结直肠癌化疗患者经外周静脉置入中心静脉相关性血栓的列线图构建及验证

Construction and verification of a nomogram for peripherally inserted central venous thrombosis in patients with colorectal cancer chemotherapy

目的探讨结直肠癌化疗患者经外周静脉置入中心静脉(PICC)相关性血栓(PRT)的危险因素并建立列线图模型。方法选取2020年2月至2022年1月福建省肿瘤医院收治的648例PICC置管的结直肠癌患者作为研究对象,按照3∶1的比例随机分为训练组(n=486)和验证组(n=162)。采用LASSO回归筛选变量和多因素Logistic回归分析训练组患者发生PRT的独立危险因素,并建立列线图模型对两组进行内部和外部验证。结果发生PRT患者与未发生PRT患者糖尿病、置管前D-二聚体水平、置管位置、导管型号比较差异有统计学意义(P<0.05)。使用LASSO交叉验证法计算最佳λ参数,并在最佳λ参数处选择回归系数非零的变量作为潜在的预测因子,结果显示糖尿病、置管前D-二聚体水平、置管位置、导管型号是发生PRT的影响因素。多因素Logistic回归分析结果显示,糖尿病、置管前D-二聚体>0.5mg/L、左侧置管、5Fr型号导管均是发生PRT的独立危险因素(P<0.05)。对训练组和验证组进行内部和外部验证,结果显示,两组的一致性指数分别为0.754和0.774;Hosmer-Lemeshow检验结果分别为χ^(2)=0.744,P=0.384和χ^(2)=6.199,P=0.401;绘制校准曲线显示两组的预测结果和实测结果一致性均良好。结论本研究建立的列线图模型对判断PRT高危人群并采取针对性的预防措施具有重要的指导价值。

Objective To explore the risk factors of peripherally inserted central catheter(PICC)-related thrombosis(PRT)in patients with colorectal cancer undergoing chemotherapy and establish the nomogram model.Methods 648 colorectal cancer patients with PICC catheterization admitted to Fujian Cancer Hospital from February 2020 to January 2022 were were selected as the study subjects,and they were randomly divided into the training group(n=486)and the verification group(n=162)according to the ratio of 3∶1.LASSO regression was used to screen variables and multivariate Logistic regression was used to analyze the independent risk factors of PRT in the training group,and the nomogram model was es-tablished to verify the internal and external validation of the two groups.Results There were significant differences in diabetes mellitus,D-dimer level before catheterization,catheterization position and catheter model between PRT patients and non-PRT patients(P<0.05).The LASSO cross-validation method was used to calculate the optimalλparameter,and variables with non-zero regression coefficients were selected at the optimalλparameter as potential predictors,the results showed that diabetes,D-dimer level before catheterization,catheterization position and catheter model were the influencing factors for PRT.Multivariate Logistic regression analysis showed that diabetes mellitus,D-dimer>0.5 mg/L before catheter-ization,left catheterization and 5Fr catheter were independent risk factors for PRT(P<0.05).The internal and external validation of the training group and the validation group was carried out.The results showed that the consistency indexes of the two groups were 0.754 and 0.774,respective-ly,the results of Hosmer-Lemeshow test wereχ^(2)=0.744,P=0.384 andχ^(2)=6.199,P=0.401,respectively.The calibration curve shows that the predict-ed results of the two groups are in good agreement with the measured results.Conclusion The nomogram model established in this study has im-portant guiding value for judging high-risk groups of PRT and taking targeted preventive measures.