In accordance with the principles of the Declaration of Helsinki, we prospectively registered 182 serial patients who consulted the memory clinic of the Mie University Hospital. All procedures followed the clinical study guidelines of the ethics committee of the Mie University hospital and were approved by the internal review board. All procedures were described to the patients, and informed consent was obtained from them or their caregivers in the written form. Neurologists with sufficient experience in examining patients with dementia comprehensively examined each patient in our study. We collected data from the patients who fulfilled the following inclusion criteria: (1) patients who consulted the Memory Clinic of the Mie University Hospital from October 2013 to August 2016 and were diagnosed with AD based on pre-established criteria, fulfilling the criteria for probable AD of the National Institute of Neurologic Disorders and Stroke/Alzheimer Disease and Related Disorders Association (NINCDS-ADRDA) (41); (2) patients who received a neuroimaging examination on a 3T-MRI; (3) patients who received a ^99mTc-ECD SPECT examination; and (4) patients who underwent neuropsychological assessments. The exclusion criteria were as follows: (1) patients who were not examined using MRI and ^99mTc-ECD SPECT; (2) patients who were not administered neuropsychological assessments; (3) patients who were diagnosed with dementia other than AD; and (4) patients who had normal cognitive function.

Magnetic resonance imaging studies were performed using two different 3T-MRI scanners (Achieva and Ingenia; Philips Health Care, Best, the Netherlands). We used 3D-FLAIR and T1-weighted images. The parameters for 3D FLAIR were as follows: repetition time (TR), 6,000 ms; echo time (TE), 310 ms; inversion time, 2,000 ms; turbo factor, 203; sensitivity encoding factor, 3; field of view (FOV), 25 cm; matrix size, 480 × 256; and section thickness, 1.14 mm. The parameters for T1-weighted images were as follows: TR, 7.6 ms; TE, 3.6 ms; flip angle, 8°; FOV, 250 mm × 250 mm; in-plane resolution, 1.04 mm × 1.04 mm; and slice thickness, 0.7 mm.

Tissue quantification was performed using a novel in-house software (FUsed Software for Imaging Of Nervous system: FUSION) that yielded an individualized volumetric profile of brain tissue. The obtained T1-weighted and FLAIR images were imported from DICOM format files for processing. To increase the accuracy of segmentation, we used the Lesion Segmentation Tool for lesion filling (42). Lesion filling was applied to T1-weighted images that were in alignment with the lesion probability map. For the preprocessing level, T1-weighted images were coregistered to FLAIR images. Next, to separate WM, segmentation was performed by using the T1-weighted images and a mask of cerebral ventricles. The preprocessing function was based on SPM 8 (Wellcome Trust Centre for Neuroimaging, UCL). Second-level tissue segmentation was performed to separate WMHs from WM, using a semiautomated operation that extracted the pixels falling within predetermined value as WMHs. The WMH volume, which appeared as hyperintense areas on FLAIR images, was quantified for each area. The brain tissue was classified into four areas based on the division of the longitudinal fissure of the cerebrum and central sulcus. WMHs were automatically classified as periventricular hyperintensity (PVH) or deep WMH (DWMH), and their corrected volumes were calculated in cubic centimeters (cc).

Intravenous radionuclide angiography was performed by bolus injection of the reconstituted 99mTc-ethyl cysteinate dimer (ECD) (600 MBq). Passage of the tracer from the aortic arch to the brain was monitored in a 128 × 128 format for 120 s at 1-s intervals using a three-head gamma camera system (GCA-9300A/DI, Toshiba, Tokyo, Japan) equipped with low-energy high-resolution fanbeam collimators. SPECT images were reconstructed by filtered back-projection using a ramp filter follower and postprocessing with a Butterworth filter. The triple-energy window technique was employed for scatter correction. ROIs were placed manually over the aortic arch and bilateral cerebral hemispheres. Time activity curves of these two ROIs were plotted and the brain perfusion index (BPI) was determined as described previously. BPI was then converted to the mean CBF (mCBF) value. The rCBF values were obtained by the conversion of total counts in brain SPECT into mCBF, using Lassen’s correction (43).

