【作者】 赵莉芸；

【导师】 周纯武；

【作者基本信息】 北京协和医学院 ， 影像医学与核医学（专业学位）， 2014， 博士

【摘要】 第一部分动态增强MRI预测乳腺癌新辅助化疗疗效目的探讨动态对比增强MRI(DCE-MRI)早期预测乳腺癌新辅助化疗(NAC)疗效的可行性。资料与方法收集2011年5月～2014年1月计划在我院行NAC的乳腺癌患者共112例,年龄23～67岁,中位年龄48岁,于化疗前和NAC2周期后行DCE-MRI扫描,分别采用ROIw,ol。法和ROIhs法测量肿瘤三个方位最大径、半定量参数最大上升斜率(MSI)、第二期强化程度(SI2%)、峰值强化程度(SIpeak%)、正性增强积分(PEI)、最大下降斜率(MSD)、达峰时间(TTP)以及R0Iwhole法测量定量参数容量转移常数(Ktrans)、速率常数(Kep)、细胞外血管外间隙容积比(Ve)。按照Miller&Payne分级系统,将肿瘤反应性分为组织学显著反应(MHR)和组织学非显著反应(NMHR)。分别比较MHR组及NMHR组化疗前与NAC2周期后的量化参数,并比较MHR组和NMHR组NAC2周期后各参数的变化值以及基线参数值,正态分布者行两独立(或配对)样本t检验,非正态分布者行两独立样本(或配对资料)的非参数检验,绘制受试者工作曲线(ROC),找到早期预测乳腺癌NAC疗效的最佳参数及其诊断阈值,并比较两种ROI选取方法的预测效能。结果最终入组行半定量分析者72例,NMHR组53例,MHR组19例；行定量分析者62例,NMHR组45例,MHR组17例。无论MHR组还是NMHR组,NAC前与化疗2周期后的肿瘤大小均有统计学差异(P<0.001)；MHR组NAC前与化疗2周期后的半定量参数及定量参数均有统计学差异(P<0.05)；NMHR组除MSD(ROIwhole)、MSD(ROIhs)、TTP(ROIhs)、KepVe外(P值分别为0.615,0.205,0.085,0.118,0.236),其余DCE-MRI参数在NAC前与化疗2周期后的差异均有统计学意义(P<0.05)；NAC2周期后肿瘤各径线消退率、半定量参数变化值及定量参数变化值在MHR组和NMHR组间的差异均有统计学意义(P<0.05),即MHR组参数下降较NMHR组明显；而化疗前的各参数在两组间的差异无统计学意义(P>0.05)。ROC分析显示以肿瘤三径线几何均值的变化值(△G)预测NAC疗效的曲线下面积、敏感度、特异度和诊断阈值分别为0.908、84.2%、92.5%和-43.7%,以K。。的变化值预测的曲线下面积、敏感度、特异度和诊断阈值分别为0.890、76.5%、93.3%和-54.8%。比较ROIwhole法和R0Ihs法测得的半定量参数MSI、SI2%、SIpeak%、PEI和MSD变化值预测NAC疗效的ROC曲线下面积,P值分别为0.769、0.588、0.490、0.807和0.793。结论DCE-MRI量化参数可于化疗2周期预测乳腺癌NAC的疗效,而化疗前的参数尚不能预测化疗的最终疗效,其中△G、△Kep是较好的预测指标,两种ROI选取方法预测乳腺癌NAC疗效的效能相当。第二部分动态增强MRI参数与乳腺癌分子亚型的关系研究目的探讨动态对比增强MRI (DCE-MRI)量化参数与乳腺癌不同分子亚型及预后因子的关系。资料与方法以第一部分中收集的2011年5月-2014年1月计划在我院行新辅助化疗(NAC)的112例乳腺癌患者为研究对象,行DCE-MRI检查,分别采用ROIwhole法和ROIhs法测量半定量参数最大上升斜率(MSI)、第二期强化程度(SI2%)、峰值强化程度(SIpeak%)、正性增强积分(PEI)、最大下降斜率(MSD)、达峰时间(TTP)以及ROIwhole法测量定量参数容量转移常数(Ktrans)、速率常数(Kep)、细胞外血管外间隙容积比(Ve)。采用SP法检测免疫组化指标ER、PR、HER-2及Ki-67,对HER-2(++)者行免疫荧光原位杂交法(FISH)检测,以此将乳腺癌分为Luminal A、Luminal B、HER-2+和三阴性乳腺癌(TNBC)四个亚型。采用两独立样本t检验或Mann-WhitneyU检验比较ER、PR和HER-2不同表达状态时各参数的差别,行单因素方差分析比较不同分子亚型间的半定量参数,行Mann-Whitney U检验对不同分子亚型间定量参数两两比较,P≤0.05为差异有统计学意义。结果最终入组行半定量分析的101例患者,有4例未行FISH检测,Luminal A型15例,Luminal B型56例,HER-2+型12例,TNBC14例。PR+者的MSD高于PR-者(P=0.002,0.016),HER-2+者的MSI高于HER-2-者(P=0.021,0.011),PR-者的SI2%(ROI,,hole)高于PR+者(P=0.031),HER-2-者的TTP(ROIhs)高于HER-2+者(P=0.029),其余半定量参数在不同受体表达状态时的差异无统计学意义(P>0.05)。除MSD(ROI,,hole)外(P=0.045),其余半定量参数在不同分子亚型间的差异无统计学意义,但进一步两两比较,各组间差异均无统计学意义。最终入组行定量分析的82例患者,有3例未行FISH检测,Luminal A型13例,Luminal B型42例,HER-2+型11例,TNBC13例。不同受体表达状态时定量参数的差异无统计学意义(P>0.05)。Lumianl A型和TNBC、LumianlA型和HER-2+型的Ktrans值有统计学差异(P=0.026,0.047),Luminal B型和TNBC的Kep值有统计学差异(P=0.013),其余任何两型间的定量参数值均无统计学差异。结论不同PR表达状态时的MSD不同,PR+者高于PR-者,HER-2过表达者的MSI较正常表达者显著性增高；TNBC和HER-2+型的Ktrans值高于Luminal A型,TNBC的K。。值高于Luminal B型,即TNBC和HER-2+型的局部血流灌注较高,Luminal A和Luminal B型则相反,揭示了各分子亚型的不同生物学特性。第三部分不同分子亚型对动态增强MRI预测乳腺癌NAC疗效准确性的影响目的探讨乳腺癌不同分子亚型对动态对比增强MRI(DCE-MRI)量化参数预测乳腺癌新辅助化疗(NAC)疗效准确性的影响。