Arterial spin labeling magnetic resonance perfusion study to evaluate the effects of age and gender on normal cerebral blood flow
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.178037
Source of Support: None, Conflict of Interest: None
Purpose: Arterial spin labeling (ASL) is a noninvasive magnetic resonance (MR) perfusion technique to detect changes in blood flow. This study was undertaken to obtain a reference set of normal values of cerebral blood flow (CBF) in different age groups using three-dimensional pseudocontinuous ASL (3D PCASL) technique. The existence of an age-related decline in the gray matter (GM) and white matter (WM) CBF was evaluated. The gender-related CBF was also analyzed.
Keywords: Global and regional cerebral blood flow; magnetic resonance perfusion imaging; pseudocontinous arterial spin labeling
Continuous and sufficient cerebral blood flow (CBF) is vital to neural function and an important measure in the understanding of brain pathophysiology. CBF reflects the amount of blood perfusion in the brain, measured in ml of blood per 100 g of brain tissue per minute. Its value decreases with age in both sexes. A higher CBF value has been reported in women than in men probably due to the effect of estrogen. ,, Kety and Schmidt were the first to measure mean CBF, using nitrous oxide (Fick's formula) and reported a global mean CBF level of 54 ± 12 ml/100 g/min, in healthy young men, which is still regarded as a reasonable value.  Many pathological conditions such as acute stroke, brain tumors, neurodegenerative diseases, and epilepsy are associated with abnormal CBF values. CBF is also important in the monitoring of treatment of many brain diseases and also in the functional brain imaging. , Various methods for the quantitative measurement of CBF have been developed such as single photon emission computerized tomography, positron emission tomography (PET), computed tomography, and magnetic resonance imaging (MRI). Almost all of these methods employ a model of tracer kinetics which requires the contrast agent to quantify the CBF. These techniques are expensive and most require radiation exposure. The authors of these studies have shown a significant negative correlation between age and CBF in both gray matter (GM) and white matter (WM) and have given possible explanations for the changing CBF based on development and also regional variation. ,,,,, Thus, it is important to establish a robust noninvasive method, suitable for longitudinal and cross-sectional studies of various neurovascular and neurodegenerative diseases. Arterial spin labeling (ASL) is a noninvasive MR perfusion technique reported first by Detre et al.,  in 1992, based on the use of magnetically labeled blood-water protons as an endogenous tracer. Due to the noninvasiveness and easy availability of the ASL technique, there is revival of interest in reinvestigating the effects of age and gender on CBF in the normal population. In the past, a number of studies have evaluated the age- and gender-related effect on CBF using the ASL technique. ,,, In the current study, we have used the fast three-dimensional pseudocontinuous ASL (3D PCASL) technique to quantify age-related changes in the whole and regional CBF and have also investigated the effect of gender in influencing CBF. Studies with larger subjects that use the PCASL technique for determining the normal mean CBF values in global as well as regional GM and WM in all the different age groups (children to elderly) and that also investigate the impact of gender differences are scant, and are urgently required.
This was a prospective study approved by the institutional review board and was undertaken at the Department of Radiodiagnosis, Sanjay Gandhi Postgraduate Institute of Medical Sciences, Lucknow (conducted between June 2010-December 2012) after obtaining informed consent. The study included 160 normal individuals (age range: 6-72 years, 95 male and 65 female subjects) with no history of any neurological disease, substance abuse, injury, and nonspecific headache. The recruited patients were found to be normal on routine imaging. Patients with known neurological disorders, structural abnormality, and any other disorder with a known effect on the CBF were excluded from the study. The subjects were divided into four different age groups [Table 1] that included children (0-12 years), adolescents (12-18 years), adults (18-60 years), and elderly (>60 years) subjects.
