Structural correlates of mild cognitive impairment: A clinicovolumetric study
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.227298
Source of Support: None, Conflict of Interest: None
Keywords: Alzheimer's disease, amnestic mild cognitive impairment, automated regional volumetry, voxel-based morphometry
Mild cognitive impairment (MCI) is considered to be a transitional stage between normal aging and Alzheimer's Disease (AD), and its rate of progression varies between individuals. The longitudinal risk of developing AD in amnestic MCI (a-MCI) can be estimated by measures of memory performance and subjective level of cognitive functioning.,, Neuropsychology tests such as vernacular adaptations of Addenbrooke's Cognitive Examination (ACE) and Rey Auditory Verbal Learning Test (RAVLT), have been found to be effective screening and definitive measures of cognitive impairment in a-MCI. Learning and retention measures, as assessed in RAVLT, have been shown to be among the most comprehensive clinical predictors of conversion from a-MCI to AD. While it is believed that roughly 12% of patients with MCI annually run the risk of conversion to AD, with approximately 80% converting to dementia by 6 years, other studies debate this conversion rate. It is also well-known that a large subset of patients remains stable over the years or may clinically revert to normal cognitive status.
The most well-known approaches employed in the analysis of gray matter (GM) loss in MCI and AD is voxel-based morphometry (VBM) and automated regional volumetry (ARV). Not surprisingly, prominent studies till date have focused on MTL and amygdala volumes. Little is known regarding the relationship between GM volume in other cortical regions and performance on the neuropsychological tests of patients with MCI or AD. Therefore, the aim of the present cohort study was to quantify the structural changes in cognitively stable a-MCI and AD patients compared to individuals with no cognitive impairment (NCI) in South India, employing quantitative volumetric magnetic resonance imaging (MRI) analysis tools such as ARV and VBM. The study also aimed to correlate the neuropsychological test performances using screening tests and definitive tools for memory impairment with morphometric variables in patients with MCI in comparison with a diseased group of patients suffering from an early AD.
Sixty-two participants (NCI: 25, MCI: 24, AD: 13) were prospectively recruited from Memory and Behavioral Neurology clinic at a reputed centre in the city of Trivandrum, located in the southern state of Kerala, India. Early AD patients were selected using the standard NINCDS–ADRDA (National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer's Disease and Related Disorders Association) diagnostic criteria with a clinical dementia rating (CDR) score of <2. Patients with a-MCI, aged between 55–80 years, diagnosed as per the modified Petersen's criteria  with a CDR of ≤0.5, who were still functioning independently in the community, and having a normal general cognition (i.e., MMSE >24) with normal performance on tests of other cognitive domains, viz., language, executive functions, visuospatial functions, praxis, and perception were included. The minimum duration of follow up of MCI patients required for inclusion into the study was 1.5 years. It was ensured prior to their inclusion that none of them had neuropsychological or CDR evidence of progression to dementia. The objective evidence of memory impairment required that an individual scored less than mean −1.5 standard deviation (SD) from the norm on at least two tests in the memory domain. This method has been validated previously. The inclusion criteria for the cognitively normal healthy controls included an age range of 55–80 years, with formal education of more than 8 years, with no history of subjective memory complaints, and no major neurological, psychiatric, or medical comorbidities. All participants underwent a detailed, clinical, neuropsychological, and radiological evaluation after providing a written informed consent. The study was approved by the Institutional Ethics Committee.
The participants underwent cognitive screening by the vernacular adaptation of ACE, which has been validated previously. This battery has a measure of global cognition (mini-mental state examination, MMSE), and also includes tests for memory (immediate and delayed recall of a seven-item address list), verbal fluency (initial letter P and categories of animals), confrontation naming (ten items), and constructional praxis (copying two line-drawings). It also assesses executive functions and constructional ability (clock-drawing), remote memory, and language. Registration/learning is scored on a 24-point scale which has 3 points for registration of 3 words and 21 points for 3-trial learning of an address. The recall score was drawn from a 10-point scoring which included a 5-min recall of the three items presented previously and a 7-point recall of the address. Participants were required to have a depression score on the Hospital Anxiety Depression Scale (HADS) of less than 7. The measures of learning and recall studied in this cohort were the cumulative learning trials and delayed recall scores in the vernacular adaptation of Rey Auditory Verbal Learning Test (RAVLT) and memory subsets of ACE. Multidomain involvement in MCI was excluded as diagnosis of a-MCI necessitated normal performance on the language, executive, and visuospatial domains. A semantic battery employing confrontation naming for language; trail making A and B, and Wisconsin card sorting test (WCST) for executive functions; and, visual objective space perception battery, and judgement of line orientation for visuospatial functions were the tests used to exclude the presence of patients with either a multidomain involvement, or having the presence of non-a-MCI.
