Comparison of arterial spin labeling perfusion with dynamic susceptibility contrast perfusion in Moyamoya disease
How to cite this article: Yadav N, Pendharkar H, Gupta AK, Prasad C, Shukla D, Kandavel T, et al. Comparison of arterial spin labeling perfusion with dynamic susceptibility contrast perfusion in Moyamoya disease. J Neurosci Rural Pract 2023;14:286-92.
Moyamoya disease (MMD) leads to frequent ischemic/hemorrhagic manifestations. Our aim was to compare findings of arterial spin labeling (ASL) with dynamic susceptibility contrast (DSC) perfusion in patients of MMD
Materials and Methods:
Patients diagnosed as MMD underwent magnetic resonance imaging with ASL and DSC perfusion sequences. Perfusion in bilateral anterior cerebral artery and middle cerebral artery territories at two levels (level of thalami and centrum semiovale) was graded as normal (score 1), or reduced (score 2) on DSC and ASL cerebral blood flow (CBF) maps by comparison with normal cerebellar perfusion. Time to peak (TTP) maps of DSC perfusion were also qualitatively scored as normal (score 1), or increased (score 2) similarly. Correlation between scores of ASL, CBF, DSC, CBF, and DSC, TTP maps was assessed by using Spearman’s rank correlation.
Among the 34 patients, we did not find any significant correlation between the ASL CBF maps and DSC CBF maps (r = −0.028, P = 0.878), mean matching index 0.39 ± 0.31, whereas significant correlation was noted between the ASL CBF maps and DSC TTP maps (r = 0.58, P = 0.0003), mean matching index 0.79 ± 0.26. ASL CBF underestimated the perfusion compared to DSC perfusion.
ASL perfusion CBF maps do not match the DSC perfusion CBF maps and rather match the TTP maps of DSC perfusion. This is explained by inherent problems in estimation of CBF in these techniques because of delay in arrival of label (in ASL perfusion) or contrast bolus (in DSC perfusion) due to the presence of stenotic lesions.
Arterial spin labeling
Cerebral blood flow
Moyamoya disease (MMD) is characterized by stenotic and occlusive changes involving the supraclinoid segment of internal carotid artery along with basal fine net-like vascular network, which are known as “Moyamoya” vessels. Clinical manifestations of MMD involve ischemic manifestations such as transient ischemic attacks (TIA) or stroke with cerebral infarction, intracerebral hemorrhage, headache, seizures, involuntary movements, cognitive deficits and mental retardation, visual disturbance, and psychiatric manifestations. Treatment is done by surgical revascularization procedures which are divided as: Direct, indirect, and combined bypass.
Perfusion studies are important to evaluate hemodynamic information necessary for further management. Various methods of assessing perfusion on magnetic resonance imaging (MRI) are available, including contrast agent based methods (dynamic susceptibility contrast - [DSC] perfusion/ T2* perfusion) and non-contrast perfusion methods (arterial spin labeling - [ASL] perfusion). DSC perfusion utilizes the drop of T2/T2* signal intensity of brain parenchyma occurring during the passage of gadolinium contrast through the tissue capillary bed during first pass. It detects perfusion changes even in morphologically normal brain in patients with MMD. It provides multiple useful parameters such as the “time of the maximum observed concentration (Tmax),” “Time to Peak (TTP),” and “Mean Transit Time (MTT),” which are important hemodynamic markers in MMD and are also helpful in determining success of revascularization surgery. Arterial spin-labeling (ASL) perfusion uses the labeled protons (endogenous tracer) in arterial blood which are labeled using radiofrequency pulses. For calculating cerebral blood flow (CBF), the amount of labeled blood reaching the parenchyma after a fixed time (after labeling) is measured by subtracting it from a control image obtained without labeled blood. As injection of contrast agent is not required in this modality, it is very advantageous for patients with kidney disease and also for imaging in Children.
This study was carried out to evaluate the utility of ASL perfusion imaging in patients with MMD and to compare ASL perfusion with DSC perfusion in providing relevant hemodynamic information necessary for management.
