Article Text
Abstract
Objective To determine the course of cognitive deficits and the regional progression of brain atrophy in patients with Parkinson's disease (PD) with and without visual hallucinations (VH).
Methods The authors performed MRI and neuropsychological assessment at entry to the study and at follow-up (mean±SD=29.91±5.74 months) in a sample of initially non-demented 12 PD patients with VH, 14 PD patients without VH and 12 healthy controls (HC). Grey-matter changes over time were assessed by means of voxel-based morphometry (VBM) and cognitive changes by an extensive neuropsychological battery.
Results At follow-up, 75% of patients with VH developed dementia. The greatest decline was observed in verbal memory, semantic fluency, language comprehension and visuoperceptive functions. None of the patients without VH met criteria for dementia and did not show any worsening in cognitive functions over time. Patients with VH showed widespread limbic, paralimbic and neocortical grey-matter loss, whereas in the PD without VH group, grey-matter loss was restricted to a small region in the frontal cortex and cerebellum. The authors also found significant correlations between the changes in several cognitive functions and grey-matter loss over time in PD patients with VH.
Conclusion The presence of VH in PD determines a different cognitive outcome and a different pattern of progressive brain atrophy. PD patients with VH, unlike PD without VH, frequently develop dementia and show a widespread atrophy involving limbic, paralimbic and neocortical areas.
- Parkinson's disease
- dementia
- MRI
- hallucinations
- longitudinal
- neuropsychology
Statistics from Altmetric.com
Introduction
Visual hallucinations (VH) have been reported to occur in 50% of patients with Parkinson's disease (PD).1 Generally, VH occur during the second half of the disease's course1 and have a persistent and progressive nature.2 Although VH can be exacerbated by dopaminergic treatment, several studies did not find a direct relationship between antiparkinsonian agents and presence of VH.1 2 The presence of VH has been associated with Lewy Body (LB) pathology. While VH are considered characteristic features of PD and dementia with Lewy bodies, they are rarely reported in other parkinsonian disorders such as corticobasal degeneration or supranuclear palsy.1
Cross-sectional studies have reported that non-demented PD patients with VH present greater neuropsychological impairment than those without VH in domains such as verbal3–5 and visual memory,6 language comprehension,4 and visuospatial5 and visuoperceptive functions.4 6 Frontal dysfunction has also been described in PD with VH including deficits in verbal fluency,3 4 7 8 sustained attention9 and inhibition.7 8 Moreover, longitudinal studies have shown the presence of VH as a significant predictor of dementia in PD,10–12 and it is associated with a more rapid general cognitive decline assessed by Mini-Mental State Examination (MMSE).13 14 In a previous longitudinal study, we reported that 45% of the PD patients with VH developed dementia after 1 year's follow-up and showed a significant progressive decline in visual memory and visuoperceptive functions.15 The combination of degraded visual information from the environment and failing visual memory together with deficits in executive functions7 8 have been pointed out as important causal factors for the occurrence of VH.16
The brain mechanisms underlying VH in PD are not completely understood. Structural and functional abnormalities within the primary visual system and visual association areas, including ventral and dorsal pathways, have been reported in PD with VH.16 We previously reported that non-demented PD patients with VH showed areas of grey-matter reduction in parieto-occipital areas in comparison with non-hallucinating PD patients.17 Functional imaging studies also reported perfusion abnormalities in occipital, temporal and parietal areas.18–21 In addition to the dysfunction in posterior areas, dysfunctional frontal activation was reported in fMRI studies in response to simple22 and complex23 visual stimuli. The involvement of the frontal lobe in VH has also been confirmed by PET studies showing a pattern of frontal hypermetabolism in patients with PD and VH.24 Neuropathological studies showed the association between VH and the presence of high densities of LBs in medial temporal areas.25 Interestingly, we have recently reported hippocampal head reductions in PD with VH in comparison with healthy controls.26
The involvement of all these areas (visual associative, frontal and medial temporal areas) is indicative of brain abnormalities in PD patients with VH already when they are not demented. In PD patients with dementia, the pattern of brain dysfunction involves these and other areas, and it is more severe and extended. Specifically, VBM studies have revealed that PD patients with dementia present basal ganglia,27 28 limbic,27 29 paralimbic27 and widespread neocortical atrophy,28 29 and functional imaging studies have reported an extensive hypoperfusion in anterior in addition to posterior brain regions.30
As previously mentioned, if patients with VH are at high risk of developing dementia,10 11 it is probable that the brain atrophy in these patients will progressively affect all the areas described to be involved in dementia in PD including limbic, paralimbic and neocortical areas. To test this hypothesis, longitudinal MRI studies are needed, but so far there are only two longitudinal studies using VBM in PD. The first one showed limbic and temporo-occipital areas of grey-matter reduction after 25 months' follow-up,31 while the other study did not show any areas of grey-matter loss in PD patients after a follow-up period of 1.4 years.32 These studies did not differentiate between patients with and without hallucinations. The present study used VBM to assess progression of regional brain atrophy and cognitive deficits over time in patients with and without VH. We expected that, unlike patients without VH, hallucinating PD patients will show a pattern of progressive cortical atrophy parallel to the development of dementia.
