- Top of page
Aims: The purpose of the present study was to investigate whether there were correlations between atrophy of the entorhinal cortex and individual regional cerebral blood flow (rCBF) in patients with Alzheimer's disease (AD) and amnestic mild cognitive impairment (MCI) to better clarify the relationships between morphological and functional changes in AD.
Methods: Twenty-six patients including sixteen AD and 10 amnestic MCI patients were enrolled. Z scores of voxel-based specific regional analysis system for AD (VSRAD) were determined to assess the degree of atrophy of the entorhinal cortex. Single-photon emission computed tomography (SPECT) and 3-D stereotaxic region of interest template (3DSRT) were used to quantify absolute rCBF.
Results: The Z scores of the entorhinal cortex were found to have significant negative correlations with the absolute rCBF in the bilateral hippocampus, thalamus and temporal regions. A negative correlation between Z scores and rCBF of the cerebellum region, especially on the right side, was also noted.
Conclusions: Atrophy of the entorhinal cortex had an obvious functional relationship with rCBF changes in the hippocampus, thalamus, temporal lobe and cerebellum in AD and MCI patients, which was attributed to their close anatomical and physiological connections.
ALZHEIMER'S DISEASE (AD) is a neurodegenerative disorder characterized by progressive cognitive functional decline and behavioral disturbances, and remains the most common type of dementia. The entorhinal cortex, a portion of the anterior parahippocampus gyrus, is considered to be the prime site of neuropathological changes in the earliest stages of AD.1 The size of the entorhinal cortex was shown to be a useful index to discriminate patients with mild AD from healthy controls, as well as being a sensitive predictor of conversion to AD in traditional regions of interest (ROI) studies of magnetic resonance imaging (MRI).2 The atrophy rate in the entorhinal cortex has been found to be higher than that of the hippocampus, consistent with the view that AD pathology begins in the entorhinal cortex.3
Previous MRI studies were based on ROI analysis, which is, however, both time-consuming and observer dependent. Recently, a new method of voxel-based morphometry (VBM) has been developed and has overcome many of the limitations of ROI analysis, automatically mapping any gray matter loss on a voxel-by-voxel basis after anatomic standardization based on statistical parametric mapping (SPM).4 VBM confirmed temporal lobe atrophy in AD, and it was reported to be more accurate than ROI-based analysis for discriminating mild to moderate AD patients from controls.5,6 Subsequently, Matsuda used VBM to develop an automated software program, voxel-based specific regional analysis system for Alzheimer's disease (VSRAD).7,8 Z score of VSRAD became an indicator reflecting the degree of atrophy of the entorhinal cortex for the diagnosis of early AD. Hirata et al. found a high accuracy (87.8%) for discriminating patients with very early AD at the mild cognitive impairment (MCI) stage from normal control subjects by means of the Z scores of VSRAD.9
In contrast, Takeuchi et al. developed a fully automated regional cerebral blood flow (rCBF) quantification software, 3-D stereotaxic ROI template (3DSRT), which utilized SPM and allowed more objective assessment of rCBF by setting the ROI identically on anatomically standardized single-photon emission computed tomography (SPECT) images.10 The automated SPECT analysis software could contribute to increase the accuracy of diagnosis of AD, especially in the early stage.11
An animal study demonstrated that bilateral lesions of the entorhinal and perirhinal cortices caused a persistent decline of glucose metabolism in the remote brain regions, the posterior cingulate gyrus, the inferior parietal lobe, posterior temporal lobe, associative occipital cortices, and posterior hippocampus.12 Presumably, this decline of glucose metabolism is due to functional inactivation secondary to the organic lesions of the entorhinal and perirhinal cortices. The aim of the present study was to determine whether a correlation existed between Z scores of VSRAD and absolute rCBF assessed on SPECT with technetium-99m-l, l-ethyl cysteinate dimmer (99mTc-ECD) in patients with AD in order to examine relationships between morphological and functional changes in AD. To the best of our knowledge no previous studies have focused on such relationships between the MRI and SPECT findings.
MCI is a clinical term describing the transitional state between normal aging and dementia. In particular, amnestic MCI is considered to be preclinical AD because in up to 80% of cases it is known to convert to AD during an approximately 6-year period.13 To detect the earlier changes in AD, we included amnestic MCI patients in the present study.
