International Journal of Psychophysiology 92 (2014) 1–7

Contents lists available at ScienceDirect

International Journal of Psychophysiology journal homepage: www.elsevier.com/locate/ijpsycho

EEG network connectivity changes in mild cognitive impairment — Preliminary results Brigitta Tóth a,c,⁎, Bálint File f, Roland Boha a, Zsófia Kardos a,c, Zoltán Hidasi d, Zsófia Anna Gaál a, Éva Csibri d, Pál Salacz d, Cornelis Jan Stam b, Márk Molnár a,e a

Institute of Cognitive Neuroscience and Psychology, RCNS, HAS, Hungary Department of Clinical Neurophysiology, VU University Medical Centre, Amsterdam, Netherlands Department of Cognitive Science, Institute of Psychology, Eötvös Loránd University, Budapest, Hungary d Department of Psychiatry and Psychotherapy, Faculty of Medicine, Semmelweis University, Budapest, Hungary e Psychophysiology Unit, Institute of Psychology, Eötvös Loránd University, Budapest, Hungary f Faculty of Information Technology, Pázmány Péter Catholic University, Budapest, Hungary b c

a r t i c l e

i n f o

Article history: Received 6 August 2013 Received in revised form 31 January 2014 Accepted 1 February 2014 Available online 6 February 2014 Keywords: Mild cognitive impairment Functional connectivity Phase synchronization EEG Follow-up study

a b s t r a c t Resting state EEGs were compared between patients with amnestic subtype of mild cognitive impairment (aMCI) and matched elderly controls at two times over a one year period. The study aimed at investigating the role of functional connectivity between and within different brain regions in relation to the progression of cognitive deficit in MCI. The EEG was recorded in two sessions during eyes closed and eyes open resting conditions. Functional brain connectivity was investigated based on the measurement of phase synchronization in different frequency bands. Delta and theta synchronization characteristics indicated decreased level of local and large-scale connectivity in the patients within the frontal, between the frontal and temporal, and frontal and parietal brain areas which was more pronounced 1 year later. As a consequence of opening the eyes connectivity in the alpha1 band within the parietal lobe decreased compared to the eyes closed condition but only in the control group. The lack of alpha1 band reactivity following eye opening could reliably differentiate patients from controls. Our preliminary results support the notion that the functional disconnection between distant brain areas is a characteristic feature of MCI, and may prove to be predictive in terms of the progression of this condition. © 2014 Elsevier B.V. All rights reserved.

1. Introduction The term “mild cognitive impairment” (MCI) conventionally applies to a condition in which the decline of cognitive abilities is more apparent than that seen in normal aging, but it still does not satisfy the criteria of dementia. Individuals with predominating memory problems are referred to as amnestic MCI (aMCI) (Petersen et al., 2001). While MCI may be regarded as a transitional state from which Alzheimer disease (AD) may develop (“MCI-converters”), this conversion does not occur in all MCI patients (“MCI-nonconverters”, or “stable MCI” patients) (Decarli, 2003). Eckerström et al. (2008) suggested that the left hippocampal volume loss was predictive for the possible development of AD and non-AD dementia in individuals with MCI. The practical importance of a predictive biomarker by which conversion from aMCI to AD is emphasized by the fact that the rate of conversion to AD is much higher in MCI than in normal aging individuals (Petersen et al., 2001; Petersen, 2004). Therefore, the investigation of a neurophysiological marker which is sensitive to the progression of MCI or conversion into AD is of crucial importance. Despite the obvious importance of the functional disconnection of brain ⁎ Corresponding author at: H-1117 Budapest, Magyar Tudósok körútja 2., P.O.B. 286 H1519, Budapest, Hungary. Tel.: +36 1 382 6812. E-mail address: [email protected] (B. Tóth). 0167-8760/$ – see front matter © 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.ijpsycho.2014.02.001