资料与方法收集2011年5月～2014年1月计划在我院行NAC的乳腺癌患者共112例,分别于化疗前和NAC2周期后行DCE-MR T扫描,分别采用R0Iwhole法和R0Ihs法测量肿瘤三个方位最大径、半定量参数最大上升斜率(MSI)、第二期强化程度(SI：%)、峰值强化程度(SIpeak%)、正性增强积分(PEI)、最大下降斜率(MSD)、达峰时间(TTP)以及ROIwhole法测量定量参数容量转移常数(Ktrans)、速率常数(Kep)、细胞外血管外间隙容积比(V。)。按照Miller&Payne分级系统,将肿瘤反应性分为组织学显著反应(MHR)和组织学非显著反应(NMHR)。采用SP法检测免疫组化指标ER.PR.HER-2及Ki-67,对HER-2(++)者行免疫荧光原位杂交法(FISH)检测,以此将乳腺癌分为Luminal A、Luminal B、HER-2+和三阴性乳腺癌(TNBC)四个亚型。行二分类Logi stic回归,从患者年龄、肿瘤形态、化疗方案和分子亚型中筛选出影响病理与影像预测结果一致性的因素；计算DEC-MRI参数预测乳腺癌不同分子亚型NAC疗效的敏感度、特异度、阳性预测值、阴性预测值和准确度。结果最终入组69例,MHR组19例,NMIIR组50例；Luminal A型9例,Luminal B型39例,HER2+型8例,TNBC13例。经过筛选,分子分型是影响I)CEMRl预测乳腺癌NAC疗效准确性的因素,对SIpenk%(ROIwhole)、MSD(ROIwhone)及SI：%(ROIts)三个参数的影响较大,比值比分别为2.580、0.437和2.569。定量参数(Ktrans、Kep、Vec)及左右径预测不同分子亚型乳腺癌NAC疗效的准确性均较高；对于肿瘤前后径和MSD(ROIwhole),Lumianl的准确度较高,分别为87.5%和89.6%,高于HER-2+型(75.0%和87.5%)和TNBC(76.9%和61.5%)；对于上下径及半定量参数MSI(ROIwhole)、SI2%(ROIwhols)、SIpeak%(ROIwhole)、PEI(ROIwhole)、MSI(ROIhs)、SI2%(R0Ihs)、SIpeak%(ROIh、)、PEI(ROIhs)、MSD(ROIhs),Luminal型乳腺癌预测的准确性较差(分别为68.7%,77.1%,70.8%,62.5%,73.0%,79.2%,64.6%,64.6%,66.7%,72.9%),HER-2+(分别为100.0%,87.5%,87.5%,87.5%,87.5%,100.0%,100.0%,87.5%,87.5%,100.0%)和TNBC的准确性较高(分别为92.3%,100.0%,100.0%,100.0%,100.0%,100.0%,100.0%,100.0%,92.3%,84.6%)。结论定量参数和在延迟MRI图像(高分辨率)上测得的肿瘤左右径预测NAC疗效的准确性较高,不受分子亚型影响；通过大多数半定量参数预测TNBC和HER-2+型NAC疗效的准确性较高,Luminal型则较低。更多还原

【Abstract】 Part I Dynamic contrast-enhanced MRI to predict response for breast cancer patients undergoing neoadjuvant chemotherapyObjective To investigate whether dynamic contrast-enhanced MRI (DCE-MRI) could predict final pathologic response in primary breast cancer patients undergoing neoadjuvant chemotherapy (NAC).Materials and Methods112patients (age range,23-67years; median age,48years) with primary breast carcinoma and scheduled to receive NAC were recruited from May2011to January2014. DEC-MRI examination was performed before NAC and after2cycles of treatment. The longest diameters of three dimension, the semi-quantitative parameters maximum slope of increase (MSI), signal enhancement ratio at the second phase (SI2%) and the peak (SIpeak%), positive enhancement integral (PEI), maximum slope of decrease (MSD), time to peak (TTP) with ROIwhole and ROIhs, and the quantitative parameters volume transfer constant (Ktrans), rate constant (Kep) and extravascular extracellular space fractional volume (Ve) with ROIWhole were measured. Histological response was categorized as major histological response (MHR) and non-major histological response (NMHR). The baseline parameters and the ones after2cycles of NAC of both MHR and NMHR were compared. Moreover, parameter changes after2cycles of NAC, as well as the baseline parameters were compared between MHR and NMHR. Independent-samples (or paried-sampels) T test was used for the normal distribution and independent-samples (or related-sampels) nonparametric