- Top of page
The HDS-R scores of all 16 AD patients ranged from 9 to 24 (mean ± SD, 14.94 ± 4.91), reflecting severe, moderate and mild disease in three, five and eight patients, respectively. Meanwhile the scores of all 10 amnestic MCI patients were from 21 to 29 (mean ± SD, 23.70 ± 3.09). In addition, the full intelligence quotient (FIQ) for the WAIS-R and the General Memory scores of the WMS-R for all the AD patients averaged 70.40 ± 13.50 and 55.47 ± 6.95, respectively, whereas those of all amnestic MCI patients averaged 92.33 ± 18.97 and 71.57 ± 12.29. Although HDS-R scores of two AD patients were >20 (22 and 24), they were diagnosed as having probable AD according to the consensus criteria; their FIQ and General Memory scores were, respectively, 62 and 67, and 64 and 67.
All 26 patients underwent VSRAD and SPECT except fort one probable AD and one amnestic MCI patient who missed the 3DSRT analysis, although the mean rCBF on the left and right hemispheres were achieved. MRI indicated various degrees of hippocampal atrophy and diffuse cortical atrophy in all patients on visual analysis, while obvious infarction or high signals around the lateral ventricle were not found. Z scores of all the patients averaged 1.94 ± 1.24 (range, 0–4.69). Figure 1 shows typical changes of VSRAD and SPECT in one probable AD patient.
Figure 1. Voxel-based specific regional analysis system for Alzheimer's disease (VSRAD) and single-photon emission computed tomography (SPECT) with 3-D stereotaxic region of interest template (3DSRT) in a patient with probable AD. A 76-year-old woman was diagnosed as having probable AD on a Hasegawa's Dementia Scale–Revised score of 11. (a) Axial VSRAD and its enlarged image at −20 mm. The severe atrophy of the entorhinal cortex (regions enclosed by pink lines) was found. Z score was 4.69. (b) 3DSRT: regional cerebral blood flow diffusely decreased in almost all regions, especially in the frontal and temporal lobes.
Download figure to PowerPoint
To clarify the relationship between atrophy of the entorhinal cortex and decreased rCBF, we performed correlation analysis. A significant negative correlation was noted between Z scores and rCBF of the right thalamus (r = −0.548, P = 0.0086) and the left (r = −0.593, P = 0.0045) and right hippocampal (r = −0.630, P = 0.0025) regions (P < 0.01). Negative correlations were also shown between Z scores and rCBF of the right hemisphere (r = −0.398, P = 0.0464), the left thalamus (r = −0.497, P = 0.0173), the left (r = −0.457, P = 0.0286) and right temporal (r = −0.503, P = 0.0159), and right cerebellar (r = −0.421, P = 0.0435) regions (P < 0.05). The results indicated that Z scores increased with reduction of rCBF on those regions. Z scores tended to positively and negatively correlate with age (r = 0.370, P = 0.0658) and rCBF of the left cerebellum (r = −0.365, P = 0.0799), respectively, although the values did not reach full statistical significance (P > 0.05; Table 1).
