表观扩散系数直方图在新辅助化疗前预测luminal型乳腺癌疗效的价值  

Magnetic resonance imaging with apparent diffusion coefficient histogram analysis: evaluation of luminal type breast cancer prior to neoadjuvant chemotherapy

作　　者：耿小川[1,2] 华佳[1] 庄治国[1] 陈洁[1] 张科蓓[1] 张庆[1] 成芳[1] GENG Xiaochuan;HUA Jia;ZHUANG Zhiguo;CHEN Jie;ZHANG Kebei;ZHANG Qing;CHENG Fang(Department of Radiology,Renji Hospital,School of Medicine,Shanghai Jiao Tong University,Shanghai 200127,China;Department of Radiology,Renji Hospital South Campus,School of Medicine,Shanghai Jiao Tong University,Shanghai 201112,China)

机构地区：[1]上海交通大学医学院附属仁济医院放射科,上海200127 [2]上海交通大学医学院附属仁济医院南院放射科,上海201112

出　　处：《肿瘤影像学》2019年第6期377-383,共7页Oncoradiology

基　　金：上海申康医院发展中心专科疾病临床“五新”转化项目(16CR3024A);上海市科学技术委员会高新技术领域重点项目(18511102900;18511102901)

摘　　要：目的:探讨表观扩散系数(apparentdiffusioncoefficient,ADC)直方图在新辅助化疗(neoadjuvant chemotherapy,NAC)前预测luminal型乳腺癌疗效的临床价值。方法:回顾性分析NAC前穿刺病理及免疫组织化学证实为luminal型乳腺癌、完善磁共振成像(magnetic resonance imaging,MRI)检查[包括扩散加权成像(diffusion-weighted imaging,DWI)序列]、NAC后手术的患者57例,其中有效组17例。记录乳腺癌的ADC直方图参数,包括平均值、最小值、标准差、最大值、10%、25%、50%、75%及90%百分位数、偏度和峰度。用组内相关系数(intraclass correlation coefficient,ICC)分析观察者间和观察者内一致性。有效组和无效组间各参数的比较采用独立样本t检验。应用受试者工作特征(receiver operating characteristic,ROC)曲线评价ADC直方图参数预测NAC疗效的诊断效能。结果:ADC直方图参数观察者间ICC(0.089~0.845)均低于观察者内(0.306~0.916)。最小值、10%、25%及50%的一致性较好。有效组的最小值、10%和25%的ADC值分别为(0.54±0.36)×10^-3、(0.90±0.23)×10^-3、(1.02±0.23)×10^-3mm^2/s。无效组的最小值、10%和25%的ADC值分别为(0.29±0.28)×10^-3、(0.71±0.20)×10^-3、(0.85±0.19)×10^-3mm^2/s。有效组的最小值、10%和25%的ADC值均高于无效组,差异有统计学意义(P<0.05)。10%的ROC曲线的曲线下面积最大(0.746),以0.746×10^-3 mm^2/s为最佳界值,灵敏度、特异度分别为82.35%、65.00%。但10%的诊断效能未明显高于25%(ROC曲线的曲线下面积0.724,P=0.505)。结论:ADC直方图在NAC前预测luminal型乳腺癌有一定的诊断价值,10%、25%的诊断价值较高且一致性较好。Objective: To determine response prior to neoadjuvant chemotherapy(NAC) in patients with luminal type breast cancer by using magnetic resonance imaging(MRI) apparent diffusion coefficient(ADC) histogram analysis. Methods: Fifty-seven patients with luminal type breast cancer confirmed by core needle biopsy underwent MRI(including diffusion-weighted imaging sequence) before NAC. Surgery was performed after NAC. Seventeen patients were responders. ADC histogram parameters were recorded, including mean, minimum, standard deviation, maximum, 10%, 25%, 50%, 75%, and 90% percentiles, skewness and kurtosis. Inter-observer and intra-observer agreement were analyzed using intraclass correlation coefficient(ICC). ADC histogram parameters between responders and non-responders were compared using an independent sample t test. The receiver operating characteristic(ROC) curve was used to evaluate the diagnostic performance of ADC histogram parameters in the response prediction. Results: The inter-observer ICCs(0.089~0.845) of the ADC histogram parameters were lower than that of the intra-observer(0.306~0.916). The agreement of minimum, 10%, 25% and 50% were excellent. The minimum, 10% and 25% of responders were(0.54±0.36)×10^-3,(0.90±0.23)×10^-3 and(1.02±0.23)×10^-3 mm^2/s. The minimum, 10% and 25% of non-responders were(0.29±0.28)×10^-3,(0.71±0.20)×10^-3 and(0.85±0.19)×10^-3 mm^2/s. The minimum, 10% and 25% of responders were higher than non-responders(P<0.05), and the area under curve(AUC) of the 10% percentile was the largest(0.746). When a cut-off value of 0.746×10^-3 mm^2/s in 10% percentile was used, the sensitivity of 82.35% and specificity of 65.00% were achieved. However, the diagnostic performance of 10% percentile was not significantly higher than 25%(AUC was 0.724, P=0.505). Conclusion: In patients with luminal type breast cancer, the ADC histogram analysis was useful in the response prediction prior to NAC, and 10% and 25% percentiles had better diagnostic performance and excellent agreement.

关 键 词：乳腺癌 新辅助化疗 直方图 磁共振成像 

分 类 号：R737.9[医药卫生—肿瘤] R445.2[医药卫生—临床医学]