内镜下≤20 mm直肠神经内分泌瘤非治愈性切除风险模型的建立与验证

Establishment and Validation of a Risk Model for Non-curative Resection of Rectal Neuroendocrine Tumors ≤20 mm Under Endoscopy

摘要:

目的 评估内镜下＜10mm与10-20mm直肠神经内分泌瘤治疗效果差异。建立≤20mm直肠神经内分泌瘤的非治愈性切除预测模型并验证。方法 回顾分析苏州大学附属第一医院、苏州第九人民医院2013年1月至 2023 年12月收治的直肠神经内分泌肿瘤患者数据，分析镜下观、手术方式、术中术后出血率、病理结果特点；对比分析<10mm与10mm-20mm R-NET的术后出血率、治愈性切除率、临床经济学指标、随访结果。同时建立≤20mm直肠神经内分泌瘤的非治愈性切除logistic回归模型，绘制列线图、ROC曲线、校准曲线及决策曲线。结果 在纳入的213例患者中，年龄为50.53±11.42岁，47.9%为男性。距肛缘的中位距离为7.11±2.79cm。肿瘤直径大小为8.24±3.75 mm。直径＜10mm 患者133例 （62.4%）；10mm-20mm患者80例（37.6%）。治疗方式：EMR25例，ESD179 例 ，EFR 5例，EMR后追加ESD 2例；病理分析：G1期 170例 （79.8%），G2 43例 （20.2%）；CgA 阳性率138例 （64.8%）；Syn 138例 97.2%；Ki67 2.37±2.82%。病例均无远处淋巴血管或神经周围浸润；内镜下切除<10mm与10mm-20 mm R-NET在术中术后出血率，治愈性切除，长期随访后转移及复发率方面均无明显差异。以此为前提，建立以镜下表现（表面凹陷）、Ki-67、CgA阳性为变量的预测≤20mm R-NET非治愈性切除logistic回归模型。模型AUC=0.766，95%置信区间 0.696-0.837，并建立列线图，通过临床决策分析证实其有良好的临床净收益。模型预测概率和观测概率一致性良好。结论 内镜下切除<10mm与10mm-20 mm R-NET在治疗效果上无明显差异;可以10mm-20 mm R-NET考虑将内镜切除作为一线治疗方法；表面凹陷、Ki-67指数高、CgA阳性是影响内镜下非R0切除的危险因素。以此建立的预测模型具有较好的检验效能。

Abstract:

Purpose: Assessment of treatment efficacy differences between rectal neuroendocrine tumors <10 mm and 10-20 mm nnder endoscopy. Establishment and validation of a predictive model for non-curative resection of rectal neuroendocrine tumors ≤20 mm. Method: Analyze the data of rectal neuroendocrine tumor patients admitted to suzhou university affiliated first hospital and suzhou ninth people's hospital from january 2013 to december 2023, including endoscopic findings, surgical methods, intraoperative and postoperative bleeding rates, and pathological results. Compare the postoperative bleeding rates, curative resection rates, clinical economic indicators, and follow-up results between <10mm and 10mm-20mm R-NETs. Establish a logistic regression model for non-curative resection of rectal neuroendocrine tumors ≤20mm, and create nomograms, roc curves, calibration curves, and decision curves. Result: Among the 213 patients included, the average age was 50.53±11.42 years, and 47.9% were male. The median distance from the anal verge was 7.11±2.79 cm. The tumor diameter was 8.24±3.75 mm. There were 133 patients (<10mm, 62.4%) and 80 patients (10mm-20mm, 37.6%). Treatment methods: EMR 25 cases, ESD 179 cases, EFR 5 cases, EMR followed by ESD 2 cases; pathological analysis: G1 stage 170 cases (79.8%), G2 43 cases (20.2%); CgA positive rate 138 cases (64.8%); Syn 138 cases 97.2%; Ki67 2.37±2.82%. no cases had distant lymphatic, vascular, or perineural infiltration. There were no significant differences in intraoperative and postoperative bleeding rates, curative resection rates, or long-term metastasis and recurrence rates between <10mm and 10mm-20mm R-NETs. based on this, a logistic regression model predicting non-curative resection of ≤20mm R-NETs was established using endoscopic features (surface depression), Ki-67, and CgA positivity as variables. the model had an AUC of 0.766, with a 95% confidence interval of 0.696-0.837, and a nomogram was created. clinical decision analysis confirmed its good clinical net benefit. the predicted probabilities of the model were consistent with observed probabilities. Conclusion: There is no significant difference in treatment outcomes between endoscopic resection of <10mm and 10mm-20mm R-NETs. For 10mm-20mm R-NETs, endoscopic resection should be considered as a first-line treatment method. Surface depression, high Ki-67 index, and positive CgA are risk factors for non-R0 resection under endoscopy. The predictive model established based on these factors demonstrates good performance in testing.