Methods
Participants
Patients were recruited from an outpatient movement disorders clinic (Parkinson's disease and Movement Disorders Unit, Department of Neurology, Hospital Clinic, Barcelona) in collaboration with the Department of Psychiatry and Psychobiology (University of Barcelona). The participants were part of a previously studied sample,4 15 17 26 and they were invited by telephone for a follow-up assessment.
Twelve PD patients with VH, 14 PD patients without VH and 12 controls participated in the follow-up assessment. The average follow-up period was 29.91 months (SD=5.74; range=26–32). The inclusion criterion for patients with VH was the presence of well-formed VH. Patients with only minor forms consisting of a sensation of presence, a sideways passages or illusions were not included in the study.
All participants included in the present study had neuropsychological and MRI evaluation both at baseline and at follow-up. At baseline 18 patients with VH and 20 without VH participated in the study, and at follow-up there was an attrition of 33.3% and 30% in each group respectively. Two participants from the original PD sample with VH died. In another four cases, the next of kin decided not to give consent for participation due to severe deterioration of the patients' condition. From the original PD without VH group, three subjects declined to participate. In three others cases, brain MRI could not be obtained due to severe motor impairment.
The study was approved by the institutional ethics committee. Written informed consent was obtained from the patients or their care giver in the case of patients with overt cognitive impairment, and from healthy controls after full explanation of the procedures involved in the study.
The baseline and follow-up examinations included assessment of the following.
Hallucinations
To evaluate the phenomenology of hallucinations, we used a structured interview designed at our hospital which comprised items covering modality (visual, auditory, tactile and olfactory), content (animals, people and objects) and temporal aspects (time of the day, frequency and duration). The severity of the VH was rated using the Spanish version of the neuropsychiatric inventory (NPI) (subscale hallucinations). The VH in the PD sample consisted of well-formed images of people, faces or animals. Hallucinations occurred only in the visual modality while patients were alert and had their eyes open. None of the VH patients experienced auditory hallucinations. Insight into the hallucinatory nature of the phenomenon was maintained in 58.3% of the patients. Associated delusions were present in 33.3% of the patients. These delusions were primarily paranoid in type and involved elementary misbelieves concerning infidelity or theft.
Dementia
The diagnosis of dementia was based on an interview with the patient and the care giver using the Diagnostic and Statistical Manual of Mental Disorders, Revised, Fourth Edition (DSM IV-TR), and the Movement Disorder Society's diagnostic criteria for Parkinson's Disease Dementia.33 The algorithm for the diagnosis of dementia in PD (level I) requires: (1) a diagnosis of PD, (2) the development of PD prior to the onset of dementia, (3) MMSE below 26; (4) presence of cognitive deficits severe enough to impact daily living, (5) impairment in more than one cognitive domain including attention, executive function, visuo-constructive ability and memory.33
Clinical and demographical variables
Age, education, duration of illness and medication were recorded. Illness severity was staged according to Hoehn and Yahr, and motor subscale from the Unified Parkinson's Disease Rating Scale (UPDRS). Mood was assessed using the Hamilton Depression Rating Scale (HDRS). All patients were treated with levodopa alone or a combination of levodopa and dopamine-agonist (pramipexole, ropinirole or pergolide). In order to take into account the amount of all dopaminergic drugs taken, we calculated a levodopa-equivalent dose for each patient. Four (33.3%) PD patients with VH were taking atypical antipsychotics (quetiapine and clozapine). None of the patients was treated with anticholinesterases. All patients were highly l-dopa responsive. Both baseline and follow-up clinical and cognitive examinations were performed in the on phase.