MATERIALS AND METHODS
This was a prospective observational study. Ethics approval letter number: nimh/do/ethics sub-committee (BS & NS) 7th meeting/2017, date 22.11.2017, place: Nimhans, Bangalore. Inclusion criteria patient’s demographic, clinical information and biochemical investigation reports were collected from the patient’s outpatient/inpatient file and e-hospital portal of the institute. Information about symptoms and their duration and other demographic profile was taken in a predesigned pro forma.
Confirmation of diagnosis of MMD was based on magnetic resonance angiography/cerebral angiography diagnostic criteria (based on Guidelines for Diagnosis and Treatment of MMD).
Cooperative patients were taken up for MRI examination without sedation. For children <10 years, imaging/DSA procedure was done under anesthesia.
MRI were performed on 1.5T Aera (Siemens) and 1.5T Optima MR 450 w (GE) system. The pulse sequences and their parameters are mentioned in [Table 1].
|1.5 T Aera (Siemens)||1.5 T Optima MR450w (GE)|
|T1-W Axial||TR/TE: 500 ms/11 ms; Slice thickness: 5 mm, 2D||T1-W Axial:||TR/TE: 2004.24 ms/8.76 ms; Slice thickness: 5 mm, 2D|
|T2 Axial, Coronal, Sagittal||TR/TE: 4500 ms/96 ms; Slice thickness: 5 mm, 2D||T2 Axial, Coronal, Sagittal||TR/TE: 5938.45 ms/110.7 ms; Slice thickness: 5 mm, 2D|
|FLAIR Axial||TR/TE: 9000 ms/88 ms; Slice thickness: 5 mm, 2D||FLAIR Axial||TR/TE: 12000 ms/98.7 ms; Slice thickness: 5 mm, 2D|
|DWI Axial||TR/TE: 3600 ms/83 ms; Slice thickness: 5 mm, 2D||DWI Axial||TR/TE: 8000 ms/101.6 ms; Slice thickness: 5 mm, 2D|
|SWI Axial||TR/TE: 49 ms/30 ms; Slice thickness: 2 mm, 2D||SWI Axial||TR/TE: 78.2 ms/48.8 ms; Slice thickness: 10 mm, 2D|
|TOF-MRA -multislab 3D||TR/TE: 32 ms/7 ms; Flip angle 25°, FOV read 200 mm Slice thickness: 0.7 mm, Voxel size : 0.4×0.4×0.7 mm,||TOF-MRA -multislab 3D||TR/TE : 22 ms/2.6 ms; Slice thickness: 1.4 mm|
|T1 MPRAGE 3D||TR/TE: 1800 ms/2.6 ms; Slice thickness: 1 mm, 3D||T1 BRAVO 3D||TR/TE: 9.5 ms/3.7 ms; Slice thickness: 1 mm, 3D|
|DSC perfusion||T2* perfusion, TR/TE: 1610 ms/30 ms, 20 slices, 50 measurements, Flip angle 90°, Voxel size: 1.8×1.8×5 mm, FOV read 230 mm, Slice thickness: 5 mm, EPI factor 128||DSC perfusion||T2* perfusion, TR 2000 ms, 20 slices, Flip angle 60°, FOV read 220 mm, Slice thickness: 5 mm|
|ASL perfusion||Pulsed 3D ASL, TR/TE: 4000 ms/16.38 ms, Flip angle 180°, FAIR QII mode, Inversion time 1800 ms, Voxel size: 3×3×3 mm, FOV read 192 mm, Slice thickness: 3 mm, Turbo factor 18, EPI factor 21||ASL perfusion||Pulsed continuous 3D ASL, TR/TE: 4766 ms/10.7 ms, Flip angle 180°, Inversion time 2025 ms, Voxel size: 3×3×3 mm, FOV freq 240 mm, Section thickness: 3 mm, Number of sections 40, matrix 512×8, Labelling time 1.5 s, Post label delay 1525|
|Post contrast T1 MPRAGE 3D||TR/TE: 1800 ms/2.6 ms; Slice thickness: 1 mm, 3D||Post contrast T1 BRAVO 3D||TR/TE: 9.5 ms/3.7 ms; Slice thickness: 1 mm, 3D|
Analysis of ASL and DSC perfusion
Cerebral perfusion was assessed in bilateral anterior cerebral artery (ACA) and middle cerebral artery (MCA) territories at two different levels (one at the level of the thalami, and another at level of the centrum semiovale, same level as Alberta Stroke Programme Early Computed Tomography Score locations).
