Differential Cortical Atrophy in Subgroups of Mild Cognitive Impairment

Objective  To compare gray matter brain volumes in patients diagnosed with subtypes of mild cognitive impairment (MCI) (those with a focal amnestic disorder and those with more diffuse cognitive dysfunction) with those of elderly controls.

Design  Magnetic resonance imaging volumetric study of MCI subgroups (MCI-amnestic [MCI-A], and MCI-multiple cognitive domain [MCI-MCD]) using a whole brain voxel-based analysis.

Setting  Referral dementia clinic.

Patients  Thirty-seven patients with MCI (age range, 49-85 years; MCI-A, n = 9; MCI-MCD, n = 28) and 47 control subjects (age range, 55-81 years).

Main Outcome Measures  Volumetric anatomical magnetic resonance imaging differences between MCI subgroups and normal controls, and between patients with MCI who progressed to dementia. Magnetic resonance imaging scans were analyzed using statistical software SPM99.

Results  Overall, the patients with MCI had significantly decreased volume in the hippocampus and middle temporal gyrus, bilaterally, compared with control subjects. Compared with patients with MCI-MCD, patients with MCI-A had significant volume loss of the left entorhinal cortex and inferior parietal lobe. Compared with patients with MCI-A, patients with MCI-MCD had significantly reduced volume of the right inferior frontal gyrus, right middle temporal gyrus, and bilateral superior temporal gyrus. Patients with MCI who progressed to Alzheimer disease during follow-up (mean interval 2 years, maximum 4.5 years), showed greater atrophy in the left entorhinal cortex, bilateral superior temporal gyri, and right inferior frontal gyrus compared with those who did not progress.

Conclusions  These data provide evidence of distinct brain structural abnormalities in 2 groups of patients with MCI. While both have mesial temporal and cortical volume loss, those with a focal memory deficit have more involvement of the mesial temporal structures and less involvement of the neocortical heteromodal association areas than those patients with MCI with diffuse cognitive dysfunction. Thus, MCI may represent a more heterogeneous group than currently conceived, possibly reflecting 2 different etiological processes to dementia. These data also suggest that these structural abnormalities precede the development of Alzheimer disease.

The transitional state between normal aging and Alzheimer disease (AD), mild cognitive impairment (MCI), has become a focus of research owing to the development of effective pharmacotherapy aimed at altering the natural history of the disease.^1,2 A number of brain structural abnormalities have been identified among patients with MCI with abnormal memory, including significant reduction in the volume of the hippocampus,^3,4 medial occipitotemporal lobe,⁵ parahippocampal gyrus, entorhinal cortex, superior temporal gyrus, and anterior cingulate gyrus.^6,7 These morphological abnormalities are particularly severe among those patients with MCI who progress to AD compared with those who do not.^6,8

Even though MCI diagnosis relies primarily on the presence of memory dysfunction, a growing number of studies have concluded that performance in other cognitive domains is often not entirely normal.^1,9 While some patients exhibit an isolated memory problem, others can have altered neuropsychological test performance in multiple cognitive areas.^1,9 The purpose of this study was to compare regional gray matter brain volumes in 2 subtypes of patients with MCI using a whole brain voxel-based analysis.¹⁰ This approach is not biased to a specific brain region and permits identification of potential unsuspected brain structure abnormalities,¹⁰ allowing for a more comprehensive description of the differences between MCI subtypes.

Thirty-seven patients from a group of 200 who met criteria for MCI, as described below, underwent a volumetric spoiled gradient-recalled magnetic resonance imaging (MRI) scan. Each patient received an extensive evaluation,¹¹ and was reevaluated on an annual basis with regard to neuropsychiatric status to determine whether there was a change in diagnosis.

Volumetric MRI scans were obtained on 47 older comparison subjects from 3 ongoing studies, including the Alzheimer's Disease Research Center (n = 28),¹¹ the University of Pittsburgh’s Mental Health Intervention Research Center for Late-Life Mood Disorders,¹² and a study of cognitive and cerebrovascular consequences of hypertension (n = 19).¹³ None of the controls converted to dementia or MCI within 5 years of the scan.