Neuropsychological functions
The evaluation included assessment of general cognitive ability with information and similarities subtest from WAIS-III, verbal memory with the Rey Auditory Verbal Learning Test (RAVLT); visual memory by means of the Warrington Recognition Memory for Faces; visuoperceptive functions with the Visual Form Discrimination Test and Benton Facial Recognition Test; frontal functions with phonetic and semantic fluencies; language comprehension with the Token Test and Boston Naming Test for naming. The neuropsychological battery was administered by a trained neuropsychologist (BR-R) and completed in a single session lasting 1–2 h. The neuropsychologist in charge of the assessment was the same at baseline and follow-up examinations, but she was not blinded to the presence of VH.
Neuropsychological and clinical statistical analysis
Statistical analysis was carried out using SPSS 14.0 (SPSS Inc, Chicago (IL), USA). A multivariate analysis of variance was performed on demographic, clinical and neuropsychological variables to examine the differences between groups at baseline. A post-hoc pairwise Tukey test was performed where appropriate. The analysis for changes in clinical and neuropsychological variables between baseline and follow-up was performed using the general linear mixed model (GLM) for repeated measures to test whether these variables differed across time in the groups. A two-factor (3×2) ANOVA was performed for each of the variables. Factors were: group (PD with VH, PD without VH and controls) and time (baseline and follow-up), and we obtained the effect of time (FT), effect of group (FG) and interaction between group and time (FG×T) for each of the variables. In addition, to adjust for baseline performance in comparing the rate of decline between groups, we calculated a percentage of loss for each of the cognitive variables with a significant interaction between time and group using the following formula:
MRI acquisition and voxel-based morphometry analysis
All scans were obtained from a 1.5 T GE Nvi/Cvi 8.4 machine (GE, Milwaukee, Wisconsin). The imaging protocol included an axial 3D IR Prep SPGR (Inversion Recovery Prepared Spoiled Gradient-echo) sequence of the entire brain and the following parameters: TR (repetition time)=17; TE (echo time)=5; TI (inversion time)=300; 1.5 mm thickness; FoV (field of view)=24×24; 256×256; 1 NEX (number of excitations). No hardware or software upgrades were made over the period of assessment, since longitudinal techniques are very sensitive to any change in image acquisition.
A VBM analysis34 was carried out on SMP5 (Statistical Parametric Mapping, Wellcome Department of Cognitive Neurology, University College London, London; http://www.fil.ion.ucl.ac.uk.spm) running in Matlab 7.0 (Mathworks, Natick, Massachusetts). First, images were reoriented according to the anterior–posterior commissure and were then segmented into grey matter, white matter and cerebrospinal fluid using the unified model that integrates segmentation and normalisation. The spatial normalisation involves registering each of the images onto the SPM T1 template, whereas the segmentation step uses a priori probability maps to segment different tissues. A separate ‘modulation’ step was used to control for deformations from the spatial normalisation step, and the grey-matter modulated images obtained were smoothed with an 8 mm full-width at half-maximum isotropic Gaussian kernel. The smoothed grey-matter images were analysed in a factorial design (3×2) ANOVA. MMSE score at baseline was also introduced as a covariate in an ANCOVA model.
Our main interest was the comparison between baseline and follow-up for each of the groups, specifically for the progression of volume loss in PD patients with and without VH over time.
We also carried out correlation analyses in SPM5 between the progression of cerebral atrophy (image at baseline minus image at follow-up) and the decline in neuropsychological tests (score at baseline minus score at follow-up). The correlation analyses were performed with those neuropsychological variables that showed a significant decline over time.
All results were thresholded at a cluster and voxel level of p<0.05 corrected for multiple comparisons by false discovery rate (FDR) which controls the expected proportion of the rejected hypothesis that are falsely rejected.
Results
Neuropsychological and clinical variables
There were no differences in age, gender, years of education and years of disease duration between groups (see table 1), but groups were not matched for global cognitive function, and the MMSE scores were significantly lower in those with VH. Neuropsychological and clinical data both at baseline and at follow-up are presented in table 2. None of the patients in either group met criteria for dementia at baseline examination, but the PD patients with VH were already more cognitively impaired than PD without VH. Specifically, at baseline, patients with VH showed impairment in verbal memory and facial recognition in comparison with PD without VH.