regions in MCI and AD, there are only few longitudinal studies in which the relationship between the progression of MCI and changes of functional connectivity characteristics was considered from this perspective (Giannakopoulos et al., 2009; Fernández et al., 2012). The present study aimed for the first time at investigating the role of functional connectivity of brain regions in relation to the progression of cognitive deficit in MCI. Since neurodegenerative diseases such as AD or aMCI are considered to be disconnection syndromes, functional connectivity analysis would seem to be an optimal approach for the purpose of their investigation (Missonnier et al., 2007; Buscema et al., 2007). Functional connectivity is defined as the temporal interdependence of neuronal activity of anatomically separated brain regions (reflected by hemodynamic and/or electrophysiological responses), and its analysis enables the quantification of the interaction between and within different neural systems (Rodriguez et al., 1999). Recent findings indicate disturbed network organization in aMCI and AD: reduced level of functional communication between distant brain regions and altered patterns of functional brain organization were observed, already apparent during resting state, and not just under high cognitive load (Bokde et al., 2009). Resting state (Raichle et al., 2001) denotes a state in which an individual is awake and alert, but is not actively involved in an attention demanding or goal directed task (“psychological baseline”). Neuroimaging studies

2

B. Tóth et al. / International Journal of Psychophysiology 92 (2014) 1–7

up till now described around eight networks of anatomically distinct brain regions that show a high level of functional connectivity during rest, including the so-called default mode network (DMN), consisting of the precuneus, the medial frontal, inferior parietal and temporal regions. The DMN is assumed to be exclusively crucial for the maintenance of cognitive functioning and is presumably altered in mental disorders (Broyd et al., 2009). In functional neuroimaging (fMRI — functional magnetic resonance imaging) studies peculiar features of the DMN were observed in MCI subjects distinguishing this condition from healthy aging (Qi et al., 2010; Liu et al., 2012). A consistent pattern of deactivation was observed in DMN during cognitive processes (Raichle et al., 2001). Since the EEG signal represents directly the ongoing neural activity from which both global and local properties characterizing physiological and/or pathological function can be extracted this method is appropriate for the investigation of resting state networks related to various oscillatory frequency bands. EEG oscillations are generated locally in different brain regions and mediate coordinated interactions within and between different neuronal systems (Rodriguez et al., 1999). Functional connections are reflected by the temporal correlation – synchrony – between the oscillatory firing patterns of neuronal assemblies. Numerous EEG studies reported lower EEG synchrony (indicating decreased functional connectivity) in MCI and AD patients in resting condition compared to age matched control subjects (Koenig et al., 2005; Stam, 2010; Park et al., 2008). Babiloni et al. (2010) found characteristic changes in MCI and AD patients with respect to sources of the delta, theta and alpha1 and alpha2 bands when compared to healthy elderly. Delta oscillation (related to large-scale cortical integration with homeostatic processes and also to attentional processes) sensitively reflects brain structural damage (lesions) and a wide range of neurodegenerative disorders (Parkinson's disease, AD, and schizophrenia depression (for review see Knyazev, 2012)). Pathological changes of theta oscillation are mainly reported in association with memory deficits (important for a variety of cognitive functions such as declarative memory and attentional control processes). Low alpha rhythm as a characteristic oscillation of resting state (the “idling rhythm”, but which may also have a role in inhibiting neural task irrelevant regions) is shown to be abnormal in dementia and AD and MCI (Scheeringa et al., 2012). The observed loss of synchrony was interpreted as functional disconnection between different cortical regions which cannot simply be due to the loss of cortical neurons (Jeong et al., 2001; Schliebs and Arendt, 2011). Interestingly, in the few studies when the EEG of MCI and AD patients was analyzed in conditions with cognitive load such as a working memory task a frequency band dependent increase of EEG synchrony was found in the patient groups (Jiang and Zheng, 2006; Pijnenburg et al., 2004). The results obtained by the computational neural mass model of de Haan et al. (2012) used for the investigation of the relation between the level of neural activity and hub vulnerability in Alzheimer's disease supported these latter findings. The model predicted a range of AD hallmarks (loss of spectral power and long-range synchronization, hub vulnerability, disrupted functional network topology) and reproduced the transient increase of functional connectivity in preclinical AD patients followed by subsequent breakdown of functional connections in definite AD. In healthy adults, corroborating the above mentioned neuroimaging findings on DMN sensory activation, attentional focusing was found to be associated with decreases in alpha power in the corresponding sensory area (Niedermeyer, 1999). In the present preliminary study functional connectivity was investigated with respect to the progression of aMCI status as an attempt to identify reliable electrophysiological markers that are able to capture the decline of MCI patients. As a putative electrophysiological marker the spatial distribution of EEG phase synchronization (phase lag index, a method that eliminates the distorting bias of volume conduction), (Stam et al., 2007) was analyzed to characterize the longitudinal pathophysiological changes in aMCI. EEG data of aMCI patients and age matched control subjects were recorded in two sessions (one year in

