Abstract
Objective: The authors studied mild patients with AD with a visual learning paradigm to determine whether activations of medial temporal regions on fMRI differ in AD compared to nondemented individuals.
Background: Changes in activation patterns of medial temporal lobe regions may serve as a biologic marker of altered brain function early in the course of AD.
Methods: The authors studied eight healthy young subjects, eight late middle-age nondemented volunteers, and seven patients with mild AD. All subjects underwent fMRI scanning in which they viewed a set of geometric designs for 45 seconds. Changes in blood flow were analyzed by comparing the prestimulus fMRI signal with that present during the stimulus presentation.
Results: Patients with AD, who had very poor recall of the geometric designs subsequently, showed increased blood flow (activation) during stimulus presentation only in a visual association area. Both the young and older nondemented subjects, all of whom had good recall of the designs, showed activations during stimulus presentation of the right entorhinal cortex, right supramarginal gyrus, right prefrontal regions, and left anterior-inferior temporal lobe. The younger and older nondemented subjects did not differ in fMRI activation patterns.
Conclusions: Failure of activation in AD of either temporal lobe or prefrontal regions is consistent with established clinical-pathologic correlations in AD. fMRI may be useful in confirming a memory disorder diagnosis and also may be useful in detecting individuals with incipient dysfunction in learning as a result of disorders such as AD.
As individuals with AD become symptomatic, learning and recall are very frequently impaired. The early and extensive pathologic changes in the hippocampal formation and entorhinal cortex in AD anticipate the clinical appearance of memory dysfunction by many years.1,2⇓ Structural MRI quantitative volumetric measurements of the hippocampal formation and entorhinal cortex have moderately good ability to distinguish patients with AD from nondemented elderly persons.3-6⇓⇓⇓ However, in a neurodegenerative disease such as AD, atrophic changes in the brain almost always appear after symptoms have begun to develop.7 An imaging technique that could identify hippocampal–entorhinal region dysfunction before the development of hippocampal and entorhinal atrophy may allow for earlier detection of the pathologic features of AD than is feasible with structural MR or neuropsychological assessment.
fMRI may be such a technique, but the activation patterns in normal and abnormal states must first be clarified. fMRI activation of the hippocampal–entorhinal regions has been demonstrated in normal subjects,8-11⇓⇓⇓ but only a few studies have examined hippocampal activation strategies in either patients with AD12 or individuals at risk for AD.13-16⇓⇓⇓ We previously studied a simple cognitive task using auditory stimuli that elicited reliable activation of the hippocampal formation and entorhinal cortex in young normal subjects.8,10,17⇓⇓ After developing a visually presented version using the same principles for use in the elderly in whom hearing impairment is common, we tested the paradigm to determine the fMRI activation patterns in normal individuals and in those with mild but certain dementia the result of probable AD.
Methods.
Subjects.
Three groups of right-handed18 subjects were studied: 1) eight normal younger subjects (21–29 years of age; mean ± SEM, 24.1 ± 1.2 years; 3/8 women), 2) eight normal elderly subjects (56–72 years of age; 65.1 ± 1.8 years; 2/8 women), 3) seven patients with AD (58–82 years of age; 73.6 ± 2.9 years; 4/7 women). The AD subjects were older than the normal elders (p < 0.001, Student t-test). The patients were diagnosed with probable AD according to National Institute of Neurological and Communicative Disorders and Stroke–Alzheimer’s Disease and Related Disorders Association criteria,19 except for one patient who would be classified as possible AD by the criteria because she had only memory problems. None of the patients were taking antipsychotic, antidepressant, or antianxiety medications, but three were taking donepezil. All of the patients with AD scored 23 or higher on the Mini-Mental State Examination.20
Each participant signed an informed written consent form, including the patients with AD, who were selected to be mildly impaired enough to understand the nature of the experiment as well as the risks and obligations. The young and elder subjects were paid $25 for their participation, whereas the patients with AD were not paid.
Task design.
The task design is depicted in figure 1. Subjects received detailed instructions and had a brief practice session before entering the scanner. Before entering the scanner, subjects were instructed to memorize the figure that they were going to be shown.
