恶性梗阻性黄疸患者经ERCP后出血的危险因素分析及模型预测

Analysis of risk factors for bleeding in patients with malignant obstructive jaundice after endoscopic retrograde cholangiopancreatography

目的探讨恶性梗阻性黄疸(MOJ)患者经内镜逆行胰胆管造影术(ERCP)后出血的危险因素分析及模型预测。方法回顾性分析2015年1月—2021年6月于西南医科大学附属三六三医院就诊的ERCP治疗的MOJ患者302例,收集患者一般临床资料,并对患者胰胆管生化指标进行检测。出院后对患者进行随访,依据随访患者ERCP术后是否出血分为出血组(n=47)和对照组(n=255),比较两组患者的一般资料和临床数据,包括年龄、性别、血小板数量、是否具有胆管结石、急性胆管炎、急性胰腺炎、结石数量、术中出血、胰腺癌、胆管癌、结石直径大、结石嵌顿、十二指肠乳头憩室,手术预切开的情况。正态分布的计量资料用均数±标准差(±s)表示,组间比较采用两独立样本t检验;非正态分布的计量资料用M(Q_(1),Q_(3))表示,组间比较采用Mann-Whitney U检验;计数资料组间比较采用χ^(2)检验。采用Logistic多因素回归分析ERCP术后出血的独立危险因素,依据患者术后出血的独立危险因素建立列线图预测模型并验证。结果两组患者在年龄、性别、血小板数量、胆管结石、急性胆管炎、急性胰腺炎、结石数量、术中出血等方面比较,差异无统计学意义(P>0.05)。出血组患有胰腺癌、胆管癌、结石直径大、结石嵌顿、十二指肠乳头憩室、手术预切开的百分率分别为12.77%、17.02%、19.15%、51.06%、59.57%、14.89%,对照组的百分率分别为3.92%、5.10%、9.02%、19.22%、17.65%、5.88%,两组相比差异有统计学意义(P<0.05);以患者术后是否出血为因变量,以单因素分析中差异具有统计学意义的指标作为自变量进行多因素Logistic回归分析,结果显示,患有胰腺癌(OR=1.838,95%CI:1.524~4.613,P=0.041)、胆管癌(OR=2.548,95%CI:1.870~5.116,P=0.015)、结石嵌顿(OR=3.078,95%CI:2.374~6.012,P<0.001)、十二指肠乳头憩室(OR=1.140,95%CI:1.045~1.628,P<0.001)、手术预切开(OR=1.640,95%CI:1.321~1.928,P<0.001)是MOJ患者ERCP术后出血的独立危险因素。十二指肠乳头憩室的预测能力最大;结石嵌顿和胆管癌的预测能力次之,且相互间无太大差别;胰腺癌、结石直径、预切开对MOJ患者ERCP术后出血的预测能力较小。胰腺癌、胆管癌、结石直径大、结石嵌顿、十二指肠乳头憩室、预切开的积分分别为42、63、28、65、76、34分,总分308分对应列线图模型的预测能力为61.6%,总体而言,列线图预测效能良好。采用Harrell concordance index分析法和ROC曲线对模型区分度进行评价,C-index计算结果为0.826(95%CI:0.771~0.847),ROC曲线AUC为0.843(95%CI:0.801~0.884),ROC预测值和C-index计算结果较为接近,模型区分度应用于本次研究中,具有一定的预测效果。Calibration曲线中列线图模型预测MOJ患者ERCP术后出血的概率与实际发生概率具有较高的一致性。结论ERCP对大多数MOJ患者安全可行,但对于患有胰腺癌、胆管癌、结石直径较大、结石嵌顿、十二指肠乳头憩室的患者,应谨慎操作,且要避免手术预切开的情况,以降低术后出血的风险。另外,列线图模型预测MOJ患者ERCP术后出血的预测能力较强,临床研究中值得借鉴。