test for the abnormal distribution. Receiver operating characteristic curve (ROC) was used to determine the best predictor and cutoff value and the predictive efficiency of the two ROI methods was compared.Results Among the72patients included in the semi-quantitative study,19are grouped into MHR and53are NMHR.17are enrolled in MHR and53are NMHR for quantitative analysis. The longest diameters before NAC and the ones after2cycles of NAC are significantly different in both MHR and NMHR (P<0.001). In MHR, there is significant difference between DCE-MRI parameters of the baseline and the ones after2cycles of NAC (P<0.05). In NMHR, the DEC-MRI parameters are significantly different except MSD(ROIWhole), MSD(ROIhs), TTP(ROIhs), Kepand Ve (P value are0.615,0.205,0.085,0.118and0.236, respectively). The changes of the longest diameters and DCE-MRI parameters show significantly different between MHR and NMHR (P<0.05). There is more dramatic decline in MHR than in NMHR. Pretreatment parameters have no significant difference between the two groups (P>0.05). The ROC curve analysis indicates that area under the curve(AUR), sensitivity, specificity, accuracy and the cutoff value are0.908,84.2%,92.5%,-43.7%for the change of geometric mean of the three longest diameters (ΔG), and are0.890.16.5%,93.3%,-54.8%for the change of KeP. AUR of MSI, SI2%, SIPeak%, PEI and MSD changes for predicting NAC between the two ROC methods are compared with P value0.769,0.588,0.490,0.807and0.793, respectively.Conclusion DCE-MRI parameters can predict final pathologic response in primary breast cancers after2cycles of NAC. However, the baseline parameters haven’t possessed the predictive ability. The changes of G and Kep are the better predictors. The two ROI methods have the comparable predictive efficiency. Part Ⅱ Association of of dynamic contrast-enhanced MRI parameters with molecular subtypes of breast carcinomaObjective To explore dynamic contrast-enhanced MRI (DCE-MRI) in patients with different subtypes of breast carcinoma, and the association between DCE-MRI parameters and prognostic factors.Materials and Methods112patients from part I with primary breast carcinoma and scheduled to receive neoadjuvant chemotherapy (NAC) were recruited between May2011and January2014. DEC-MRI examination was performed before treatment. The semi-quantitative parameters maximum slope of increase (MSI), signal enhancement ratio at the second phase (SI2%) and the peak (SIpeak%), positive enhancement integral (PEI),1maximum slope of decrease (MSD), time to peak (TTP) with ROIwhole and ROIhs, and the quantitative parameters volume transfer constant (Ktrans), rate constant (Kep) and extravascular extracellular space fractional volume (Ve) with ROIwhole were