Magnetic resonance imaging acquisition
MRI was performed on a 3T MR scanner (Signa HDxt, General Electric, Milwaukee, USA) with the 12 channel head component of the 16 channel head-neck-spine. Routine imaging using axial T2-weighted propeller and T1-weighted sequences were performed with the following parameters: TE = 91 ms/TR = 3200 ms/ (number of excitations) NEX = 1.2/number of slices = 46/slice thickness = 3 mm/FOV = 24/image matrix = 320/256 and TE = 9.0 ms/TR = 1325 ms/NEX = 1/number of slices = 46/slice thickness = 3 mm/FOV (field of view) = 24/image matrix = 320/256, respectively. With adequate background suppression, PCASL was performed using the following acquisition parameters: Label duration = 1.5 s/postlabeling delay = 1. 5 s/TR = 4.4 s/TE = 9.2 ms/frequency = 512/phase = 8/NEX = 3/number of slices = 46/FOV = 24/slice thickness = 3 mm/bandwidth = 62.50/scan time 4 min, and pulse labeling plane placed just below the volume of interest.
Postprocessing and data analysis
Data postprocessing was performed on workstation using Functool Brain stat Software (ADW4.4 GE workstation) for 3D ASL with automated generation of quantitative perfusion and CBF maps [Figure 1] and [Figure 2] from each subject. The mean values of GM and WM CBF in frontal GM, frontal WM (FGM, FWM), parietal GM, parietal WM (PGM, PWM), temporal GM, temporal WM (TGM, TWM), and occipital GM, and occipital WM (OGM, OWM) regions were extracted by averaging the data obtained in 10 manually selected regions of interest (ROI) by placing elliptical ROIs (4-6 mm 2 ) on the CBF maps overlaid on T1-weighted images acquired with the same planning. In addition, CBF mean values were also measured in the thalamus, basal ganglia, and cerebellum. From these individual subject data, mean values of global GM and WM CBF were calculated by averaging the above-mentioned regional values [(GM CBF = FGM + PGM + TGM + OGM/n) and (WM CBF = FWM + PWM + TWM + OWM/n)] for each age group (n = 4).
Based on a pilot study in 10 subjects (unpublished), it was found that age and CBF have an inverse relationship with Pearson correlation r = −0.6. Thus, to show the effect of age on WM CBF and GM CBF including the regional differences, 28 subjects were required in each group using 80% power, 5% level of significance, and design effect of 1.5. Assuming a design effect of 1.5, the modified sample size comes out to be 39 subjects in each group. Thus, a total of 160 subjects were included in the study, i.e. 40 subjects in each group. Descriptive statistics was shown as mean ± standard deviation. One-way ANOVA test was used to analyze the anatomical regions within different age groups. All statistical analyses were performed using SPSS 22.0 (SPSS, Chicago, IL, USA). The sample sizes were estimated using PASS 6.0.
The mean CBF values calculated in the whole GM and WM for each subgroup are reported in [Table 1]. The mean value of GM CBF is higher in comparison to WM CBF in each age group, with age-related decrease in mean values. A significant difference (P = 0.001) in the mean GM and WM CBF between different age groups, with progressive age-related decline, was found using one-way ANOVA test. The individual subject's GM CBF and WM CBF data are plotted with present to their age [Figure 3]. On bivariate analysis using Pearson's correlation, a significant negative correlation was observed between an increasing age, and mean GM CBF and mean WM CBF values [(r = −0.80, P = 0.001) and (r = −0.59, P = 0.001)], respectively.
