Zonari Paolo, Baraldi Patrizia, Crisi Girolamo
Neuroradiologia, Dipartimento Integrato di Neuroscienze, Ospedale "B. Ramazzini", AUSL Modena, Via G. Molinari 2, 41012 Carpi, Modena, Italy.
Neuroradiology. 2007 Oct;49(10):795-803. doi: 10.1007/s00234-007-0253-x. Epub 2007 Jul 10.
Diffusion-weighted imaging (DWI), perfusion-weighted imaging (PWI) and MR spectroscopy (MRS) provide useful data for tumor evaluation. To assess the contribution of these multimodal techniques in grading glial neoplasms, we compared the value of DWI, PWI and MRS in the evaluation of histologically proven high- and low-grade gliomas in a population of 105 patients.
Independently for each modality, the following variables were used to compare the tumors: minimum apparent diffusion coefficient (ADC) and maximum relative cerebral blood volume (rCBV) normalized values between tumor and healthy tissue, maximum Cho/Cr ratio and minimum NAA/Cr ratio in tumor, and scored lactate and lipid values in tumor. The Mann-Whitney and Wilcoxon tests were employed to compare DWI, PWI and MRS between tumor types. Logistic regression analysis was used to determine which parameters best increased the diagnostic accuracy in terms of sensitivity, specificity, and positive and negative predictive values. ROC curves were determined for parameters with high sensitivity and specificity to identify threshold values to separate high- from low-grade lesions.
Statistically significant differences were found for rCBV tumor/normal tissue ratio, and NAA/Cr ratio in tumor and Cho/Cr ratio in tumor between low- and high-grade tumors. The best performing single parameter for group classification was the normalized rCBV value; including all parameters, statistical significance was reached by rCBV tumor/normal tissue ratio, NAA/Cr tumor ratio and lactate. From the ROC curves, a high probability for a neoplasm to be a high-grade lesion was associated with a rCBV tumor/normal tissue ratio of >1.16 and NAA/Cr tumor ratio of <0.44.
Combining PWI and MRS with conventional MR imaging increases the accuracy of the attribution of malignancy to glial neoplasms. The best performing parameter was found to be the perfusion level.
扩散加权成像（DWI）、灌注加权成像（PWI）和磁共振波谱（MRS）为肿瘤评估提供了有用的数据。为了评估这些多模态技术在神经胶质瘤分级中的作用，我们比较了DWI、PWI和MRS在105例经组织学证实的高低级别神经胶质瘤患者群体评估中的价值。
对于每种模态，独立使用以下变量比较肿瘤：肿瘤与健康组织之间的最小表观扩散系数（ADC）和最大相对脑血容量（rCBV）归一化值、肿瘤中的最大胆碱/肌酸（Cho/Cr）比值和最小N-乙酰天门冬氨酸/肌酸（NAA/Cr）比值，以及肿瘤中的乳酸和脂质评分值。采用曼-惠特尼检验和威尔科克森检验比较不同肿瘤类型之间的DWI、PWI和MRS。使用逻辑回归分析确定哪些参数在敏感性、特异性、阳性和阴性预测值方面能最佳提高诊断准确性。针对具有高敏感性和特异性的参数绘制ROC曲线，以确定区分高低级别病变的阈值。
低级别和高级别肿瘤之间在rCBV肿瘤/正常组织比值、肿瘤中的NAA/Cr比值和肿瘤中的Cho/Cr比值方面存在统计学显著差异。用于组分类的最佳单一参数是归一化rCBV值；纳入所有参数后，rCBV肿瘤/正常组织比值、NAA/Cr肿瘤比值和乳酸达到统计学显著性。从ROC曲线来看，肿瘤为高级别病变的高概率与rCBV肿瘤/正常组织比值>1.16和NAA/Cr肿瘤比值<0.44相关。
将PWI和MRS与传统磁共振成像相结合可提高神经胶质瘤恶性程度归因的准确性。发现最佳性能参数是灌注水平。