Two and a half years after the baseline evaluation, nine out of 12 PD with VH presented with dementia, whereas none of the PD without VH patients met criteria for dementia. The neuropsychological functions showing a significant interaction between group and time are presented in figure 1. In the PD with VH group, we observed a 6.5-point decline in the MMSE (FG=13.8, p<0.001; FT=19.6, p<0.001; FG×T=9.2; p<0.001). In the specific cognitive domains, PD with VH group showed a faster and steeper cognitive decline than PD without VH and controls in verbal learning (FG=12.7, p<0.001; FT=0.4, p<0.5; FG×T=12.7, p<0.001), delayed recall (FG=34.7, p<0.001; FT=0.8, p<0.4; FG×T=6.5; p<0.004), semantic fluency (FG=11.4, p<0.001; FT=13.6, p<0.001, FG×T=3.6; p<0.04), language comprehension (FG=10.9; p<0.001; FT=2.5; p<0.13; FG×T=6.5; p<0.004) and visuoperceptive function (facial recognition) (FG=20.3; p<0.001; FT=16.7; p<0.001; FG×T=4.7; p<0.016). When adjusting for baseline performance, the percentage of loss in each of these neuropsychological variables remained statistically significant (see table 3).
Regarding the clinical scales, PD patients with VH were at a more advanced stage of the disease at baseline (see table 2). At follow-up, patients from both groups showed a significant motor deterioration measured using the UPDRS-III motor scale (FG=3.1; p<0.097; FT=17.5; p<0.001; FG×T=2.1; p<0.164) and a progression of the illness rated by means of the Hoehn and Yahr scale (FG=6.5; p<0.018; FT=9.1; p<0.007; FG×T=0.4; p<0.527) with no significant interactions between time and group. l-dopa dosage was only increased from baseline to follow-up in the PD without VH group and not in the VH sample in an attempt to reduce the VH.
Voxel-based morphometry results
From baseline to follow-up, PD with VH patients showed grey-matter loss in the bilateral parietal cortex (mostly precuneus and supramarginal gyrus), bilateral insula, bilateral superior and inferior temporal gyrus, bilateral superior and inferior frontal gyrus, bilateral anterior (ventral and dorsal) and left dorsal posterior cingulate gyrus, bilateral thalamus and limbic areas (nucleus accumbens, amygdala, and hippocampus) (figure 2, table 4). In the PD without VH group only small clusters of grey-matter loss were observed in the right frontal areas including the primary motor, premotor, supplementary motor areas, and the anterior and posterior areas of the cerebellum (see table 5, figure 2). Healthy controls did not show any clusters of significant grey-matter loss.
Correlation analyses between cognitive decline and progression of atrophy
We looked for correlations between grey-matter loss over time and those cognitive functions that showed a significant decline in PD patients with VH. The decline in learning was correlated with hippocampal head atrophy, and delayed recall worsening was related to grey-matter loss in the left prefrontal cortex. The decline in semantic fluency was related to progressive atrophy in the thalamus, and language comprehension was related to atrophy in the amygdala (see table 6). The correlations between the maxima of each of the significant clusters and cognitive decline are plotted in figure 3.
Discussion
The present study provides evidence of progression to dementia associated with distinct brain atrophy in PD patients with VH. These patients presented progressive and extensive grey-matter loss involving limbic, paralimbic and neocortical areas, whereas PD patients without VH only showed small clusters of progressive atrophy in frontal areas and cerebellum.
The cognitive domains that showed a significant decline over time in PD patients with VH compared with patients without hallucinations were verbal memory, semantic fluency, language comprehension and visuoperceptive functions. In our cross-sectional study,4 we already reported impairment in these cognitive domains. Interestingly, semantic fluency35 and verbal memory36 have previously been reported to predict dementia in PD. After 1 year of follow-up, nearly half of these patients met criteria for dementia.15 In the present study, 2 years and 6 months after the initial evaluation, 75% of patients with VH developed dementia.
The cognitive decline in PD patients with VH was accompanied by progressive brain atrophy involving limbic, paralimbic and neocortical areas. Grey-matter reductions were extensive and bilaterally symmetrical, involving both anterior and posterior cortical regions. This pattern of grey-matter atrophy is comparable with that reported in PD with dementia. Specifically, cross-sectional VBM studies found that PD patients with dementia present basal ganglia,27 28 limbic,27 29 paralimbic27 and widespread neocortical atrophy.28 29 Functional imaging studies showed extensive hypoperfusion billateraly involving posterior but also anterior regions.30 In our previous cross-sectional VBM study, we found that non-demented PD patients with VH showed atrophy of the visual association areas in comparison with non-hallucinating PD patients.17 Functional imaging studies have also reported perfusion abnormalities in visual association areas in non-demented hallucinating patients.18–21 These findings could suggest that in PD patients with VH, there is an initial involvement of posterior neocortical areas extending to all associative neocortical areas when dementia is developed.
