591
Journal of Alzheimer’s Disease 50 (2016) 591–604 DOI 10.3233/JAD-150612 IOS Press
Different Patterns of Correlation between Grey and White Matter Integrity Account for Behavioral and Psychological Symptoms in Alzheimer’s Disease Elena Makovaca , Laura Serraa , Barbara Span`oa , Giovanni Giuliettia , Mario Torsoa , Mara Cercignania,b , Carlo Caltagironec,d and Marco Bozzalia,∗ a Neuroimaging
Laboratory, IRCCS Santa Lucia Foundation, Rome, Italy and Sussex Medical School, Clinical Imaging Sciences Centre, University of Sussex, Brighton, Falmer, UK c Department of Clinical and Behavioural Neurology, IRCCS Santa Lucia Foundation, Rome, Italy d Department of Neuroscience, University of Rome ‘Tor Vergata’, Rome, Italy b Brighton
Handling Associate Editor: Daniela Galimberti Accepted 26 October 2015
Abstract. Behavioral disorders and psychological symptoms (BPSD) in Alzheimer’s disease (AD) are known to correlate with grey matter (GM) atrophy and, as shown recently, also with white matter (WM) damage. WM damage and its relationship with GM atrophy are reported in AD, reinforcing the interpretation of the AD pathology in light of a disconnection syndrome. It remains uncertain whether this disconnection might account also for different BPSD observable in AD. Here, we tested the hypothesis of different patterns of association between WM damage of the corpus callosum (CC) and GM atrophy in AD patients exhibiting one of the following BPSD clusters: Mood (i.e., anxiety and depression; ADmood ), Frontal (i.e., dishinibition and elation; ADfrontal ), and Psychotic (delusions and hallucinations; ADpsychotic ) related symptoms, as well as AD patients without BPSD. Overall, this study brings to light the strict relationship between WM alterations in different parts of the CC and GM atrophy in AD patients exhibiting BPSD, supporting the hypothesis that such symptoms are likely to be caused by characteristic patterns of neurodegeneration of WM and GM, rather than being a reactive response to accumulation of cognitive disabilities, and should therefore be regarded as potential markers of diagnostic and prognostic value in AD. Keywords: Alzheimer’s disease, behavioral, grey matter, magnetic resonance imaging, white matter
INTRODUCTION Alzheimer’s disease (AD) is a chronic neurodegenerative brain disorder clinically characterized by a variety of cognitive deficits within the memory, language, and abstract thinking domains [1]. In addition to cognitive decline, behavioral and psychological symp∗ Correspondence
to: Dr. Marco Bozzali, Via Ardeatina 306, 00179 Rome, Italy. Tel.: +39 06 5150 1324; Fax: +39 06 5150 1213; E-mail:
[email protected].
toms of dementia (BPSD) are commonly observed over the progression of AD [2, 3]. BPSD have adverse consequences for AD patients and caregivers and are more strongly associated with quality of life than cognition or functional limitation [4]. BPSD cause often a deterioration of the quality of life, a greater impairment in activities of daily living, an earlier institutionalization of patients and caregiver depression [5, 6]. The negative impact of BPSD is also reflected in the increased costs of patients’ care [7]. The most common BPSD in AD are apathy [8], depression [9], anxiety [10],
ISSN 1387-2877/16/$35.00 © 2016 – IOS Press and the authors. All rights reserved
592
E. Makovac et al. / Brain Tissue Damage and BPSD in AD
and delusions [11], which can be observed since the early [12] or even preclinical stages of disease [13]. For instance, mood disorders, anxiety, and agitation are described in both patients with amnestic mild cognitive impairment (MCI) and AD, while psychotic symptoms are mainly observed in AD patients [14, 15]. BPSD are typically associated with a faster functional and cognitive decline [8, 14, 16] and with a higher likelihood of transition from MCI to AD [17]. Neurological associations with discrete BPSD symptoms have been traced from the earliest phases of AD [18], highlighting a crucial role of the fronto-striatal circuitry in the pathogenesis of BPSD. AD patients with apathy are more likely to show damage to the medial frontal and anterior cingulate regions [19, 20], but also to the head of the caudate nucleus and putamen [21]. Higher neurofibrillary tangle burden has been described in the orbito-frontal cortex of AD patients exhibiting agitation [22] and apathy [23], when compared to those without BPSD. The dorsolateral prefrontal area is also significantly involved in the pathogenesis of depressive symptoms in AD [24]. Delusions are associated with decreased grey matter (GM) volumes in the left frontal lobe, in the right frontoparietal cortex, and in the left claustrum [21] or in the right hippocampus [14], whereas disinhibition