To examine gender differences in GM CBF and WM CBF, the analysis was also performed on the male and female subgroups in different age groups. The female subjects have a higher mean GM CBF in ml/100 g/min (group 1 = 76.57 ± 10.9, group 2 = 64.2 ± 4.4, group 3 = 60.7 ± 5.2, and group 4 = 48.54 ± 3.1) in comparison to the male subjects (the mean GM CBF in group 1 = 72.77 ± 10.8, group 2 = 62.7 ± 5.2, group 3 = 60.7 ± 5.2, and group 4 = 48.15 ± 4.2). Similarly, the WM CBF mean values (in ml/100 g/min) were higher in the female subjects (group 1 = 22.18 ± 1, group 2 = 20.79 ± 2.2, group 3 = 19.31 ± 1.4, and group 4 = 19.34 ± 1.7) as compared with the male subjects (group 1 = 21.98 ± 1.7, group 2 = 20.80 ± 2.1, group 3 = 18.98 ± 1.2, group 4 = 18.27 ± 0.8). However, no statistically significant gender-related effect was found in any age group (P > 0.05) in both GM and WM CBF.
Finally, to create a regional reference for the range of normal perfusion values as a function of age, the analysis of regional CBF values was also performed in regions that included the frontal cortex, parietal cortex, occipital cortex, deep gray matter structures and cerebellum [Table 2] and [Table 3]. One-way ANOVA was used to test the mean differences in the regional CBF based on different age groups in all the GM regions (FGM, PGM, TGM, OGM, basal ganglia, thalamus, and cerebellum [Table 2] and [Table 3] [P = 0.001]).
In regional WM CBF also, a significant difference was seen between the different age groups (P = 0.001) [Table 4]. A bivariate analysis of mean regional CBF values was done with age in all of the investigated brain regions. The results showed the same general pattern of negative correlation with age in all of the investigated brain regions (FGM r = −0.86, PGM r = −0.87, TGM r = −0.81, OGM r = −0.78, thalamus r = −0.65, BG r = −0.71, FWM r = −0.08, PWM r = −0.40, TWM r = −0.36, OWM r = −0.38, cerebellum r = −0.51) [Figure 4] and [Figure 5].
The determination of the changing perfusion patterns accompanying normal brain development, and the creation of a reference set of normal values of CBF in different age groups is a necessary prelude to the use of CBF measurement in the diagnosis and management of brain disorders. This study reports the age- and gender-related changes in cerebral perfusion on 160 normal subjects of varying age (6-72 years) measured by the PCASL technique at 3T MRI scanner. Xu et al.,  compared the reliability and accuracy of PCASL with 15-O-water PET in normal children as well as young, and elderly subjects. The recently published studies also showed that compared to the previous ASL and PET perfusion studies, PCASL provides better reliability in repeated measurements for the entire age group from children to elderly subjects. ,,,
The whole brain GM CBF, WM CBF, and cerebellum CBF values obtained in our study [Table 1] and [Table 3] are consistent with those obtained in the previously reported studies done with PCASL, CASL, pulsed arterial spin labeling (PASL) in children and adults, ,,, and through other techniques that have used PET and dynamic susceptibility (DSC) MRI. ,,, The whole GM CBF and WM CBF values in children (0-12 years) [Table 1] are similar to the PASL study done by Wang et al., who performed a study on 9 children (1 month-10 years) [Table 5].  In the adolescent group (12-18 years), our whole GM CBF and WM CBF values [Table 1] are in concordance with the values obtained with CASL by Biagi et al., performed on 8 adolescents (mean age 16 ± 2 years) [Table 5]. 
In adults (18-60 years), our whole GM CBF and WM CBF values [Table 1] are consistent with the values obtained with pulsed arterial spin labeling (PASL) by Chen et al., with continuous arterial spin labeling (CASL) by Biagi et al., with PCASL by Xu et al., and, with dynamic susceptibility contrast (DSC) MRI by Bonekamp [Table 5]. ,,,, The values obtained in the elderly age group in whole GM CBF and WM CBF [Table 1] are consistent with values obtained with PASL by Chen et al., and with PCASL by Xu et al. [Table 5]. ,
Effect of age on whole gray matter and white matter cerebral blood flow
A significant difference (P = 0.001) was found in the GM and WM CBF among different age groups [Table 1]. On regression analysis, a significant negative correlation was found between age and GM and WM CBF (r = −0.8, r = −0.59). These findings were similar to that reported in previous studies conducted using ASL, PET, and Xenon 133. ,,,, This result were also in agreement with the few pediatric studies conducted both with MRI and with different imaging techniques, which show differences in the cerebral perfusion between children and adults. ,,,, On comparing the regional CBF values in adults and the elderly group, the values for FGM, TGM, OGM, and thalamus found in our study were consistent with the values obtained with PASL by Chen et al. [Table 5]. 