between) in eyes closed and eyes open resting conditions. First, aMCI patients and elderly controls were compared to assess the differences of functional connectivity between and within different brain regions. The difference of these connectivity characteristics between the two recording sessions was used to identify pathophysiological changes of aMCI over time. Discrimination analysis was applied in order to determine which of the electrophysiological connectivity characteristics the best predictors of aMCI status are. Furthermore, the effect of sensory stimulation (due to opening the eyes) on resting state functional connectivity was investigated by the comparison of eyes closed and eyes open resting conditions. It was hypothesized that connectivity measures will be sensitive indices of 1) deterioration of connectivity and 2) decline of reactivity in aMCI which, compared to healthy controls, was supposed to increase with elapsing time. 2. Methods 2.1. Participants Elderly adults (N = 14; women 8; age: 64.8, ± 2.5) and patients (N = 9; women 6; age: 67.5, ±3.2) with the diagnosis of amnestic subtype of MCI (aMCI) took part in the study. The participants signed a written informed consent form and received financial compensation for taking part in the study that was approved by the relevant institutional ethical committee. None of the control subjects had any neurological or mental disorders. Dementia, sedative medication and antipsychotic-based medical treatment were exclusion criteria in both groups. The aMCI patients were recruited from the Department of Psychiatry and Psychotherapy in Budapest. The diagnosis of aMCI was based on neurological and psychiatric examination and neuroimaging scans, including subjectively reported and neuropsychologically assessed memory impairment (Petersen's criteria standard clinical protocol, Petersen, 2004). The test results in the patients were the following: Mini Mental State Examination (MMSE) mean 1 session: 27.4; SD: ± 1.8, 2. session: 26.9; SD: ± 2.2; Addebrooke's Cognitive Test mean: 83.0; SD: ± 8.6; Global Deterioration Scale mean: 3.0; SD: ± 0. The MCI diagnosis of the patients was confirmed one year later by the same clinical department. Follow-up electrophysiological and behavioral data collection was performed in the patients and matched (age, sex) elderly controls in the Institute of Cognitive Neuroscience and Psychology. Prior to the EEG recordings in all participants the IQ (Wechsler Adult Intelligence Scale [WAIS]) was tested and the Mini Mental State Examination was performed in the aMCI patients. The IQ and MMSE results were used to assess possible cognitive decline over time in the aMCI patients by taking it two times over a one-year period. The participants were seated in an acoustically attenuated and electrically shielded room. The EEG was recorded with 33 Ag/AgCl electrodes (positioned according to the international 10–20 system) using Neuroscan software and amplifiers (Scan 4.3., Nuamps, bandpass: DC70 Hz, FIR, sampling rate: 1000 Hz). Vertical and horizontal eye movements were recorded. The tip of the nose was used as reference and an electrode placed between Cz and Fz as ground. The EEG data of elderly adults and patients with aMCI were recorded two times over a oneyear period. The time between the two recording sessions in the elderly group was 13.2 (SD: 2.1) months, and in the case of the aMCI group 12.6 (SD: 4.1) months. 4 min of spontaneous EEG was recorded both in eyes closed (EC) and also in eyes open (EO) conditions. 2.2. Data analysis The EEG epochs recorded at the two sessions (session 1, session 2) in EC and EO conditions were analyzed separately. A single epoch length was 2048 data points (2048 ms). The EEG epochs were filtered in six frequency bands (delta: 0.5–4 Hz, theta: 4–8 Hz, alpha1: 8–10 Hz, alpha2: 10–13 Hz, beta: 13–30 Hz, gamma: 30–45 Hz). Visual screening, and ICA (using EEGLab 10.2.5.8b ADJUST plugin) were used to exclude blinks