is strongly associated with GM volumes in the bilateral cingulate and right middle frontal gyri [14]. White matter (WM) microstructural damage plays also a crucial role in the development and severity of BPSD in patients with AD, and can be investigated in vivo using diffusion tensor imaging (DTI) [25]. Fractional anisotropy (FA), one of the most popular DTI indices of WM integrity, is known to be altered in the anterior cingulum [26] and in the genu of corpus callosum (CC) [27] of apathetic as compared to non-apathetic AD patients. Moreover, both mild AD and MCI patients with lower FA values in the cingulum are more likely to exhibit irritability, agitation, dysphoria, apathy, and night-time behavioral disturbances [28]. It is thus clear that clinical manifestations of AD are associated with both GM damage and abnormal integration between cortical regions by disconnection mechanism [29]. While the investigation of GM and WM modifications provide complementary windows into the disease, it is important to assess the degree to which abnormalities in the two brain tissues are correlated to each other. Recent studies have described the extent of WM damage and its relationship with GM atrophy in AD [30, 31], reinforcing the interpretation of AD pathophysiology in light of a disconnection syndrome [29]. Within this scenario, the CC is particularly relevant for the pathophysiology of AD. The reduction
of FA in the anterior and posterior CC of patients with AD [32, 33] is consistent with the loss of interhemispherically projecting neurons. Moreover, FA values in the anterior CC correlate with GM volumes in the bilateral frontal cortex of AD patients, whereas FA values of the posterior CC correlate with GM volumes in the left parietal lobe, reflecting a simultaneous neurodegeneration along functional systems in AD [31]. Since alterations of the CC has been frequently described in AD patients exhibiting BPSD symptoms [27], the present study was designed to establish the nature of correlations between regional GM atrophy and WM abnormalities in the CC of AD patients exhibiting various neuropsychiatric symptoms, such as psychotic, depressive, or frontal symptoms.
MATERIALS AND METHODS Subjects Fifty-eight AD patients (F/M = 37/21; mean age = 71.9, SD = 7.2; mean years of formal education = 9.6, SD = 4.4) were recruited from the Specialist Dementia Clinic of Santa Lucia Foundation (Rome, Italy) and from the Catholic University of Rome (Rome, Italy). Local Ethical Committee approval and written informed consent (either from the patients or from their responsible guardians if incapable) were obtained before study initiation. All patients were right handed, in order to reduce any potential source of variability due to hemispheric dominance. Based on clinical interview, patients with a history of major psychiatric disorders (e.g., major depression, psychosis, mania) were excluded from the study. Patients with probable AD were diagnosed according to the clinical criteria established by the National Institute of Neurological and Communicative Disorders and Stroke Alzheimer’s Disease and Related Disorders Association (NINCDS-ADRDA) [34]. Patients with signs of concomitant cerebral vascular disease on conventional magnetic resonance imaging (MRI) scans (as detailed below) were excluded from the analysis. Forty-six AD patients out of 58 were under treatment anticholinesterase inhibitors (27 with donepezil at a dosage of 5 mg or 10 mg/daily, 15 with rivastigmine at a dosage of 3 mg or 6 mg/ daily). Additionally, 19 patients were administered with antidepressants, 9 with anxiolytics, and 5 with neuroleptics. Following the approach described by Preti et al. [35], 25 healthy subjects (F/M = 12/13; mean age = 63.4; years of formal education = 13.7, SD = 3.5) were recruited for
E. Makovac et al. / Brain Tissue Damage and BPSD in AD
the purpose of constructing an average atlas of the CC. Neuropsychological assessment All recruited patients underwent an extensive neuropsychological battery administered by two trained neuropsychologists on the same day of the MRI acquisition. Global cognitive function was assessed by means of the Mini-Mental State Examination [36, 37]. Patients underwent also a complete battery of tests specific for each cognitive domain: 1) Verbal episodic long-term memory: Immediate and Delayed recall of a 15-Word List [38], Short Story Recall [39]; 2) Visuospatial episodic long term memory: Delayed recall of Complex Rey’s Figure [40]; 3) short-term memory: Digit span and Corsi Block Tapping task [41]; 4) executive functions and attention: Phonological Word Fluency [38]; Language: Naming subtest of Aachener Aphasia test [42]; and 5); Problem-solving: Raven’s Coloured Progressive