基于CT检查列线图模型预测急性胰腺炎合并胰周液体积聚行经皮穿刺置管引流术效果的应用价值

石仕晗; 阮志兵; 陈坤; 文凤; 陈慧淋; 唐戈雅; 徐茂丽; 张宜杰; 梅永

doi:10.3760/cma.j.cn115610-20250902-00567

基于CT检查列线图模型预测急性胰腺炎合并胰周液体积聚行经皮穿刺置管引流术效果的应用价值

Application value of a CT‑based nomogram in predicting the efficacy of percutaneous catheter drainage for acute pancreatitis complicated with peripancreatic fluid collection

摘要

摘要:
探讨基于CT检查列线图模型预测急性胰腺炎（AP）合并胰周液体积聚（PFCs）行经皮穿刺置管引流术（PCD）效果的应用价值。
采用回顾性队列研究方法。收集2023年1月至2025年2月贵州医科大学附属医院收治的60例AP合并PFCs患者的临床和CT检查资料；男43例,女17例；年龄为42（37,56）岁。60例患者均行PCD,其中33例引流成功,设为引流成功组；27例引流失败,设为引流失败组。观察指标：（1）引流成功组和引流失败组患者临床和CT检查资料比较。（2）影响PCD失败的因素分析。（3）预测模型的构建及评价。正态分布的计量资料组间比较采用独立样本t检验,偏态分布的计量资料组间比较采用Mann‑Whitney U检验。计数资料组间比较采用χ²检验或Fisher确切概率法。单因素分析和多因素分析采用Logistic回归模型。根据多因素分析结果构建列线图预测模型,绘制受试者工作特征曲线（ROC）评估模型的预测效能,采用校准曲线评估模型的拟合效能,决策曲线评估模型的临床应用价值。
（1）引流成功组和引流失败组患者临床和CT检查资料比较。引流成功组和引流失败组患者住院时间、PFCs均质性、PFCs体积、胰腺坏死体积比较,差异均有统计学意义（Z=2.654,χ²=10.909、8.148、4.949,P<0.05）,胰腺出血比较,差异有统计学意义（P<0.05）,改良CT严重指数（MCTSI）评分比较,差异有统计学意义（χ²=10.884,P<0.05）。（2）影响PCD失败的因素分析。单因素分析结果显示：PFCs均质性、PFCs体积、胰腺坏死体积及MCTSI评分是影响AP合并PFCs患者行PCD失败的相关因素（优势比=8.500,4.750,3.333,6.250,95%可信区间为2.137~33.815,1.584~14.245,1.134~9.801,2.021~19.324,P<0.05）。多因素分析结果显示：PFCs均质性、PFCs体积、MCTSI评分是AP合并PFCs患者行PCD失败的独立影响因素（优势比=4.818,4.159,4.755,95%可信区间为1.023~22.700,1.149~15.050,1.061~21.316,P<0.05）。（3）预测模型的构建及评价。纳入PFCs均质性、PFCs体积、MCTSI评分构建AP合并PFCs患者行PCD效果的列线图预测模型。3项因素评分总和越高,PCD失败概率越大。ROC结果显示：曲线下面积为0.829（95%可信区间为0.725~0.933,P<0.05）,灵敏度、特异度分别为88.9%、66.7%。校准曲线结果显示：模型的预测概率与实际概率拟合度良好。决策曲线结果显示：阈值为0.4,模型的总体净收益较好。
PFCs均质性、PFCs体积、MCTSI评分是AP合并PFCs患者行PCD失败的独立影响因素。基于此构建的列线图模型可用于预测AP合并PFCs患者行PCD的效果。

Abstract:
Objective To explore the application value of a computed tomography (CT)⁃based nomogram in predicting the efficacy of percutaneous catheter drainage (PCD) for acute pancrea⁃titis (AP) complicated with peripancreatic fluid collections (PFCs).
Methods The retrospective cohort study was conducted. The clinical and CT data of 60 patients of AP complicated with PFCs who were admitted to The Affiliated Hospital of Guizhou Medical University from January 2023 to February 2025 were collected. There were 43 males and 17 females, aged 42(37,56)years. All 60 patients underwent PCD, among whom 33 cases with successful drainage were set as the successful drainage group, and 27 cases with failed drainage were set as the failed drainage group. Observation indicators: (1) comparison of clinical and CT imaging data between patients of the successful drainage group and failed drainage group; (2) analysis of factors influencing PCD failure; (3) construction and evaluation of the prediction model. Comparison of measurement data with normal distribution between groups was conducted using the independent sample t test. Comparison of measurement data with skewed distribution between groups was conducted using the Mann-Whitney U test. Comparison of count data between groups was conducted using the chi‑square test or Fisher exact probability. Univariate and multivariate analyses were conducted using the Logistic regression model. A nomogram prediction model was constructed based on the results of multi-variate analysis. Receiver operating characteristic (ROC) curve was drawn to evaluate the predictive efficacy of the model. The calibration curve was used to assess the fitting effect, and decision curve was used to determine the clinical application value.
Results (1) Comparison of clinical and CT imaging data between patients of the successful drainage group and failed drainage group. There were significant differences in duration of hospital stay, PFCs homogeneity, PFCs volume and pancreatic necrosis volume between patients of the successful drainage group and failed drainage group (Z=2.654, χ²=10.909, 8.148, 4.949, P<0.05). There was a significant difference in pancreatic hemorrhage between patients of the successful drainage group and failed drainage group (P<0.05), and a signifi-cant difference in modified CT severity index (MCTSI) score between patients of the successful drainage group and failed drainage group (χ²=10.884, P<0.05). (2) Analysis of factors influencing PCD failure. Results of univariate analysis showed that PFCs homogeneity, PFCs volume, pancreatic necrosis volume, and MCTSI score were influencing factors for PCD failure in patients of AP complicated with PFCs (odds ratio=8.500, 4.750, 3.333, 6.250, 95% confidence interval as 2.137-33.815, 1.584-14.245, 1.134-9.801, 2.021-19.324, P<0.05). Results of multivariate analysis showed that PFCs homogeneity, PFCs volume, and MCTSI score were independent factors for PCD failure in patients of AP complicated with PFCs (odds ratio=4.818, 4.159, 4.755, 95% confidence interval as 1.023-22.700, 1.149-15.050, 1.061-21.316, P<0.05). (3) Construction and evaluation of the prediction model. The nomogram model for predicting PCD efficacy in patients of AP complicated with PFCs was constructed by incor-porating PFCs homogeneity, PFCs volume, and MCTSI score. A higher total score of the three factors indicated a higher probability of PCD failure. ROC curve analysis showed an area under the curve of 0.829 (95% confidence interval as 0.725-0.933, P<0.05), with a sensitivity of 88.9% and specificity of 66.7%. Calibration curve demonstrated good consistency between the predicted probability of the model and the actual probability. Decision curve analysis revealed that the model achieved a favor-able overall net benefit when the threshold probability was 0.4.
Conclusions PFCs homogeneity, PFCs volume, and MCTSI score are independent factors for PCD failure in patients of AP complicated with PFCs. The nomogram model constructed based on these factors can be used to predict the efficacy of PCD in patients of AP complicated with PFCs.

HTML全文

参考文献(37)

施引文献

资源附件(0)