Participants were scanned with a 1.5T (Seimens Magnetom-Avanto SQ engine, Erlangen, Germany) MRI scanner in the Department of Radiology. The three-dimensional (3D) T1 weighted anatomical scans were acquired using a flash-spoiled gradient echo sequence with 1mm slice thickness, 176 sagittal slices, 1 × 1 × 1mm voxel size, 11 ms repetition time, 4.95 ms echotime, 15° flip angle, matrix size of 256 × 256 with a total scan time of 6 min 21 s.
Whole brain analysis
VBM8 toolbox in Statistical Parametric Mapping (SPM8- Wellcome Department of Imaging Neuroscience, London) was applied to investigate the cortical GM changes across the entire brain without any prior knowledge of the region of interest (ROI). All the images were oriented in the anterior commissure-posterior commissure (AC-PC) line. The images were segmented into GM, white matter (WM), and cerebrospinal fluid (CSF), and spatially normalized into the Montreal Neurological Institute (MNI) space using VBM8 DARTEL procedure with custom settings. The segmentation was followed by modulation for preserving the volume of a particular tissue in a fixed voxel, and volumes of GM, WM, CSF, and total intracranial volume (TIV) were calculated. The resulting images were smoothed with an isotropic Gaussian kernel of 8mm full width half maximum (FWHM).
Group comparison among MCI, AD, and controls were performed by one-way analysis of variance (ANOVA) within the Statistical Parametric Mapping 8 (SPM8) general linear model. Group comparisons were tested with a corrected threshold of P < 0.001, which determines the clusters with significant differences in GM concentration. Three separate contrasts (NCI vs MCI, NCI vs AD, and MCI vs AD) with age, sex, and total intracranial volume (TIV) as covariates were used to compare the GM density between participants. The significant atrophic regions were overlaid on T1-weighted standard brain images, allowing the localization of areas of significant GM loss. The atrophic regions are reported in Montreal Neurological Institute (MNI) coordinates with the help of xjview toolbox (http://www.alivelearn.net/xjview/).
Volumes of cortical gray matter structures using automated regional volumetry
The segmentation algorithm in SPM8 and Automated Anatomic Labelling (AAL), (http://www.cyceron.fr/web/aal) template was used to estimate the bilateral volumes of eight anterior and medial temporal structures, namely, inferior temporal gyrus, superior temporal gyrus, entorhinal cortex, middle temporal pole, superior temporal pole, parahippocampus, hippocampus, and amygdala, one deep gray matter structure (thalamus), and two parietal structures (precuneus and posterior cingulate). After segmentation, the volume of each region was obtained by writing scripts in MATLAB window.
Demographic data, neuropsychological measures, and cortical GM volumes between study groups were analyzed using the Statistical Package for the Social Sciences Statistics for Windows version 21 (SPSS, IBM. Armonk, NY, USA). The demographic and cognitive variables were compared across the groups, NCI, MCI, and AD, using univariate analysis of variance (ANOVA). Bonferroni post-hoc procedure with age correction was applied to compare means of the independent variables of volume with diagnosis. Partial correlation was used to assess the relationship between independent volumes and neuropsychological test measures in each study group. In all comparisons, the level of statistical significance was set at P < 0.05.