B. Tóth et al. / International Journal of Psychophysiology 92 (2014) 1–7

and other kind of artifacts. There was no significant difference between the average numbers of epochs in the different conditions categories. The average number of EEG epochs included in the final analysis in the aMCI group was 97.44 (SD: 29.9), and in the elderly control group was 89.77 (SD: 31.02). 2.3. Functional connectivity measures Characteristics of functional brain networks were analyzed by measuring the phase lag index (PLI) in five frequency bands calculated in all the EEG epochs. PLI is a measure of the asymmetry of the distribution of instantaneous phase differences between two signals quantifying the consistency of phase relation of the signals (Stam et al., 2007). PLI was calculated by using the Brainwave 0.9.58. software (http://home.kpn. nl/stam7883/brainwave.html). PLIs are expressed as values between 0 (symmetric phase distribution) and 1 (distribution with constant, non-zero phase difference indicating maximum strength of functional connectivity). The assessment of PLI between and within brain regions provides information about the large-scale and local connectivity strength. For the regional analysis of PLI the EEG channels were grouped into regions of interest (ROIs): frontal, temporal, parietal (right and left, respectively). The average PLIs were calculated for all channels within a region (local connectivity) or between two regions (long distance connectivity: left and right fronto-temporal, left and right fronto-parietal, left and right temporo-parietal). Electrodes making up the different ROIs were the following: right frontal (FP2, F8 and F4); left frontal (FP1, F7 and F3); right temporal (T8, FT10 and TP10); left temporal (T7, FT9 and TP9); right parietal (CP2, CP6, P4 and P8); left parietal (CP1, CP5, P7 and P3). In the present study the priority was for the selection of EEG channels into ROIs to measure changes of functional connectivity characteristics not only in each cerebral lobe (frontal, temporal, parietal, and occipital) but also to differentiate these characteristics between right and left hemispheres (for example left and right frontal lobes). Therefore, the EEG signals recorded by fronto-central electrodes were excluded from further analysis. 2.4. Statistical analysis The statistical analyses of the PLI data between 1) control and aMCI, 2) EO and EC conditions and 3) first and second recording sessions were performed by ANOVA for repeated measures by using the Statistica 9.1 software for each frequency band and ROIs separately where group was the between subject factor, session and conditions were the within subject factors. No condition (EC, EO) effects were found for the delta and theta bands, therefore, the statistical analyses of the PLI data between 1) control and aMCI, and 2) first and second recording sessions were performed separately for the EO and EC conditions. The Tukey's test was used for post hoc analyses. All test variables considered had approximately normal distribution, as verified by the Kolmogorov–Smirnov test. Discriminant function analysis (DA which has the same assumptions as linear regression analysis but is appropriate for categorical predictor variables) was applied to determine whether a set of functional connectivity parameter variables was effective in predicting category membership (aMCI and control). Discriminant function analysis (DA) was used to determine 1) the overall predictive power of functional connectivity parameters which discriminate between the aMCI and control groups and 2) which of these were the best predictors of aMCI status. 3. Results