Matrices [38]; Praxis: Copy of drawings [39]; Copy of Complex Rey’s Figure [40]. For all administered tests, we used the Italian normative data for score adjustment (gender, age, and education) and to define cut-off scores of normality, determined as the lower limit of the 95% tolerance interval for a confidence level of 95%. For each test, normative data are reported in the corresponding references. Group comparisons were estimated by one-way ANOVAs. Behavioral assessment For the behavioral assessment, AD patients’ caregivers were required to complete the NPI12 [43]. The NPI is a validated, caregiver-based behavioral rating system that allows quantification of the presence and severity of delusions, hallucinations, agitation/aggression, dysphoria/depression, anxiety, euphoria/elation, apathy, disinhibition, irritability/lability, aberrant motor behavior, and sleep and eating disturbances. Each item’s score ranges from 0 to 12, and reflects both, ratings of severity and frequency of each behavioral symptom, with 0 corresponding to the absence of behavioral symptom and 12 corresponding to its maximum frequency and severity. An index of severity is created for each behavioral variable by multiplying the frequency and severity scores, creating a frequency by severity product [44]. Several studies have tried to identify neuropsychiatric sub-groups by clustering BPSD symptoms, which contingently co-occur during the course of dementia. Following these studies, and based on NPI scores, we
593
grouped patients exhibiting delusions and hallucinations in a “psychosis” cluster (ADpsychotic ); depression and anxiety in a “mood” cluster (ADmood ); agitation and disinhibition in a “frontal” cluster (ADfrontal ) (for a review on the clustering procedure see [3]). Patients who were administered with antidepressants or anxiolytics entered the ADmood sub-group, while AD patients under treatment with neuroleptics entered the ADpsychotic sub-group. Patients exhibiting more than one symptom were classified according to the symptom with the highest score. A group of AD patients without BPSD symptoms (ADnoBPSD ) was identified and represented the main group of comparison in both neuropsychological and imaging data analyses. Although apathy is an important symptom within the AD pathology, its clustering is still not well described in literature [3], and for this reason apathy was not considered in current analyses. The correlation between behavioral and neuropsychological scores was estimated by using the Pearson correlation coefficient along with the p value performed in SPSS statistical package (SPSS Inc., Chicago IL). MRI acquisition All subjects (patients and controls) underwent an MRI examination at 3T (Magnetom Allegra, Siemens, Erlangen, Germany), including the following acquisitions: (1) Dual-echo turbo spin echo (TSE) (repetition time [TR] = 6.190 ms, echo time [TE] = 12/109 ms); (2) fast-FLAIR (TR = 8.170 ms, TE = 96 ms, TI = 2.100 ms); (3) 3D Modified-Driven-EquilibriumFourier-Transform (MDEFT) scan (TR = 1338 ms, TE = 2.4 ms, Matrix = 256 × 224 × 176, in–plane FOV = 250 × 250 mm2 , slice thickness = 1 mm), (4) a diffusion weighted Spin-Echo Echo Planar Imaging (SE EPI) (TR = 7000 ms, TE = 85 ms, maximum b factor = 1000 s.mm2 , isotropic resolution = 2.3 mm3 ). DTI data were obtained along 61 non-collinear directions, with a maximum b value of 1000. Seven volumes with b = 0 were also acquired resulting in a total of 68 volumes, with 45 contiguous slices and a 2.3 mm3 isotropic reconstructed voxel size. Diffusion tensor MR imaging and probabilistic tractography Data were corrected for head movements and eddy currents using affine registration. Then FA and mean diffusivity (MD) were computed from the diffusion tensor fitted with weighted linear least-square using
594
E. Makovac et al. / Brain Tissue Damage and BPSD in AD
Camino [45]. The CC was reconstructed in native space for every healthy control (who was not used for comparison purposes in the later analysis) with multi-fiber probabilistic tractography, carried out using 10000 iterations of the probabilistic index of connectivity (PICo) algorithm [46] applied to fiber orientation distribution functions estimated with PAS MRI [47]. Five principal parts of the CC were reconstructed separately, using as seeds the regions resulting from parcellation of the midsagittal section suggested by Hofer et al. [48] (see Fig. 1): genu, anterior midbody, posterior midbody, isthmus, and splenium. Probabilistic atlas construction of the corpus callosum AD brains are known to present with diffusion irregularities which can interfere with an effective reconstruction of individual white matter tracts based on diffusion indices [49]. For this reason, here we adopted an atlas-based method, as previously proposed by Preti et al. [35]. The tractographic atlas is obtained from data acquired using the same acquisition protocol employed for AD patients, in a group of agematched healthy controls. The group-averaged atlas is warped onto each individual patient’s brain to extract the relevant quantities. This method removes the need to individually reconstruct the CC from pathological brains. In order to create the tractographic atlas, the following steps were implemented. For each healthy control, FA images were nonlinearly registered to the Montreal Neurological
