Abstract
Keywords
INTRODUCTION
Amnestic mild cognitive impairment (aMCI) is a syndrome characterized by below normal memory performance for a given age, whereas intellectual functioning and activities of daily living are otherwise unimpaired [1]. Five to fifteen percent of patients with aMCI develop clinically diagnosable Alzheimer’s disease (AD) within 1 year [2]. Patients with AD show various neuropsychiatric symptoms (NPSs). Apathy, which is characterized by social withdrawal and loss of goal-oriented behavior, interest, and motivation, is the most common NPS in patients with AD [3, 4] and aMCI [5]. The prevalence of apathy in aMCI patients is from 11.1% to 55% [5–9]. In addition, aMCI patients with apathy are more likely to progress to AD. Sixty percent of aMCI patients with apathy progressed to AD during an observation period of up to 2 years [10], a higher proportion than expected based on the 5–15% annual progression rate for aMCI patients stated above. Furthermore, aMCI patients with apathy have an approximately seven times greater risk of progressing to AD than aMCI patients without apathy [11]. Thus, as with memory impairment, apathy in aMCI patients is an important symptom that is closely associated with AD.
Apathy in AD is correlated with reduced metabolism and cerebral blood flow (CBF) in the anterior cingulate cortex [12–16] and the orbitofrontal cortex [12–14, 16], with decreased CBF in the right temporoparietal cortex [17], and prefrontal cortex and anterior temporal cortex [18]. Apathy in AD has also been correlated with atrophy in the bilateral anterior cingulate cortex and left medial frontal cortex [19] and cortical thinning of the bilateral anterior cingulate cortex, bilateral frontal cortex, head of the left caudate nucleus, and bilateral putamen [20]. In addition, in moderate to severe AD, apathy is associated with greater neurofibrillary tangle burden in the anterior cingulate cortex [21]. Thus, apathy in AD involves extensive damage to the anterior cingulate and medial prefrontal cortices and the surrounding regions.
Apathy in aMCI patients may also be associated with damage to the brain. Reduced glucose metabolism in the posterior cingulate cortex (PCC) has been reported in early-stage AD patients [22]. Glucose metabolism in the PCC is lower in aMCI patients with apathy than in aMCI patients without apathy [23]. In aMCI patients, the association between apathy and local white matter microstructural deficit in four regions (the right temporal portion of the uncinate, middle longitudinal and inferior longitudinal fasciculi, and the medial white matter area including parathalamic white matter, fornix, and the posterior cingulum of the right hemisphere) was shown by diffusion tensor imaging (DTI) and using Statistical Parametric Mapping (SPM) [24]. In a study of four regions of interest (ROIs) [25], apathy was shown to be negatively related to cortical thickness in the inferior temporal cortex and positively related to cortical thickness in the anterior cingulate cortex. The latter result was interpreted as a compensatory or inflammatory response to AD pathology and might have preceded the atrophy that was observed in the anterior cingulate cortex. However, it is unknown whether apathy in aMCI patients is associated with atrophy in other brain regions. The association between apathy and regional CBF (rCBF) in aMCI patients is also unknown.
The aim of the present study was to identify the brain regions in which apathy in aMCI patients was associated with 1) atrophy using magnetic resonance (MR) images and voxel-based morphometry (VBM) and 2) reduced rCBF using the single photon emission computed tomography (SPECT) images and SPM.
METHODS
This study was a retrospective observational study without an intervention and was performed in compliance with national legislation and the Declaration of Helsinki. All patient information was anonymized as unlinked data prior to analysis to prevent the identification of personal information. The study was undertaken with approval of the ethics committee at Osaka University Hospital.