Table 1. Correlations between VSRAD Z scores and age and absolute rCBF
|Items||n||Mean ± SD (mL/100 g per min)||r||P|
|Age||26||74.54 ± 8.51||0.370||0.0658|
|Right mean rCBF||26||40.50 ± 5.58||−0.398||0.0464*|
|Left mean rCBF||26||40.03 ± 4.99||−0.299||>0.05|
|Right callosomarginal rCBF||24||49.05 ± 9.81||−0.177||>0.05|
|Left callosomarginal rCBF||24||48.98 ± 9.79||−0.157||>0.05|
|Right precentral rCBF||24||51.26 ± 10.41||−0.291||>0.05|
|Left precentral rCBF||24||49.78 ± 9.53||−0.276||>0.05|
|Right central rCBF||24||54.68 ± 10.44||−0.143||>0.05|
|Left central rCBF||24||54.85 ± 9.57||−0.039||>0.05|
|Right parietal rCBF||24||51.44 ± 10.83||−0.171||>0.05|
|Left parietal rCBF||24||50.74 ± 10.13||−0.222||>0.05|
|Right angular rCBF||24||56.76 ± 12.30||−0.221||>0.05|
|Left angular rCBF||24||57.40 ± 12.16||−0.205||>0.05|
|Right temporal rCBF||24||47.13 ± 9.43||−0.503||0.0159*|
|Left temporal rCBF||24||45.63 ± 8.44||−0.457||0.0286*|
|Right posterior cerebral rCBF||24||56.87 ± 9.83||−0.295||>0.05|
|Left posterior cerebral rCBF||24||57.44 ± 10.31||−0.257||>0.05|
|Right pericallosal rCBF||24||50.24 ± 10.32||−0.237||>0.05|
|Left pericallosal rCBF||24||50.45 ± 10.42||−0.242||>0.05|
|Right lenticular nucleus rCBF||24||48.53 ± 8.74||−0.353||>0.05|
|Left lenticular nucleus rCBF||24||47.50 ± 7.66||−0.293||>0.05|
|Right thalamus rCBF||24||40.77 ± 9.04||−0.548||0.0086**|
|Left thalamus rCBF||24||41.00 ± 8.61||−0.497||0.0173*|
|Right hippocampus rCBF||24||34.15 ± 7.78||−0.630||0.0025**|
|Left hippocampus rCBF||24||34.71 ± 7.28||−0.593||0.0045**|
|Right cerebellum rCBF||24||64.19 ± 12.12||−0.421||0.0435*|
|Left cerebellum rCBF||24||63.62 ± 11.43||−0.365||0.0799|
|Right posterior cingulate rCBF||24||52.60 ± 11.92||−0.445||0.0327*|
|Left posterior cingulate rCBF||24||52.60 ± 12.20||−0.434||0.0374*|
|Right precuneus (lower) rCBF||24||65.55 ± 12.56||−0.400||0.0551|
|Left precuneus (lower) rCBF||24||67.44 ± 13.26||−0.357||>0.05|
|Right precuneus (upper) rCBF||24||60.06 ± 13.67||−0.035||>0.05|
|Left precuneus (upper) rCBF||24||60.48 ± 13.45||−0.141||>0.05|
|Right parahippocampal rCBF||24||39.22 ± 8.06||−0.517||0.0131*|
|Left parahippocampal rCBF||24||39.44 ± 7.79||−0.525||0.0118*|
Furthermore, Z scores had a negative correlation with rCBF of the left (r = −0.525, P = 0.0118) and right parahippocampus regions (r = −0.517, P = 0.0131), left (r = −0.434, P = 0.0374) and right posterior cingulated regions (r = −0.445, P = 0.0327; P < 0.05). No correlations were found between Z scores and the precuneus regions (lower and upper; P > 0.05).
In addition, the Z scores of the MCI and AD patients were not significantly different (P > 0.05).
- Top of page
As in previous VSRAD studies, all AD and MCI patients in the present study had atrophy of the entorhinal cortex except for one, who fulfilled the diagnosis criteria of probable AD, but whose Z score was 0. This exceptional case may be explained by a dissociation between anatomy and function, or the existence of some compensatory mechanisms.
The present study found obvious negative correlations between Z scores and rCBF changes of the bilateral hippocampus, thalamus and temporal regions. The entorhinal cortex itself is the origin of the perforant pathway, which provides the largest source of input to the hippocampal formation and its major cortical input. It receives the input of the perirhinal, parasubicular, and ectorhinal cortices (parahippocampal cortex), and secondarily projects to the subicular/CA1 region of the hippocampus.22 The perirhinal and parahippocampal cortices in turn receive projections from polymodal and unimodal association cortices and from supramodal cortices.23 The entorhinal cortex relays the main input and output of the hippocampus.
The entorhinal cortex also connects with the thalamus by the Papez circuit (entorhinal cortex–hippocampus–mamillary body–anteromedial thalamus–posterior cingulate cortex–entorhinal cortex). The Papez circuit was initially considered to be related to emotions in humans.24 More recently, its involvement in memory function has been proposed. Lesions in any component of this circuit could cause dysfunctions of memory. The input originating in the nucleus reuniens and nucleus centralis medialis of the thalamus has recently been recognized. The rhomboid, paraventicular and parataenial nuclei also contributed minor projections.25
The entorhinal cortex receives input from the neocortex, and its efferents project to the corresponding cortical areas. A quantitative analysis of the organization of cortical inputs to the rat entorhinal cortex found that the lateral and medial entorhinal areas received approximately equal proportions from the temporal, frontal and piriform regions. The lateral entorhinal area received more inputs from the insular cortex. In contrast, the medial area received more inputs from the cingulate, parietal and occipital regions.26 The inputs from the temporal region were more numerous than those from other areas except the piriform cortex.