Brain perfusion in bilateral ACA and MCA territories at both the levels was qualitatively graded as normal (score 1), or reduced (score 2) by comparison with normal cerebellar perfusion.
The imaging analysis was performed on IntelliSpace Portal 9.0 workstation (by Philips Healthcare). The subtracted ASL images were overlaid on the 3D T1 images and colored relative ASL perfusion maps were generated after motion correction. The fusion images were evaluated. Corresponding DSC perfusion images were similarly analyzed using IntelliSpace Portal 9.0 workstation (Philips Healthcare) and gamma variate processing was used to generate perfusion maps. CBF maps of ASL perfusion and DSC perfusion were analyzed for qualitative perfusion changes. TTP maps of DSC perfusion were also analyzed and qualitatively scored as normal (score 1), or increased (score 2) by comparison with normal cerebellar TTP. TTP maps were chosen because of the advantages of TTP in case of ischemia, that is, lesions in TTP maps are easily conspicuous and usually well-defined. In addition, TTP is a reproducible technique, as measurements obtained by different centers from a single data set are similar, with less technique-dependent artifacts to complicate interpretation.
Data were expressed using descriptive statistics such as mean ± standard deviation for continuous variables and frequency percentages for categorical variables. Correlation between ASL CBF and DSC CBF was assessed using Spearman’s rank correlation for the qualitative perfusion scores.
There were 34 patients, including 21 females and 13 male patients, with female to male ratio being 1.61. The mean age was 16.3 years, with age range from 2 to 58 years. There were 22 children (<18 years) and 12 adults (≥18 years). Thirty two patients (94.1%) had ischemic type of MMD, and 2 patients (5.9%) had hemorrhagic type. All children had ischemic type of MMD and 83.3% adults (10/12) had ischemic type and 16.7% of adult patients (2/12) had hemorrhagic type MMD.
Patients with ischemic type of Moyamoya disease (n = 32) presented with complaints of headache (n = 10, 31.5%), weakness (n = 27, 84.3%), seizures (n = 11, 34.3%), cognitive decline (n = 3, 9.3%), and personality change and psychosis (n = 1, 3.1%), in varying combinations.
Two patients who had hemorrhagic type of MMD presented with sudden onset of headache, vomiting, and loss of consciousness. Among patients presenting with weakness (n = 27), 20 patients had stroke and seven patients had TIA. MRI findings included cortical infarcts in n = 10 patients (29.4%), cortical watershed infarcts in n = 17 patients (50%), deep watershed infarcts in n = 15 patients (44.1%), and deep lacunar infarcts in n = 1 patient (2.9%). No infarct was seen in n = 6 (17.6%) patients. Two patients with hemorrhagic presentation had basal ganglia bleed with intraventricular hemorrhage. Demographic details, symptoms and MRI findings are tabulated in [Table 2].