The neuropsychological evaluation included the Mini-Mental State Examination (MMSE),¹⁴ the Mattis Dementia Rating Scale,¹⁵ and measures of 4 cognitive domains: memory, language, visuospatial/visuoconstructional, and attention/executive functions. Details of the neuropsychological battery have been described elsewhere.¹⁶ The results of the cognitive tests were classified normal or abnormal (>1.5 below that of subjects of comparable age and education) based on normative data obtained from the Alzheimer's Disease Research Center normal control sample.

Patients with MCI-amnestic (MCI-A) (n = 9) required memory deficits, with otherwise normal cognitive function. These patients must have impairments in delayed recall verbal memory, nonverbal memory, or both.¹⁶

Patients with MCI-multiple cognitive domain (MCI-MCD) (n = 28) required deterioration in at least 1 cognitive domain (not including memory), without sufficiently severe cognitive impairment or loss of daily living skills to constitute dementia, or 2 abnormal tests in 2 different domains.¹⁶

Magnetic resonance imaging scans were conducted using a Signa 1.5 Tesla scanner (GE Medical Systems, Milwaukee, Wis). The MRI of the brain was done within 6 months after the initial evaluation. The spoiled gradient-recalled sequence was designed to maximize contrast between gray and white matter (echo time = 5 milliseconds, repetition time = 25 milliseconds, 1.5-mm section, 0-mm intersection interval, 40° flip angle).

All MRI data were processed using Statistical Parametric Mapping (SPM99; Wellcome Department of Cognitive Neurology, London, England) running in MATLAB (Mathworks, Sherborn, Mass). The spoiled gradient recalled images were spatially normalized (Montreal Neurological Institute coordinate system; McGill University, Montreal, Quebec), and the tissue segmented using a modified mixture model cluster analysis technique.¹⁷ The segmented gray matter images were then smoothed using an 8-mm isotropic gaussian kernel. A more complete description of voxel-based morphometry method can be found in Good et al¹⁸ and Ashburner et al.¹⁷

The demographic characteristics of all subjects and MCI subgroups are shown in Table 1. Fourteen patients with MCI (38%) converted to AD during follow-up (mean ± SD follow-up: 45.7 ± 26.5 months). The proportion of patients with MCI-A (44%) and MCI-MCD (36%) who converted to AD was similar between groups (χ² = 0.65, P = .41). The baseline demographic characteristics of those who converted to AD are shown in Table 2. Overall, those who converted to AD did have lower MMSE (t = 3.2, P = 0.01) and Mattis Dementia Rating Scale (t = 4.15, P = .004) scores at the time of study entry compared with nonconverters.

Patients with MCI, as a group, had significantly decreased volume in the hippocampus and middle temporal gyrus, bilaterally. In addition, the left inferior parietal, left middle frontal, and right superior frontal volumes were also reduced in patients with MCI compared with control subjects (Table 3).

Patients with MCI-A had significantly reduced volume of the mesial temporal lobe on the right, including the hippocampus, entorhinal cortex, and amygdala (Figure, A) compared with control subjects. In addition, reduced volume was observed in the left inferior parietal, inferior and middle frontal, and superior temporal gyri. Patients diagnosed with MCI-MCD had significant bilateral volume loss of the hippocampus, middle and superior temporal, and inferior frontal gyri compared with controls (Figure, B). In addition, the left inferior parietal gyrus and the right superior frontal gyrus were significantly decreased in patients with MCI-MCD compared with controls (Table 3).

Compared with patients with MCI-MCD, patients with MCI-A had significantly greater volume loss in the left inferior parietal lobe (P<.001) and the left entorhinal/perirhinal cortex (P<.01). By contrast, compared with patients with MCI-A, patients with MCI-MCD had significantly greater volume loss in the right inferior frontal gyrus, right middle temporal gyrus, and superior temporal gyrus, bilaterally (Table 4).

Among all patients with MCI who progressed to AD, there was greater baseline atrophy in the left entorhinal cortex, bilateral superior temporal gyri, and right inferior frontal gyrus (Table 5). Partial correlation analyses, controlling for age, showed significant positive correlations between the MMSE score and the volumes of the left entorhinal cortex (r = 0.35; df = 34; P = .04) and right inferior frontal gyrus (r = 0.36; df = 34; P = .03).