Neuropathological studies in PD propose limbic and/or neocortical Lewy bodies as the main substrate of dementia in PD.37 The cause of VH in PD, even if for many years it was thought to be a complication of antiparkinsonian treatment, is now thought to be nerve-cell loss and Lewy-body pathology in the ventral-temporal regions of the brain.1 25 We recently reported hippocampal head reductions in PD with VH in comparison with healthy controls.26 In this longitudinal study, we found that the atrophy of medial temporal areas increases from baseline to follow-up, suggesting that both VH and dementia may form part of the same degenerative process, VH being the first expression and dementia later. In contrast, patients without VH did not show a significant decline in the cognitive domains explored, and the grey-matter loss only affected small regions in the frontal areas and cerebellum.
In our study, both PD with and without VH groups had the same age and disease duration, although PD patients with VH showed a more advanced stage of the illness and a lower cognitive status. These characteristics, together with the severity of motor symptoms, have been consistently associated with the presence of dementia in PD patients.38 The remarkable 6.5-point decline in the MMSE of PD patients with VH was even greater than that reported in demented patients.13 This could suggest that PD patients with VH were already at the starting-point of a more general cognitive decline which later led to dementia. However, the different pattern of brain atrophy in PD patients with and without VH cannot be explained by differences in general cognitive function at baseline, since the grey-matter loss in specific brain areas remained significant after MMSE was introduced as a covariate in the VBM analysis.
We found that the progressive atrophy in specific brain areas was associated with a decline in several cognitive domains but not with the MMSE. The correlation analyses revealed that the higher the decline in verbal memory, the higher the atrophy in medial temporal and frontal areas. Specifically, the learning decline was related to hippocampal head atrophy, and the worsening in free delayed recall performance was related to progressive prefrontal cortex. We previously reported a correlation between hippocampal head atrophy and learning in PD,26 but this new finding gives further evidence of progressive hippocampal atrophy as a neuroanatomical substrate for learning decline in PD with VH. On the other hand, delayed recall was related to grey-matter atrophy in the prefrontal cortex, showing that both encoding and retrieval deficits are responsible for memory dysfunction in PD. The decline in semantic fluency showed a correlation with progressive thalamic atrophy, which is in agreement with the fact that vascular lesions in the thalamus impair verbal fluency.39 Finally, language comprehension also showed a great decline over time and was related to the progressive atrophy of medial temporal structures, specifically, the amygdala. These results supported the evidence that not only does the dopamine deficiency affecting corticostriatal-cortico information exchange in PD have an impact on cognitive dysfunction, but structural grey-matter changes may account for it.
One of the methodological limitations of the present work was that the neuropsychologist involved in the cognitive assessment at baseline and follow-up was not blind to the presence of VH, which might have had an influence on the neuropsychological results but was unlikely to have had an influence on the neuroimaging results. Another limitation was that our neuropsychological battery was designed to evaluate mainly temporal lobe functions. However, frontal dysfunction has been found to play an important role in the presence of VH.3–5 7 8 In our study, we only used verbal fluencies to assess executive dysfunction, but they have been shown to be an index of the progressive deterioration of executive functions in PD.40 In addition, following the diagnostic procedures recommended by the Movement Disorder Society Task Force, we considered demented patients to be only those PD patients with an MMSE below 26. We appreciate that this is a strict criterion because patients with an MMSE score above this might also present with dementia. Finally, the small number of patients in each group and the attrition of 33.3% in the VH group and 30% in the non-VH sample are further methodological limitations of the present study.
In conclusion, this 2.5-year follow-up of a cohort of PD patients shows that, unlike PD patients without VH, PD patients with VH frequently developed dementia, which is associated with a significant grey-matter loss in limbic, paralimbic and neocortical areas. Our present work provides a new insight into the study of progression of brain atrophy in PD showing a different pattern of regional atrophy in PD patients with and without VH.
References
Footnotes
Funding This work was supported by Generalitat de Catalunya (2005SGR0000836 to ET, 2005SGR00855 to CJ, UNI/2001/2003 to ET, Uni/2001/2004 to CJ); Ministerio de Educación y Ciencia (AP 2005-019 to NI-B) and de Investigación Biomédica en red sobre Enfermedades Neurodegenerativas (CIBERNED).
Competing interests None.
Patient consent Obtained.
Ethics approval Ethics approval was provided by the Hospital Clinic.
Provenance and peer review Not commissioned; externally peer reviewed.