difference was observed regarding the level of IQ and MMSE in the aMCI group between the two recording sessions (Table 1). 3.2. Electrophysiological data Results with regard to significant group differences of connectivity strength are presented. Consistent group differences in PLI were observed in the delta, theta, and alpha1 bands. Significant differences are shown in Fig. 1 (delta band), Fig. 2 (theta band) and Fig. 3 (alpha1 band). 3.2.1. Delta band 3.2.1.1. Control–aMCI group differences in PLI (main effects of groups) 3.2.1.1.1. Intra-regional connectivity. A significant decrease of PLI within regions was present in aMCI patients. Compared to healthy controls, aMCI patients showed significantly lower level of local connectivity within the left and right frontal regions (F1. 22 = 4.25, p = 0.052; F1. 22 = 6.97, p = 0.015, respectively) in the EO condition. 3.2.1.1.2. Inter-regional connectivity. A difference of PLI between regions was found between the control and aMCI group. Compared to healthy controls, aMCI patients showed significantly lower level of connectivity between the left frontal and temporal areas (F1. 22 = 5.22. p = 0.033), and also between frontal and parietal areas in both hemispheres (left F1. 22 = 8.16. p = 0.009; right F1. 22 = 7.25. p = 0.014) but only in the EO condition. 3.2.1.2. Group and recording session interaction effects 3.2.1.2.1. Intra-regional connectivity. A significant difference with respect to recording session was found only in aMCI. PLI was lower in the EC condition within the left and right frontal regions in the second session compared to the first (left: F1. 22 = 10.94, p = 0.003, post hoc p = 0.005; right: F1. 22 = 5.66, p = 0.027, post hoc p = 0.013). 3.2.1.2.2. Inter-regional connectivity. Fronto-temporal PLI tended to decrease as a function of time only in the aMCI group. This difference was more pronounced in the right hemisphere (left F1. 22 = 3.82, p = 0.064, post hoc p = 0.012; right F1. 22 = 5.65, p = 0.027, post hoc p = 0.046). Fronto-parietal PLI showed a similar decline as a function of time in the aMCI patients in the right (F1. 22 = 4.52. p = 0.046, post hoc p = 0.017) and left (F1. 22 = 4.38. p = 0.049, post hoc p = 0.039) hemisphere as well. 3.2.2. Theta band 3.2.2.1. Group and recording session interaction effects 3.2.2.1.1. Intra-regional connectivity. Interaction of recording session and group was found in the right parietal region (F1. 22 = 6.29, p = 0.022, post hoc p = 0.065) in the EO condition: PLI decreased as a function of time only in the aMCI group. 3.2.2.1.2. Inter-regional connectivity. Fronto-parietal PLI decreased as a function of time in the aMCI patients in the right (F1. 22 = 13.46, p = 0.002, post hoc p = 0.015) and left (F1. 22 = 14.12, p = 0.002, post hoc p b 0.001) hemispheres as well. Table 1 Results of the psychological tests (Wechsler Adult Intelligence Scale [WAIS], Mini Mental State Examination, [MMSE]). Patient

3.1. Behavioral assessment The patients showed decreased performance on the subtests of performance IQ such as Symbol-Coding and Picture Completion test (p = 0.044; p = 0.032 respectively) compared to controls. No significant

3

IQ Verbal IQ Performance IQ

Elderly

1. session

2. session

1. session

2. session

109.5 110.7 107.1

110.8 113.5 108.5

118.2 115.9 119.9

122.1 120.25 121.5

±9.8 ±9.1 ±12.5

±8.62 ±8.52 ±12.57

±9.2 ±8.6 ±9.1

±6.7 ±9.2 ±6.1

4

B. Tóth et al. / International Journal of Psychophysiology 92 (2014) 1–7

Fig. 1. Delta band functional network characteristics in aMCI. PLI measured within the frontal (a) and between the fronto-temporal (b) and fronto-parietal regions (c) in eyes closed (EC) and eyes open (EO) conditions (first and second recording sessions). Horizontal lines corresponding to the diagrams indicate significant differences.

3.2.3. Alpha1 band 3.2.3.1. Control–aMCI group differences in PLI (main effects of groups) 3.2.3.1.1. Intra-regional connectivity. A significant group difference was found in the right parietal region but only in the EO condition. Higher level of connectivity was observed in the aMCI group compared to the healthy elderly group (F1. 22 = 4.65, p = 0.042). 3.2.3.2. Difference between EC and EO conditions 3.2.3.2.1. Intra-regional connectivity. A trend of condition–group interaction was found in the apha1 band (F1. 22 = 3.737, p = 0.066). PLI decreased within the parietal regions in the EO condition compared to the EC condition (p = 0.001) but only in the control group. This difference was evident in both the left (p b 0.001) and the right parietal areas (p b 0.001). 3.2.4. Predictive validity of functional connectivity strength The variables (inter- and intra-regional connectivity strength) of the second sessions showing significant group differences were used as predictor variables in DA. The discriminant function revealed a significant association between groups and all predictors (Wilks' Lambda = 0.027, p = 0.002) accounting for 97.2% of between group

