Neurophysiological signature(s) of visual hallucinations across neurological and perceptual
Dauwan, Meenakshi
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2019
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Dauwan, M. (2019). Neurophysiological signature(s) of visual hallucinations across neurological and perceptual: and non-invasive treatment with physical exercise. Rijksuniversiteit Groningen.
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
CHAPTER
4
1 Brain Center Rudolf Magnus, University Medical Center Utrecht, The Netherlands
2 Department of Clinical Neurophysiology and MEG Center, VU University Medical Center, Amsterdam, The Netherlands
3 University of Groningen, University Medical Center Groningen, The Netherlands
4 Alzheimer Center, VU University Medical Center, Amsterdam, The Netherlands
5 Department of Biological and Medical Psychology, University of Bergen, Norway
Meenakshi Dauwan1,2,3 Mascha M.J. Linszen1,3,4 $ÀQD:/HPVWUD4 Philip Scheltens4 Cornelis J. Stam2,* Iris E. Sommer1,3,5,*
* These authors are joint senior authors
EEG-based
neurophysiological
indicators of hallucinations
in Alzheimer’s disease:
comparison with dementia
ABSTRACT
We studied neurophysiological indicators of hallucinations in Alzheimer’s disease patients with hallucinations (ADhall+), and compared them with non-hallucinating AD (ADhall-) and dementia with Lewy bodies (DLBhall+) patients. 36 matched ADhall+ and 108 ADhall-, and 29 DLBhall+ patients were selected from the Amsterdam Dementia Cohort. EEG spectral and functional connectivity (FC) analysis (phase lag index) were performed. Quantitative and visual EEG measures were combined in a random forest algorithm to determine which EEG-based variable(s) play a role in hallucinations. $'KDOOVKRZHGORZHUSHDNIUHTXHQF\ YV +]S DDQGESRZHUDQG a2-FC, but higher d-power compared to ADhall-. ADhall+ showed lower d-power, higher b-power, and a1-FC than DLBhall+, but did not differ in peak frequency (7.26 vs. 6.95Hz), q- or a-power. ADhall+ patients could be differentiated from ADhall- and DLBhall+ with a weighted accuracy of 71% with a1-power and 100% with b-FC, the two most differentiating features. In sum, EEG slowing and decrease in a1- and b-band DFWLYLW\IRUPSRWHQWLDOQHXURSK\VLRORJLFDOLQGLFDWRUVRIXQGHUO\LQJFKROLQHUJLFGHÀFLHQF\ in ADhall+ and DLBhall+.
1. INTRODUCTION
Hallucinations are one of the common neuropsychiatric symptoms in Alzheimer’s disease (AD), with prevalence rates ranging widely from 6% to 41% (Zhao et al., 2016a). Hallucinations in AD are associated with faster disease progression, higher caregiver burden, institutionalization, and increased mortality (Scarmeas et al., 2005; Wadsworth et al., 2012; Yaffe et al., 2002). However, it remains unclear why some AD patients develop hallucinations while others remain unaffected. In neuropathological studies, a strong correlation is found between reported hallucinations in AD and the presence of /HZ\ERGLHVEXWQRWZLWKWKHSUHVHQFHRIQHXULWLFSODTXHVDQGQHXURÀEULOODU\WDQJOHV (Ballard et al., 2004; Fischer et al., 2015; Jacobson et al., 2014).
Dementia with Lewy bodies (DLB) is a form of dementia in which hallucinations are IUHTXHQWO\HQFRXQWHUHG+DOOXFLQDWLRQVWRJHWKHUZLWKSDUNLQVRQLVPDQGÁXFWXDWLRQV in cognition, attention and alertness, form the core features of DLB, and are generally present early in the course of disease (McKeith et al., 2017). However, mainly in early VWDJHVRI'/%QRWDOOFRUHV\PSWRPVPD\EHSUHVHQWDQGWKHFOLQLFDOSURÀOHPD\ overlap with AD, resulting in a diagnostic dilemma (Gouw AA; Stam CJ, 2016; McKeith et al., 2016). One solution to this diagnostic dilemma is electroencephalography (EEG), which directly measures neural activity, and has a high discriminative value in the GLIIHUHQWLDWLRQEHWZHHQ$'DQG'/%VKRZLQJGLVWLQFW((*SDWWHUQVVSHFLÀFIRU$' and DLB. (Bonanni et al., 2008; Cromarty et al., 2015; Meenakshi Dauwan et al., 2016; Gouw AA; Stam CJ, 2016; Lee et al., 2015; Roks et al., 2008). Furthermore, EEG is a low-cost and noninvasive technique that is widely available in clinical practice. In this study, we aim to explore possible neurophysiological indicators of hallucinations in AD and compare them to DLB patients with hallucinations (DLBhall+) to gain insight in common underlying mechanism(s) of hallucinations in these disorders. To investigate this, we performed EEG-based spectral and functional connectivity analysis between early-stage AD patients with (ADhall+) and without (AD hall-) hallucinations. Subsequently, we combined different quantitative and visual EEG variables in a random forest algorithm, an ensemble-learning method developed by L. Breiman (Breiman, WR ÀQG RXW ZKLFK ((*EDVHG YDULDEOH V ZDVZHUH PRVW GLVWLQFW EHWZHHQ
ADhall+ and AD hall- patients, and could, therefore, aid in making inferences about underlying mechanisms of hallucinations. Finally, we compared all analyses with a group RIHDUO\VWDJH'/%KDOOWRGHWHUPLQHZKHWKHUWKH((*SURÀOHRI$'KDOOSDWLHQWV shows resemblance with DLBhall+, indicating common underlying mechanisms of hallucinations in these disorders.