Subjects
This study was conducted in the neuropsychological clinic of the Department of Psychiatry of Osaka University Medical Hospital. In the clinic, all patients underwent standard neuropsychological examinations and routine laboratory tests, and cranial MR imaging, except for patients who were prohibited from undergoing the MRI. Because we can run only 8 SPECT scans per week, we were unable to use it on all patients. The aMCI patients who were most likely to be selected for SPECT were those who were suspected to be at the pre-dementia stage of some dementia diseases, especially AD. The evaluations were based on clinical history and the results of neuropsychological and neurological examinations by dementia specialists in our clinic. The cerebrospinal fluid (CSF) of some patients was also examined. The clinical and investigative data were collected prospectively in a standardized manner and were entered into our dementia registry.
The subjects of this study were 98 aMCI patients who were entered in our dementia registry between March 1, 2009 and April 30, 2015 and met Petersen’s criteria for aMCI [26]: (1) a memory complaint documented by the patient or collateral source; (2) a score of 1.5 standard deviations (SDs) below the education-adjusted normal value in the story A recall task in the logical memory II subtest of the Wechsler Memory Scale-Revised; (3) a score ≥24 in the Mini-Mental State Examination (MMSE); (4) a total Clinical Dementia Rating (CDR) score of 0.5; (5) no symptoms of dementia based on clinical examination and an extensive interview with a knowledgeable informant; and (6) normal activities of daily living. Exclusion criteria were: (1) subjects who had a diagnosis of cerebrovascular disease (CVD), as confirmed by brain imaging, or who had a history of CVD diagnosis; (2) subjects who were diagnosed with neurological, psychiatric, or physical disease, which could influence cognitive decline; (3) subjects who had no data for the neuropsychiatric inventory (NPI) or geriatric depression scale (GDS); and (4) subjects who had no MR images and SPECT data.
Procedure
Neuropsychological examinations
In addition to the MMSE, all patients were examined with the Alzheimer’s disease Assessment Scale-cognitive component-Japanese version (ADAS) and the short version of the Wechsler Adult Intelligence Scale Third Edition, which included digit span, block design, digit symbol, and information subtests. Higher scores indicate better performance in WAIS-III and worse performance in ADAS.
Clinical evaluations
Memory complaints by the patient and caregiver were assessed in our clinic with the Everyday Memory Checklist (EMC) [27, 28], which is a questionnaire in which subjects were asked whether a memory failure had occurred in each of 13 areas of daily life (e.g., “Do you forget things you were told yesterday or a few days ago and have to be reminded of them?”). Each question has four possible answers: never (0 points), sometimes (1 point), usually (2 points), or always (3 points). Thus, higher scores indicate more severe impairment. The EMC was prepared in two forms, which were designed to be answered by the patient and the caregiver, respectively.
The caregiver’s physical, emotional, social, and financial aspects in relation to caring for the patient were assessed with the Japanese version of the Zarit Caregiver Burden Interview (ZBI) [29]. Higher scores indicate more severe caregiver burden.
Neuropsychiatric assessments
Neuropsychiatric symptoms were assessed with the Japanese version of the NPI [3] in our clinic. The NPI involves 10 items: delusions, hallucinations, agitation, depression, anxiety, euphoria, apathy, disinhibition, irritability, and aberrant motor behavior. Caregivers of patients are asked the severity and frequency of each symptom. The severity of each item is classified into grades 0 to 3 and the frequency of each item is classified into grades 0 to 4 : 0 indicating absence of symptom and higher scores indicate worse symptom or higher frequency. A composite score is obtained for each symptom by multiplying severity and frequency scores, with a maximum score of 12 for each item. A total NPI score was generated by adding the scores for each item, with a higher total NPI score indicating greater severity of NPS. We used the composite score of the apathy item of the NPI as the apathy score in thisstudy.
We also assessed the subjective depressive symptoms with 15-item short form of geriatric depression scale (GDS), which was answered by the patient [30]. A score greater than 5 suggests depression and higher scores indicate more severe depression.