1/2
1/2 1/3 1/4
1/6 (c) (b)
(d) (e)
(a)
Institute (MNI) standard space using the following procedure: first, the affine transformation that matches each subject’s FA map to the template was computed using FLIRT [50]; next, FNIRT [51] was used to compute the deformation field that warps the original FA map to the template, setting as starting estimate the previously computed affine transformation. The estimated warping transformation between each patient’s FA map and the template was applied to the correspondent tract density maps of the five segmented CC portions. The tract density maps were then binarized by setting a threshold of 0.1 to the PICo maps and averaged separately for each CC portion, in order to obtain images indicating the groupwise probability of each voxel to belong to the considered tract portion. With the aim of increasing the confidence of belonging to the tract of interest, probability maps were thresholded to retain only voxels with more than 10% probability of belonging to the tract. DTI analysis of the corpus callosum using atlas-based tractography In order to extract average FA and MD values in each separate portion of the CC, we applied the following atlas-based method, which was shown by others [35] to be more informative when investigating patients with AD or MCI than other approaches such as tract-based spatial statistics (TBSS). The warping field matching the FA map of each patient to the atlas space (in MNI coordinates, 2 mm resolution) was estimated in the same way as for the healthy subjects. The transformation was then inverted and applied to the CC portion template images obtained from the healthy subjects. For each CC atlas portion and for each subject, we computed the average FA weighted by the probability of every voxel to belong to a considered CC portion. The mean FA and the average MD of each portion were estimated and used for the correlation analysis described in the next section. Moreover, between-group statistics was performed using SPSS Statistics v17.0 (http://www.spss.com). An ANOVA model was employed to test for between group differences in mean FA and MD for each considered portion of the CC. Bonferroni’s correction was used to correct for multiple comparisons. Voxel-based morphometry analysis and statistics
Fig. 1. Method for corpus callosum segmentation. Delineation of five regions of interest within the CC: (a) Genu; (b) anterior midbody; (c) posterior midbody; (d) isthmus; (e) splenium (Hofer & Frahm [48]).
The T1 weighted (MDEFT) volumes from all participants were visually reviewed to exclude the presence of macroscopic artifacts. T1 weighted
E. Makovac et al. / Brain Tissue Damage and BPSD in AD
volumes were preprocessed for voxel-based morphometry (VBM) using the VBM8 toolbox implemented in SPM8 (Statistical Parametrical Mapping, http://www.fil.ion.ucl.ac.uk), to produce a GM probability map [52, 53] in standard space (MNI coordinates) for every subject. In order to compensate for compression or expansion which might occur during warping of images to match the template, GM maps were “modulated” using the “non-linear only” option of VBM8, which adjusts every voxel’s signal intensity by multiplying it by the amount of non-linear deformation only. This procedure allows to compare the absolute amount of tissue corrected for individual brain sizes (as global head size is accounted for by the linear scaling). GM volumes were computed from these probabilistic images for every subject. All data were then smoothed using a 12 mm FWHM Gaussian kernel. Statistical analysis was performed on smoothed GM maps within the framework of the general linear model. We performed our VBM analysis, adopting a fullfactorial design implemented in SPM8, with group (i.e., ADnoBPSD , ADmood , ADfrontal , ADpsychotic ) as the main factor, age, years of education, gender, and MMSE scores as variables of no interest and MD and FA values of each of the 5 CC anatomical parts as variables of interest. Therefore, five different full-factorial analyses were run separately (one for each part of CC). Results were accepted as significant at p values