2. Method
2.1 Study population
36 early-stage (median disease duration of 3.0 years) ADhall+ and 108 ADhall- patients were retrospectively selected from the Amsterdam Dementia Cohort (Van Der Flier et al., 2014). Presence of hallucinations was assessed with the caregiver-based Neuropsychiatric Inventory (NPI) (Cummings et al., 1994). ADhall+ and ADhall- patients were matched on individual level for age, gender and educational level, with a 1:3 ratio. For comparison, 29 early-stage (median disease duration of 3.1 years) probable DLBhall+ patients were selected. The three groups were matched on group level for disease duration and dementia severity measured with the Clinical Dementia Rating (CDR) scale (Morris, 1993). All patients had been referred to the Alzheimer Center of the VU University Medical Center (VUmc, Amsterdam, The Netherlands) between April 2001 and December 2013. Standardized dementia screening included medical history, standard neurological examination, neuropsychological assessment, standard laboratory tests, brain MRI, resting state EEG, and CSF sampling in a subsample (see VXSSOHPHQWDU\PDWHULDOIRUWKH&6)ELRPDUNHUSURÀOHRIWKHWKUHHJURXSV $OOSDWLHQWV gave written informed consent for storage and use of their clinical data for research purposes. This procedure was approved by the Medical Ethics Committee of the VUmc. Patients were diagnosed during a multidisciplinary consensus meeting (Van Der Flier et al., 2014). Probable AD was diagnosed according to the (revised)
NINCDS-2.2 EEG
2.2.1 EEG acquisition and preprocessing
All patients underwent a 20-minutes no-task resting-state EEG recording with OSG digital equipment (Brainlab and BrainRT®; OSG B.V. Belgium). Twenty-one electrodes were placed on the scalp according to the international 10-20 system. Patients were seated in a slightly reclined chair in a sound attenuated room and kept awake by EEG technicians with sound stimuli if necessary (Van Der Flier et al., 2014). The sample IUHTXHQF\ZDV+](OHFWURGHLPSHGDQFHZDVEHORZNZLWKDWLPHFRQVWDQWRI VDQGDORZSDVVÀOWHUDW+] %UDLQODE RU+] %UDLQ57
For each patient, four artifact-free epochs of each 4096 samples (i.e. 4*8.192 sec EEG GDWDSHUVXEMHFWVXIÀFLHQWWRSHUIRUPT((*DQDO\VHV *DVVHUHWDO YDQ'LHVVHQ et al., 2015)) were visually selected (MD/CJS). Data were converted to American Standard Code for Information Interchange (ASCII) format, and loaded into the BrainWave software for further analysis (BrainWave version 0.9.152.4.1, C. J. Stam; available at http:/home.kpn.nl/stam7883/brainwave.html).
2.2.2 Spectral analysis
5HODWLYHSRZHULQÀYHIUHTXHQF\EDQGV GHOWD +]WKHWD +]DOSKD +] alpha2: 10-13 Hz, and beta: 13-30 Hz), and peak frequency (i.e. dominant frequency in 4-13 Hz range) were calculated at each channel with the Fast Fourier Transformation. Considering the spatial brain heterogeneity and its effect on different functions of the lower (associated with attentional processes) and upper (associated with cortical memory processes) alpha frequency band, we used the alpha1 and alpha2 instead of the broad alpha (8-13 Hz) band for all the analysis (Klimesch, 2012; Olejarczyk et al., 2017). The relative power and peak frequency values were averaged over the four epochs per patient to obtain one value per subject.
2.2.3 Functional connectivity
The Phase Lag Index (PLI) was used to calculate functional connectivity strength per channel per epoch for all the above-mentioned frequency bands. EEG source derivation was used as reference (Hjorth, 1975). PLI is based on the Hilbert-transformed
instantaneous phase differences and estimates synchronization between two time series based on the consistency in nonzero phase lag between signals. PLI is relatively insensitive to the effects of volume conduction. The PLI ranges between 0 (no coupling) and 1 (perfect phase locking) (Stam et al., 2007). Per frequency band, an average over all PLI values was calculated and compared between the three groups.
2.2.4 Random forest
2.2.4.1 Quantitative EEG (qEEG)
As previously described, we used the machine learning module of Brainwave to compute, for each patient, a standard set of features for each epoch. This standard set of features included lowest, mean, and highest relative power in the delta (0.5-4 Hz), theta (4-8 Hz), alpha1 (8-10 Hz), alpha2 (10-13 Hz), and beta (13-30 Hz) band, lowest, mean, and highest peak frequency in the range between 4-13 Hz, theta/alpha ratio, functional connectivity measure (PLI) per EEG channel per frequency band, and Minimum Spanning Tree (MST) measures per frequency band (i.e. highest degree, leaf number, and tree hierarchy; Table 1) (Meenakshi Dauwan et al., 2016).
The MST was calculated from the weighted adjacency values of the PLI matrix and is used to characterize network topology across conditions and groups. The MST represents a subgraph of the network containing only the strongest connections without forming loops (Stam et al., 2014).
We also added average PLI per frequency band and three visual EEG measures (see 2.2.4.2. and Table 1) to the set of features.
Table 1: Overview selected features Feature number Feature name Quantitative EEG
1- 3 Lowest, Mean, Highest Delta power
4 - 6 Lowest, Mean, Highest Theta power
7 - 9 Lowest, Mean, Highest Alpha1 power
10 - 12 Lowest, Mean, Highest Alpha2 power
13 - 15 Lowest, Mean, Highest Beta power
16 - 18 Lowest, Mean, Highest peak frequency
19 Theta/alpha ratio
20 - 40 PLI per channel in delta band
41 - 61 PLI per channel in theta band
62 - 82 PLI per channel in alpha1 band
83 - 103 PLI per channel in alpha2 band
104 - 124 PLI per channel in beta band
125 Average PLI in delta band
126 Average PLI in theta band
127 Average PLI in alpha1 band
128 Average PLI in alpha2 band
129 Average PLI in beta band
130 - 132 MST highest degree, leaf number, tree hierarchy in delta band
133 - 135 MST highest degree, leaf number, tree hierarchy in theta band
136 - 138 MST highest degree, leaf number, tree hierarchy in alpha1 band
139 - 141 MST highest degree, leaf number, tree hierarchy in alpha2 band
142 - 144 MST highest degree, leaf number, tree hierarchy in beta band
Visual EEG
145 Severity of EEG abnormalities
146 Diffuse abnormalities
147 Focal abnormalities
Power is the relative power per frequency band (delta (0.5-4 Hz), theta (4-8 Hz), alpha1 (8-10 Hz), alpha2 (10-13 Hz), beta (13-30 Hz)). Peak frequency is the frequency with highest power in range between 4-13 Hz. Theta/alpha ratio is calculated as theta/(theta + alpha1+alpha2). MST highest degree is the maximum degree (i.e. number of links for a given node) within the MST. MST leaf number is the number of nodes in the MST with only one link (i.e. degree). MST tree hierarchy is a measure of optimal network organization.
EEG: electroencephalography; MST: Minimum Spanning Tree; PLI: Phase Lag Index
2.2.4.2 Visual EEG (vEEG)
A standard visual rating scheme was used to assess all EEG recordings (Liedorp et al., 7KHYLVXDOUDWLQJLQFOXGHGSUHVHQFHRIIRFDODEQRUPDOLWLHVGHÀQHGDV WUDQVLHQWV RI VORZRUVKDUSZDYHDFWLYLW\LQRUPRUH((*OHDGVGLIIXVHDEQRUPDOLWLHVGHÀQHGDV a dominant frequency of rhythmic background activity below 8 Hz, diffuse slow-wave activity or diminished reactivity of the rhythmic background activity to the opening of the eyes, and severity of EEG abnormalities on a 4-point rating scale (Liedorp et al., 2009; Van Der Flier et al., 2014).