|S. No.||Age||Sex||Ischemic/hemorrhagic||Symptoms||Stroke/TIA symptom||Infarct in MRI|
|1.||10||F||Ischemic||Right weakness, seizures, headache||Stroke||Left MCA cortical|
|2.||24||M||Ischemic||Psychosis, personality change||Right frontal watershed cortical watershed|
|3.||11||M||Ischemic||Left weakness, focal seizures||Stroke||Left centrum semiovale infarct deep watershed|
|4.||2||F||Ischemic||Seizures, B/L limb weakness||Stroke||Infarcts B/L watershed cortical and deep watershed|
|5.||32||F||Ischemic||Headache, anxiety depression||Left frontal watershed cortical watershed|
|6.||6||M||Ischemic||Right weakness||TIA||No infarct|
|7.||6||M||Ischemic||Left weakness, headache||TIA||No infarct|
|8.||24||F||Ischemic||Headache, cognitive dysfunction, Right weakness||Stroke||Infarct B/L watershed cortical and deep watershed|
|9.||4||M||Ischemic||Left weakness||Stroke||Right MCA cortical and deep lacunar|
|10.||14||F||Ischemic||Headache, B/L weakness, seizures||TIA||Right frontal watershed cortical watershed|
|11.||21||F||Hemorrhagic||Headache, vomiting, Loss of consciousness||Hemorrhagic||Right Basal Ganglia bleed with IVH|
|12.||11||F||Ischemic||Right weakness||Stroke||Left MCA cortical|
|13.||9||F||Ischemic||Right weakness||Stroke||Left centrum semiovale infarct deep watershed|
|14.||17||F||Ischemic||Left weakness, headache||TIA||Right PCA, posterior MCA infarct cortical and cortical watershed and deep watershed|
|15.||14||F||Ischemic||Left weakness||Stroke||Infarct B/L watershed cortical, cortical watershed and deep watershed|
|16.||5||F||Ischemic||Right weakness||Stroke||Left frontal corona deep watershed|
|17.||22||F||Hemorrhagic||Right weakness, headache, vomiting||Hemorrhagic||Left BG bleed with IVH with chronic Left MCA-PCA infarct|
|18.||11||M||Ischemic||Left weakness||TIA||Deep watershed|
|19.||25||F||Ischemic||Seizures, cognitive decline||No infarct|
|20.||58||F||Ischemic||Left weakness, seizures||Stroke||Right watershed ACA cortical, cortical watershed and deep lacunar|
|21.||13||M||Ischemic||Right weakness||TIA||No infarct|
|22.||33||M||Ischemic||Left weakness||Stroke||Right ACA, watershed infarct cortical, cortical watershed and deep watershed|
|23.||27||F||Ischemic||Right weakness, headache||Stroke||Left MCA cortical|
|24.||15||M||Ischemic||Right weakness||Stroke||Left CS infarct deep watershed|
|25.||9||M||Ischemic||Left weakness||TIA||No infarct|
|26.||21||M||Ischemic||Right weakness||Stroke||Left MCA-PCA corttical watershed and deep watershed|
|27.||4||F||Ischemic||Right focal seizure||No infarct|
|28.||12||F||Ischemic||Right focal seizure, Right weakness, cognitive decline||Stroke||Left frontal, parieto-occipital watershed cortical watershed and deep watershed|
|29.||19||M||Ischemic||Left weakness, headache||Stroke||B/L watershed cortical, cortical watershed and deep watershed|
|30.||18||F||Ischemic||Right weakness||Stroke||Left MCA, watershed cortical and cortical and deep watershed|
|31.||17||F||Ischemic||Headache||B/L watershed deep watershed|
|32.||15||M||Ischemic||Heaadache, seizure, Right weakness||Stroke||Left MCA-PCA watershed cortical watershed|
|33.||12||F||Ischemic||Seizure, Left weakness||Stroke||B/L watershed cortical watershed and deep watershed|
|34.||15||F||Ischemic||Right weakness, giddiness||Stroke||Left MCA-PCA watershed cortical watershed|
On qualitative analysis of perfusion maps of ASL and DSC perfusion the matching index between DSC CBF maps and ASL CBF maps showed poor correlation (mean matching index 0.39 ± 0.31). On Spearman correlation analysis, we did not find any significant correlation between the CBF perfusion maps of ASL perfusion and CBF maps of DSC perfusion (r = −0.028, P = 0.878), as shown in [Table 3]. ASL CBF underestimated the perfusion compared to DSC perfusion.
However, there was significant correlation between CBF maps of ASL perfusion and TTP maps of DSC perfusion (mean matching index 0.79 ± 0.26). On Spearman correlation analysis, we found significant correlation between the CBF perfusion maps of ASL perfusion and TTP maps of DSC perfusion (r = 0.58, P = 0.0003), as shown in [Table 4]. Examples of qualitative analysis of perfusion maps are shown in [Figure 1].
We observed that perfusion maps of ASL perfusion were qualitatively similar to TTP maps of DSC perfusion. Wolf et al. reported similar results using ASL and demonstrated that TTP maps of DSC perfusion correlated best with ASL perfusion MR CBF values (both subjectively as well as objectively) when all patients in the study were considered. However, when cases with a major transit delay were excluded, then DSC relative CBF (rCBF) values correlated best with ASL CBF values.