SPECT imaging acquisition
N-isopropyl-p-(123I) iodoamphetamine (123I-IMP)-CBF SPECT scan was conducted using an integrated SPECT/computed tomography system (Symbia T-6; Siemens Healthcare, Erlangen, Germany). Images were obtained in a resting state with the patient’s eyes closed and ears unplugged. The SPECT scan was initiated 15 min after injection of 167 MBq of 123I-IMP, and data acquisition was performed for 30 min in the dynamic mode with circularly rotating gamma cameras over a 360° range in four angular steps (90° views). Images were reconstructed by filtered-back projection method and Chang’s method for attenuation correction. For the attenuation coefficient, 0.15 cm–1 was used based on the previously obtained pooled phantom data. Scatter component in the projection data was corrected using triple-energy window method.
MR imaging acquisition
MR imaging was performed on a 1.5-T system (Signa Excite HD 12.x; General Electric Medical Systems, Milwaukee, WI, USA). A three-dimensional volumetric acquisition of a T1-weighted gradient echo sequence produced a gapless series of thin sagittal sections that covered the whole calvarium. The operating parameters were: field of view = 240 mm, matrix = 256×256, 124×1.40 mm contiguous sections, TR = 12.55 ms, TE = 4.20 ms, and flip angle = 15°.
Data analyses
Analysis of SPECT data
To identify whether the SPECT images were suggestive of AD, the Z-score summation analysis method [31] was used in this study. The software assessed whether hypoperfusion in the bilateral parietal lobes, posterior cingulate gyrus, and precuneus was less than 1.64 SDs below the mean of healthy elderly subjects, and whether perfusion in occipital lobes was normal. aMCI patients in this study were classified into two groups; one was those with the SPECT images suggesting of AD (amnestic MCI patients with SPECT evidence of AD: aMCI-AD+) and another was those without the SPECT images suggesting of AD (aMCI-AD–).
Anatomical normalization and statistical processing of the SPECT images were performed using SPM version 12 (SPM 12; Wellcome Department of Cognitive Neurology, London, United Kingdom). The calculations and image matrix manipulations were performed using MATLAB R2015a (MathWorks Inc., Natick, MA, USA). Each SPECT image firstly co-registered to the MR image of the subject was transformed into a stereotactic anatomical space with transformation parameters that were created in the DARTEL (diffeomorphic anatomic registration through exponentiated lie algebra) transformation process of MR imaging. Further, each image was smoothed using an isotropic 8-mm Gaussian kernel to increase the signal-to-noise ratio and compensate for differences in gyral anatomy between individuals.
The association between rCBF and the NPI apathy score was examined by voxel-by-voxel multiple regression analysis using gender, age, education level, GDS, and MMSE as covariate parameters. We examined SPM t-maps with a threshold of p = 0.005 uncorrected and voxel extent threshold set to 300 voxels. The analyses were conducted in each of the aMCI-AD+ and aMCI-AD–.
Analysis of MRI data
The MR images were anatomically normalized and statistically processed with DARTEL toolbox implemented in SPM12. For MRI, all individual images were segmented using “New segment” implemented in SPM12b, and then transformed into a standard stereotaxic anatomical space with the brain template from the Montreal Neurological Institute (voxel size: 1.5 mm×1.5 mm×1.5 mm) using DARTEL flow created in the DARTEL process.
The association between regional gray matter volume and the NPI apathy score was examined with a voxel-by-voxel multiple regression analysis using gender, age, education level, total intracranial volume, GDS, and MMSE as covariate parameters. We examined SPM t-maps with a threshold of p = 0.005 uncorrected and voxel extent threshold set to 300 voxels. The analyses were conducted for each group.
Statistical analysis
Differences between aMCI-AD+ and aMCI-AD- groups were examined by using Mann-Whitney’s U test, t test or Fisher’s exact test. Correlations between the NPI apathy score and scores of the neuropsychological assessments were evaluated by using the Spearman rank correlation coefficient in each group. The significance level for all comparisons was set at p < 0.05. Statistical analysis was done using IBM SPSS 22.0 software (IBM, Armonk, NY, USA).