Demographics, cognitive function, and memory
Demographic and neuropsychometric measures between the three groups are depicted in [Table 1]. The mean duration of subjective memory complaints in the MCI group was 5.8 ± 3.7 years and the mean disease duration in the AD group was 4.2 ± 1.5 years. The mean duration of follow-up since the time of diagnosis to the time of inclusion into the study was 3.5 ± 1.9 years in the MCI group, and 2.2 ± 0.7 years in the AD group. After adjusting for age, patients with both MCI and AD were significantly impaired on ACE recall, ACE total, RAVLT cumulative learning, and delayed recall compared to individuals with NCI (P< 0.001). Similarly, the AD group performed poorly in comparison to the MCI group on all components of cognitive and memory scores [ACE recall, ACE total, RAVLT recall (P< 0.001) and RAVLT learning scores (P = 0.013)]. None of the MCI patients had progressed to early AD during the minimum 1.5-year period of follow-up at the time of inclusion into the study.
Whole brain findings
The participant groups were contrasted to reveal the patterns of GM atrophy on VBM group analysis. Compared to controls, MCI patients revealed predominant atrophy in bilateral superior temporal gyri. Significant atrophic patterns were also observed in the hippocampus, inferior temporal gyrus, middle frontal gyrus, cuneus and lingual gyrus in the left hemisphere, the middle occipital gyrus and cingulate gyrus in the right hemisphere [Figure 1] and [Supplementary Table 1]. For AD patients, most significant GM loss was detected in the left parahippocampal gyrus. In addition, AD patients had significant GM loss in bilateral hippocampi, bilateral precunei, bilateral fusiform gyri, bilateral middle temporal gyri, bilateral anterior cingulate regions, right inferior temporal gyri, right superior temporal pole, right posterior cingulate, right rolandic operculum, right thalamus, right inferior parietal lobule, right middle occipital gyri, left middle frontal, and orbitofrontal gyri [Figure 2] and [Supplementary Table 2]. Moreover, a direct comparison of the patient groups revealed greater cerebral atrophy in AD patients [Figure 3] and [Supplementary Table 3] in bilateral parahippocampi, hippocampi, amygdalae, fusiform gyri, superior and middle temporal poles, inferior temporal gyri, and medial orbitofrontal regions along with atrophy in the right precuneus and left angular regions.
Gray matter volumes using automated regional volumetry
The regional volumes of GM structures estimated by using ARV are summarized in [Table 2]. Post-hoc Bonferroni procedure showed significant volume differences for the left hippocampus, right precuneus, and bilateral superior temporal gyri in the MCI patients compared to individuals with NCI. In patients with AD, bilateral hippocampi, precunei, superior temporal gyri, parahippocampi, right amygdala, and right superior temporal pole revealed significant GM volume loss compared to controls. Only the total GM and left hippocampal volume discriminated the patient groups, with lower volumes detectable in AD patients.
Correlation between neuropsychology parameters and automated lobar volumetry
Partial correlation analysis with age as the controlling variable showed that RAVLT delayed recall positively correlated with reduced volumes of left (rp= 0.42, P < 0.04) and right (rp= 0.38, P < 0.05) hippocampi in MCI patients. A similar observation was noted between RAVLT total learning and the hippocampal volume (rp= 0.42, P < 0.04) in the left hemisphere; whereas, in AD, a significant correlation was observed between RAVLT total learning and the right hemisphere volume of the middle temporal pole (rp= 0.66, P < 0.01) and superior temporal pole (rp= 0.84, P < 0.001). In addition, ACE total score significantly correlated with volume of the right superior temporal pole (rp= 0.73, P < 0.006).
Our study was designed to quantify the GM differences in stable “non-progressor” a-MCI patients compared to early AD patients and NCI individuals using VBM group analysis and ARV. Further, the study compared the cognitive measures that were deemed clinically relevant for tracking the progression of MCI to AD. Previous studies have demonstrated variable results using these two methods, predominantly favouring VBM over ARV for assessing GM volume differences between the groups., The features unique to a-MCI patients who are clinically stable are crucial to determining structural measures of stability in what is usually considered to be a pre-AD status. To the best of our knowledge, this is the first study in the Indian population that utilized automated software tools to discriminate MCI and AD population from individuals with NCI, as well as to correlate the GM loss in specific anatomical regions with cognitive measures. A recent Indian study addressed the relationship between regional morphometric brain changes and cognitive deficits in AD patients in relation to NCI.