measured. ER, PR, HER-2and Ki-67were obtained by immunohistochemistry (IHC) SP method and those of HER-2(++) were further tested by fluorescence in situ hybridization (FISH) analysis. Then four subtypes of Luminal A, Luminal B, HER-2+and triple-negative breast cancer (TNBC) were established. Independent-samples T test or Mann-Whitney U test was used to compare the DCE-MRI parameters between different receptors status. The semi-quantitative parameters of different subtypes were compared by one way ANOVA and pairwise comparision of the quantitative parameters was performed by Mann-Whitney U test. Results Among the101available cases for semi-quantitative analysis,4are excluded forno FISH test.15are Luminal A,56are Luminal B,12are HER-2+, and14are TNBC.MSD of PR+is significantly higher than that of PR-(P=0.002,0.016). MSI of HER-2+is significantly higher than that of HER-2-(P=0.021,0.011). SI2%(ROIwhole) of PR-is significantly higher than that of PR+(P=0.031). TTP(ROIhs) of HER-2-is significantly higher than that of HER-2+(P=0.029). The other semi-quantitative parameters have no significant difference between different receptors status (P>0.05). There is no significant difference of the semi-quantitative parameters between different subtypes except MSD(ROIwhole)(P=0.045). However, futher pairwise comparision for MSD(ROIwhole) doesn’t showed any significant difference. Among the82patients included in the quantitative study,3are excluded for no FISH test.13are Luminal A,42are Luminal B.11are HER-2+, and13are TNBC. There is no significant difference between different receptors status (P>0.05). Ktrans for Luminal A showes significant difference from TNBC and HER-2+(P=0.026,0.047). Kep between Luminal B and TNBC haven’t significant difference (P=0.013). There is no significant difference for any other two subtypes.Conclusion MSD of PR+is significantly higher than those of PR-. MSI of HER-2overexpression is significantly higher than the nomal one’s. Ktrans for TNBC and HER-2+is significantly higher than Luminal A. Kep for TNBC is significantly higher than Luminal B. On the other hand, TNBC and HER-2+are high perfusion tumors, Luminal A and Luminal B are just the opposite. Every molecular subtype of breast carcinoma reveals unique biological characteristics. Part III Dynamic contrast-enhanced MRI predicting response for breast cancer after neoadjuvant chemotherapy in various subtypesObjective To evaluate the relevance of breast cancer subtypes for dynamic contrast-enhanced MRI (DCE-MRI) parameters predicting response of breast cancer after neoadjuvant chemotherapy (NAC).Materials and Methods112patients with primary breast carcinoma and scheduled to receive NAC were recruited from May2011to January2014. DEC-MRI examination was performed before