RESULTS
Subjects
A total of 1,698 patients were newly registered in our dementia registry during the 6-year period. Of these, 243 patients met the inclusion criteria of aMCI (Fig. 1). Of these 243 aMCI patients, 145 patients were excluded (40 for having CVD, 56 for having neurological, psychiatric, or physical disease, which could influence cognitive decline, and 49 for lack of SPECT, MRI, NPI, or GDS data). The remaining 98 patients (39 men and 59 women) were used for the analysis. Ninety-five of the 98 patients were right-handed.
The Z-score summation analysis with the SPECT images was used to detect hypoperfusion which is suggestive of AD. Patients with and without hypoperfusion were classified as AD+ and AD–, respectively. Accordingly, we classified the 98 aMCI patients into 31 aMCI-AD+ patients and 67 aMCI-AD–patients (Table 1).
Results of clinical evaluations
In the EMC, the patient self-rating scores were significantly lower (indicating less impairment) than the caregiver rating scores (n = 98 for each score, p < 0.0001, Wilcoxson signed rank test). Thirty-five of the 98 patients (35.7%) had GDS scores of 6 or more, which suggests that they suffered from depression.
The aMCI-AD+ patients were significantly younger than the aMCI-AD–patients. The WAIS-III digit span raw score and block design raw score were significantly higher in the aMCI-AD+ group than in the aMCI-AD- group. The two groups did not significantly differ in any other evaluations (Table 1).
Results of Neuropsychiatric Inventory
Of the 98 aMCI patients, 75.5% had one or more NPSs. Apathy was the most prevalent NPS (58.2%), followed by irritability (23.5%), agitation (19.4%), delusions (16.3%), and depression (15.3%). The percent of patients suffering from each item in the NPI did not significantly differ between the two groups. In addition, the composite score for the items of the NPI did not significantly differ between the two groups. The percentages of patients with apathy in the aMCI-AD+ group (61.3%) and the aMCI-AD–group (56.7%) were not significantly different (Table 2).
Correlations between NPI apathy score and other evaluations
The NPI apathy score was significantly and positively correlated with the GDS score, NPI depression score, ZBI total and personal strain scores, and CDR sum of the box score (Table 3). The NPI apathy score was also weakly and positively correlated with the ADAS recall score, EMC caregivers score, and ZBI role strain score, and negatively correlated with WAIS-III digit symbol raw score.
In the aMCI-AD+ group, the NPI apathy score was significantly and positively correlated with the EMC caregiver score and weakly correlated with the ADAS total score, GDS score, and CDR sum of the box score. In the aMCI-AD–group, the correlations between the NPI apathy score and clinical evaluations were similar to those for all of the patients: the NPI apathy score was significantly and positively correlated with the NPI depression score, ZBI total and personal strain scores, and CDR sum of the box score and weakly correlated with GDS score. The NPI apathy score was also significantly and negatively correlated with the ADAS recognition score.
Correlations between apathy and gray matter volume/rCBF
In voxel-based correlation analyses with MRI data (Table 4), the NPI apathy score was significantly and negatively correlated with gray matter volume in one region in the aMCI-AD+ patients (the right caudate nucleus) (Fig. 2), with five regions in the aMCI-AD–patients (right middle globus pallidus, right putamen, left fusiform gyrus, left occipital lobe, and right occipital lobe) (Fig. 3) and with one region in all patients (the right caudate nucleus).
In SPM analyses with SPECT data (Table 4), the NPI apathy score was significantly and negatively correlated with decreased rCBF in five regions in the aMCI-AD+ patients (left posterior-medial frontal cortex, right superior frontal gyrus, right culmen-fusiform gyrus, left culmen-fusiform gyrus, and left occipital lobe) (Fig. 4), and with three regions in all patients (left posterior-medial frontal cortex, right medial frontal gyrus, and right middle frontal gyrus). However, in the aMCI-AD–patients, the NPI apathy score was not significantly correlated with rCBF in any of the regions.