Objective To investigate the risk factor analysis and model prediction of bleeding after endoscopic retrograde cholangiopancreatography in patients with malignant obstructive jaundice(MOJ).Methods A retrospective analysis was performed on 302 patients with MOJ treated with ERCP who were treated in the No.363 Hospital Affiliated to Southwest Medical University from January 2015 to June 2021.The general clinical data of the patients were collected,and the biochemical indicators of the pancreatic and bile ducts were detected.The patients were followed up after discharge,and the patients were divided into a bleeding group(n=47)and a control group(n=255)according to whether the follow-up patients were bleeding after ERCP.Compared the general and clinical data of the two groups of patients,including age,gender,platelet count,presence of bile duct stones,acute cholangitis,acute pancreatitis,number of stones,intraoperative bleeding,pancreatic cancer,cholangiocarcinoma,large stone diameter,stone incarceration,duodenal papillary diverticulum,and pre-surgical incision.The measurement data that obey the normal distribution were represented by the mean±standard deviation(±s),and the two independent sample t test was used for the comparison between groups;the data that do not conform to the normal distribution were represented by M(Q1,Q3),and the comparison between groups was used Mann-Whitney U test.The comparison of enumeration data between groups adopted chi-square test.Logistic multivariate regression was used to analyze the independent risk factors of postoperative bleeding after ERCP,and a nomogram prediction model was established and verified according to the independent risk factors of postoperative bleeding.Results The two groups of patients were compared in age,gender,platelet count,bile duct stones,acute cholangitis,acute pancreatitis,the number of stones,intraoperative bleeding and other aspects,the difference was not statistically significant(P>0.05).The percentages of pancreatic cancer,cholangiocarcinoma,large stone diameter,stone incarceration,duodenal papillary diverticulum,and surgical pre-incision in the bleeding group were 12.77%,17.02%,19.15%,51.06%,59.57%,and 14.89%,respectively.,the percentages of the control group were 3.92%,5.10%,9.02%,19.22%,17.65%,and 5.88%,and the difference was statistically significant between the two groups(P<0.05).Taking postoperative bleeding as the dependent variable,and using the indicators with statistical differences in univariate analysis as independent variables,multivariate Logistic regression analysis showed that the patient had pancreatic cancer(OR=1.838,95%CI:1.524-4.613,P=0.041),cholangiocarcinoma(OR=2.548,95%CI:1.870-5.116,P=0.015),stone incarceration(OR=3.078,95%CI:2.374-6.012,P<0.001),duodenum Intestinal papillary diverticula(OR=1.140,95%CI:1.045-1.628,P<0.001),surgical pre-incision(OR=1.640,95%CI:1.321-1.928,P<0.001)were associated with postoperative bleeding in MOJ patients after ERCP independent risk factors.The predictive ability of duodenal papillary diverticulum was the highest;the predictive ability of stone incarceration and cholangiocarcinoma was the second,and there was no significant difference between them;the predictive ability of pancreatic cancer,stone diameter,and pre-incision on bleeding after ERCP in MOJ patients smaller.Pancreatic cancer,cholangiocarcinoma,large stone diameter,stone incarceration,duodenal papillary diverticulum,and pre-incision scores were 42,63,28,65,76,and 34 points respectively,and the total score was 308 points corresponding to the nomogram model.The predictive power of the nomogram was 61.6%,and overall,the nomogram had good predictive performance.Harrell concordance index analysis and ROC curve were used to evaluate the model discrimination,the C-index calculation result was 0.826(95%CI:0.771-0.847),the ROC curve AUC was 0.843(95%CI:0.801-0.884),and the ROC prediction The value and the calculation result of C-index are relatively close.The model discrimination is applied in this study and has a certain prediction effect.The nomogram model in the Calibration curve predicted the probability of postoperative bleeding after ERCP in MOJ patients with high consistency with the actual probability.Conclusion ERCP is safe and feasible for most patients with MOJ,but for patients with pancreatic cancer,bile duct cancer,large stone diameter,stone incarceration,and duodenal papillary diverticulum,it should be performed with caution,and preoperative incision should be avoided,to reduce the risk of postoperative bleeding.In addition,the nomogram model has a strong predictive ability in predicting bleeding after ERCP in patients with MOJ,which is worthy of reference in clinical research.