There are a number of possible explanations for the high values of CBF during the development phase. During its growth and development, the brain produces a vast excess of neurons, synapses, and dendritic spines and achieves its maximum volume by the middle of the second decade of life (for both males and female subjects ), reaching a plateau by the 12 th year. 
Several factors have been proposed to account for the reduction in CBF in the older age groups,. Firstly, there is a loss of brain substance and hence decreased cerebral metabolism. It has been reported that populations of cerebral cortical neurons progressively decrease in number with age, especially in the superior temporal and precentral regions.  A gradual increase in the ventricular size with advancing age also correlates well with reduction in the CBF and metabolism.  Chen et al., demonstrated that reduction in the CBF in the older age is independent of cortical thinning and atrophy.  In our study, we did not consider the effect of atrophy on CBF.
Effect of age on regional cerebral blood flow
We found a substantial spatial nonuniformity in the various regional GM and WM CBF values that were independent of age, with a higher CBF being found in the frontal lobe compared to other regions [Table 2] and [Table 4]. The results correspond to the previous values found in the 15-O PET and CASL studies, that included the frontal and frontal-temporal regions. ,,,,,,,,
Significant age-associated regional CBF reduction was widely observed throughout the cortex, and the current findings may indicate a distinct pattern characteristic of the normal aging process. Our measurements of the CBF-age relationship in the frontal and parietal regions were in excellent agreement with the observations made by Biagi et al;  while those in the deep gray matter nuclei are in agreement with the longitudinal findings by Beason-Held et al. ,, The zone of FGM hyperperfusion extended backwards approximately to the Rolandic fissure in accordance with the data reported by Ingvar and Schwartz. , According to the modern neurophysiologic concepts, developed in particular by Luria,  the frontal lobes play a fundamental role in programming behavior, whereas the parietal, temporal, and occipital regions are involved in analyzing and memorizing information. It, therefore, seems logical that during the state of rest, while the subject is receiving a low level of sensory stimulation, only the regions involved in thought process and behavioral programming remain active. This explains the finding of hyperperfusion in the frontal region.
Effect of gender on cerebral blood flow
Global GM CBF and WM CBF showed disparities between the genders, with women being associated with relative higher CBF values. This is in accordance with the prior published results. ,,,,, However, the results were not statistically significant (P > 0.05) in our study.
Despite the manual region of interest (ROI)-based analysis of CBF, our results were consistent with that obtained in the previous studies that have used a voxel-based morphometric analysis.
Various factors such as end-tidal CO 2 , heart rate, respiratory rate, wakefulness, usage of caffeine, and drugs that can have an impact on the quantitative perfusion measurements were not considered. A manual region of interest (ROI) was used on the perfusion maps overlaid on T1-weighted images.The voxel-based morphometry, using statistical parametric mapping between various age groups, could have reduced the error that may have been induced in the placement of ROIs and would have also helped to provide cluster analysis on a voxel level. Time was kept constant for the pulse labeling duration (1.5-2.0 s) in all age groups and this might have affected the signal obtained.
We report a reference dataset of normal values of CBF in different age groups using PCASL. Our results demonstrate a significant age-related decline in various regional and global GM and WM CBF values with increasing age. Knowledge of the patterns of age-associated CBF decline is invaluable for distinguishing normal changes from the more detrimental disease-related degeneration. Regional variability of CBF helps in the assessment of local cerebral disease.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5]
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5]