DISCUSSION
In this study, we classified 98 aMCI patients into 31 aMCI-AD+ patients and 67 aMCI-AD–patients based on the results of a Z-score summation analysis of the SPECT images. The sensitivity of the Z-score summation analysis for discriminating AD from non-AD is high (88.0%) [31], indicating that the aMCI-AD+ patients in this study have a high probability of being in at the aMCI stage of AD. In the aMCI-AD+ group, apathy was significantly and negatively associated with gray matter volume of the right caudate nucleus and with rCBF in five regions (the left posterior-medial frontal gyrus, right superior frontal gyrus, bilateral culmen-fusiform gyri, and left occipital lobe) after adjusting for gender, age, education level, total intracranial volume, GDS score, and MMSE score.
Although gray matter density loss in the caudate nucleus has been shown to be involved in apathy in AD patients [20], this is the first report of such an association in aMCI patients. So far, four studies have examined the neurobiological correlates of apathy in MCI [24, 33]. One of these studies [24], using DTI to examine white matter microstructure in 20 aMCI patients, found negative correlations between apathy and white matter microstructural deficit in four regions (the right temporal portion of the uncinate, middle longitudinal and inferior longitudinal fasciculi, and the medial white matter area including parathalamic white matter, fornix, and the posterior cingulum of the right hemisphere). This is the only study in the Stella review [34] that examined the associations between apathy and the white matter microstructure in MCI patients. In another study of 47 aMCI patients [25], apathy was negatively related to thickness in the inferior temporal cortex and positively related to thickness in the anterior cingulate cortex. The greater thickness in the anterior cingulate cortex was interpreted as a compensatory or inflammatory response to AD pathology and might have preceded the atrophy that was observed. However, another study [33] that examined 389 aMCI patients found no relationship between apathy and cortical thickness in either of these two regions. The fourth study [32] examined 334 aMCI patients and found no association between apathy and cortical thickness in four regions (entorhinal cortex, middle frontal gyrus, orbitofrontal cortex, and anterior cingulate cortex). Although the caudate nucleus is known to be damaged at the aMCI stage [35] and early stage of AD [36, 37], it was not examined in three of these studies [25, 33]. We were able to identify the caudate nucleus as an apathy-associated region by measuring whole gray matter using VBM.
The neural circuits traversing the caudate nucleus are involved in frontal-subcortical circuits, which serve the motivational and executive functions [38]. Anhedonia, the condition in which pleasant stimuli elicit only poor responses, can be caused by damage to the caudate head [39, 40] and is also closely associated with apathy [41]. These findings are consistent with our finding of an association between apathy and the caudate nucleus. As mentioned above, only one study has found an involvement of the caudate nucleus in apathy in AD patients [20]. In that study, apathy in 31 patients with mild AD (as measuredby the NPI subscale score) was associated with decreased gray matter volumes (as measured by VBM) in the caudate head, as well as in three other regions (the anterior cingulate cortex, bilateral frontal cortices, and the putamen). The finding of decreased gray matter in regions other than the caudate head in those patients might be because their cognitive impairment and apathy (MMSE score 23.3±2.8 and NPI composite score 3.2±3.1, respectively) were more severe than those in our patients (26.4±1.8 and 2.5±2.7).
In the aMCI-AD+ group, apathy was significantly associated with reduced rCBF in the left posterior-medial frontal gyrus and right superior frontal gyrus in this study. The medial frontal lobe has a role in producing motivation in frontal-subcortical circuits [38, 42]. Therefore, damage in the medial frontal lobe can cause apathy. The superior frontal gyrus plays a role in executive function in frontal-subcortical circuits. Apathy in AD has been suggested to be related to impairment of executive function as part of the functioning of the dorsolateral prefrontal region [43]. The medial frontal and superior frontal lobes are reported to be associated with apathy in MCI and AD patients [34]. However, the present study found no relationship between the NPI apathy score and the results of WAIS-III digit span, block design, or digit symbol subtests in the aMCI-AD+ group. The absence of a correlation between apathy and performances on the tests in the present study may be because our tests were not appropriate for measuring executive functions. The digit span, which measures verbal working memory, and the block design, which measures visuospatial reasoning, do not measure switching or attention/inhibition, which are the core executive functions. In aMCI patients, performance on a letter fluency test [44] or a verbal fluency test and psychomotor tracking [45] was found to be correlated with apathy. The WAIS-III digit symbol test may require the executive functions and has characteristics similar to those of the psychomotor tracking test. Thus, the WAIS-III digit symbol raw score tended to be associated with the NPI apathy score in the combined aMCI group in this study.
