Ho C Y, Cardinal J S, Kamer A P, Lin C, Kralik S F
From the Department of Radiology, Indiana University School of Medicine, Indianapolis, Indiana.
AJNR Am J Neuroradiol. 2016 Mar;37(3):544-51. doi: 10.3174/ajnr.A4559. Epub 2015 Nov 12.
The pattern of contrast leakage from DSC tissue signal intensity time curves have shown utility in distinguishing adult brain neoplasms, but has limited description in the literature for pediatric brain tumors. The purpose of this study is to evaluate the utility of grading pediatric brain tumors with this technique.
A retrospective review of tissue signal-intensity time curves from 63 pediatric brain tumors with preoperative DSC perfusion MR imaging was performed independently by 2 neuroradiologists. Tissue signal-intensity time curves were generated from ROIs placed in the highest perceived tumor relative CBV. The postbolus portion of the curve was independently classified as returning to baseline, continuing above baseline (T1-dominant contrast leakage), or failing to return to baseline (T2*-dominant contrast leakage). Interobserver agreement of curve classification was evaluated by using the Cohen κ. A consensus classification of curve type was obtained in discrepant cases, and the consensus classification was compared with tumor histology and World Health Organization grade.
Tissue signal-intensity time curve classification concordance was 0.69 (95% CI, 0.54-0.84) overall and 0.79 (95% CI, 0.59-0.91) for a T1-dominant contrast leakage pattern. Twenty-five of 25 tumors with consensus T1-dominant contrast leakage were low-grade (positive predictive value, 1.0; 95% CI, 0.83-1.00). By comparison, tumors with consensus T2*-dominant contrast leakage or return to baseline were predominantly high-grade (10/15 and 15/23, respectively) with a high negative predictive value (1.0; 95% CI, 0.83-1.0). For pilomyxoid or pilocytic astrocytomas, a T1-dominant leak demonstrated high sensitivity (0.91; 95% CI, 0.70-0.98) and specificity (0.90, 95% CI, 0.75-0.97).
There was good interobserver agreement in the classification of DSC perfusion tissue signal-intensity time curves for pediatric brain tumors, particularly for T1-dominant leakage. Among patients with pediatric brain tumors, a T1-dominant leakage pattern is highly specific for a low-grade tumor and demonstrates high sensitivity and specificity for pilocytic or pilomyxoid astrocytomas.
DSC组织信号强度时间曲线的对比剂渗漏模式已被证明有助于鉴别成人大脑肿瘤,但在儿科脑肿瘤方面的文献描述有限。本研究的目的是评估该技术对儿科脑肿瘤分级的实用性。
2名神经放射科医生独立对63例术前行DSC灌注磁共振成像的儿科脑肿瘤的组织信号强度时间曲线进行回顾性分析。组织信号强度时间曲线由置于最高相对脑血容量的肿瘤区域(ROI)生成。曲线的团注后部分被独立分类为恢复至基线、持续高于基线(T1主导的对比剂渗漏)或未恢复至基线(T2*主导的对比剂渗漏)。使用Cohen κ评估评估观察者间曲线分类的一致性。在存在分歧的病例中获得曲线类型的共识分类,并将该共识分类与肿瘤组织学和世界卫生组织分级进行比较。
组织信号强度时间曲线分类的总体一致性为0.69(95%CI,0.54 - 0.84),T1主导的对比剂渗漏模式的一致性为0.79(95%CI,0.59 - 0.91)。25例具有共识性T1主导对比剂渗漏的肿瘤均为低级别肿瘤(阳性预测值为1.0;95%CI,0.83 - 1.00)。相比之下,具有共识性T2*主导对比剂渗漏或恢复至基线的肿瘤主要为高级别肿瘤(分别为10/15和15/23),阴性预测值较高(1.0;95%CI,0.83 - 1.0)。对于毛细胞黏液样或毛细胞型星形细胞瘤,T1主导的渗漏显示出高敏感性(0.91;95%CI,0.70 - 0.98)和特异性(0.90,95%CI,0.75 - 0.97)。
儿科脑肿瘤DSC灌注组织信号强度时间曲线的分类在观察者间具有良好的一致性,特别是对于T1主导的渗漏。在儿科脑肿瘤患者中,T1主导的渗漏模式对低级别肿瘤具有高度特异性,对毛细胞型或毛细胞黏液样星形细胞瘤具有高敏感性和特异性。