&ODVVLÀFDWLRQDOJRULWKP
$UDQGRPIRUHVWFODVVLÀHULQFOXGHGLQWKHPDFKLQHOHDUQLQJPRGXOHRI%UDLQZDYH was used to differentiate between ADhall+ and ADhall-, ADhall+ and DLBhall+, and ADhall- and DLBhall+ patients. The number of trees (i.e. nTree parameter) grown in the random forest was set at 500. The number of variables included at each split of each individual tree (nTry) was chosen as the square root of number of features %UHLPDQ 7KHGLVFULPLQDWLYHDELOLW\RIWKHUHVXOWLQJFODVVLÀHUZDVHYDOXDWHGLQ WHUPVRIDFFXUDF\ UDWLREHWZHHQFRUUHFWO\FODVVLÀHGSDWLHQWVDQGWRWDOQXPEHURI SDWLHQWV VHQVLWLYLW\DQGVSHFLÀFLW\7KHLPSRUWDQFHRIHDFKIHDWXUHZDVFDOFXODWHGRQ a score between 0 and 1 and visualized as variable importance (VIMP) score. For a full description of data processing and random forest algorithm, see (Meenakshi Dauwan et al., 2016).
Primary analysis involved the original imbalanced group of 36 ADhall+ and 108 ADhall-, 36 ADhall+ and 29 DLBhall+, and 108 ADhall- and 29 DLBhall+ patients. Because of imbalanced data, we computed a weighted accuracy using the metric described by Chen et al. (Chen et al., 2004):
Weighted accuracy = ȕ x Acc+ + (1-ȕ) x Acc
-with
ȕ = sample size of cases (ADhall+ or DLBhall+) / total sample size (i.e. for ADhall+ vs.
ADhall-: 36/144)
True Positives
Acc+ = True Positive Rate (i.e. sensitivity) =
True Positives + False Negatives
True Negatives
Acc- 7UXH1HJDWLYH5DWH LHVSHFLÀFLW\
True Negatives + False Positives
Secondary analysis included groups with balanced data of 36 ADhall+ and 36 ADhall- patients. For this, we randomly selected 36 unique ADhall- patients. To verify these results, we performed 30 iterations to test multiple combinations of ADhall+ and ADhall- patients. ADhall- patients were randomly selected with replacement and without duplications within an iteration (see supplement for results).
2.3 Statistical analysis
Statistical analyses were performed using IBM SPSS statistics 24.0. Baseline characteristics, spectral measures and PLI values were compared between the three groups. Categorical data were compared using the chi-square test. Continuous data was tested for normality using the Shapiro-Wilk test. Normally distributed variables were compared using one-way ANOVA test. Data that did not follow a Gaussian distribution were compared using nonparametric Kruskal-Wallis test. The False Discovery Rate (FDR) approach with adjusted p value (i.e., q-value) of .05 was used to correct for multiple comparisons. Post-hoc analyses were performed using the chi-square test, independent samples T-test or pair wise Mann-Whitney U test. A p-value RI ZDVFRQVLGHUHGVLJQLÀFDQW
3. RESULTS
3.1 Baseline characteristics
ADhall+ and ADhall- patients were matched on individual level for age, gender, and educational level, with a 1:3 ratio. ADhall+ patients mostly experienced YLVXDOKDOOXFLQDWLRQV Q ,QWKLVJURXSÀYHSDWLHQWV Q DOVR experienced auditory hallucinations, while one (n=1/25, 4%) patient experienced both auditory and olfactory hallucinations. Three (n=3/30, 10%) patients experienced only auditory hallucinations, while one (n=1/30, 3%) patient experienced auditory, tactile, and olfactory hallucinations. One (n=1/30, 3%) patient experienced only olfactory hallucinations. In six (n=6/36, 17%) ADhall+ patients, the modality of hallucinations was unknown. All DLBhall+ patients experienced visual hallucinations. ADhall+ patients VKRZHGDVLJQLÀFDQWO\KLJKHUWRWDO13,VFRUHWKDQWKH$'KDOOJURXSEXWGLGQRW GLIIHUFRPSDUHGWR'/%KDOOSDWLHQWV S 7DEOH '/%KDOOSDWLHQWVSHUIRUPHG EHWWHUWKDQ$'SDWLHQWVRQPRVWQHXURSV\FKRORJLFDOWHVWV S 7KHWKUHHJURXSV did not differ in terms of dementia severity measured with the CDR scale and disease duration (Table 2).
