MRI列线图模型对软组织肉瘤病理分级预测价值

VALUE OF PREOPERATIVE NOMOGRAM MODEL BASED ON MRI FEATURES FOR PREDICTING THE PATHOLOGICAL GRADE OF SOFT TISSUE SARCOMA

目的探讨基于MRI影像学特征的列线图模型在术前预测软组织肉瘤(STS)病理分级的价值。方法回顾性收集137例经术后病理证实的STS病人的术前MRI资料。根据法国癌症中心联合会组织学分级系统,低级别STS(Ⅰ~Ⅱ级)82例,高级别STS(Ⅲ级)55例。采用单因素和多因素Logistic回归分析筛选STS病理分级预测因子,纳入并建立预测模型,生成列线图。基于10折交叉验证对模型性能进行评估,采用中位受试者工作特征曲线下面积(AUC)及中位准确度评估模型的预测效能。结果单因素和多因素Logistic回归分析显示,N分期、深度、T_(2)WI信号异质性和瘤周强化为STS病理分级预测因子。预测模型的中位AUC为0.898,中位准确度为82.1%。结论基于术前MRI影像学特征的列线图模型可有效预测STS病理分级。

Objective To investigate the value of the preoperative magnetic resonance imaging(MRI)features-based nomogram model for predicting the pathological grade of soft tissue sarcoma(STS).Methods Preoperative MRI data from 137 patients with STS confirmed by postoperative pathology were retrospectively collected.According to the French Federation of Cancer Centers Histologic Grading System,82 patients were defined as low-grade STS(gradesⅠ-Ⅱ),and 55 patients were defined as high-grade STS(gradeⅢ).Univariate and multivariate Logistic regression analyses were applied to screen out the predictors for the pathological grade of STS.The MRI features-based predictive nomogram model was established based on the predictors.The 10-fold cross validation was applied to evaluate the model performance,and the median area under the receiver operating characte-ristic curve(AUC)and median accuracy were used to evaluate the prediction efficiency of the model.Results According to the result of univariate and multivariate Logistic regression analyses,N-stage,depth,heterogeneous signal intensity at T_(2)WI,and pe-ritumoral enhancement were identified as predictors for the pathological grade of STS.The predictive nomogram model showed the median AUC value of 0.898 and median accuracy of 82.1%.Conclusion The preoperative MRI-features-based nomogram model can predict the pathological grade of STS effectively.