On qualitative analysis of perfusion maps in our study, there was poor correlation between DSC CBF maps and ASL CBF maps, when we qualitatively graded perfusion as normal (score 1) and reduced (score 2). In a study by Goetti et al., they compared qualitative perfusion scores between ASL perfusion and DSC perfusion and showed a strong and significant correlation between the two. In addition, ASL perfusion had an accuracy, sensitivity and specificity of 93%, 94%, and 93%, respectively, for detecting decreased perfusion per territory while considering DSC as standard of reference. They used qualitatively graded perfusion in each territory as normal (score 1), reduced (score 2), or severely reduced (score 3). However, the results of our study were different from this study and qualitatively there was a lot of difference in perfusion maps of ASL perfusion and DSC perfusion. Our results are similar to study by Mutke et al. who found a weak correlation for DSC-relCBF versus ASL-relCBF (r = 0.24) and found a large spread of values in the Bland-Altman-plot attributed to unreliable CBF-measurement.
A delay in arrival times of the labeled arterial water protons or the gadolinium bolus, as noted in cerebrovascular diseases and also in MMD, affects the CBF estimation in both the modalities. For DSC perfusion, this bolus arrival delay causes bolus dispersion and thus leads to underestimation of rCBF and the overestimation of MTT. For ASL perfusion, delayed transit times do not allow blood to reach tissue parenchyma at the time of image acquisition leading to loss of perfusion signal and artificially low CBF values. On the other hand, at the time of acquisition, the labeled blood may be present in the collaterals leading to false high intra-arterial signal (Arterial Transit Artifact) which can be misinterpreted as hyperperfusion. These inherent problems in DSC and ASL perfusion methods are responsible for the complex interpretation of their maps and thus leading to various incongruence in various studies, all of which use different method of acquisition, post processing methods, labeling duration, post label delay, etc.
To overcome the deleterious effect of delayed transit time on estimation of CBF in ASL perfusion, nowadays use of multiphase ASL is being promoted. Recently, Yun et al. evaluated the usefulness of transit time corrected CBF maps using multiphase ASL-MR perfusion imaging and they concluded that the use of transit time corrected CBF maps can overcome the effect of delayed transit time on conventional ASL perfusion maps. In a study by Fan et al., they noted that standard-delay conventional ASL underestimated rCBF. Transit times correction using multidelay ASL acquisitions led to improved correlation with positron emission tomography (PET), while still underestimating CBF in the cases with long transit delays. However long-label long-delay ASL scans had the strongest correlation with PET. They concluded that post-label delay times of ≥4 s are required in correct estimation of CBF in MMD using ASL perfusion, and this may be combined with using multidelay strategies for accurate ASL assessment of CBF in these patients.
Our results show that on qualitative visual analysis, ASL perfusion CBF maps do not match the DSC perfusion CBF maps and rather match the TTP maps of DSC perfusion. This is explained by inherent problems in estimation of CBF in these techniques because of delay in arrival of label (in ASL perfusion) or contrast bolus (in DSC perfusion) due to the presence of stenotic lesions.
Our study has few limitations. The studies were performed in two machines using pulsed ASL and pseudocontinuous ASL techniques. Children <10 years (n = 8) underwent MRI under sedation and effect of anesthetic medication might have altered the measured CBF. Mutidelay ASL sequence was not available which is more accurate for the estimation of CBF as transit-time corrected CBF maps are provided using this method. Perfusion parameters using Gold standard perfusion estimation methods such as Single Photon Emission Computed Tomography/O−15 PET were not evaluated.
ASL CBF maps correlate well with TTP maps of DSC perfusion and do not correlate with CBF maps of DSC perfusion. The information regarding perfusion given by these two modalities (ASL perfusion and DSC perfusion) is different and should be interpreted keeping these incongruences in mind. Multi-delay ASL and long delay ASL may be required for better estimation of CBF in patients of MMD.
Declaration of patient consent
Ethical committee approval was obtained.
Conflict of interest
There is no conflict of interest.
Financial support and sponsorship