胰十二指肠切除术后临床相关胰瘘风险预测模型构建及验证  

Construction and validation of a risk prediction model for clinically relevant pancreatic fistula after pancreaticoduodenectomy

作　　者：张丹阳 雷鹏[1] 张宇波 杨刚 张伟[1] ZHANG Danyang;LEI Peng;ZHANG Yubo;YANG Gang;ZHANG Wei(Department of Hepatobiliary Surgery,General Hospital of Ningxia Medical University,Yinchuan 750004,China)

机构地区：[1]宁夏医科大学总医院肝胆外科,宁夏银川750004

出　　处：《中国普通外科杂志》2024年第3期366-375,共10页China Journal of General Surgery

基　　金：宁夏回族自治区财政厅重点研发计划基金资助项目(2018BEG03001)。

摘　　要：背景与目的:随着外科技术的进步以及临床医师不断完善手术方式,胰十二指肠切除术(PD)的手术死亡显著减少,但术后并发症发生率仍居高不下,其中最常见和最严重的是术后胰瘘(POPF)。因此,本研究探讨PD后发生临床相关胰瘘(CR-POPF)的危险因素并开发风险预测模型。方法:回顾性收集2015年1月—2021年7月宁夏医科大学总医院肝胆外科行PD治疗的365例患者临床资料。通过计算机产生随机数的方法以7∶3比例随机分为建模组和验证组,对建模组采用单因素及多因素Logistic回归分析确立CR-POPF的独立危险因素,构建临床预测模型,以列线图进行可视化呈现;利用受试者工作特征(ROC)曲线评价区分度,并通过Bootstrap重复自抽样法进行内部验证,绘制校准曲线评价校准度;将验证组代入模型绘制ROC曲线和校准曲线,验证模型的预测效能。结果:单因素分析结果显示,性别、BMI、腹部手术史、主胰管直径、胰腺质地、中性粒细胞计数、单核细胞计数、淋巴细胞计数与单核细胞计数比值、术后乳酸与CR-POPF明显有关(均P<0.05);多因素二元Logistic回归分析显示,男性(OR=2.896,95%CI=1.368~6.390)、术后高乳酸(OR=3.593,95%CI=2.211~6.172)、主胰管直径≤3mm (OR=0.243,95%CI=0.102~0.552)、胰腺质软(OR=0.146,95%CI=0.061~0.331)是CR-POPF的独立危险因素(均P<0.05)。结合回归系数建立数学模型,构建列线图实现模型可视化。模型ROC曲线下面积(AUC)为0.897 (95%CI=0.857~0.936);校准度评估结果显示,模拟曲线和实际曲线走势基本一致,平均绝对误差(MAE)=0.014。将验证组数据代入预测模型,绘制验证组预测CR-POPF风险的ROC曲线,结果显示,AUC为0.901 (95%CI=0.844~0.959);校准曲线显示,验证组的模拟曲线和实际曲线走势基本一致(MAE=0.019)。结论:男性、胰腺质软、主胰管直径≤3 mm、术后高乳酸与PD术后发生POPF密切相关。依据这四项指标建立的术后早�Background and Aims:With the advancement of surgical techniques and clinicians' continuous refinement of surgical approaches,the surgical mortality of pancreaticoduodenectomy(PD) has significantly decreased.However,the incidence of postoperative complications remains high,with postoperative pancreatic fistula(POPF) being the most common and severe.Therefore,this study was conducted to investigate the risk factors for clinically relevant POPF(CR-POPF) after PD and develop a risk prediction model.Methods:The clinical data of 365 patients who underwent PD in the Department of Hepatobiliary Surgery of General Hospital of Ningxia Medical University from January 2015 to July 2021 were retrospectively collected.Patients were randomly divided into modeling and validation groups at a ratio of 7∶3 based on a random number generator.Univariate and multivariate Logistic regression analyses were conducted on the modeling group to determine independent risk factors for CR-POPF.A clinical prediction model was constructed and visualized using a nomogram.The discriminative ability was evaluated using the ROC curve,and Bootstrap drew the calibration curve and repeated the self-sampling method for internal validation.The validation group was incorporated into the model to verify the predictive performance of the model by drawing the ROC curve and calibration curve.Results:Univariate analysis showed that sex,BMI,history of abdominal surgery,main pancreatic duct diameter,pancreatic texture,neutrophil count,monocyte count,lymphocyte count,the ratio of monocytes to lymphocytes,and postoperative lactate were significantly associated with CR-POPF(all P<0.05).Multivariate Logistic regression analysis revealed that male sex(OR=2.896,95% CI=1.368-6.390),high postoperative lactate(OR=3.593,95% CI=2.211-6.172),main pancreatic duct diameter ≤3 mm(OR=0.243,95% CI=0.102-0.552),and soft pancreatic texture(OR=0.146,95% CI=0.061-0.331) were independent risk factors for CR-POPF(all P<0.05).A mathematical model was established based on regression

关 键 词：胰十二指肠切除术 手术后并发症 胰腺瘘 危险因素 

分 类 号：R657.5[医药卫生—外科学]