The major findings of this study were as follows: First, we demonstrated patterns of hypoperfusion in subjects with mild to moderate AD by using arterial spin-labeling MR imaging. The brain regions involved were similar to those in previous FDG PET and HMPAO SPECT studies of similar populations. Specifically, we found regional hypoperfusion in the parietal association cortices and posterior cingulate gyri of subjects with AD as compared with the perfusion in those structures in CN subjects. Second, these patterns of hypoperfusion in subjects with AD were largely independent of underlying cortical gray matter atrophy. Third, we observed hypoperfusion of the right inferior parietal lobe in the MCI group compared with that in the CN group. Considered together, these results suggest that arterial spin-labeling MR images, analyzed together with coregistered structural MR images, may depict patterns of reduced brain function in subjects with AD and subjects at risk for AD; this finding is similar to results previously observed with FDG PET or HMPAO SPECT but without the disadvantages of these techniques.
Our first finding of hypoperfusion in subjects with AD—specifically, reductions in the parietal association cortices and posterior cingulate gyri—is consistent with a large volume of PET and SPECT research showing that regional hypoperfusion is most prominent in the sensory and multimodal association cortices, as well as in the posterior cingulate gyri (4
). Results of previous studies have shown a temporal relationship for this pattern, with milder AD focused more in the cingulate gyri and temporoparietal cortices and frontal hypoperfusion seen with more advanced disease (39
). This is consistent with our findings in a group of subjects with mild to moderate AD, in whom the greatest hypoperfusion was shown in the parietal lobes and cingulate gyri and a smaller effect was seen in the frontal lobes. These results are also consistent with those of a previous arterial spin-labeling MR study involving subjects with AD, in which a similar voxel-based approach that showed hypoperfusion most substantially in the temporoparietal cortices was used (21
We also found that nonnormalized measurements of perfusion showed more extensive regional hypoperfusion in subjects with AD compared with those in CN subjects than did perfusion measurements adjusted for a global measure of perfusion for each subject. It is difficult to determine whether this is due to confounding effects of instrumental errors specific to the group or whether there are global disease effects in addition to regional effects. The fact that similar regions remained significantly hypoperfused even after accounting for inter-subject variability of motor strip perfusion does not help separate these effects, but it does suggest that there are regional effects independent of global differences. In contrast to results of the AD group versus CN group comparison, results in the MCI group showed the greatest hypoperfusion relative to the CN group only after normalization. This may result from early disease effects being more localized while the magnitudes of biologic variability and disease effects are more similar at this early stage.
In light of the incomplete understanding of sex effects on regional perfusion in normal aging and dementia, a conjunction analysis for group and sex was done to determine the extent to which men and women with AD show similar patterns of hypoperfusion. We found very similar regions in the parietal lobes, cingulate gyri, and frontal lobes with both the conjunction analysis and the overall group comparison. These results suggest that the regional hypoperfusion seen in AD is largely independent of sex.
Our second finding was that patterns of hypoperfusion in AD were not caused by PVEs of cerebrospinal fluid and white matter in the setting of cortical gray matter atrophy. The separation of underlying structural changes from hypoperfusion is important in understanding the onset and progression of dementia, since these two processes may differ in temporal as well as spatial extent. Results of PET and SPECT studies with asymptomatic subjects at high risk for development of dementia suggest that perfusion deficits may exist in presymptomatic stages before substantial atrophy is present (7
). Furthermore, studies combining both functional and structural measurements demonstrate that metabolic and structural changes associated with AD have a complex relationship both spatially and temporally (41
). PVE correction is particularly important in light of large voxel size in the perfusion image relative to gray matter thickness, which results in greater PVEs. Our results are consistent with those of previous PET and SPECT studies that include corrections for PVEs of cerebrospinal fluid and show that hypoperfusion in the parietal lobes and cingulate are largely independent of atrophy in mild to moderate AD (22
). Additionally, our atrophy correction technique corrects for the PVEs of white matter, thereby accounting for the differential perfusion to gray matter and white matter, a correction that is usually not performed in PET and SPECT studies in which there is correction for atrophy.
Our third finding was that there was hypoperfusion of the right inferior parietal lobe in the MCI group compared with perfusion in the CN group. Notably, the region of greatest difference between the MCI and CN groups was in a region similar to that most affected in the AD group. The AD versus MCI analysis showed that the AD group had significant hypoperfusion in the cingulate gyri and portions of the inferior parietal lobes compared with perfusion in the MCI group, but these regions did not overlap the region most affected in the MCI group compared to perfusion in the CN group. While the significance of the regional hypoperfusion seen in the MCI group relied on a lower voxel-level statistical threshold, its location in the region most affected in the AD group and the lack of a difference between AD and MCI in this same region was consistent with the hypothesis that subjects with MCI may demonstrate regional hypoperfusion in the regions most affected in AD prior to the onset of clinical AD. Our study may have lacked the power to depict more extensive differences between the MCI and CN groups because of the heterogeneity of our MCI group since, presumably, our MCI group included both subjects whose conditions would convert to AD and those whose conditions would not convert. In fact, published results of most studies showing statistically significant regional hypoperfusion in a population with MCI with use of PET or SPECT have included subjects with MCI whose conditions were known to have subsequently converted to AD or those who were at higher risk because of strong genetic risk factors (45
). While it is interesting to note that our MCI group showed a trend toward hypoperfusion in a region severely affected in the AD group, the relevance of this finding depends on future analyses in which the clinical outcome of the subjects with MCI is known.
This study had several limitations. The most important limitation was that data were acquired only for the superior cerebral cortex because of technical limitations that required spin labeling at the circle of Willis, thus excluding major portions of the temporal lobes and inferior frontal lobes from the perfusion image. Consequently, interpretation of our results as they relate to the characterization of the diagnostic groups studied must be tempered by the understanding that the brain regions with the greatest pathologic changes in early disease (ie, median temporal lobe structures such as hippocampal regions and entorhinal cortex) were not included. A second limitation was that transit time and T1 relaxation times of labeled water in blood and brain tissue were not measured because they require longer imaging times, which may have increased the attrition rate. Therefore, increased transit times in imaging of AD and MCI because of cerebrovascular factors or decreased T1 relaxation may have mimicked hypoperfusion. While we attempted to minimize group differences due to technical factors and to account for intersubject variations in global perfusion when comparing groups, the more reliable way to study effects of disease is by using quantitative arterial spin-labeling MR imaging that also takes into account effects of transit times and blood and tissue water relaxations.
In conclusion, regional hypoperfusion measured with arterial spin-labeling MR imaging in a population with AD is similar to that seen with other functional modalities and is, to a large extent, independent of underlying gray matter atrophy. These results suggest that perfusion MR imaging is a useful tool for the functional characterization of AD and may be useful for early detection of the disease.