摘要目的:探讨乳腺癌患者术前经常规动态对比增强MRI（DCE-MRI）扫描后获得的肿瘤血流动力学及形态学特征预测术后复发时间的价值。方法:回顾性分析2012年11月至2014年12月辽宁省肿瘤医院术后复发的乳腺癌患者58例，依据复发时间分为早期复发（术后≤2年）组33例，晚期复发（术后>2年）组25例。所有患者均在术前行常规DCE-MRI扫描，通过计算机提取肿瘤三维容积内的血流动力学特征及在每一个期相下的肿瘤形态学特征及纹理特征。早期复发组和晚期复发组患者间计数资料和计量资料的比较分别采用Fisher精确概率法和Mann-Whitney U检验，绘制受试者操作特征（ROC）曲线，应用多元logistic回归计算特征联合预测早期复发与晚期复发的诊断效能。应用Kaplan-Meier法分析生存预后，并用Log-Rank检验比较各组生存曲线的差异。 结果:早期复发组和晚期复发组患者背景实质强化、病灶边缘、病灶内部强化特征、病灶形态、时间信号曲线类型和全乳血管增加程度的差异均无统计学意义（ P均>0.05）；2组患者的对比剂最大浓度值（Max Conc）、时间信号曲线下面积（AUC）及时间信号曲线最大斜率值（Max Slope）值差异有统计学意义（ P<0.05）。对比分析8期DCE-MRI影像组学特征参数，第3期形态特征参数球度（sphericity）在早期复发与晚期复发组间差异有统计学意义（ P=0.03）。AUC、Max Conc、Max Slope和第3期形态特征参数球度预测早期与晚期复发的ROC曲线下面积分别为0.664、0.659、0.684、0.670，上述4个参数联合预测的ROC曲线下面积为0.765，特异度为63.6%，灵敏度为84.0%，预测效能高于单变量。58例患者随访时间17~64个月，中位随访时间47个月。早期复发组的无进展生存和总生存均明显低于晚期复发组，差异有统计学意义（ P<0.05）。 结论:基于术前无创性常规DCE-MRI获取的肿瘤血流动力学特征联合形态学特征预测乳腺癌患者的术后复发时间有一定的价值。

abstractsObjective:To investigate the value of tumor hemodynamics and morphological features from conventional dynamic contrast-enhanced MRI (DCE-MRI) scan before surgery in predicting postoperative recurrence time in breast cancer patients.Methods:A retrospective analysis of 58 patients with breast cancer who had recurred after operation from November 2012 to December 2014 in Liaoning Cancer Hospital was performed. According to the recurrence time, the patients were divided into early recurrence group (≤2 years after surgery, 33 cases) and late recurrence group (>2 years after surgery, 25 cases). All patients underwent routine DCE-MRI scans before surgery, and hemodynamic features of the three-dimensional volume of the tumor and the morphological and textural features of the tumor in each phase were extracted by computer. The counts and measurement data of patients in early recurrence group and late recurrence group were compared by Fisher′s exact probability method and Mann-Whitney U test, and receiver operating characteristic (ROC) curves were drawn. The multivariate logistic regression was used to calculate the combined efficacy in predicting early recurrence and late recurrence. Kaplan-Meier method was used to analyze the survival prognosis, and Log-Rank test was used to compare the differences in survival curves between groups. Results:There was no significant difference in background parenchymal enhancement, lesion margin, lesion internal enhancement characteristics, lesion morphology, time-signal intensity curve type and the degree of whole-breast vascularity increase between early recurrence and late recurrence groups ( P>0.05).There were significant differences in the maximum concentration of contrast (Max Conc), the area under the time signal curve (AUC) and the maximum slope value of the time signal curve (Max Slope) ( P<0.05). Comparative analysis of the radiomics parameters of 8 phases DCE-MRI found that the sphericity of morphological characteristic parameters in the phase 3 was statistically different between the early recurrence and late recurrence groups ( P=0.03). Area under the ROC curve of AUC, Max Conc, Max Slope and parameter sphericity of phase 3 morphological characteristics for predicting early and late recurrence were 0.664, 0.659, 0.684 and 0.670, respectively. The area under the ROC combined with the above four parameters for prediction was 0.765, with a specificity of 63.6% and a sensitivity of 84.0%; the predictive efficacy was higher than that of univariate. Fifty-eight patients were followed up for 17 to 64 months with a median follow-up of 47 months. The disease-free survival and overall survival in the early recurrence group were significantly lower than those in the late recurrence group, and the difference was statistically significant ( P<0.05). Conclusion:It is of certain value to predict the postoperative recurrence time of breast cancer based on the tumor hemodynamic characteristics combined with morphological characteristics from preoperative non-invasive conventional DCE-MRI.