Donepezil Hydrochloride Preserves Regional Cerebral Blood Flow in Patients with Alzheimer’s Disease

Subjects

Thirty-five patients (22 men, 13 women; age range, 57–80 y; mean age, 70.4 y) who complained of memory impairment were recruited from an outpatient memory clinic of the National Center Hospital for Mental, Nervous, and Muscular Disorders, National Center of Neurology and Psychiatry, between 1998 and 1999. The clinical diagnosis of AD was based on the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) (3), and criteria of the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association (4). The patients had a Mini-Mental State Examination score > 16. Dementia was staged using the clinical dementia rating and ranged from 0.5 to 2. The patients were randomized to receive daily oral doses of either donepezil (5 mg/d) or placebo. Each patient underwent baseline SPECT and MRI at the time of the initial evaluation and was clinically followed by a second SPECT and MRI examination after 12 mo. Demographic information is shown in Table 1.

A subset of neuropsychological tests from the standard clinical protocol was selected a priori (Table 2) to profile multiple cognitive abilities, including frontal lobe functioning, memory, visuospatial ability, and language skills. Raven’s Coloured Progressive Matrices (5) was used to measure visuoperceptual, attentional, and problem-solving skills. In addition, the digit span task (5) was used to measure attention. The learning of a list of 10 words and story recall (5) were used to measure short-term verbal memory and language ability. The Ray-Osterrieth complex figure test (5) was used to evaluate both visuoconstructional ability and visual memory. Measures of frontal lobe functioning included the Stroop test (6). Impaired performance on the Stroop test is common in patients with frontal lobe lesions and is also associated with a specific pattern of rCBF in the anterior cingulate cortex (7). All tests were administered and scored according to standard protocols. Age- and education-adjusted norms were applied when available. No focal brain lesions on MRI were observed except for age-related hyperintensities on T2-weighted images.

Global and Regional CBF Measurements

Before SPECT scanning was performed, an intravenous line was established in the right cubital vein of all patients while they were supine with their eyes closed in a dimly lit, quiet room. Each patient received a 600-MBq intravenous injection of ^99mTc-ethyl cysteinate dimer (ECD). The global cerebral blood flow (CBF) was noninvasively measured using graphic analysis as described in previous reports, without any blood sampling (8–10). Passage of the tracer from the aortic arch to the brain was monitored in a 128 × 128 format for a short period of less than 2 min at 1-s intervals using the rectangular gamma camera of a triple-head SPECT system (MULTISPECT 3; CTI, Knoxville, TN/Siemens Medical Systems, Inc., Hoffman Estates, IL) equipped with high-resolution fanbeam collimators. Regions of interest were hand-drawn over the aortic arch and both brain hemispheres. The hemispheric brain perfusion index was determined before the start of the initial backdiffusion of the tracer from brain to blood using the method described for a previous study (8): Eq. 1 where BPI is the brain perfusion index, ROI_aorta is the region of interest drawn over the aortic arch, ROI_brain is the region of interest drawn over the brain, and ku is the unidirectional influx rate of the tracer from blood to brain, as determined by the slope of the line in graphic analysis within the first 30 s after injection. Then, brain perfusion index (x) was converted to global CBF values (y) obtained by ¹³³Xe-inhalation SPECT studies (y = 2.60x + 19.8) (8).

Ten minutes after the injection of ^99mTc-ECD, brain SPECT imaging was performed. The projection data were obtained in a 128 × 128 format for 24 angles in 120° for each camera with 50 s per angle. A Shepp and Logan filter was used for SPECT image reconstruction at 0.75 cycle per centimeter. Attenuation correction was performed using Chang’s method. The following linearization algorithm (11) of a curve-linear relationship between brain activity and blood flow was applied to calculate rCBF and to correct for incomplete retention of ^99mTc-ECD in the brain: Eq. 2 where Fi and Fr represent CBF values for region i and a reference region, respectively, and Ci and Cr are the SPECT counts for region i and a reference region, respectively. The cerebral hemisphere was used as the reference region, and global CBF obtained from graphic analysis was substituted for the CBF value for the reference region. The linearization factor α was set at 2.59, which was the value proposed by Friberg et al. (11).

