肝细胞癌术后严重并发症发生线上预测模型建立与评价

Constructing and validating an online predictive model of severe postoperative complication for hepatocellular carcinoma

目的探讨肝细胞癌(HCC)术后严重并发症发生的高危因素,并构建线上预测模型以帮助临床医务工作者快速识别高危人群并予以干预措施,从而促进患者术后早期康复。方法回顾性分析中山大学孙逸仙纪念医院2009年12月至2017年12月347例接受肝癌切除术的HCC患者的临床病历资料,基于Logistic回归分析明确术后出现严重并发症的高危因素,并以此为基础构建线上预测模型,通过一致性指数(C指数)及校正曲线等验证模型预测效能。结果共39例患者术后出现严重并发症,其术后平均住院时长(19±10)天。经过多因素回归分析提示伴发慢性病,吸烟酗酒史,术前血清ALT及GGT水平,癌灶数目与手术时间是HCC术后出现严重并发症的危险因素(P<0.05)。以多因素分析结果为基础构建线上预测模型。经内部验证,该预测模型C指数为0.845,且校正曲线提示该模型具有良好的预测性能。结论HCC术后严重并发症的发生是围术期死亡的重要原因,通过本研究所建立的线上预测模型,临床医务工作者可快速简便识别高危人群并予以干预措施,可有效缩短术后康复时间。

Objective To construct and validate an online predictive model of severe postoperative complications for hepatocellular carcinoma(HCC).Methods The clinical data of 347 HCC patients who underwent resection from December 2009 to December 2017 in Sun Yat-sen Memorial Hospital of Sun Yat-sen University were retrospectively analyzed.The significantly independent factors of severe postop-erative complications were identified according to the logistic regression analysis,and then an online model was constructed.A concordance index(C-index)and calibration curve were used to validate the performance and clinical utility of the model.Results A total of 39 cases had severe complications after hepatectomy and the mean postoperative hospital stay was 19±10 days.Multivariate regression analysis suggested that concomitant disease,smoking/drinking,preoperative serum ALT and GGT levels,surgery time,and the number of tumors were risk factors for severe complications(P<0.05).Based on the results of the multi-factor analysis,an online prediction model was developed.After internal verification,the C-index of the developed model was 0.845.Conclusion Severe postoperative complications tend to increase the risk of perioperative mortality for HCC.Clinicians can figure out those patients at high risk of severe postoperative complications more efficiently by using the constructed model,which can effectively shorten postoperative hospital stay and improve short-term outcomes.