MRI时间依赖扩散诊断指标判断胶质母细胞瘤MGMT启动子甲基化  

Feasibility of time-dependent diffusion MRI-based indicators for identifying MGMT promotermethylation in glioblastomas

作　　者：梁晓晶 张云 符永瑰 谢雁 魏倩倩 程流泉[1] 王庆军[1] LIANG Xiaojing;ZHANG Yun;FU Yonggui;XIE Yan;WEI Qianqian;CHENG Liuquan;WANG Qingjun(Department of Radiology,the Sixth Medicine Center of PLAGH,Beijing 100048,China;Department of Radiology,Peking University International Hospital,Beijing 102206,China)

机构地区：[1]解放军总医院第六医学中心放射诊断科,北京100048 [2]北京大学国际医院放射诊断科,北京102206

出　　处：《磁共振成像》2024年第11期67-74,共8页Chinese Journal of Magnetic Resonance Imaging

基　　金：2021年北京市海淀区卫生健康发展科研培育计划立项项目(编号:HP2021-32-80601)。

摘　　要：目的使用MRI时间依赖扩散诊断指标判断新诊断胶质母细胞瘤O6-甲基鸟嘌呤-脱氧核糖核酸甲基转移酶(O^(6)-methylguanine deoxyribonucleic acid methyltransferase,MGMT)启动子甲基化状态。材料与方法对22例甲基化MGMT启动子(methylated MGMT,mMGMT)和29例非甲基化MGMT启动子(unmethylated MGMT,uMGMT)新诊断胶质母细胞瘤患者进行诊断分析以及对14例mMGMT和14例uMGMT新诊断胶质母细胞瘤患者进行应用验证。使用3 T MRI震荡梯度回波和脉冲梯度回波进行时间依赖扩散扫描。基于两室模型获取MRI时间依赖扩散微结构诊断指标,包括细胞内容积分数(f_(in))、细胞外扩散率(D_(ex))、细胞直径(d)、细胞密度(cellularity)、不同频率的扩散率(D_(0Hz)、D_(15Hz)和D_(30Hz))。比较mMGMT和uMGMT两组胶质母细胞瘤之间以上指标的差异。首先使用单因素逻辑回归分析各指标区分两组胶质母细胞瘤的能力,然后使用多因素逻辑回归分析确定是否存在构建逻辑回归联合诊断模型的可能。两两比较各诊断指标区分两组胶质母细胞瘤的能力。结果相对于uMGMT胶质母细胞瘤,mMGMT胶质母细胞瘤表现出更高的f_(in),D_(ex)和cellularity(P<0.05)以及更低的D0 Hz(P=0.018)。f_(in)对于区分两组胶质母细胞瘤诊断能力最佳,曲线下面积(area under the curve,AUC)值为0.95,敏感度和特异度分别为95%和90%,且与其他诊断指标两两比较差异均具统计学意义(P<0.05)。多因素逻辑回归分析显示f_(in)为独立影响变量,因此无联合诊断模型的构建。以f_(in)>0.16作为诊断阈值进行应用验证,准确度为82.14%。结论MRI时间依赖扩散诊断指标f_(in)对于判断新诊断胶质母细胞瘤MGMT启动子甲基化状态表现出较好的应用价值。Objective:To investigate the feasibility of time-dependent diffusion MRI based diagnostic indicators for identifying O^(6)-methylguanine deoxyribonucleic acid methyltransferase(MGMT)promoter methylation status in newly diagnosed glioblastomas.Materials and Methods:We enrolled 22 glioblastomas with methylated MGMT promoter(mMGMT)and 29 glioblastomas with unmethylated MGMT promoter(uMGMT)for diagnostic analysis and then 14 mMGMT glioblastomas and 14 uMGMT glioblastomas for validation application.Time-dependent diffusion MRI data was acquired using pulsed and oscillating gradient sequences on a 3 T scanner.Microstructural diagnostic indicators,including intracellular volume fraction(f_(in)),extracellular diffusivity(D_(ex)),cell diameter(d),cellularity,and diffusivities at different frequencies(D_(0Hz),D_(15Hz),and D_(30Hz)),were estimated using a two-compartment model.These indicators were compared between mMGMT and uMGMT glioblastomas,and their discriminative performance was assessed with univariate logistic regression analysis.Significant variables were identified via multivariate logistic regression to construct a combined diagnostic model.Pairwise comparisons were used to evaluate diagnostic abilities.Results:mMGMT glioblastomas showed higher f_(in),D_(ex)and cellularity(all P<0.05)and lower D_(0Hz)(P=0.018)compared to uMGMT glioblastomas.Among these indicators,f_(in)had the highest discriminant power with area under curve(AUC)was 0.95,sensitivity was 95%,specificity was 90%,and showed differences compared to other indicators(all P<0.05).No combined diagnostic model was constructed because f_(in)was the independent influence variable in the multivariate logistic regression analysis.The accuracy was 82.14%using f_(in)>0.16 as the diagnostic threshold for validation.Conclusions:Time-dependent diffusion MRI–based f_(in)show promise for characterizing MGMT promoter methylation status in newly diagnosed glioblastomas.

关 键 词：胶质母细胞瘤 时间依赖扩散磁共振成像 磁共振成像 微结构 震荡梯度回波 MGMT启动子 诊断 

分 类 号：R445.2[医药卫生—影像医学与核医学] R730.264[医药卫生—诊断学]