variability. The structure matrix revealed that the best predictors of aMCI status could be identified in the EO condition, namely frontal connectivity in the delta band (left: 0.442, right: 0.214), fronto-parietal connectivity in the theta band (left: 0.276, right: 0.341) and parietal connectivity in the alpha1 band (left: − 0.372, right: − 0.309). The cross validated classification showed that in 69.7% of the overall cases the participants were correctly identified as one that belonged to the aMCI or to the normal control group. 4. Discussion The present longitudinal preliminary study attempted to explore resting state EEG functional connectivity characteristics in aMCI as a function of time. Although the results must be taken as preliminary ones, it seems likely that the analysis of connectivity strength may provide new insights as to which brain regions and what EEG frequency bands are important to be considered from this perspective. The results suggest that cortical functional coupling mechanisms of delta, theta and alpha rhythms are impaired in aMCI patients. The observed changes in connectivity represent evidence of altered resting state neural networks changing as a function of time in these patients. In the aMCI patient group decreased level of intra and inter-regional connectivity was observed, as reflected by PLI changes in the delta and theta bands within the frontal, between the frontal and temporal, and between the frontal and parietal areas. Although the use of 2048 ms epochs was not optimal

B. Tóth et al. / International Journal of Psychophysiology 92 (2014) 1–7

5

Fig. 2. Theta band functional network characteristics in aMCI. PLI measured within the frontal (a) and between and fronto-parietal regions (b) in eyes open (EO) conditions (first and second recording sessions). Horizontal lines corresponding to the diagrams indicate significant differences.

for the study of the delta band (see later in Limitations) the results (although not particularly sensitive for the phase synchronization changes below 2 Hz) obtained by these means proved to be especially informative with respect to fronto-temporal connectivity changes. Our preliminary findings are consistent with the results of earlier EEG studies using coherence and synchronization likelihood measures in which a decreased level of synchronization was found between cortical regions in AD patients (Stam, 2010), particularly evident between distant areas (Babiloni et al., 2006; Stam and Reijneveld, 2007). Only a few studies compared phase synchronization patterns between MCI patients and controls, but no significant differences were found in one study (Pijnenburg et al., 2004) and contradictory results were reported in another (Koenig et al., 2005). It should be noted that the behavioral measurements (IQ) did not show change as a function of time, on the contrary a decline of

functional connectivity of patients was already apparent after one year. It is possible that the explicit behavioral decline was not observable because the dysfunction in the neural activity became overt on behavioral outcome of only after it reaches its critical quantitation. Functional connectivity characteristics could be seen as a highly sensitive tool for the investigation of neurodegenerative progressions in MCI. On the other hand, the lack of behavioral decline also could be attributed to compensatory mechanisms in the aMCI group, for example semantic knowledge due to the practice effects on the IQ tasks which were completed twice. 4.1. Fronto-temporal inter-regional connectivity characteristics in aMCI In AD and aMCI the medial temporal lobe (MTL) is usually involved earlier than other areas (Buscema et al., 2007) and therefore the

Fig. 3. Alpha1 band functional network characteristics in aMCI. PLI measured within the parietal region in eyes closed (EC) and eyes open (EO) conditions (first and second recording sessions). Horizontal lines corresponding to the diagrams indicate significant differences.

