Am J Neuroradiol. 2017 Apr;38(4):696-702. doi: 10.3174/ajnr.A5068.

Noninvasive Evaluation of CBF and Perfusion Delay of Moyamoya Disease Using Arterial Spin-Labeling MRI with Multiple Postlabeling Delays: Comparison with 15O-Gas PET and DSC-MRI.




BACKGROUND AND PURPOSE: Arterial spin-labeling MR imaging with multiple postlabeling delays has a potential to evaluate various hemodynamic parameters. To clarify whether arterial spin-labeling MR imaging can identify CBF and perfusion delay in patients with Moyamoya disease, we compared arterial spin-labeling, DSC, and 15O-gas PET in terms of their ability to identify these parameters.

MATERIALS AND METHODS: Eighteen patients with Moyamoya disease (5 men, 13 women; ages, 21-55 years) were retrospectively analyzed. CBF values of pulsed continuous arterial spin-labeling using 2 postlabeling delays (short arterial spin-labeling, 1525 ms; delayed arterial spin-labeling, 2525 ms) were compared with CBF values measured by 15O-gas PET. All plots were divided into 2 groups by the cutoff of time-based parameters (the time of the maximum observed concentration, TTP, MTT, delay of MTT to cerebellum, and disease severity [symptomatic or not]). The ratio of 2 arterial spin-labeling CBFs (delayed arterial spin-labeling CBF to short arterial spin-labeling CBF) was compared with time-based parameters: time of the maximum observed concentration, TTP, and MTT.

RESULTS: The short arterial spin-labeling-CBF values were significantly correlated with the PET-CBF values (r = 0.63; P = .01). However, the short arterial spin-labeling-CBF value dropped in the regions with severe perfusion delay. The delayed arterial spin-labeling CBF overestimated PET-CBF regardless of the degree of perfusion delay. Delayed arterial spin-labeling CBF/short arterial spin-labeling CBF was well correlated with the time of the maximum observed concentration, TTP, and MTT (p = 0.71, 0.64, and 0.47, respectively).

CONCLUSIONS: Arterial spin-labeling using 2 postlabeling delays may detect PET-measured true CBF and perfusion delay in patients with Moyamoya disease. Provided its theoretic basis and limitations are considered, noninvasive arterial spin-labeling could be a useful alternative for evaluating the hemodynamics of Moyamoya disease.



Supplementary information:

Arterial spin labeling MRI (ASL) is a purely noninvasive technique to quantify cerebral blood flow (CBF) in clinically feasible scan time (~5 min). This is pretty useful in clinical settings since it requires no radiation exposure nor injection of contrast dyes, and could be obtained with routine MRI sequences at a single exam (within 30 min). The main problem of this technique is its CBF values affected by perfusional delay and the choice of postlabeling delay (PLD). Moreover, the interscan variability related to spin efficacy was frequently observed, even when the patient’s clinical condition was stable.

As clinical neurosurgeons, we believe it is better to find different significance of ASL images other than the accuracy of CBF measurement, especially for the patients with perfusional delay as in moyamoya disease. In this study, we acquired ASL images with two PLD (short ASL: sASL: PLD=1525ms and delayed ASL: dASL: PLD=2525ms) and compared them with the gold standard 15O gas positron emission tomography (PET) and dynamic susceptibility contrast images (DSC). We found the correlation of ASL-measured CBF and PET-measured CBF was significant with r=0.63 when we choose PLD of 1525ms. Moreover, we found the ASL ratio (dASL-CBF/sASL-CBF) correlated well with time-based parameters of DSC (r=0.47-0.71). After the report of this study, we also found similar results from ASL images in different moyamoya patients or using another method to evaluate regional values (e.g. normalize parametrical maps and calculate atlas-based regional values using SPM12 or 3DSRT). Using different patient group and/or different methodology also reproduced significant correlation between sASL-CBF and PET-CBF with r=0.4-0.6, and better correlation between ASL ratio and time-based parameters (r=0.6-0.8).

Our current project is to find out the relationship of ASL images with other hemodynamic parameters other than CBF (i.e. cerebral blood volume and oxygen extraction fraction). Now we started to add ASL ratio images (created in in-house MATLAB program) to our clinical practice in the hospital. So far, it seems working, correlate well with time-based parameters of DSC images, and aid our understanding of hemodynamic compromise, especially for neurosurgeons not specialized in ASL images (see Figure 2).



Figure 1 (reprinted and modified from Figure 1 of the original paper). Images of a 21-yr-old female with transient weakness of the right hand. The magnetic resonance angiography (MRA, A) shows occlusion of the bilateral terminal internal carotid arteries (arrow), absence of the bilateral middle cerebral arteries, and stenosis of the left posterior cerebral artery (arrowhead). T2 weighted images reveals a small ischemic change of the left hemisphere (B). C: Sample regions of interests used for the quantification of parameters in this study. Color maps of PET-measured cerebral blood flow (CBF), sASL (ASL with PLD=1525ms) – measured CBF, dASL (ASL with PLD=2525ms) – CBF, dynamic susceptibility contrast imaging (DSC)-measured Time to maximum concentration (Tmax) and time to peak (TTP) are shown respectively from D to H, and reveal hemodynamic stress on the left hemisphere. Two ASL-CBF maps were visually comparable with the PET-CBF maps, but showed generally higher values than those of PET. However, the sASL-CBF (E) shows lower values than dASL-CBF (F) on the affected side (arrows). Conversely, sASL-CBF shows higher values than dASL-CBF (arrowheads) on the less-affected side.




Figure 2. The two ASL images (A, sASL: ASL of PLD=1525ms and B, dASL: ASL of PLD=2525ms) and the ASL ratio image of the same patient in Figure 1. The increase of ASL ratio in the left hemisphere was clearly shown in the ASL ratio image (C). The area with increased ASL ratio also show increased Tmax and TTP (Figure 1 G and H).




Figure 3. The coauthors and coworkers of our research team.