Department of Radiology, QingPu Branch of Zhongshan Hospital Affiliated to Fudan University, No. 1158 Park East Road, Qingpu District, ShangHai, China.
Department of Radiology, QingPu Hospital of Traditional Chinese Medicine, No. 95 Qing'an Road, Qingpu District, ShangHai, China.
Clin Radiol. 2024 Feb;79(2):e256-e263. doi: 10.1016/j.crad.2023.10.034. Epub 2023 Nov 17.
To assess the association of quantitative computed tomography (CT) features on admission with acute pancreatitis (AP) severity, and to explore the performance of combined CT and laboratory markers for predicting severe AP (SAP).
Data from 208 AP patients were reviewed retrospectively. Pancreas volume, the area of extrapancreatic inflammation, extrapancreatic fluid collection volume, and number were calculated based on CT images on admission. Laboratory biomarkers within 24 h of admission were collected. Interobserver agreement for CT measurements was measured by calculating interclass correlation coefficient (ICC). The associations of quantitative CT features with AP severity were evaluated. Predictive models for SAP were constructed based on CT and laboratory markers. Performances of single marker and the models were evaluated using receiver operating characteristic (ROC) curve and area under the ROC curve (AUC).
Pancreas volume, area of extrapancreatic inflammation, extrapancreatic fluid collection volume, and number were significantly different between severe and non-severe AP groups. In predicting SAP, the AUCs of quantitative CT indicators ranged from 0.72 to 0.79; the AUCs of laboratory biomarkers were between 0.53 and 0.66. The combined model of area of extrapancreatic inflammation, serum calcium, and haematocrit yielded an AUC of 0.84, significantly higher than that of the laboratory model, single CT, or laboratory marker. Interobserver agreements for quantitative CT indicators were excellent, with ICC ranging from 0.91 to 0.98.
Quantitative CT features on admission were significantly associated with AP severity; the combination of extrapancreatic inflammation area, serum calcium, and haematocrit could be taken as a new method for predicting SAP.
评估入院时定量计算机断层扫描（CT）特征与急性胰腺炎（AP）严重程度的相关性，并探讨 CT 与实验室标志物联合预测重症急性胰腺炎（SAP）的效能。
回顾性分析 208 例 AP 患者的数据。根据入院时的 CT 图像计算胰腺体积、胰外炎症面积、胰外积液量和数量。收集入院 24 小时内的实验室生物标志物。采用组内相关系数（ICC）评估 CT 测量的观察者间一致性。评估定量 CT 特征与 AP 严重程度的相关性。基于 CT 和实验室标志物构建 SAP 预测模型。采用受试者工作特征（ROC）曲线和 ROC 曲线下面积（AUC）评估单标记物和模型的性能。
胰腺体积、胰外炎症面积、胰外积液量和数量在重症和非重症 AP 组之间存在显著差异。在预测 SAP 时，定量 CT 指标的 AUC 范围为 0.72 至 0.79；实验室标志物的 AUC 范围为 0.53 至 0.66。胰外炎症面积、血清钙和红细胞压积联合模型的 AUC 为 0.84，明显高于实验室模型、单一 CT 或实验室标志物。定量 CT 指标的观察者间一致性极好，ICC 范围为 0.91 至 0.98。
入院时定量 CT 特征与 AP 严重程度显著相关；胰外炎症面积、血清钙和红细胞压积的联合可作为预测 SAP 的新方法。