T a b le 2 : P ati ent cha ra c teristi cs ADha ll+ (N = 3 6 ) ADha ll - (N = 1 0 8 ) D LBha ll+ (N = 2 9 ) A ge , y rs 6 9 .4 1 ( 7. 7 4) 6 9 .4 1 ( 7. 3 7 ) 7 0 .7 6 (9 .5 1) S e x , f e m al e 1 9 (5 2 .8 % ) 5 7 (5 2 .8 % ) 9 ( 31 .0 % ) E d u c at io n l e ve l e, g 5 ( 4 – 5) n= 35 5 ( 4 – 5) n= 1 0 6 5 (5 – 7 ) n= 19 D is e as e d u ra ti o n , y rs 3 .0 ( 2 .0 - 5 .0 ) 3 .0 ( 2 .0 – 4 .0 ) 3 .1 (2 .1 – 3 .6 ) n= 26 C D R 1 .0 ( 1 .0 – 2 .0 ) n= 29 1 .0 ( 1 .0 – 1 .0 ) n= 97 1 .0 ( 1 .0 – 1 .0 ) n= 1 6 U se o f h al lu ci n at io n -a ss o ci at e d m e d ic at io n 2 3 ( 6 3 .9 % ) 51 ( 4 7. 2% ) 12 ( 4 1 .4 % ) H is to ry o f h al lu ci n at io n -a ss o ci at e d d is e as e 12 ( 3 3 .3% ) 2 8 ( 2 5 .9 % ) 4 ( 13 .8 % ) E x tr a p yr a m ida l si gns Ri gi d it y d, g, e, h 3 ( 8 .3 % ) n= 29 8 ( 7. 4 % ) n= 1 0 3 12 ( 4 1 .4 % ) n= 2 2 Bra d y k in e sia d, e, h 2 (5 .6 % ) n= 3 0 5 ( 4 .6 % ) n= 1 0 2 14 ( 4 8 .3% ) n= 2 2 T re m o r 1 ( 2 .8 % ) n= 3 0 1 0 (9 .3 % ) n= 1 0 2 4 ( 13 .8 % ) n= 2 1 NP I De lusions c, h ,e, f 11 ( 3 0 .6 % ) n= 3 6 7 ( 6 .5 % ) n= 1 0 8 4 ( 13 .8 % ) n= 12 Ag it at io n c, g 1 5 ( 4 1 .7 % ) n= 3 6 1 6 ( 14 .8 % ) n= 1 0 8 2 ( 6 .9 % ) n= 12 D e p re ss io n 8 ( 2 2 .2% ) n= 3 6 2 7 ( 2 5 .0 % ) n= 1 0 8 5 ( 17 .2% ) n= 12 A n x ie ty 1 5 ( 4 1 .7 % ) n= 3 6 26 ( 2 4 .1 % ) n= 1 0 8 5 ( 17 .2% ) n= 12 E u p h o ri a 3 ( 8 .3 % ) n= 3 6 7 ( 6 .5 % ) n= 1 0 7 0 ( 0 .0 % ) n= 12 A p at hy 29 ( 8 0 .6 % ) n= 3 6 7 6 ( 7 0 .4 % ) n= 1 0 7 1 0 ( 3 4 .5% ) n= 12 D is in h ib it io n 7 ( 19 .4 % ) n= 3 6 9 ( 8 .3 % ) n= 1 0 8 1 ( 3 .4 % ) n= 12 Irrit abilit y c, d ,g 2 7 ( 7 5 .0 % ) n= 3 6 5 0 ( 4 6 .3% ) n= 1 0 8 4 ( 13 .8 % ) n= 12 M o to r d is tu rb an ce 1 0 ( 2 7. 7 % ) n= 35 1 6 ( 14 .8 % ) n= 1 0 7 1 ( 3 .4 % ) n= 12
4
ADha ll+ (N = 3 6 ) ADha ll - (N = 1 0 8 ) D LBha ll+ (N = 2 9 ) 1 6 ( 4 4 .4 % ) n= 3 6 2 8 ( 2 5 .9 % ) n= 1 0 8 7 ( 2 4 .1 % ) n= 12 1 5 ( 4 1 .7 % ) n= 3 6 2 8 ( 2 5 .9 % ) n= 1 0 8 2 ( 6 .9 % ) n= 12 2 3 .0 ( 1 0 .0 – 3 1 .0 ) n= 35 9 .0 ( 3 .7 5 – 1 4 .0 ) n= 1 0 6 11 (5 .5 – 2 1 .7 5) n= 12 tests 19 ( 13 .5 – 2 2 .0 ) n= 3 3 2 1 ( 17 .0 – 2 4 .0 ) n= 1 0 8 2 3 ( 1 8 .0 – 26 .0 ) n= 2 4 3 .0 ( 0 .0 – 9 .0 ) n= 3 3 4 .0 ( 2 .0 – 8 .0 ) n= 1 0 0 1 0 .0 ( 6 .0 – 1 1 .3 ) n= 1 8 8 4 .0 (5 3 .0 – 1 4 6 .0 ) n= 2 7 7 4 .0 ( 4 8 .0 – 1 1 0 .0 ) n= 9 1 7 9 .0 (5 7. 0 – 1 3 9 .5) n= 17 1 0 .0 ( 6 .5 – 1 1 .0 ) n= 3 3 1 0 .0 ( 8 .0 – 1 2 .0 ) n= 1 0 0 12 .0 ( 1 0 .0 – 1 3 .0 ) n= 19 f 5 .0 ( 3 .7 5 – 7 .0 ) n= 3 0 6 .0 ( 4 .0 – 8 .0 ) n= 97 7. 0 (5 .7 5 – 9 .2 5 ) n= 1 8 5 0 8 .75 ( 1 56 .7 4) n= 2 8 4 8 9 .1 2 ( 14 8 .2 9 ) n= 78 6 0 8 .4 0 ( 19 8 .3 3 ) n= 2 0 7 6 6 .0 0 ( 4 1 0 .0 5) n= 2 8 8 0 4 .2 8 ( 4 8 9 .6 5) n= 78 35 0 .7 0 ( 2 26 .8 8 ) n= 2 0 8 7. 0 0 ( 3 7. 1 5 ) n= 2 8 9 3 .2 8 ( 4 1 .2 8 ) n= 78 5 4 .8 0 ( 2 7. 4 1) n= 2 0 .3 5 ( .1 2 ) .2 8 ( .1 0 ) .4 4 ( .1 4 ) .2 5 ( .1 0 ) .2 2 ( .1 0 ) .2 8 ( .0 9 ) .1 1 (. 0 6 ) .1 4 (. 0 7 ) .10 (. 0 6 ) .0 7 ( .0 4 ) .1 0 ( .0 6 ) .0 6 ( .0 3 ) .1 5 ( .0 8 ) .1 8 ( .0 7 ) .0 9 ( .0 4 ) 7. 2 6 ( 1 .4 2 ) 7. 9 4 ( 1 .1 5 ) 6 .9 5 ( 0 .9 3 ) .5 7 (. 1 8 ) .4 8 (. 17 ) .6 4 (. 1 5 ) 3 ( 8 .3 % ) 6 (5 .6 % ) 9 ( 31 % )
at a a re m e an ( S D ), m e d ia n (i n te rq u ar ti le r an ge ), o r n (% ). E d u c at io n l e ve l w as a ss e ss e d w it h t h e 7 -i te m V e rh ag e c o d in g s ys te m f o r e d u c at io n ( V e rh ag e , 1 9 6 4 ). D is e as e u ra ti o n m e as u re d a s ye ar s s in ce o n se t o f c o m p la in ts . C D R i s a 5 -p o in t s c al e u se d t o e v al u at e t h e s e ve ri ty o f d e m e n ti a . H al lu ci n at io n -a ss o ci at e d m e d ic al h is to ry s c hiz ophr e ni a spec tr um dis o rd er , mood dis o rd er , s u bst an ce a b u se , p e rs o n al it y d is o rd e r, P T S D , a n x ie ty d is o rd e r, P ar k in so n dis e as e , he aring imp air me nt , is u al i m p ai rm e n t, e p il e p sy , s y st e m ic l u p u s e ry th e m at o su s a n d /o r a u ti sm s p e c tr u m d is o rd e r. H al lu c in at io n -a ss o c ia te d m e d ic at io n s ( i. e . dr ug s wit h h allucin at ions ted a s side -e ff e c t) included: be nzodi a ze pine s, ant ide pr e ss ant s, ant ipsy c hot ic dr ug s, or al ant icholine rgic dr ug s, choline st e ra se inhibit o rs , dop amine rgic dr ug s, or al EH WD E ORFN HU V R SL DW HV DQW L HSLOH SW LF G UX JV PH P DQW LQH P HW K\ OSKH Q LG DW H P RG DÀQLO DQW LPLJU DLQH G UX JV OLW K LXP E HW DKLV WL QH D Q WL K LV WD P LQ HG UX JVD Q G R US UR WR Q S X P S h ib it o rs ( F ar m ac o th e ra p e u ti sc h K o m p as , w w w .f ar n ac o th e ra p e u ti sc h ko m p as .n l) . E x tr ap y ra m id al s ig n s we re q u al it at iv e ly a ss e ss e d o n t h e ir p re se n ce o r a b se n ce a t WK H À UV W FO LQ LF DO S UH VH Q WD WL R Q ) D] HN DV VF R UH LV D P HD VX UH R IZ K LW H P DW WH U K\ S HU LQ WH Q VL WL HV R Q 7 Z HL JK WH G Á X LG D WW HQ X DW HG LQ YH UV LR Q UH FRYH U\ ) / $ ,5 LP DJ LQ J 3 RZH U t h e r e la ti ve p o we r p e r f re q u e n c y b an d ( d e lt a [ 0 – 4 H z] , t h e ta [ 4 – 8 H z] , a lp h a1 [ 8 –1 0 H z] , a lp h a 2 [ 1 0 – 1 3 H z] , a n d b e ta [ 1 3 – 3 0 H z] ). Pe ak f re q u e n c y i s t h e f re q u e n c y it h h ig h e st p o we r i n r an ge b e tw e e n 4 a n d 1 3 H z. T h e ta /a lp h a r at io i s a n i n d e x t h at s h ow s t h e p e rc e n ta ge o f t h e ta ve rs u s a lp h a sp e c tr al p o te n ti al d u ri n g r e st in g s ta te , m p u te d a s t h e ta /( th e ta 1 + al p h a1 + al p h a 2 ). ȕ42 : a m yl o id -ȕ 1 -42 ; A D h al l+ : A lz h e im e r’ s d is e as e w it h h al lu ci n at io n s; A D h al l-: A lz h e im e r’ s d is e as e w it h o u t h al lu ci n at io n s; A P O E : a p o lip o p ro te in , e 4 alle le; C D R : Clinic al e m e n ti a R at in g S c al e ; C S F : C e re b ro sp in al F lu id ; D L B h al l+ : d e m e n ti a w it h L e w y b o d ie s w it h h al lu ci n at io n s; F IR D A : F ro n ta l I n te rm it te n t R h y th m ic D e lt a A c ti vi ty ; G C A : lo b al C o rt ic al A tr o p h y ; M M S E : M in i M e n ta l S ta te E x am in at io n ; M T A : M e d ia l T e m p o ra l l o b e A tr o p h y ; N P I: N e u ro p sy c h ia tr ic I n ve n to ry; p -T au : t au phospho ryl at ed a t re o n in e 1 8 1 ; q E E G : Q u an ti ta ti ve E E G ; s e c: s e co n d s; T au : t o ta l T au ; T M T -A : T ra il -M ak in g T e st p ar t A ; V A T : V is u al A ss o ci at io n T e st ; y rs : ye ar s 6 HHV X S S OH P HQ WD U\P DW HU LD OI R UW K H& 6)E LR P DU NH US UR À OHR IW K HW K UH HJ UR X S V 6 LJ Q LÀ FD Q WO \G LI IH UH Q WE HW ZH HQD OOJ UR X S V 6 LJ Q LÀ FD Q WO \G LI IH UH Q WE HW ZH HQ$ ' K DO OD Q G$ ' K DO O 6 LJ Q LÀ FD Q WO \G LI IH UH Q WE HW ZH HQ$ ' K DO OD Q G' /% K DO O 6 LJ Q LÀ FD Q WO \G LI IH UH Q WE HW ZH HQ$ ' K DO OD Q G' /% K DO O S S S
4
3.2 Spectral analysis
Figure 1 shows, for all three groups, the mean power spectrum. The peak frequency GLIIHUHGVLJQLÀFDQWO\EHWZHHQ$'KDOO +] DQG$'KDOO +] SDWLHQWV S DQGEHWZHHQ$'KDOODQG'/%KDOO +] SDWLHQWV S EXWQRWEHWZHHQ ADhall+ and DLBhall+ patients (Table 2). Relative delta and beta band power differed VLJQLÀFDQWO\DPRQJWKHWKUHHJURXSVZLWK'/%KDOOSDWLHQWVVKRZLQJKLJKHVWGHOWDDQG lowest beta power followed by ADhall+ and ADhall- group (Table 2). ADhall+ patients VKRZHGORZHUDOSKDSRZHUDQGKLJKHUWKHWDDOSKDUDWLRWKDQ$'KDOOSDWLHQWV S ADhall+ patients did not differ from DLBhall+ patients in theta, alpha1 and alpha2 SRZHUDQGWKHWDDOSKDUDWLR$'KDOOSDWLHQWVVKRZHGVLJQLÀFDQWO\KLJKHUDOSKDDQG alpha2 power, and lower theta power and theta/alpha ratio compared to DLBhall+ patients (Table 2).
3.3 Functional connectivity
7KHDYHUDJH3/,ZDVVLJQLÀFDQWO\KLJKHULQWKHDOSKDEDQGLQ$'KDOO PHGLDQ IQR 0.185-0.231) and ADhall- (median 0.214, IQR 0.187-0.253) patients compared to the DLBhall+ group (median 0.188, IQR 0.177-0.203; Table 3). ADhall+ patients PHGLDQ,45 VKRZHGVLJQLÀFDQWO\ORZHUDYHUDJH3/,LQWKHDOSKD EDQGWKDQWKH$'KDOOJURXS PHGLDQ,45 7KHUHZHUHQRVLJQLÀFDQW differences between the three groups in the other frequency bands.
