多模态脑图与失语性手术缺损的相关性研究
该研究纳入71例语言功能区胶质瘤切除患者,将术前fMRI、TMS及DTI定位与术后长期语言结局相关分析。结果显示切除术前皮质语言区不能预测长期失语恶化,而切除经TMS点和瘤周1cm网格在75%标准化FA阈值定位的纤维束显著预测语言结局,fMRI BOLD信号定位的纤维束则无预测作用。
要点
- 切除术前fMRI/TMS定位的皮质语言区不预测长期失语恶化。
- 切除TMS点定位纤维束(75% FA阈值)显著预测语言结局,OR 51,p<0.001。
- 切除瘤周1cm网格定位纤维束亦预测语言结局,OR 5.0,p=0.04。
核心结论
- 切除术前fMRI与TMS定位的皮质语言区不预测失语性手术缺损。
- 切除TMS点和瘤周皮质网格定位的白质纤维束显著预测失语恶化。
- 以fMRI BOLD信号为种子定位的纤维束不预测语言结局。
研究设计与样本
- 单中心回顾性研究,纳入2017年7月1日至2023年8月1日连续71例语言功能区胶质瘤切除患者。
- 采用非线性配准算法将术后切除腔叠加至术前MRI,分析术前皮质与皮质下定位的切除与长期语言结局的相关性。
- 以TMS点、fMRI BOLD信号及瘤周1cm网格为种子行DTI纤维束示踪,使用多个FA阈值并归一化至MNI空间。
主要发现
- 切除TMS点定位纤维束在75%标准化FA阈值显著预测语言结局(OR 51,p<0.001)。
- 切除瘤周1cm网格定位纤维束在75%标准化FA阈值显著预测语言结局(OR 5.0,p=0.04)。
- fMRI BOLD信号定位纤维束在任何FA阈值均不预测语言结局(OR 0.29-5.32,p=0.42-0.94)。
- 三种皮质种子结构定位的纤维束均与人类连接组计划语言相关规范纤维束共定位并共享组水平特征。
关键数值与推荐
- TMS点定位纤维束切除预测语言结局:OR 51,p<0.001(75%标准化FA阈值)。
- 瘤周1cm网格定位纤维束切除预测语言结局:OR 5.0,p=0.04(75%标准化FA阈值)。
- fMRI BOLD信号定位纤维束:OR 0.29-5.32,p=0.42-0.94,各FA阈值均不显著。
临床怎么用
- 术前规划应重视TMS点及瘤周网格种子定位的白质纤维束,而非仅依赖皮质语言区定位。
- fMRI BOLD信号定位的纤维束不宜单独作为预测术后失语风险的依据。
- DTI纤维束示踪的FA阈值选择可能影响预测效能,需标准化。
局限性
- 为单中心回顾性研究,样本量71例。
- 结论基于特定FA阈值与归一化处理流程,外推需谨慎验证。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
OBJECTIVE: For patients with language-eloquent brain tumors, surgeons must maximize the extent of resection while minimizing neurological morbidity. Despite the extensive use of noninvasive brain mapping techniques, it remains unclear how to preoperatively localize brain regions essential for sustaining language function in a data-driven manner. Here the authors investigate the clinical utility and functional relevance of regions mapped by functional MRI (fMRI), transcranial magnetic stimulation (TMS), and diffusion tensor imaging (DTI). METHODS: This study included 71 consecutive patients undergoing resection of language-eloquent gliomas from July 1, 2017, to August 1, 2023. A nonlinear coregistration algorithm was used to overlay the postoperative resection cavity onto the preoperative MRI. The resection of preoperative cortical and subcortical localizations was correlated with long-term linguistic outcomes. Fiber tracts were localized by seeding DTI tractography with different cortical features: TMS points (TMS tracts), the fMRI blood oxygen level-dependent (BOLD) signal (fMRI tracts), and a peritumoral 1-cm grid (grid tracts). Multiple fractional anisotropy (FA) thresholds were used for each cortical seed structure. Fiber tracts were normalized to Montreal Neurological Institute space to identify group-level characteristics. RESULTS: The majority of patients were male (n = 42, 59%) and younger than 60 years of age (n = 46, 65%). Forty-one patients (58%) had high-grade gliomas, while 30 patients (42%) had low-grade gliomas. The authors found that the resection of preoperatively identified cortical language regions does not predict long-term aphasic decline. However, the resection of tracts localized by the TMS points and grid at the 75% normalized FA threshold significantly predicted language outcomes (OR 51, p < 0.001 and OR 5.0, p = 0.04, respectively). Tracts localized by the fMRI BOLD signal did not predict language outcomes at any FA threshold (OR 0.29-5.32, p = 0.42-0.94). Tracts localized by all three cortical seed structures recapitulated normative patterns of subcortical connectivity by colocalizing with and sharing group-level features with language-associated normative tracts derived from the Human Connectome Project. Functional tracts that predicted linguistic outcomes (TMS and grid tracts) demonstrated shared features at a group-level distinct from the nonpredictive fMRI tracts. CONCLUSIONS: The resection of cortical language regions mapped preoperatively by fMRI and TMS did not predict aphasic surgical deficits. In contrast, resecting white matter tracts localized by cortical TMS points and a peritumoral cortical grid significantly predicted aphasic decline, while resecting tracts localized by the fMRI BOLD signal as seeds did not predict linguistic outcomes.
引用
- Muir MT, Noll K, Patel R, et al. Toward data-driven surgical planning: multimodal mapping correlates of aphasic surgical deficits. J Neurosurg. 2025. PMID: 41135113. doi:10.3171/2025.6.JNS242301
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。