6

B. Tóth et al. / International Journal of Psychophysiology 92 (2014) 1–7

disconnection of the temporal lobe might be responsible for the reduced connectivity between the frontal and the temporal regions, appearing in the delta band. This finding is in line with results of previous reports for impaired resting state connectivity of temporal regions both in terms of electrophysiological and hemodynamic data (Bokde et al., 2009; Greicius, 2008). Our findings indicate progressively declining frontotemporal connectivity measured in delta band oscillatory activity, which supports the hypothesis that disconnection of the temporal lobe leads to an increased vulnerability of fronto-temporal connections that progress with time. Among several regions hypoactivity of the medial temporal lobe in resting state has been suggested in fMRI studies as a candidate marker that may aid to detect and to monitor the progression of AD (Broyd et al., 2009) which, according to the present results, is evident in early stages of aMCI as well. Consistent with the present results in the longitudinal study of Knyazev et al. (2012) a significant progressive reduction of synchronization was found in the left temporal cortex in AD patients. 4.2. Fronto-parietal inter-regional connectivity characteristics in aMCI The effect of sensory activation on connectivity measures may be more adequate to discriminate between the aMCI and control groups than resting state characteristics. Decreased connectivity between frontal and parietal regions as reflected by PLI in the delta and theta frequency bands in aMCI patients was evident mainly in the EO condition, i.e. during visual activation. Greater inter-individual variability of PLI can be seen in EC than in EO conditions which might be responsible for not observing the same differences in these two conditions. The decrease of fronto-parietal connectivity in the theta band, related to visual activation might be also associated with cognitive decline in MCI patients. In the present study, the decline of connectivity strength between frontal and parietal lobes was more pronounced 1 year later in the aMCI group, as shown by both delta and theta PLI results. The present findings are in line with previous reports which showed that the decrease of delta coupling was most apparent between the frontoparietal regions in both MCI and AD patients (Babiloni et al., 2006). In MCI and AD patients correlations were shown between the source power of the cortical delta rhythm and neuropsychological measures of immediate memory (Babiloni et al., 2007). According to the few studies addressing the same issues as the present one, the most affected frequencies in AD and aMCI during cognitive tasks were found to be the delta and theta bands, showing lower delta amplitude in the central areas and lower phase locking in the theta band in the frontal areas (Pijnenburg et al., 2004; Rossini et al., 2008; Babiloni et al., 2007). Fronto-parietal connectivity decline was assumed to reflect the presence of early deficits in attention-related neural circuits in patients with MCI (Babiloni et al., 2006). 4.3. Frontal intra-regional connectivity characteristics in aMCI The frontal midline theta has been associated with several cognitive mechanisms such as focused attention and effective stimulus processing. Decreased induced frontal theta activity was shown during a working memory task in progressive MCI group compared to stable MCI patients which was interpreted as an increasing deficit of attention function (Deiber et al., 2009). Cummins et al. (2008) also reported reduced theta power of fronto-central regions under memory load in aMCI patients compared to controls. Taken all together, our results with respect to the theta band may reflect some aspects of cognitive deficit in the aMCI patients. Decreased local connectivity was observed in the delta band in aMCI exclusively in the frontal regions, which became more pronounced over time. Taking into account the results of other studies it can be assumed that the frontal and central areas in AD and aMCI patients are those regions of the brain that show consistent evidence of disconnection associated with such neurodegenerative disorders (Yener et al., 2007).

Increased delta band functional connectivity in association with increased memory performance was found following successful drug treatment in patients with mild AD (Scheltens et al., 2012). This observation supports the present findings regarding the functional role of delta oscillations in cognitive functions.

4.4. The effect of sensory activation on alpha band functional connectivity of parietal lobe differentiates control subjects from aMCI patients As a consequence of opening the eyes connectivity in the alpha1 band within the parietal lobe decreased compared to the EC condition but only in the control group. Thus the lack of alpha1 band reactivity following eye opening differentiates the patient group from the controls. Previous studies have reported that cortico-cortical coupling in the lower alpha band was increased in AD and MCI patients compared to controls (Stam, 2010; Qi et al., 2010). It seems reasonable to interpret the lack of alpha1 band reactivity seen in MCI in that the brain fails to respond appropriately to unspecific visual stimulation. In an earlier study we observed decreased reactivity elicited by eye opening in the alpha1 band in healthy elderly subjects compared to young ones (Gaál et al., 2010), indicating that the flexibility of neural circuits involved in this process shows an age-dependent decline which may become even more conspicuous in MCI. Cortical sources of resting alpha rhythms were reported to correlate with neuropsychological measures of immediate memory in MCI and AD groups (Babiloni et al., 2007). The finding of failures in neural phase coding between thalamic and cortical alpha-EEG sources in MCI patients indicates that MCI related abnormalities in phase synchronization might extend to thalamo-cortical circuits (Cummins et al., 2008). Various findings indicate that the lower and upper alpha bands can be separated by their topographical distributions and task related changes (Klimesch, 1999). The lower alpha band does not show any task-, or stimulus-specificity, whereas the upper alpha band is related to semantic memory functions. In the present study, however, no task was given to the participants.