Table 3: Average PLI value per frequency band
ADhall+ (n=36) ADhall- (n=108) DLBhall+ (n=29) Delta 0.151 (0.139-0.163) 0.150 (0.143-0.165) 0.158 (0.142-0.168)
Theta 0.151 (0.132-0.163) 0.148 (0.132-0.171) 0.151 (0.142-0.167)
Alpha1b,d,c,e 0.210 (0.185-0.231) 0.214 (0.187-0.253) 0.188 (0.177-0.203)
Alpha2a,e 0.152 (0.139-0.165) 0.158 (0.149-0.181) 0.155 (0.141-0.165)
Beta 0.071 (0.065-0.077) 0.070 (0.066-0.078) 0.073 (0.068-0.080)
ADhall+: Alzheimer’s disease with hallucinations; ADhall-: Alzheimer’s disease without hallucinations; DLBhall+: dementia with Lewy bodies with hallucinations
a6LJQLÀFDQWO\GLIIHUHQWEHWZHHQ$'KDOODQG$'KDOO b6LJQLÀFDQWO\GLIIHUHQWEHWZHHQ$'KDOODQG'/%KDOO c6LJQLÀFDQWO\GLIIHUHQWEHWZHHQ$'KDOODQG'/%KDOO dS eS 3.4 Random forest
3.4.1 Primary analyses - Imbalanced groups
&ODVVLÀHUSHUIRUPDQFHIRUDOOWKUHHFODVVLÀFDWLRQV $'KDOOYV$'KDOO$'KDOOYV DLBhall+, and ADhall- vs. DLBhall+) is shown in Table 4. Using both qEEG and vEEG IHDWXUHV Q WKHFODVVLÀHUGLVFULPLQDWHG$'KDOODQG$'KDOOSDWLHQWVZLWKD ZHLJKWHGDFFXUDF\RIVHQVLWLYLW\RIDQGVSHFLÀFLW\RI+LJKHVWUHODWLYH alpha1 power (higher in ADhall- group) was the most important discriminating feature, followed by PLI value of channel C3 in delta band (higher in ADhall- group), and highest relative delta power (higher in ADhall+ group) (Figure 2). Evaluation of highest UHODWLYHDOSKDSRZHUDVDVLQJOHFODVVLÀHUUHVXOWHGLQDQDFFXUDF\RIVHQVLWLYLW\ DQGVSHFLÀFLW\ZHUHDQGUHVSHFWLYHO\
Differentiation between ADhall+ and DLBhall+, and ADhall- and DLBhall+ was possible with an accuracy of 100% (Table 4). For ADhall+ vs. DLBhall+, the PLI value of channel T4 in beta band (higher in ADhall+ group) was the most important feature, followed by the PLI value of channel A2 and C3 in beta band (both higher in ADhall+ group). The PLI value of channel T4 in the beta band, solely, resulted in an accuracy of 95%, VHQVLWLYLW\RI DQGVSHFLÀFLW\RI
For ADhall- vs. DLBhall+, severity of EEG abnormalities (more severe in DLBhall+ group) showed the highest VIMP score, followed by the PLI value of channel A2 and T4 in beta band (both higher in ADhall- group). Using severity of EEG abnormalities as WKHRQO\IHDWXUHUHVXOWHGLQDQDFFXUDF\RI ZLWKDVHQVLWLYLW\RI DQGVSHFLÀFLW\ of 100%.
Table 4:5DQGRPIRUHVWFODVVLÀHUUHVXOWVIRULPEDODQFHGDQGEDODQFHGJURXSV
Primary analysis (Imbalanced groups: 36 ADhall+, 108 ADhall-, 29 DLBhall+) Group and feature selection Weighted accuracy
(%) Sensitivity (%) Specif icit y (%) ADhall+ vs.
ADhall-Quantitative and visual EEG 71 20 88
Only quantitative EEG 70 20 86
ADhall+ vs. DLBhall+
Quantitative and visual EEG 100 100 100
Only quantitative EEG 100 100 100
ADhall- vs. DLBhall+
Quantitative and visual EEG 100 99 100
Only quantitative EEG 97 84 100
Secondary analysis (Balanced groups: 36 ADhall+ and 36 ADhall-) Group and feature selection Accuracy
(%)
Sensitivity (%)
Specif icit y (%)
Figure 2: 9DULDEOHLPSRUWDQFH 9,03 VFRUHV RQDVFDOH LQWKHWKUHHFODVVLÀFDWLRQVXVLQJ quantitative and visual EEG features. VIMP scores show the relative importance of a feature for discrimination between ADhall+ and ADhall-, ADhall+ and DLBhall+, and ADhall- and DLBhall+. )RUHDFKFODVVLÀFDWLRQWKHWKUHHPRVWLPSRUWDQWIHDWXUHV 7DEOH DUHOLVWHG
ADhall+: Alzheimer’s disease with hallucination; ADhall-: Alzheimer’s disease without hallucinations; DLBhall+: dementia with Lewy bodies with hallucinations; PLI: Phase Lag Index
3.4.2 Secondary analysis - Balanced AD groups
&ODVVLÀFDWLRQRI$'KDOODQG$'KDOOEDVHGRQEDODQFHGJURXSVGHFUHDVHGWKHRYHUDOO DFFXUDF\WRZLWKVHQVLWLYLW\RIDQGVSHFLÀFLW\RIUHVSHFWLYHO\ 7DEOH Highest relative delta power was the most discriminating feature, followed by lowest relative alpha2 power and MST tree hierarchy in the alpha1 band. Highest delta power, DVDVLQJOHIHDWXUHUHVXOWHGLQDQDFFXUDF\VHQVLWLYLW\DQGVSHFLÀFLW\RI
4. DISCUSSION
0DLQÀQGLQJV7KLVVWXG\LVWKHÀUVWWRH[SORUHQHXURSK\VLRORJLFDOLQGLFDWRUVRIKDOOXFLQDWLRQVLQ $'WRDSSO\DPDFKLQHOHDUQLQJFODVVLÀHUWRÀQGPRVWGLIIHUHQWLDWLQJ((*YDULDEOH V between ADhall+ and ADhall- that might play a role in possible underlying mechanisms of hallucinations in AD, and to investigate possible similarities with DLBhall+. First, compared to ADhall- patients, more severe EEG slowing was seen in ADhall+ patients together with a lower peak frequency, lower alpha2 and higher delta power, and higher theta/alpha ratio. In contrast, EEG slowing was similar in ADhall+ and DLBhall+ group ZLWKQRVLJQLÀFDQWGLIIHUHQFHLQSHDNIUHTXHQF\DOSKDDQGWKHWDSRZHUDQGWKHWD alpha ratio, although delta power was higher in the DLBhall+ group. Second, functional connectivity was lower in the alpha2 band in ADhall+ patients compared to the ADhall- patients, whereas in the alpha1 band functional connectivity was lower in DLBhall+ patients than in the ADhall+ and ADhall- group. Third, differentiation was possible between the ADhall+ and ADhall- group with an accuracy of 71%. It was possible to differentiate between both ADhall+ and ADhall- patients and DLBhall+ patients with an accuracy of 100%. When studying variable importance, the highest alpha1 power was found to be most discriminating between ADhall+ and ADhall-, functional connectivity in the beta band between DLBhall+ and ADhall+, and the severity of EEG abnormalities
4.2 The role of neurotransmitters in hallucinations
In AD, EEG power is increased in lower frequency bands (i.e. delta and theta band) and decrease in higher frequency bands (i.e. alpha and beta band), with late stages of the disease being marked by further increase in delta activity and decrease in alpha DFWLYLW\ -HRQJ 1LPPULFKHWDO 7KHVHÀQGLQJVKDYHDOVREHHQFRQÀUPHG by magnetoencephalography (MEG), which has higher spatial resolution than EEG, measuring neuronal activity with even greater detail (Engels et al., 2017). Slowing of EEG activity has been associated with loss of central cholinergic function (Riekkinen et al., 1991), which presumably emerges due to loss of basal forebrain neurons in the nucleus basalis of Meynert (NBM) (D.H. Hepp et al., 2017; Nimmrich et al., 2015). As ERWKQHXURLPDJLQJDQGQHXURSDWKRORJLFDOVWXGLHVKDYHVKRZQFKROLQHUJLFGHÀFLWVDUH more severe in DLB and already present in the earliest stages of the disease, while in $'WKHVHGHÀFLWVDUHOHVVVHYHUHDQGRFFXUODWHULQWKHFOLQLFDOFRXUVH /LXHWDO Shimada et al., 2015; Tiraboschi et al., 2002).