Cortical thickness in the lower inferior temporal cortex was found to be associated with apathy in aMCI patients [25, 33]. Because the inferior temporal cortex is next to and functionally connected to the fusiform gyrus, our finding that apathy was associated with reduced rCBF in the culmen-fusiform gyrus in aMCI-AD+ group appears to be consistent with these findings. Apathy was also associated with reduced rCBF in the occipital lobe in the aMCI-AD+ group. The occipital lobe was reported to be indirectly associated with apathy: microstructural variation of the white matter areas that connect the thalami to the occipital cortex as well as the frontal cortices were indirectly associated with subclinical apathetic-like expressions in healthy females [46].
The SPECT analysis in this study did not show an association between apathy and reduced rCBF in the caudate nucleus in the aMCI-AD+ group. One possibility is that SPECT is not sufficiently sensitive to evaluate the cerebral hypoperfusion in subcortical structures. Only one of the 11 studies using SPECT in the systematic review by Stella et al. [34] showed an association between apathy and reduced rCBF in a subcortical region (the amygdala) in AD and MCI [47].
In the aMCI-AD- group, there were no regions in which rCBF correlated with apathy. However, in the aMCI-AD- group, apathy was significantly correlated with the atrophy in some regions: the right middle globus pallidus, right putamen, left fusiform gyrus, and bilateral occipital lobes. These areas have been associated with apathy in subjects with various diseases and conditions: the fusiform gyrus and occipital lobe were associated with apathy in the aMCI-AD+ group in this study, the putamen was reported to be associated with apathy in AD patients [20] and healthy aged participants [48], and the globus pallidus was reported to be associated with apathy in late-life depression [49]. These findings raise the possibility that the aMCI-AD–group included individuals with various conditions, including AD, depression, and non-neurological disease.
Limitations
Our study has a few limitations. First, the sample size was not large, although this study included the third largest sample of aMCI patients for the elucidation of neuroanatomical bases of apathy. Second, aMCI patients in this study were not consecutive ones but were limited to those who underwent the SPECT scan. The aMCI patients who underwent SPECT in this study were suspected to have a form of dementia, especially AD, based on clinical history and the results of neuropsychological and neurological examinations. Therefore, the results of this study might depend more strongly on the data of the aMCI patients with AD than the data of the aMCI patients without AD. The presence of apathy in an aMCI patient is a sign that the patient will progress to AD [10]. The prevalence of apathy was higher in the aMCI-AD+ group in this study (61.3%) than in aMCI patients in previous studies [5–9], supporting the assumption that many of the aMCI patients in this study have aMCI due to AD. Third, the NPI scale is not a fine scale so we may have failed to detect slight degrees of apathy. Fourth, the classification between aMCI-AD+ and aMCI-AD–was based on the results of SPECT images, rather than on the results of pathological examinations.
Conclusions
The present findings show that in aMCI patients that have a high probability to develop clinically diagnosable AD, atrophy in the right caudate nucleus and reduced rCBF in wider areas including the left posterior-medial frontal cortex, right superior frontal gyrus, bilateral culmen-fusiform gyri, and left occipital lobe are involved in apathy.
Footnotes
ACKNOWLEDGMENTS
The present study was supported in part by the grant provided by the Ministry of Health, Labour and Welfare (Research on Dementia) to Hiroaki Kazui (Grant Number: H25-27-Dementia-General-003) and in part by the grant provided by the Japan Agency for Medical research and Development (Science Research Grants for Dementia R&D) for Hiroaki Kazui (Grant Number: 26340601).
