检索历史 清除

目的:建立肝癌肝切除术后肝功能衰竭（PHLF）风险预测模型。方法:回顾性病例对照研究。收集2013年1月1日至2023年12月31日海军军医大学附属东方肝胆外科医院320例接受肝切除肝癌患者的临床资料及实验室结果，包括性别、年龄及患者术前18项实验室指标。根据手术时间进行分组，训练组252例，术后发生与未发生肝衰的病例分别为62和190例，验证组68例，术后发生与未发生肝衰的病例分别为34和34例。运用二元Logistic回归分析对性别、年龄及术前18项实验室指标进行单因素分析，对有意义的结果再逐步做多因素分析，确定肝癌肝切除术后肝衰竭的影响因素，建立Logistic回归模型。结果:在训练组中，与肝癌肝切除术后肝衰竭显著相关的指标包括年龄（ P=0.016）、血小板（ P=0.005）、前白蛋白（ P<0.001）和碱性磷酸酶（ P<0.001），结合这4个指标使用Logistic回归构建列线图模型并绘制校准曲线，在训练组中列线图模型在预测肝癌肝切除术后肝衰竭风险显示出良好的鉴别能力，曲线下面积为0.82（ 95%CI 0.76~0.88），以0.264 6为最佳截断值，敏感度为73%，特异度为80%。在验证组中，列线图模型的预测效能与训练组相当，曲线下面积为0.81（ 95%CI 0.71~0.92），敏感度为82%，特异度为77%。 结论:术前血小板和前白蛋白降低、碱性磷酸酶升高及高龄行肝切除术后易发生肝功能衰竭，应用术前检验数据构建的列线图模型在预测肝癌肝切除术后肝衰竭显示出良好的鉴别能力和准确性。

Objective:To establish a risk prediction model of liver failure after liver resection for hepatocellular carcinoma.Method:A retrospective case-control study was designed. Clinical data and laboratory results, including gender, age, and preoperative 18 laboratory indicators, were collected from 320 patients with hepatocellular carcinoma undergoing liver resection in Eastern Hepatobiliary Surgery Hospital Affiliated to Naval Medical University from January 1, 2013 to December 31, 2023. According to the surgical time, 252 cases in the training cohort were divided into 62 and 190 cases with and without postoperative liver failure, respectively. Of the 68 cases in validation cohort, 34 developed postoperative liver failure and 34 did not. Binary Logistic regression analysis was used to conduct univariate analysis of gender, age, and 18 preoperative laboratory indicators, and multivariate analysis was carried out for significant results to determine the influencing factors of liver failure after liver resection for hepatocellular carcinoma, and Logistic regression model was established.Result:In the training cohort, indicators significantly associated with liver failure after liver resection for hepatocellular carcinoma included age ( P=0.016), platelets ( P=0.005), prealbumin ( P<0.001), and alkaline phosphatase ( P<0.001). Logistic regression was used to construct a nomogram model and draw a calibration curve by combining these four indicators. In the training cohort, the nomogram model showed good discriminability in predicting the risk of liver failure after hepatectomy for hepatocellular carcinoma. The area under the curve of was 0.82 (95% CI 0.76-0.88), and the sensitivity was 73% and specificity was 80% when the optimal cut-off value was 0.2646. In the validation cohort, the predictive performance of the nomogram model was comparable to that of the training cohort, with an area under the curve of 0.81 (95% CI 0.71-0.92), sensitivity of 82%, and specificity of 77%. Conclusion:Preoperative platelet and prealbumin decreases, alkaline phosphatase increases, and elderly patients are prone to liver failure after liver resection. The nomogram model constructed with preoperative test data has shows good discriminatory ability and accuracy in predicting liver failure after liver resection for hepatocellular carcinoma.