67.9 to 22.5 min, p < 0.001).

Conclusion: INTELCAPE improved CD diagnosis by automating CE video analysis, thereby enhancing accuracy and efficiency, particularly for less-experienced clinicians.