Development and clinical value of a model for predicting the risk of liver metastasis based on ultrasound blood flow imaging of primary colorectal cancer lesions combined with ultrasound quantitative scoring
摘要
目的 通过检测结直肠癌原发灶的超声血流成像指标及超声量化评分,构建肝转移风险预测模型,评估该模型对结直肠癌肝转移的预测价值及临床应用意义.方法 选取2021年2月至2026年2月四川护理职业学院附属医院(四川省第三人民医院)及成都医学院第二附属医院(核工业四一六医院)收治的120例结直肠癌患者,将发生肝转移者纳入转移组(n=42),未转移者纳入未转移组(n=78),均进行超声量化评分、超声血流成像检测,分析超声血流成像指标,包括动脉舒张期末流速(EDV)、阻力指数(RI)、收缩期峰值流速(PSV)及超声量化评分;并采用Spearman法分析超声血流成像、超声量化评分与结直肠癌肝转移的关系,通过单因素及多因素logistic回归模型筛选肝转移的独立危险因素;用Bootstrap抽样验证、ROC曲线、校准曲线及决策曲线分析,评估模型临床应用价值.结果 转移组RI(0.59±0.14)低于未转移组(0.94±0.16),且转移组 PSV、EDV、超声量化评分分别为(23.07±6.32)cm/s、(4.30±0.95)cm/s、(3.83±0.91)分,高于未转移组[分别为(19.64±1.19)cm/s、(2.75±0.60)cm/s、(2.24±0.70)分],差异有统计学意义(P<0.05);多因素 logistic 回归分析显示,RI(OR=3.271)、PSV(OR=3.796)、EDV(OR=4.179)、超声量化评分(OR=2.832)是结直肠癌发生肝转移的相关因素;Spearman相关分析显示RI与肝转移呈负相关(r=-0.418,P<0.05),PSV、EDV、超声量化评分与肝转移呈正相关(r=0.527、0.347、0.409,P<0.05);ROC曲线显示,模型预测CRLM的AUC为0.844(95 CI:0.834~0.935),准确度86.67%,灵敏度83.33%、特异度88.46%;经Bootstrap法重度抽样1 000次内部验证,校正C指数为0.844,提示模型区分度良好.Hosmer Lemeshow检验(x2=2.836,P=0.085)提示列线图模型的拟合度较好;其校准曲线基本接近斜率为1的对角虚线,与理想曲线接近,表明模型的校准度较好,决策曲线显示列线图模型对结直肠癌肝转移风险预测的净获益值较高.结论 超声血流成像与超声量化评分联合检测结直肠癌原发灶可提高诊断效能,对预测肝转移风险的临床应用价值良好.
Objective By analyzing ultrasound blood flow imaging parameters and quantitative scores from primary colorectal tumors,we developed a predictive model for the risk of liver metastasis to evaluate the model's predictive value and clinical significance for colorectal cancer liver metastasis.Methods 120 patients with colorectal cancer admitted to Affiliated Hospital of Sichuan Nursing Vocational College(the Thired People's Hospital of Sichuan Province)and the Second Affiliated Hospital of Chengdu Medical College(China National Nuclear Corporation 416 Hospital)between February 2021 and February 2026 were selected.Patients who developed liver metastases were assigned to the metastasis group(n=42),while those without metastases were assigned to the non-metastasis group(n=78).All patients underwent ultrasound quantitative scoring and ultrasound blood flow imaging.The ultrasound blood flow imaging parameters,including end-diastolic velocity(EDV),resistance index(RI),peak systolic velocity(PSV)and ultrasound quantitative score,were analyzed.Spearman's correlation analysis was used to examine the relationship between these parameters and liver metastasis,and independent risk factors were identified by univariate and multivariate logistic regression.The clinical utility of the model was evaluated using Bootstrap validation,receiver operating characteristic(ROC)curves,calibration curves and decision curve analysis.Results The RI in the metastasis group(0.59±0.14)was lower than that in the non-metastasis group(0.94±0.16),while the PSV,EDV and ultrasound quantitative scores in the metastasis group were(23.07±6.32)cm/s,(4.30±0.95)cm/s and(3.83±0.91)points,respectively,higher than those in the non-metastasis group[(19.64±1.19)cm/s,(2.75±0.60)cm/s and(2.24±0.70)points].These differences were statistically significant(P<0.05).Multivariate logistic regression analysis revealed that RI(OR=3.271),PSV(OR=3.796),EDV(OR=4.179),and ultrasound quantitative score(OR=2.832)were independent factors associated with liver metastasis in colorectal cancer.Spearman's correlation analysis revealed that RI was negatively correlated with liver metastasis(r=-0.418,P<0.05),while PSV,EDV and the ultrasound quantitative score were positively correlated with liver metastasis(r=0.527,0.347,and 0.409,respectively;all P<0.05).The predictive model was presented as a nomogram,the ROC curve shows that the model's AUC for predicting CRLM was 0.844(95%CI:0.834~0.935),with an accuracy of 86.67%,a sensitivity of 83.33%,and a specificity of 88.46%.After 1 000 rounds of internal validation using the bootstrap mothod,the adjusted C-index was 0.844,indicating that the model has good discriminatory power.The Hosmer-Lemeshow test(x2=2.836,P=0.085)suggested a good fit for the nomogram model.The calibration curve closely approximated the diagonal dashed line(slope=1)and was near the ideal curve,suggesting good calibration.Decision curve analysis showed that the nomogram provided a high net benefit for predicting the risk of liver metastasis in colorectal cancer.Conclusion The combined use of ultrasound blood flow imaging and ultrasound quantitative scoring for the detection of primary colorectal cancer lesions can improve diagnostic efficacy and has good clinical value for predicting the risk of liver metastasis.