多序列MRI纹理分析预测宫颈癌新辅助化学治疗疗效

Multi-sequence MRI texture analysis for predicting efficacy of neoadjuvant chemotherapy in uterine cervical carcinoma

目的观察不同序列MRI纹理特征分析预测宫颈癌新辅助化学治疗(NACT)疗效的价值。方法回顾性分析32例接受NACT的宫颈癌患者,于NACT开始前及结束后各行1次盆腔MR,根据实体瘤疗效评价标准(RECIST)将患者分为有效组(完全缓解和部分缓解)及无效组(疾病稳定和疾病进展)。于治疗前MR T2WI、DWI及增强图像上分别勾画ROI,获得纹理参数,每组图像共采集106个纹理特征,比较2组参数差异。对每组选取诊断效能较高且相关性小的纹理特征参数进行Logistic回归分析,获得综合参数;绘制受试者工作特征(ROC)曲线,得到各序列单因素及回归模型的预测价值,并对各序列进行比较。结果治疗前T2WI、DWI和增强图像纹理特征中,分别有22、13和36个组间差异存在统计学意义(P均<0.05);T2WI、DWI和增强图像单个纹理特征预测宫颈癌NACT效果的ROC曲线下面积(AUC)分别为0.609~0.839、0.745~0.813及0.552~0.786,综合模型预测疗效的AUC分别为0.839、0.885及0.766。结论不同序列MRI纹理分析预测NACT对于宫颈癌的疗效具有较高价值,以DWI最佳。

Objective To observe the value of multi-sequence MRI texture analysis for predicting efficacy of neoadjuvant chemotherapy(NACT)in uterine cervical carcinoma.Methods A total of 32 cervical carcinoma patients underwent NACT.Pelvic MRI was performed before and after NACT,and the patients were divided into response group(complete response and partial response)and non-response group(stable disease and progressive disease)according to the standards of response evaluation criteria in solid tumors(RECIST).Texture analysis was performed on T2 WI,DWI and contrast enhanced(CE)images before NACT.Totally 106 parameters were obtained for each sequence and compared between the two groups.Multivariate Logistic regression analysis was performed using the top two significant parameters as independent variables and create formula.By drawing the receiver operating characteristic(ROC)curve,the predictive value of single factor and regression model of each sequence were obtained,and the comparison was made among sequences.Results Significant differences of 22 texture features of T2 WI,13 of DWI and 36 of CE were found between response group and non-response group(all P<0.05),and the areas under curve(AUC)using single features for predicting the effect of NACT for cervical cancer were 0.609-0.839,0.745-0.813 and 0.552-0.786,respectively.Multivariate Logistic regression analysis for each sequence showed AUC of 0.839 for T2 WI,0.885 for DWI and 0.766 for CE.Conclusion PreNACT multi-sequence MRI texture analysis has potential in predicting efficacy of NACT in cervical carcinoma,and DWI may be the best.