胰腺术后B级和C级胰瘘引流特征及最佳干预时间预测的单中心回顾性研究

A single-center retrospective study of percutaneous drainage clinical characteristics of grade B and C postoperative pancreatic fistula and determination of the optimal intervention time

目的探讨胰腺术后B级和C级胰瘘积液的分型及处理流程。方法回顾性收集2018年1月至2022年12月北京医院肝胆胰外科连续收治的232例行各种胰腺手术的患者资料,筛选诊断为B级及C级胰瘘并行超声引导下穿刺置管引流的46例患者。男性32例,女性14例;年龄(60.2±13.7)岁(范围:18~85岁)。收集患者术后CT检查的影像学资料,对纳入患者的腹腔、盆腔及腹膜后积液位置进行分型,并统计确诊胰瘘时引流管状态。应用机器学习随机森林模型构建穿刺时间与病死率关系,计算最佳穿刺时间。通过t检验、Mann-WhitneyU检验、χ^(2)检验、Fisher确切概率法等比较两组临床资料及结局指标。结果根据随机森林模型结果,最佳穿刺时间为确诊胰瘘后5.38 d内。根据最佳穿刺时间将46例患者分为早期穿刺组(21例)和晚期穿刺组(25例)。根据术后确诊胰瘘时CT检查影像显示的积液位置,将46例患者的积液分为4型:胰周型15例(32.7%)、胰外上腹型19例(41.3%)、胰外下腹型6例(13.0%)、弥漫型6例(13.0%)。分析确诊胰瘘时引流管状态,其中10例患者引流管通畅未移位(21.8%),18例患者出现引流管移位(39.1%),18例患者引流管拔除或堵塞(39.1%)。早期穿刺组围手术期病死率为19.0%(4/21),晚期穿刺组为8.0%(2/25)。晚期穿刺组的引流液培养阳性率[88.0%(22/25)比42.9%(10/21),χ^(2)=10.584,P=0.001]、二次手术率[24.0%(6/25)比0,P=0.025]、出院后90 d内再住院率[32.0%(8/25)比4.8%(1/21),χ^(2)=5.381,P=0.020]均高于早期穿刺组,且术后住院时间[M(IQR)]延长[(24(20)d)比39(53)d,Z=3.023,P=0.003]。结论胰腺手术后B级和C级胰瘘患者积液位置可分为4型,对于胰瘘患者,早期影像学评估可及时发现腹水。一旦发现积液,早期穿刺置管引流可能有利于改善此类患者的结局指标。

Objective To classified the fluid location of of grade B and C postoperative pancreatic fistula(POPF)and propose processing flow.Methods Data from 232 patients who underwent pancreatic surgery from January 2018 to December 2022 at Department of General Surgery&Hepato-billo-pancreatic,Beijing Hospital were collected retrospectively.Forty-six patients who suffered from grade B and C POPF underwent ultrasound-guided drainage.There were 32 males and 14 females,with an age of(60.2±13.7)years(range:18 to 85 years).The imaging data of postoperative CT were collected and the the fluid location was classified.Then analyzed the drainage status when patents were diagnosed as POPF.Machine learning was performed and a random forest model was applied to construct the relationship between intervention time and mortality.The optimal intervention time was calculated.The patients were then divided into early and late intervention groups and clinical data and outcomes were compared using the t test,Mann-Whitney U test,χ^(2) test or Fisher′s exact test between the two groups.Results Based on the results of the random forest model,the optimal puncture time was within 5.38 days after the diagnosis of POPF.Based on the optimal time,21 patients were subsumed into early intervention group and 25 patients were subsumed into late intervention group.The location of fluid collection was classified into four types:peripancreatic(32.7%,15/46),extra-pancreatic and epigastric(41.3%,19/46),extra-pancreatic and hypogastic(13.0%,6/46)and diffused(13.0%,6/46).The status of the drainage included normal in 10 patients(21.8%),displaced drain in 18 patients(39.1%)and drain removed or blocked in 18 patients(39.1%).The perioperative mortality rate was 19.0%(4/21)in the early intervention group and 8.0%(2/25)in the late.The late intervention group had significantly higher rates of positive drainage fluid cultures(88.0%(22/25)vs.42.9%(10/21),χ^(2)=10.584,P=0.001),secondary surgery(24.0%(6/25)vs.0(0/21),P=0.025),and readmission within 90 days(32.0%(8/25)vs.4.8%(1/21),χ^(2)=5.381,P=0.020)than the early group,and a significantly longer postoperative hospital stay(M(IQR))(24(20)days vs.39(53)days,Z=3.023,P=0.003).Conclusions The location of the POPF fluid collection is classified into four types.Early radiological evaluation can detect abdominal effusion promptly,and early puncture and drainage will be beneficial in improving outcomes in these patents.