The present study demonstrates distinct brain structural abnormalities in 2 subgroups of patients with MCI. Specifically, patients with MCI-A have atrophy in the hippocampus and entorhinal cortex, as expected,^3-7 as well as in the amygdala, and in the neocortex. Patients with MCI-MCD showed more diffuse and extensive volume loss in the neocortical heteromodal association, with less involvement of the medial temporal lobe structures compared with those diagnosed with MCI-A. The MCI subgroups share a region of atrophy in the inferior frontal cortex, specifically BA 44/45, suggesting that this region may be important in the clinical presentation of MCI, and perhaps, reflecting the impending defect in retrieval from semantic memory.

Our data also suggest that atrophy in specific cortical regions precede the development of dementia in patients with MCI. Specifically, patients who developed AD during follow-up had significantly decreased volume of the left entorhinal cortex, bilateral superior temporal gyri, and right inferior frontal gyrus at study entry. Volume reduction of the entorhinal cortex and superior temporal gyrus has been noted in recent studies;^6,19 however, our findings suggest additional structural abnormalities within the inferior frontal cortex of those patients with MCI who progress to AD. The degree of cognitive impairment, as measured by the MMSE, appears to be associated with decreased volume in both the frontal and entorhinal areas.

The structural changes associated with the MCI syndrome are more diffuse than previously thought. Although the hippocampus has been the most studied area in MCI cases, more recent studies have shown that structural lesions (eg, neurofibrillary tangles, neuritic plaques) are more widely distributed in MCI, and include the neocortex and limbic areas.^20,21 Furthermore, amyloid deposits were identified in vivo in the frontal and parietal lobes in patients with mild AD, and practically no amyloid was seen in the medial temporal lobe structures.²² Therefore, the structural abnormalities in MCI cases, especially in those who progress to AD, are not limited to the mesial temporal areas.

This study is limited by the small sample of MCI subtypes, particularly those who progressed to AD. Nevertheless, our study shows that there are at least 2 subtypes of MCI, those with the more traditionally defined memory deficit and those with more diffuse cognitive impairment, each presenting with distinct brain structural abnormalities. The therapeutic implications of our findings need to be explored, as differences in brain abnormalities may be associated with variations in disease course and treatment response. Better understanding of these subtypes may enhance our knowledge of the relationship between normal aging and dementia.

Correspondence: Oscar L. Lopez, MD, Neuropsychology Research Program, Suite 830, Oxford Bldg, 3501 Forbes Ave, Pittsburgh, PA 15213 (lopezol@upmc.edu).

Accepted for Publication: January 4, 2005.

Author Contributions:Study concept and design: Bell-McGinty, Lopez, Meltzer, Scanlon, DeKosky, and Becker. Acquisition of data: Bell-McGinty, Lopez, Meltzer, and Becker. Analysis and interpretation of data: Bell-McGinty, Lopez, Meltzer, Scanlon, Whyte, DeKosky, and Becker. Drafting of the manuscript: Bell-McGinty, Lopez, DeKosky, and Becker. Critical revision of the manuscript for important intellectual content: Bell-McGinty, Meltzer, Scanlon, Whyte, DeKosky, and Becker. Statistical analysis: Lopez and Becker. Obtained funding: Bell-McGinty and DeKosky. Administrative, technical, and material support: Bell-McGinty, Meltzer, and DeKosky. Study supervision: Lopez, Meltzer, DeKosky, and Becker.

Funding/Support: This research was supported in part by grants AG05133 and AG20098 from the National Institute on Aging, Bethesda, Md; and HL57529 from the National Heart, Lung, and Blood Institute, Bethesda. Drs Bell-McGinty, Scanlon, and Whyte are postdoctoral research fellows in the Department of Psychiatry, University of Pittsburgh Medical Center, Pittsburgh, Pa (T32 MH19986). Dr Becker is the recipient of a Research Scientist Development Award, Level-II (K02 MH01077).