多模态磁共振鉴别胶质母细胞瘤治疗后复发的风险因素分析  

Analysis of risk factors for recurrence of glioblastoma after treatment by multimodal magnetic resonance imaging

作　　者：王静[1] 师秋霞 李绍山[1] Wang Jing;Shi Qiuxia;Li Shaoshan(Department of Neurosurgery,the First Affiliated Hospital of Xinjiang Medical University,Urumqi 830000)

机构地区：[1]新疆医科大学第一附属医院神经外一科

出　　处：《卒中与神经疾病》2023年第5期474-479,共6页Stroke and Nervous Diseases

基　　金：新疆维吾尔自治区卫生健康委员会科研基金项目(20213214)。

摘　　要：目的评估多模态磁共振鉴别胶质母细胞瘤治疗后复发的性能。方法回顾性分析2020年5月-2022年5月46例经组织病理学证实的高级别神经胶质瘤患者;所有患者均经病理证实为肿瘤复发(Tumor recurrence,TR)(n=22)或放射性坏死(Radiation necrosis,RN)(n=24);进行多模态磁共振扫描,扫描方法包括横向T1加权图像(T1-weighted imaging,T1WI)、T2加权图像(T2-weighted imaging,T2WI)、T2流体衰减反转恢复(T2 fluid-attenuated inversion recovery,T2FLAIR)、对比增强T1WI(Contrast-enhanced T1 weighted imaging,CE-T1WI)、扩散加权成像(Diffusion weighted imaging,DWI)、弥散峰度成像(Diffusion kurtosis imaging,DKI)和动态磁敏感对比(Dynamic susceptibility contrast,DSC)序列;对显著变量进行受试者工作特征曲线分析以确定其诊断性能;多变量逻辑回归用于确定最佳判别预测模型。结果与RN组比较,TR组相对平均峰度(Relative mean kurtosis,rMK)、相对轴向峰度(Relative axial kurtosis,rKa)、相对脑血容量(Relative cerebral blood volume,rCBV)和相对平均通过时间(Relative mean transit time,rMTT)值显著升高(P<0.05)和相对表观扩散系数(Relative Apparent Diffusion Coefficient,rADC)值显著降低(P<0.05)。用于区分TR和RN的rADC,rMK,rKa,rCBV和rMTT值的曲线下面积(Area under curve,AUC)分别为0.750、0.891、0.694、0.760和0.674(P<0.05)。多变量分析显示,rMK(P=0.006)和rCBV(P=0.041)是区分TR和RN的重要预测因子,而rKa(P=0.524),rMTT(P=0.556)和rADC(P=0.097)则不是;rMK和rCBV组合的AUC达到了0.924(P<0.001),准确度为87.69%,敏感度为87.03%和特异度为93.46%。结论多模态磁共振参数可能是区分TR和RN的一种有价值的非侵入性工具,与单独使用任一种技术比较,DKI与DSC结合可以提高评估治疗反应的诊断性能。ObjectiveTo evaluate the performance of multimodal magnetic resonance imaging in determining the recurrence of glioblastoma after treatment.Methods Forty-six patients with histopathologically confirmed high-grade gliomas from May 2020 to May 2022 were analyzed retrospectively.All patients were histopathologically confirmed to have tumor recurrence(TR)(n=22)or radiation necrosis(RN)(n=24).Multimodal magnetic resonance scanning was performed.The scanning methods included transverse T1-weighted images(T1WI),T2-weighted images(T2WI),T2 fluid attenuation inversion recovery(T2FLAIR),contrast-enhanced T1WI(CE-T1WI),DWI,DKI and DSC sequences.ROC analysis was performed on significant variables to determine their diagnostic performance.Multivariate logistic regression was used to determine the best predictive model to distinguish between TR and RN.Results Compared with the RN group,the values of rMK,rKa,rCBV and rMTT in the TR group increased significantly(P<0.05),and the value of rADC decreased significantly(P<0.05).The AUC values of rADC,rMK,rKa,rCBV and rMTT used to distinguish TR from RN were 0.750,0.891,0.694,0.760 and 0.674,respectively(P<0.05).Multivariate analysis showed that rMK(P=0.006)and rCBV(P=0.041)were important predictors to distinguish TR from RN,while rKa(P=0.524),rMTT(P=0.556)and rADC(P=0.097)were not.The AUC of the combination of rMK and rCBV reached 0.924(P<0.001),with the accuracy of 87.69%,sensitivity of 87.03%and specificity of 93.46%.Conclusion Multimodal magnetic resonance parameters may be a valuable noninvasive tool for distinguishing TR from RN.Compared with each technique alone,DKI combined with DSC can improve the diagnostic performance in therapeutic response evaluation.

关 键 词：多模态磁共振 胶质母细胞瘤 肿瘤复发 放射性坏死 

分 类 号：R739.41[医药卫生—肿瘤]