Thus, the entorhinal cortex connects anatomically and physiologically with the hippocampus, thalamus and temporal regions. The present negative correlations indicated that rCBF in the hippocampus, thalamus and temporal regions decreased with increasing atrophy of the entorhinal cortex, presumably on account of functional inactivation based on their inter-regional connection. Numerous MRI studies, however, have demonstrated that atrophy of the medial temporal lobe including the hippocampus as well as the entorhinal cortex was associated with the onset of the symptoms of early AD,2 although the earliest pathological changes affect the entorhinal cortex. Therefore, the relationship between the atrophy of the entorhinal cortex and rCBF of the hippocampus may be partly related to concurrent pathological changes occurring in both regions. Furthermore, the present negative correlations between MRI Z scores and SPECT rCBF may be a stage-dependent finding in early AD, because mild AD and amnestic MCI patients accounted for the majority of the present subjects.
Traditional research suggested that the cerebellum is functionally devoted to motor control. Anatomic tract tracing studies indicated that there are pathways linking the cerebellum with autonomic, limbic, and associative regions of the cerebral cortex as well as with sensorimotor cortices. The cerebral cortex sends information in a precisely organized manner to the nuclei of the basilar pons, from where the information is conveyed to the cerebellum. The feedback from the deep cerebellar nuclei, notably the dentate nuclei, is relayed in the thalamus and projected to the same cortical areas in a precise arrangement. The communication with other brain areas allows the cerebellum to be involved in instinctive behaviors, mood, and the highest levels of cognition and reasoning.27,28 Parahippocampal areas, as well as the prefrontal, posterior parietal and superior temporal cortices, also have pontine afferents.29 The cerebellum, especially the right side, also had a negative correlation with the entorhinal cortex, revealing a possible relationship between the entorhinal cortex and the cerebellum too.
Functional connectivity between the entorhinal cortex and the posterior cingulate gyrus has been documented on both 18F-fluorodeoxyglucose positron emission tomography and brain perfusion SPECT.30,31 The present results also confirmed the remote effect, but the relationship was not found between the entorhinal cortex and the precuneus. The correlation between Z scores and rCBF of the parahippocampal region indicates a natural consistence of morphological and functional changes.
Thus, the present study shows the close relationships between morphological changes of the entorhinal cortex and functional changes of special cerebral regions in AD and amnestic MCI patients, which increase the reliability of VSRAD and help to diagnose AD patients at an early stage, although it was difficult to discriminate MCI from AD by means of VSRAD alone.
The negative correlation between Z scores and mean rCBF of the right hemisphere suggested a possible left–right difference in the relationship between the entorhinal cortex and cerebral regions. No age effect was obvious in the present study, although P indicated near statistical significance.
The limitations in the present study should be noted. First, the present findings must be interpreted with caution because of the small size of the sample. Second, to investigate the earlier stage of AD, amnestic MCI patients were included in the study. VBM or SPECT showed that changes of morphology and function in convert-MCI patients were similar to the alterations of AD patients, and different from those of non-converting-MCI patients. There was significantly greater gray matter loss in the medial temporal lobe, inferior and middle temporal gyri, posterior cingulated cortex, precuneus, inferior frontal gyrus and supramarginal gyri in MCI converters relative to non-converters.32–34 Possibly the effect of non-converting MCI may have affected the present results. Finally, in a few patients it was necessary to administer hypnotics or a small amount of antipsychotics, which affect the central nervous system. These drugs may have influenced the neuroimaging results, especially those of SPECT.
In conclusion, the present study showed, via correlation analysis between VSRAD and SPECT in AD and amnestic MCI patients that atrophy of the entorhinal cortex had an obvious functional relationship with the changes of blood flow in the hippocampus, thalamus and temporal lobe, which were closely attributed to their inter-regional anatomical and physiological connections. The correlation between the cerebellum and the entorhinal cortex was in agreement with that noted in previous non-human studies and requires further investigation.