Image Formatting

All subsequent image manipulation and data analysis were performed on a Windows 98– (Microsoft, Redmond, WA) and UNIX-based computer. The software for image manipulation included MATLAB (versions 4.2c and 5.3; The MathWorks, Inc., Natick, MA) and the statistical parametric mapping (SPM) programs SPM99 and SPM96 (Wellcome Department of Cognitive Neurology, London, U.K.). Transverse-slice image volumes were made compatible with SPM by the creation of usable headers for the images. Briefly, for each image a file was created that contained data on image size, number of slices, pixel depth (16-bit), maximum pixel value, and voxel size. All brain-slice images were then sampled and averaged to arrive at a mean pixel intensity value for that image. The intensity threshold was set at 80% of the whole-brain mean. The images were then spatially normalized in SPM99 to a standardized stereotactic space based on the atlas of Talairach and Tournoux (12), using 12-parameter linear and also 8 nonlinear iteration algorithms. Because the distribution of ^99mTc-ECD in the brain is different from that of ¹⁵O-water, which was used as the PET template in the SPM program, we used an original ^99mTc-ECD template, which was prepared in a previous study (13), for spatial normalization. The normalization routine also included further isotropic smoothing, to a total of 12 mm. This corresponded to almost twice the resolution (7.6-mm full width at half maximum) of the SPECT scanner. The initial parameters of the image were 128 × 128 × n, where n is the number of slices. The final image format was 16-bit, with a size of 79 × 95 × 68 and a voxel size of 2 × 2 × 2 mm.

Image Analysis

Data analysis was performed using SPM96. Statistical parametric maps are spatially extended statistical processes that were used to characterize regional specific effects in imaging data. SPM combines the general linear model and theory of gaussian fields (14,15) to make statistical inferences about regional effects. Regional CBF differences were examined as an intergroup comparison between the donepezil-treated and placebo-treated groups at baseline and in the follow-up SPECT study and as an intragroup comparison between the baseline and follow-up SPECT studies in each group. Moreover, to examine images for specific regions of differences in time-course perfusion changes from the baseline to the follow-up SPECT studies between donepezil-treated and placebo-treated patients, we compared 2 studies using 2 contrasts with study interaction (multistudy and multisubject with different conditions in SPM96). The first and second contrasts examined areas of elevation and decline, respectively, of time-course perfusion changes in the donepezil-treated patients compared with placebo-treated patients. The analysis was performed with or without the implication of changes in global CBF levels as a confounding covariate. The gray matter threshold was set at 0.8. We first used raw data (absolute rCBF parametric maps without global normalization) and then adjusted rCBF images (normalization of global CBF for each subject to 50 mL/100 g/min with proportional scaling) to compare relative rCBF distribution in 2 groups. The resulting set of values for comparison constituted a statistical parametric map of the statistic SPM{t}. The SPM{t} maps were then transformed to the unit of normal distribution (SPM{z}) and reached a threshold at P = 0.001. The resulting regions were examined using multiple comparisons. The significance of each region was estimated at a threshold of P = 0.05 using distributional approximations from the theory of gaussian fields (15). These areas of significance were best seen when overlaid on a normalized MR image to obtain a clear view of the location of the perfusion changes.

Neuropsychological Profile

Table 2 shows the mean scores of the donepezil- and placebo-treated patients on the neuropsychological tests. Compared with the donepezil-treated patients, the placebo-treated patients performed significantly worse on both forward digits (P = 0.0007) and reverse digits (P = 0.03) in the digit span test, on 30-min delayed recall in the Ray-Osterrieth complex figure test (P = 0.02), and on word-dot in the Stroop test (P = 0.01) in the follow-up SPECT study. For the placebo-treated patients, mean scores for the Mini-Mental State Examination (P = 0.003), forward digits in the digit span test (P = 0.01), and word-dot in the Stroop test (P = 0.03) declined significantly in the follow-up SPECT study. In contrast, no significant differences (P > 0.05) were found for the other neuropsychological tests.

Global Effects

The baseline and follow-up global CBF values were 35.9 ± 3.4 and 35.5 ± 2.7 mL/100 g/min, respectively, in donepezil-treated patients and 36.8 ± 4.3 and 35.4 ± 4.4 mL/100 g/min, respectively, in placebo-treated patients. There was neither a group × time interaction effect nor a main effect (group or time) on global CBF in a 2 × 2 (donepezil/placebo × baseline/follow-up) repeated-measures ANOVA.

Regional Effects

Adjusted rCBF.

Figure 1 and Table 3 show the voxels that had a significant decline in adjusted rCBF during the follow-up period among placebo-treated patients compared with donepezil-treated patients (no voxel significantly increased). Significant group × time interaction effects were observed in time-course rCBF changes between placebo-treated and donepezil-treated patients. Compared with the donepezil-treated patients, the placebo-treated patients showed a significant rCBF decline in the right and left anterior cingulate gyri, right prefrontal cortex, right inferior parietal lobules, and right middle temporal gyrus. Significant differences in adjusted rCBF were not found between the 2 groups at baseline and in the follow-up SPECT study, nor were significant differences present between the baseline and follow-up SPECT studies in either intragroup comparison.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Adjusted rCBF decrease in placebo-treated AD patients compared with donepezil-treated AD patients and from baseline study to follow-up study as shown by SPM. Significant decrease is seen as maximal intensity projection (MIP) and colored areas superimposed on standard 3-dimensional anatomic template.

View this table:

• View inline
• View popup

TABLE 3

Location and Peaks of Significant (P < 0.001) Decreases in Adjusted rCBF