As VBM analysis is a practical tool for whole-brain group comparisons, identifying regions which can be subsequently analyzed using ARV, we identified certain regions of interest (ROIs) in each comparative model. Considering the MCI patients, GM volume loss was detected in the superior temporal gyrus and left hippocampus, and variably over the mesial brain regions (cuneus, precuneus, and cingulate cortex), in comparison to individuals with NCI. The observation of hippocampal atrophy is in line with previous reports ,, and is established as a marker of future progression in MCI. Atrophy identified in the mesial cortical regions, especially the precuneus, as identified on ARV, is a unique finding in this study and represents an avenue for future research considering the role of precuneus in episodic memory retrieval and visuospatial awareness, and from a network perspective, in the default mode network. This highlights the added functional significance of the posteromedial regions of the cortex in MCI in addition to atrophy of the temporal lobe. The inferior frontal gyrus atrophy identified in the MCI group is in line with the findings of Whitwell et al., as an indicator of faster disease progression. Only a longitudinal follow-up would indicate the significance of these results in our stable MCI group.
In the AD group, as in other studies, a significant GM atrophy was noted by both the techniques in the medial temporal (MTL) structures, i.e., the hippocampus, amygdala, parahippocampal gyrus,,, and precuneus. Similar to other studies, VBM identified atrophy of the anterior cingulate,, the fusiform gyrus, and the frontal , regions in AD compared to controls. Interestingly, in contrast to the literature,, our study failed to reveal significant impairment of the posterior cingulate region on group analysis. Furthermore, our study detected greater atrophy of the hippocampus in patients with AD compared to those with MCI, in addition to the medial and neocortical temporal lobe, anterior cingulate region, insula, and precuneus identified on VBM comparison between patients with AD and MCI. As has been shown in another study, atrophy in these regions, if observed in MCI, may be indicative of transition to early AD.
As expected, patients with MCI and AD significantly differed from individuals with NCI on cognitive measures. The observations were retained even after controlling for age and regional volumes. The direct comparison between patient groups revealed significant differences in ACE recall, ACE total, and RAVLT total learning. These results would imply the existence of genuine clinical and functional differences between the 3 groups, and the independent relevance of these neuropsychological measures as predictors of cognitive decline in the elderly. Correlation between the cognitive scores and regional volume loss was very specific in the MCI and AD participants. Both RAVLT total learning and recall after 20 minutes significantly correlated with hippocampal atrophy in MCI participants; whereas, in AD, the decline in cognitive measures such as RAVLT total and ACE total significantly correlated with the volume of the superior temporal gyrus. These results indicate an association between cognitive functioning and regional morphometric brain changes, a probable epiphenomenon of pathological changes in MCI and AD considering the early deposition of amyloid plaques in neocortical areas preceding the limbic areas.,
Even though our study came up with the above-mentioned promising findings, it had some limitations. In the technique of ARV, we only selected certain regions in the temporal, parietal, and limbic regions; however, we excluded the frontal and occipital cortex. Although we included only stable MCI patients who were non-progressors during the duration of follow-up, a follow-up of greater than 5 years would have added further strength to the current study. However, the findings represent an avenue to predict conversion or stability in patients with atrophic MCI using population-based studies and to earmark patients at risk of progression.
In summary, the present study indicates that genuine neuropsychological differences exist between patients with MCI and AD, and control subjects. These differences are independent of age as well as the total GM volume and the regional cortical volumes, thereby highlighting functional differences that occur independent of structural differences between the groups. The volumetric findings of the study highlight the characteristics of stable MCI patients, with demonstration of atrophy in the temporal neocortex, hippocampus, and mesial neocortical regions as a key finding. In addition, atrophy in these regions along with frontoinsular regions could predict MCI due to AD. As we observed more findings in VBM, it is a more useful method for group comparisons. However, future cross-sectional and longitudinal studies in larger samples with comprehensive neuropsychological evaluation are needed to confirm these findings.
Financial support and sponsorship
This work was supported in part by grants from Kerala State Council for Science, Technology & Environment (19SRSHS/2011) and Department of Science and technology (SR/CSI/90/2012(G)).
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3]
[Table 1], [Table 2]