Figure 1. Experimental design in the stationary single-object memory trial. The top row depicts the visual images observed by subjects. The next row indicates the number of fMRI images acquired during each portion of the task period. The bottom row indicates the time for each portion of the experiment.
Subjects were advised to keep their eyes closed in the scanner before the fMRI experiment. The subjects were monitored using a video camera within the scanning gantry focused on their faces.
The to-be-remembered image was three stationary geometric figures that remained on the viewing screen for 45 seconds (viewing angle, 1.2°). The subjects were instructed to keep their eyes open with fixation point on the center of screen during the experiment. During the poststimulus period of 3.5 minutes, MR data were continuously acquired.
The subject’s retention of the visual image was evaluated at the end of each experiment. The subjects were asked to draw their memorized picture by freehand. As the visual image consisted of three geometric shapes with another geometric shape within them, we devised a simple scoring system (total, 0 to 20) to rate the subjects’ ability to recall the objects (0 to 10) and their relative locations (0 to 10).
Imaging data from two exposures to the visual stimulus were collected for each individual to examine the reliability and sensitivity of activations to prior exposure. In two younger subjects, one memory experiment was lost during transfer of data from one computing platform to another. In one of the elder subjects, the first memory experiment was lost. Therefore, 14 experiments in eight younger subjects, 15 experiments in eight elderly subjects, and 14 experiments in seven patients with AD were analyzed and compared. For the principal between-groups comparison, we used the first exposure data from eight younger subjects, seven elderly subjects, and seven patients with AD. For the comparison of the first and second exposures to the stimulus, we used data from six young subjects, seven elderly subjects, and seven patients with AD.
Imaging.
All experiments were conducted on a 1.5T MRI scanner (Philips ACS-NT system, Best, the Netherlands) using a volume head coil. fMRI parameters for the experiments were as follows: T2*-weighted echo planar imaging sequence; repetition time, (TR) 3.0 seconds; echo time (TE), 50 milliseconds; flip angle (FA), 90°; field of view (FOV), 21 × 21 or 22 × 22 cm2; matrix size, 64 × 64. Fourteen oblique slices covering almost the entire brain (slice thickness, 7.0 mm; slice gap, 1.0 mm) were obtained. A total of 120 temporal points were collected.
Anatomic images for these slices were obtained with a T2-weighted inversion recovery turbo-spin echo sequence (TR, 3000 milliseconds; TE, 20 milliseconds; inversion recovery time, 250 milliseconds; FOV, 22 × 22 cm2; matrix, 256 × 256; echo train length, 7; thickness, 7 mm with 1-mm gap). A multiplanar rapid acquisition with gradient echo imaging sequence was also used to acquire three-dimensional anatomic images for each subject. The image parameters in the three-dimensional sequence were as followings: TR, 15 milliseconds; TE, 7 milliseconds; FA, 8; FOV, 21 × 21 or 22 × 22 cm2; matrix, 256 × 256; 1 slab, with the spatial resolution of 2.0–2.5 mm; no gap; and oblique orientations.
Image analysis.
For the volume measurement of the hippocampal formation and the entorhinal cortex, the three-dimensional anatomic images were reconstructed from oblique orientation to coronal orientation. The entorhinal cortex was identified by the same techniques reported by others.3,21⇓ The volume of the entorhinal cortex was manually outlined and calculated using the STIMULATE software package (Center for Magnetic Resonance Research, Minneapolis, MN).22 The volume measurement of the hippocampal formation was determined from the same coronal images.