The forebrain cholinergic system is involved in arousal and attentional processes (Pepeu et al., 2013), but may also play an important role in the development of neuropsychiatric symptoms (van Dalen et al., 2016). As Lemstra et al., hypothesized, DWWHQWLRQDOGHÀFLWVDVDUHVXOWRIFHQWUDOFKROLQHUJLFGHÀFLHQF\DUHPRUHVSHFLÀFWR other neurodegenerative disorders associated with neuropsychiatric symptoms such as DLB and Parkinson’s disease (PD) than to AD (Lemstra et al., 2003). This cholinergic GHFD\ LV SURSRVHG WR XQGHUOLH D VSHFLÀF FKROLQHUJLF GHÀFLHQF\ V\QGURPH &'6 characterized by hallucinations, delusions, impairment of concentration, restlessness, agitation, anxiety, and slowing of EEG activity (Lemstra et al., 2003).
In our study, the presence of CDS in the ADhall+ group is supported by more severe scores in ADhall+ patients at the NPI sub items delusions, agitation, irritability and QLJKWWLPHEHKDYLRUVLPSO\LQJPRUHVHYHUHFKROLQHUJLFGHÀFLHQF\)XUWKHULQYLHZRI FKROLQHUJLFGHÀFLHQF\VLPLODULWLHVLQ((*VORZLQJLQ$'KDOODQG'/%KDOOEXWQRW between ADhall- and DLBhall+ patients suggest that AD patients who experience hallucinations and show severe EEG slowing form an intermediate group between AD and DLB. A similar intermediate role of the ADhall+ group was observed in alpha1 and
EHWDEDQGFRQQHFWLYLW\)XQFWLRQDOFRQQHFWLYLW\DQDO\VLVVKRZHGVLJQLÀFDQWGLIIHUHQFH in alpha1 PLI between the three groups with DLBhall+ showing lowest and ADhall- showing highest alpha1 PLI value (Table 3). As above mentioned, alpha1 frequency band has been associated with attentional processes, whereas PLI in alpha2 band, involved in cortical memory processes (Klimesch, 2012; Olejarczyk et al., 2017), was RQO\VLJQLÀFDQWO\GLIIHUHQWEHWZHHQ$'KDOODQG$'KDOO/LNHZLVHLQWKHUDQGRP forest analysis alpha1 relative power (higher in ADhall-) was the most differentiating feature between ADhall+ and ADhall- group, supporting the presence of more severe DWWHQWLRQDOGHÀFLWVDQGWKXVFKROLQHUJLFGHÀFLHQF\LQWKH$'KDOOJURXS)XUWKHUPRUH PLI values at channel T4 and A2 in the beta band were lowest in DLBhall+ and highest in $'KDOOSDWLHQWV%HWDEDQGDFWLYLW\FDQEHLQÁXHQFHGE\WKHFKROLQHUJLFV\VWHP 6ORDQ et al., 1992). Beta band rhythm has also been associated with long-range functional connectivity between brain regions (N Kopell et al., 2000), and increased activity during DWWHQWLRQDQGSHUFHSWLRQWDVNV *URVVHWDO .DPLĎVNLHWDO 7KHVHÀQGLQJV VXJJHVWWKDWEHWDEDQGDFWLYLW\SUREDEO\LQÁXHQFHGE\WKHFKROLQHUJLFV\VWHPPLJKW EHUHODWHGWRWKHSDWKRORJ\RIDWWHQWLRQDOGHÀFLWVZKLFKDUHPRUHVHYHUHLQ'/% Considering the consistency of differences in beta band activity and its differentiating ability between the groups, one might argue that PLI differences in the beta band might EHGULYHQE\GLIIHUHQFHVLQUHODWLYHEHWDSRZHU$OWKRXJKZHFRXOGQRWÀQGDSRVVLEOH correlation between PLI in beta band and relative beta power, we cannot fully rule out that relative beta power differences might be the driver of the observed PLI differences. Besides the cholinergic system, alterations in serotonin, dopamine, and norepinephrine neurotransmitter systems have been suggested to play a role in hallucinations in dementia (Ballard et al., 2013). Disruption in serotonergic neurotransmission has been associated with psychotic symptoms in DLB and PD dementia (PDD) (Ballard et al., 2013; Factor et al., 2017), whereas in AD, the role of the serotonin system on
have been associated with serotonergic neurotransmission, auditory hallucinations are much more rare (Factor et al., 2017; Lanctôt et al., 2017; Rolland et al., 2014; Tang et al., 2017). In addition, both decreased and elevated CSF-levels of dopamine and norepinephrine have been reported in AD, DLB and PDD with inconsistency in association with psychotic symptoms (Ballard et al., 2013; Strac et al., 2015; Vermeiren et al., 2014, 2013). Interestingly, Khundakar et al., found that changes in GABAergic neurotransmission correlated with visual hallucinations in DLB, while this was not the FDVHLQ$' .KXQGDNDUHWDO &RPSDUHGWRWKHFRQVLVWHQWÀQGLQJVZLWKUHJDUG to cholinergic alterations in hallucinations in AD, DLB and PD, the overall number of studies examining the associations between other monoaminergic and GABAergic V\VWHPDQGKDOOXFLQDWLRQVLQGHPHQWLDLVVPDOOZLWKHTXLYRFDOÀQGLQJVPDNLQJLWGLIÀFXOW to draw reliable conclusions.