NAC and after2cycles of treatment. The longest diameters of three dimension, the semi-quantitative parameters maximum slope of increase (MSI), signal enhancement ratio at the second phase (SI2%) and the peak (SIpeak%), positive enhancement integral (PEI), maximum slope of decrease (MSD), time to peak (TTP) with ROIwhoie and ROIhs, and the quantitative parameters volume transfer constant (Ktrans), rate constant (Kep) and extravascular extracellular space fractional volume (Ve) with ROIwhole were measured. Histological response was categorized as major histological response (MHR) and non-major histological response (NMHR). ER. PR, HER-2and Ki-67were obtained by immunohistochemistry (IHC) SP method and those of HER-2(++) were further tested by fluorescence in situ hybridization (FISH) analysis. Then four subtypes of Luminal A, Luminal B, HER-2+and triple-negative breast cancer (TNBC) were established. Factors which affect the consistency between pathology results and imaging evaluation, were screened from age, tumor morphology, chemotherapy regimen and molecular subtype by binary Logistic regression. Sensitivity, specificity, positive predictive value, negative predictive value and accuracy for predicting NAC of different breast cancer subtypes were calculated.Results Among the69patients included,19are grouped into MHR and50are NMHR.9are Luminal A,39are Luminal B,8are HER-2+, and13are TNBC. Molecular subtype is filtered as the only factor which affects the consistency, especially for SIpeak%(ROIwhole), MSD(ROIwhole) and SI2%(ROIhs). The odds ratios are2.580,0.437and2.569, respectively. The quantitative parameters (Ktrans, Kep and Ve), as well as laterolateral diameter have comparative accuracy for different subtypes. For anteroposterior diameters and MSD (ROIwhole), Luminal (accuracy:87.5%,89.6%) is more accurate than HER-2+(accuracy:75.0%,87.5%) and TNBC (accuracy:76.9%,61.5%). For craniocaudal diameters and semi-quantitative parameters MSI(ROIWhole), SI2%(ROIwhole), SIpeak%(ROIwhole), PEI(ROIwhole), MSI(ROIhs), SI2%(ROIhs), SIpeak%(ROIhs), PEI(ROIhs), MSD(ROIhs), Luminal has the poor accuracy (68.7%,77.1%,70.8%,62.5%,73.0%,79.2%,64.6%,64.6%,66.7%and72.9%, respectively) and HER-2+(accuracy:100.0%,87.5%,87.5%,87.5%,87.5%,100.0%,100.0%,87.5%,87.5%and100.0%, respectively)and TNBC (accuracy:92.3%,100.0%,100.0%,100.0%,100.0%,100.0%,100.0%,100.0%,92.3%and84.6%, respectively) are just the reverse. Conclusion The quantitative parameters, as well as laterolateral diameter obtained on delayed images (high resolution) show considerable accuracy for predicting response of breast cancer after NAC regardless of molecular subtypes. For most of the semi-quantitative parameters, TNBC and HER-2+are more accurate than Luminal breast cancer.更多还原