A three-dimensional stereotactic region of interest (ROI) template (3DSRT) program (FUJIFILM RI Pharma Co., Ltd.) was applied to assess the regional quantitative value (44–48). 3DSRT is a fully automated rCBF quantification program that can be used to examine a total of 636 ROIs. These ROIs are categorized into six brain segments on the 3DSRT template: bilateral parietal, temporal, and posterior cingulate cortices determined by previous studies (24–29). The blood flow to each ROI was quantified in mL/100 g/min.

Finally, the results of evaluation of the SPECT images were fused with WMHs on MRI scans (Figure 1).

The Mini-Mental State Examination (MMSE) (49) and Raven’s Colored Progressive Matrices (RCPM) (50) were used to quantify intellectual function. Memory was evaluated using the Rivermead Behavioural Memory Test (RBMT) (51). Assessment of visuospatial constructional ability was based on the method described by Strub and Black (52). A simple cube and a Necker cube were shown to the participants, and they were asked to draw them one by one. Each drawing was scored by assigning one of four possible grades (0: poor, 1: fair, 2: good, and 3: excellent). Frontal lobe function was assessed using two tasks, word fluency (WF) and the trail-making test A and B (TMT-A/-B) (53). The WF test consisted of category and letter domains. For the categorical WF, participants were asked to name as many animals as possible in 1 min. For the letter WF task, participants were asked to say the name of objects that begin with each of four phonemes, ka, sa, ta, and te (54).

Statistical analyses were performed with the Statistical Package for the Social Sciences, Version 20 (IBM Corp., Armonk, New York, NY, USA). Statistical analyses were conducted using one-way analysis of variance for continuous variables and the Ryan method for pairwise comparisons. We performed multivariate regression analysis using a generalized linear model. As predictor variable, we used age, and the PVH, DWMH, and CBFs of temporal, posterior cingulate, and parietal regions. As a response variable, we used MMSE, RCPM, RBMT, TMT-A/B, WF, and visuospatial constructional ability. A p-value of <0.05 was regarded as statistically significant.

A total of 182 patients were registered for the present study, and 29 patients fulfilled the inclusion criteria. Their clinical characteristics are shown in Table 1. The mean (±SD) age was 73.1 ± 7.9 years, and 18 of the participants were female. The mean (±SD) MMSE was 23.1 ± 3.0.

The participants’ WMH volumes and rCBF values are shown in Table 2 (a,b). The mean total WMH volume was 18.4 ± 19.3 cc. Analysis of variance [F(3,84) = 2.748, p < 0.05] and pairwise comparisons revealed that posterior DWMH volume was significantly larger than the anterior PVH (p = 0.038), anterior DWMH (p = 0.043), and posterior PVH volume (p = 0.009).

The effect of regional WMHs and rCBF on neuropsychological test results was evaluated with multivariate regression analysis (Table 3). The analysis showed that increased age correlated negatively with intellectual function (p = 0.008).

Multivariate regression analysis also showed that WMHs and rCBF were significantly associated with memory and intellectual, frontal, and visuospatial functions, as follows. An increased anterior PVH volume negatively affected visuospatial constructional ability (p = 0.016) and increased posterior DWMH volume negatively affected WF (category), whereas increased volumes in other areas did not have negative impacts on cognitive function. In addition, decreased rCBF in the parietal areas negatively affected RCPM (time) (p = 0.033), TMT-A (p < 0.001), and visuospatial constructional ability (p = 0.012), whereas decreased rCBF in other areas did not.

In particular, the anterior PVH, the posterior DWMH, and the parietal regions were key areas that negatively affected cognitive function. We present a representative patient with increased posterior WMH volume and decreased rCBF of the parietal region as a typical example in Figure 1.

We compared MRI and SPECT findings with neuropsychological data from outpatients of the memory clinic of our hospital. An analysis of variance revealed that the posterior DWMH volume was significantly larger than the anterior PVH, anterior DWMH, and posterior PVH volumes. Multivariate regression analysis showed that increased anterior PVH and left posterior DWMH volumes and decreased rCBF of the parietal area correlated negatively with cognitive function, whereas other areas did not have negative effects on cognitive function.