The functional images were generated using the STIMULATE software package. The fluctuation in the center of fMRI resulting from a subject’s head motion was very small (less than 0.1 pixel) in all experiments. Functional activation was defined as the differences between the prestimulus and stimulus–present time periods (p < 0.05, t-test). A brain map was constructed for stimulus presentation. The functional images were coordinated by image transformation into a standard stereotactic anatomic space followed by t-test statistical analysis in each individual to confirm the spatial consistency of positive or negative response areas in each group.23 For each subject, we computed a functional map based on the average image intensity over the whole brain and computed Z scores for each brain region as a function of whole brain activation level. For comparison between groups, the Z scores were normalized based on the global activation during stimulus presentation in 15 experiments of the elderly volunteer subjects. We identified activated areas in all nondemented elderly and younger subjects with the coordinated functional images from two nondemented groups (normal functional maps). And then, to identify systematically differences in activation between the AD group and normal groups, we subtracted the coordinated functional images of each individual AD from the functional maps of nondemented groups. Z scores in the identified areas were calculated for each individual. The spatial resolution in the transformed images was 4.0 × 4.0 × 4.0 mm3 per pixel. The mean of Z scores in all identified pixels were plotted and compared among three groups.
The memory scores, the anatomically derived volume measurements, and the magnitudes of functional activations by region were analyzed using a one-way analysis of variance (ANOVA) between younger, elder, and AD groups.
Results.
The free recall memory scores were similar for the young and elderly normal groups. With a maximum score of 20, both the young and elder normal groups performed very near ceiling levels (19.3 ± 0.3 and 18.8 ± 0.6). There was no difference in the performance of the young versus elder subjects (ANOVA, F(1,14) = 0.519; p = 0.48). The AD subjects performed worse (4.6 ± 0.9), as expected, and differed from both control groups (ANOVA, F(1,13) = 170.53 and 253.72; p < 0.0001). All of the AD subjects produced at least some recognizable fragment of the objects in the study figure, suggesting that they had understood what was expected of them during the MR scanning. In all three groups, the memory scores between first and second experiments did not differ (young, p = 0.36; elderly, p = 0.60; AD, p = 0.85, by paired t-test).
During the period of stimulus presentation, all AD and nondemented elders (see the table and figures 2 and 3⇓) showed activation of the right visual association area. In contrast, there were group differences in activation in the right entorhinal cortex, right supramarginal gyrus, right prefrontal regions, and the left anterior inferior temporal lobe. All nondemented elderly subjects but none of the AD subjects showed activation in these regions (each regional difference, p < 0.0005). There were strong correlations between the subjects’ memory performance and activations in temporal and frontal regions, but less so in the visual association area. Among the AD subjects, use of donepezil had no effect on fMRI activations.
The functional activation in elderly, young, and AD subject groups during stimulus presentation
Figure 2. The activation maps during visual stimulus presentation. The area depicted in red is the right visual association area (1), where activation occurred in all three groups. The areas in blue are left anterior-inferior temporal lobe (2), the right entorhinal cortex (3), the right lateral prefrontal regions (4), and right supramarginal gyrus (5), where there was activation in all nondemented subjects but no activation in AD subjects.
Figure 3. Scatterplots of the functional activation Z scores during stimulus presentation in selected regions identified in figure 2. Y = young subjects; E = elderly nondemented subjects; AD = AD subjects. Regions: A = right visual association area; B = right entorhinal cortex; C = right supramarginal gyrus; D = right prefrontal region; E = left anterior inferior temporal lobe.
The volume of the entorhinal formation (volume in each hemisphere combined) in the normal elderly was 4709 ± 205 mm3 and in the AD subjects was 3426 ± 173 mm3 (ANOVA, F(1,13) = 28.6; p = 0.0001). The volume of the hippocampal formation (volume in each hemisphere combined) in the normal elderly was 4709 ± 205 mm3 and in the AD subjects was 3426 ± 173 mm3 (ANOVA, F(1,13) = 22.1; p = 0.004). Whole brain volume in the AD group was also smaller (1479.8 ± 59.5 cm3) than the nondemented elderly (1654.6 ± 48.0 cm3), but with the small numbers of subjects, was not reliably different (ANOVA, F(1,13) = 4.4; p = 0.056).
The activation during stimulus presentation was larger in the young subjects than in the elderly in the right visual association area (F(1,14) = 17.2; p < 0.001; see table and figure 3). The activation in other regions was qualitatively and quantitatively similar between the young normal and elderly normal subjects in both first and second experiments.
