From the Departments of Radiology (O.M., S.L., F.Z.M., H.R., P.O.) and Gastroenterology (B.B., L.B.), CHU Toulouse Rangueil, Avenue du Professeur Jean Poulhès, 31400 Toulouse, France; and Inserm/UPS UMR 1048, I2MC Team 10, Toulouse, France (O.M.).
Radiology. 2015 Jul;276(1):119-28. doi: 10.1148/radiol.15141494. Epub 2015 Feb 2.
To determine the volume of extrapancreatic necrosis that predicts severe acute pancreatitis and to assess the reliability of this threshold in predicting severe acute pancreatitis compared with current scoring systems and C-reactive protein (CRP) levels.
This institutional review board-approved, HIPAA-compliant retrospective study included patients with acute pancreatitis who were examined with computed tomography (CT) 2-6 days after disease onset. Extrapancreatic necrosis volume, Balthazar score, and CT severity index (CTSI) were calculated. CRP levels 48 hours after the onset of symptoms were reviewed. Outcome parameters included organ failure, infection, need for surgery or percutaneous intervention, duration of hospitalization, and/or death. Receiver operating characteristic (ROC) curves were constructed to determine the optimal threshold for predicting clinical outcomes. Pairwise comparisons of areas under ROC curves (AUCs) from the different grading systems were performed. Interobserver and intraobserver agreement in the grading of extrapancreatic necrosis was assessed by using κ statistics.
In 264 patients, significant relationships were found between extrapancreatic necrosis volume and organ failure, infection, duration of hospitalization, need for intervention, and death (P < .001 for all). The optimal threshold for predicting severe acute pancreatitis was 100 mL. Sensitivity and specificity were 95% (19 of 20) and 83% (142 of 172), respectively, for predicting organ failure (vs 100% [20 of 20] and 46% [79 of 172] for the Balthazar score and 25% [five of 20] and 95% [163 of 172] for the CTSI). The extrapancreatic necrosis AUC was the highest for all systems. Interobserver and intraobserver agreement based on the 100-mL threshold was considered to be excellent.
A simple grading system based on an objective criterion such as a threshold of 100 mL of extrapancreatic necrosis provides more reliable information for predicting acute pancreatitis outcomes than do the current scoring systems.
确定预测重症急性胰腺炎所需的胰外坏死体积，并评估该阈值与当前评分系统和 C 反应蛋白（CRP）水平相比预测重症急性胰腺炎的可靠性。
本研究经机构审查委员会批准，符合 HIPAA 规定，为回顾性研究，纳入了发病后 2-6 天接受计算机断层扫描（CT）检查的急性胰腺炎患者。计算胰外坏死体积、Balthazar 评分和 CT 严重指数（CTSI）。回顾症状发作后 48 小时的 CRP 水平。观察终点包括器官衰竭、感染、需要手术或经皮介入、住院时间和/或死亡。构建受试者工作特征（ROC）曲线以确定预测临床结局的最佳阈值。通过 ROC 曲线下面积（AUC）的两两比较来比较不同评分系统的预测效能。采用κ 统计评估胰外坏死分级的观察者间和观察者内一致性。
在 264 例患者中，胰外坏死体积与器官衰竭、感染、住院时间、干预需求和死亡之间存在显著相关性（均 P <.001）。预测重症急性胰腺炎的最佳阈值为 100ml。预测器官衰竭的敏感性和特异性分别为 95%（20/21）和 83%（142/172），而 Balthazar 评分的敏感性和特异性分别为 100%（20/20）和 46%（79/172），CTSI 的敏感性和特异性分别为 25%（5/20）和 95%（163/172）。所有系统中，胰外坏死的 AUC 最高。基于 100ml 阈值的观察者间和观察者内一致性被认为是极好的。
基于客观标准（如 100ml 胰外坏死阈值）的简单分级系统可为预测急性胰腺炎结局提供比当前评分系统更可靠的信息。
Zotero 插件