Based on the above-mentioned hypothesis, ADhall+ patients might be cases of early DLB with concomitant AD pathology, who present with only one core feature, and therefore might be misdiagnosed. This hypothesis is in line with the premise of McKeith et al., that early/prodromal DLB may present as three subtypes: a mild cognitive impairment variant (DLB-MCI), a delirium onset (DLB-Del), and a psychiatric disorder onset (DLB-Psych) (McKeith et al., 2016). Post-mortem examination of the brains of $'KDOOSDWLHQWVZRXOGGHÀQLWHO\FRQÀUPRUUHMHFWWKHK\SRWKHVLVRIDQDW\SLFDOIRUP of DLB underlying hallucinations in early stage AD.
4.3 Strengths and limitations
A strength of this study is that it examines hallucinations in AD in isolation, and does not consider neuropsychiatric symptoms as a whole. Moreover, we included a DLBhall+ group to evaluate neurophysiological similarities and differences between ADhall+ and DLBhall+ patients. This has led to a new insight that AD patients with hallucinations might be cases of atypical presentation of DLB (McKeith et al., 2016). Further, we investigated both AD and DLB patients in the early stage of their illness (median 3.0 years), ruling out differences in disease stage as a major confounder between the groups. Finally, we applied a random forest algorithm, which is a highly
DFFXUDWHFODVVLÀHUZLWKDEXLOGLQFURVVYDOLGDWLRQDQGDQLQWHUQDOIHDWXUHVHOHFWLRQWKDW KHOSVWRUHGXFHPRGHOYDULDQFHDQGDYRLGVRYHUÀWWLQJ 0HHQDNVKL'DXZDQHWDO 7RDGGUHVVOLPLWDWLRQVLQWKLVVWXG\ÀUVWO\VLQFHWKHJURXSVZHUHLPEDODQFHGLQVDPSOH VL]HZKLFKPLJKWKDYHLQÁXHQFHGWKHUHVXOWVZHFRPSXWHGDZHLJKWHGDFFXUDF\WR account for the imbalanced data. Second, ADhall+ patients experienced hallucinations in more than one modality with auditory hallucinations being the second most common form of hallucinations. A possible explanation for the occurrence of auditory hallucinations in this group could also be sought in the cholinergic dysfunction. The 1%0FDQEHVXEGLYLGHGLQWRWKUHHGLYLVLRQVZLWKHDFKGLYLVLRQLQQHUYDWLQJVSHFLÀF cortical regions, i.e. the anterior division innervates the frontal and cingulate cortex, the intermediate division innervates the parietal and occipital cortex, and the posterior division innervates the temporal pole and the superior temporal gyrus (Liu et al., 2015). Involvement of the superior temporal gyrus has consistently been reported in DXGLWRU\KDOOXFLQDWLRQV ýXUĀLþ%ODNHHWDO VXJJHVWLQJWKDWPRUHVHYHUHQHXURQDO loss in the NBM-posterior division might (partly) underlie auditory hallucinations in the ADhall+ group. However, since a few patients also experienced tactile and/ or olfactory hallucinations this might not be the only explanation. Future studies investigating different modalities of hallucinations within one disorder are needed to gain insight in other potential underlying mechanisms. Third, following our hypothesis, we would expect the ADhall+ patients to form an intermediate group between ADhall- DQG'/%KDOOLQWHUPVRIFRJQLWLYHÁXFWXDWLRQVRQHRIWKHFRUHIHDWXUHVRI'/% +RZHYHUFRJQLWLYHÁXFWXDWLRQVZHUHQRWDVVHVVHGTXDQWLWDWLYHO\RUTXDOLWDWLYHO\WREH HYDOXDWHGEHWZHHQWKHJURXSV)XUWKHUWKHUHZDVQRDXWRSV\FRQÀUPHGGLDJQRVLVDVD gold standard to test our hypothesis. Finally, after diagnosis, since most subjects went back to their clinics of referral, no follow-up information was available on conversion of diagnosis during the course of the disease, or on development of hallucinations in
5. CONCLUSION
EEG shows distinct neurophysiological differences between AD patients with and without hallucinations, but at the same time, similarities between AD patients with hallucinations and DLB. EEG slowing and decrease in alpha1 and beta band activity, VXJJHVWLYHRIWKHSUHVHQFHRIFKROLQHUJLFGHÀFLHQF\IRUPSRWHQWLDOQHXURSK\VLRORJLFDO indicators of hallucinations in AD and DLB. Similarities between AD patients with hallucinations and DLB patients with hallucinations indicate that AD patients who experience hallucinations form an intermediate group between AD without hallucinations and DLB.
Supplemental Material
ResultsBased on 30 iterations in balanced ADhall+ and ADhall- group Balanced groups (36 vs. 36)
Group and feature selection Accuracy range
Quantitative and visual EEG 42-63
Only quantitative EEG 43-64
Conventional statistical analyses showed that for iterations in which the groups were GLVFULPLQDWHGZLWKKLJKDFFXUDF\JURXSVGLIIHUHGVLJQLÀFDQWO\LQT((*PHDVXUHV S as shown in Table 2 (see main text). For iterations where the groups could not be GLVFULPLQDWHGDFFXUDWHO\ LHORZDFFXUDF\YDOXH QRVLJQLÀFDQWGLIIHUHQFHZDVIRXQG in qEEG measures.
,QWHUHVWLQJO\WKHVHÀQGLQJVLQGLFDWHWKDWIRULWHUDWLRQVZKHUHWKHJURXSVVKRZHGORZ DFFXUDFLHVDQGQRVLJQLÀFDQWGLIIHUHQFHEHWZHHQPHDVXUHVDWEDVHOLQHWKHLWHUDWLRQ might include ADhall- patients with severe EEG slowing similar to the ADhall+ group. These ADhall- patients might be vulnerable to develop hallucinations and convert to DLB in the future. However, no follow-up information was available for both groups to evaluate this hypothesis.
Figure S1:&6)ELRPDUNHUSURÀOHRI$'KDOO$'KDOODQG'/%JURXS
Aȕ42: amyloid-ȕ 1-42; ADhall+: Alzheimer’s disease with hallucinations; ADhall-: Alzheimer’s disease without hallucinations; CSF: Cerebrospinal Fluid; DLBhall+: dementia with Lewy bodies with hallucinations; p-Tau: tau phosphorylated at threonine 181; Tau: total Tau