There were group differences between the young and elderly subjects in the volumes of the hippocampal formations (ANOVA, F(1,14) = 9.6; p = 0.008), but not in the entorhinal cortex (ANOVA, F(1,14) = 0.043; p = 0.84). There were no differences in whole brain volume between the young (1701.0 ± 83.7 cm3) and elderly nondemented subjects (ANOVA, F(1,13) = 0.231; p = 0.64).
When experiments 1 and 2 including all groups (n = 20) were compared (using paired t-tests), the memory scores and the activation in all regions of interest were not different (all p values, >0.5).
Discussion.
We have found a clear difference in fMRI during the stimulus presentation period between mild patients with AD and elderly control subjects. Whereas the nondemented elderly subjects uniformly demonstrated activation in the entorhinal cortex and several other functionally related regions during presentation of a stimulus that they were told to memorize, the patients with AD showed no other regional activation except in a visual association area.24 The failure of activation of the entorhinal cortex in the patients with AD is consistent with the presumption that pathologic features in this region is substantial,1,2⇓ even in the mild patients in the present study.
In our fMRI task, we measured brain activation during the period when both visual perception and encoding occur. The similarity of activation of the visual association areas in all subjects confirmed that AD subjects were processing the stimuli. The temporal and frontal lobe regions that were activated in the nondemented subjects were ones that have been identified previously as active during spatial memory and working memory tasks.25-29⇓⇓⇓⇓ Although memory performance in our subjects was strongly bimodally distributed, the correlations between memory performance and activations in temporal and frontal regions imply that these regions are involved in encoding or storage of newly acquired information. Further experiments in cognitively intact subjects are required to determine how sensitive the task activations of nonvisual processing areas are to degradation by distraction, willful intention to ignore the stimulus, or lack of effort or motivation.
There were several limitations in our study. We studied only a few subjects. Although we had a priori expectations about the nature of the fMRI activations in the AD subjects, the exploratory nature of the data analysis means that our conclusions must be considered tentative. Although our patients with AD were selected to have very mild dementia, their hippocampal volumes proved to be uniformly smaller than those of the nondemented elderly. In addition, because our subjects were not matched for age, the activation difference between groups may simply reflect the strong age effect on hippocampal volume. There were no differences between our young and elder controls, suggesting that age may not have much impact on the activations we studied.
There are only few other prior studies of fMRI in AD. One group14 found reduced hippocampal activation using faces as the visual stimuli in patients with mild AD. These authors also studied several individuals with isolated memory loss and found that four of 12 of their subjects also had reduced hippocampal activation. Without longitudinal follow-up, it is not possible to determine whether the presence or absence of activation predicted the subsequent clinical course. Several other studies have examined fMRI activation in individuals at risk for AD and found that those at risk had abnormal patterns of activation. Another group15 (p 1391) found that individuals with a positive family history and at least one APOE e4 allele had “reduced activation in the middle and posterior inferotemporal regions bilaterally during” visual naming and letter fluency tasks. In contrast, a third report noted that carriers of the APOE e4 allele had greater hippocampal activation and greater activation of other brain regions as well during the learning and recall periods of a memory task.13
It is intriguing to speculate what patterns of fMRI activation may occur in a person destined to have AD at a time before the development of overt impairment of delayed recall. Would reductions in hippocampal and entorhinal activations occur before declines in memory performance appeared? Could there be compensatory increases in activation, as suggested by others?13,30⇓ Perhaps both could occur, depending on the amount of underlying AD pathologic characteristics, which would make the analysis very difficult. Future studies should seek to study at risk subjects longitudinally. It will be important to learn whether altered activation patterns in the entorhinal cortex can precede volume loss in the region. In principle, altered activation patterns should precede regional brain volume loss.
Acknowledgments
Supported by a grant from Eisai and Pfizer Pharmaceutical, Inc.
- Received December 11, 2000.
- Accepted April 18, 2001.
References
Letters: Rapid online correspondence
REQUIREMENTS
You must ensure that your Disclosures have been updated within the previous six months. Please go to our Submission Site to add or update your Disclosure information.
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
