初发急性胰腺炎患者短期复发的预测因素及列线图模型构建

王文杰.初发急性胰腺炎患者短期复发的预测因素及列线图模型构建[J].内科急危重症杂志,2026,32(1):26-29

DOI：10.11768/nkjwzzzz20260106

英文关键词:

基金项目:

作者	单位	E-mail
王文杰		lyong133@hotmail.com

摘要点击次数: 53

全文下载次数: 62

中文摘要:

摘要目的：探讨初发急性胰腺炎（AP）患者短期复发的预测因素并构建列线图模型。方法：回顾性分析343例初发AP患者的临床资料。根据患者1年内是否复发分为复发性AP（RAP）组和非RAP组，比较2组患者临床资料。采用Spearman相关性分析及竞争风险模型筛选患者复发的预测因素，并构建列线图模型。绘制受试者工作特征（ROC）曲线，并计算曲线下面积（AUC），通过校准曲线评估列线图模型预测效能，采用决策曲线分析（DCA）评估列线图模型的临床适用性。结果：5.5%（19/343）的初发AP患者6个月内复发，15.7%（54/343）的初发AP患者1年内复发。与非RAP组患者相比，RAP患者组更年轻[44.0（34.0，59.0）岁 vs 51.0（36.0，69.0）岁，Z=1.969，P=0.049）]，饮酒[17（31.5%） vs 48（16.6%）， χ2=6.552，P=0.010]、胰腺坏死[7（13.0%） vs 13（4.5%），χ2=5.937，P=0.015]比例更高，甘油三酯（TG）水平更高[7.9（2.0，24.0）mmol/L vs 2.0（1.0，3.4）mmol/L，Z=6.009，P<0.001]。Spearman相关性分析显示，复发与年龄（r=-0.106，P=0.049）呈负相关，与饮酒（r=0.138，P=0.010）、胰腺坏死（r=0.132，P=0.015）及TG（r=0.325，P<0.001）呈正相关。竞争风险模型分析显示，饮酒（HR=2.347，95%CI：1.306～4.217，P=0.004）、胰腺坏死（HR=2.747，95%CI：1.230～6.139，P=0.014）及TG（HR=1.104，95%CI：1.084～1.124，P<0.001）为初发AP患者复发的独立预测因素。ROC曲线分析显示，列线图模型预测初发AP患者6个月内复发的AUC为0.827（95%CI：0.712～0.941），预测1年内复发的AUC为0.817（95%CI：0.748～0.886）。校准曲线分析显示，列线图模型的预测结果和实际结果之间具有良好的一致性。DCA分析表明，列线图模型具有良好的临床适用性。结论：列线图模型可作为初发AP患者复发的良好预测方法。

英文摘要:

Abstract Objective: To investigate the predictors of short-term recurrence after first-episode acute pancreatitis (AP) and construct a nomogram model. Methods: The clinical data of 343 AP patients were retrospectively analyzed. Patients were divided into recurrent AP (RAP) group and non-RAP group according to whether recurrence occurred within 1 year, and the difference of clinical data between the two groups was compared. Spearman correlation analysis and competing risk model were applied to screen the predictors of recurrence, and the nomogram model was constructed based on the predictors. The area under the curve (AUC) was calculated using receiver operating characteristic (ROC) curve to assess predictive performance of nomogram model, calibration curves were plotted to compare the predicted probability of the nomogram with the actual probability, and the clinical utility of the nomogram model was assessed using decision curve analysis (DCA). Results: 5.5% (19/343) of the first-episode AP patients experienced recurrence within 6 months and 15.7% (54/343) experienced recurrence within 1 year. Compared with the non-RAP group, patients in RAP group were younger [44.0 (34.0, 59.0) years vs 51.0 (36.0, 69.0) years, Z=1.969, P=0.049)], had high proportion of alcohol consumption [17 (31.5%) vs 48 (16.6%), χ2=6.552, P=0.010 ], pancreatic necrosis [7 (13.0%) vs 13 (4.5%), χ2=5.937, P=0.015)], and elevated triglycerides [7.9 (2.0, 24.0) mmol/L vs 2.0 (1.0, 3.4) mmol/L, Z=6.009, P< 0.001]. Spearman correlation analysis showed that recurrence was negatively correlated with age (r=-0.106, P=0.049), and positively correlated with alcohol consumption (r= 0.138, P= 0.010), pancreatic necrosis (r= 0.132, P= 0.015), and triglycerides (r= 0.325, P< 0.001). Competing risk model showed that alcohol consumption (HR=2.347, 95%CI: 1.306~4.217, P=0.004), pancreatic necrosis (HR=2.747, 95%CI: 1.230~6.139, P=0.014), and triglycerides (HR=1.104, 95%CI: 1.084~1.124, P<0.001) were independent predictors of recurrence. ROC curve analysis showed that the AUC of nomogram model for predicting recurrence within 6 months was 0.827 (95%CI: 0.712~0.941), and 0.817 (95%CI: 0.748~0.886) for predicting recurrence within 1 year. The results of risk stratification by X-tile software showed that ≤30 points were classified as low risk, 31-55 points as medium risk, and ≥56 points as high risk for recurrence. The calibration curve demonstrated a good agreement between prediction and observation. DCA revealed that the nomogram model achieved good clinical utility. Conclusion The nomogram model can be used as a good predictive method for short-term recurrence of the first-episode acute AP patients.