4.5. Limitations The present study should be considered a preliminary one because of the low number of participants. The length of the analyzed 2048 ms epochs presents problems of interpretation in case of the delta frequency band. Although the obtained results were informative with respect to connectivity changes in MCI it is likely that these were related to activities N2 Hz, and not below. 5. Conclusions In summary the results of the preliminary study suggest that in MCI the target of pathology is both local and large-scale neuronal connectivity which supports the hypothesis of a functional disconnection of resting state networks in this condition. The disconnection between frontal and distant brain areas is probably a characteristic feature of aMCI, and its measure may prove to be predictive in terms of the progression of this condition. The small sample size is certainly a limitation of the present investigation. Further follow-up studies with more patients are needed to be able to determine whether functional connectivity findings could be predictive in terms of conversion from aMCI to AD. As a future prospect the analysis of inter-individual variability of resting state functional connectivity characteristics may allow separation of various sub-types of MCI patients.

Disclosure statement The authors declare no actual or potential conflict of interest.

B. Tóth et al. / International Journal of Psychophysiology 92 (2014) 1–7

Acknowledgments This research was supported by the Hungarian Research Fund OTKA 77750 and 104332. The authors thank E. Várkonyi for the technical assistance. References Babiloni, C., Cassetta, E., Binetti, G., Tombini, M., Del Percio, C., Ferreri, F., Ferri, R., Frisoni, G., Lanuzza, B., Nobili, F., Parisi, L., Rodriguez, G., Frigerio, L., Gurzì, M., Prestia, A., Vernieri, F., Eusebi, F., Rossini, P.M., 2007. Resting EEG sources correlate with attentional span in mild cognitive impairment and Alzheimer's disease. Eur. J. Neurosci. 25, 3742–3757. Babiloni, C., Ferri, R., Binetti, G., Cassarino, A., Dal Forno, G., Ercolani, M., Ferreri, F., Frisoni, G.B., Lanuzza, B., Miniussi, C., Nobili, F., Rodriguez, G., Rundo, F., Stam, C.J., Musha, T., Vecchio, F., Rossini, P.M., 2006. Fronto-parietal coupling of brain rhythms in mild cognitive impairment: a multicentric EEG study. Brain Res. Bull. 69, 63–73. Babiloni, C., Visser, P.J., Frisoni, G., De Deyn, P.P., Bresciani, L., Jelic, V., Nagels, G., Rodriguez, G., Rossini, P.M., Vecchio, F., Colombo, D., Verhey, F., Wahlund, L.-O., Nobili, F., 2010. Cortical sources of resting EEG rhythms in mild cognitive impairment and subjective memory complaint. Neurobiol. Aging 31, 1787–1798. Bokde, A.L.W., Ewers, M., Hampel, H., 2009. Assessing neuronal networks: understanding Alzheimer's disease. Prog. Neurobiol. 89, 125–133. Broyd, S.J., Demanuele, C., Debener, S., Helps, S.K., James, C.J., Sonuga-Barke, E.J.S., 2009. Default-mode brain dysfunction in mental disorders: a systematic review. Neurosci. Biobehav. Rev. 33, 279–296. Buscema, M., Rossini, P., Babiloni, C., Grossi, E., 2007. The IFAST model, a novel parallel nonlinear EEG analysis technique, distinguishes mild cognitive impairment and Alzheimer's disease patients with high degree of accuracy. Artif. Intell. Med. 40, 127–141. Cummins, T.D.R., Broughton, M., Finnigan, S., 2008. Theta oscillations are affected by amnestic mild cognitive impairment and cognitive load. Int. J. Psychophysiol. 70, 75–81. Decarli, C., 2003. Reviews mild cognitive impairment: prevalence, prognosis, aetiology, and treatment. Lancet Neurol. 2, 15–21. Deiber, M.-P., Ibañez, V., Missonnier, P., Herrmann, F., Fazio-Costa, L., Gold, G., Giannakopoulos, P., 2009. Abnormal-induced theta activity supports early directedattention network deficits in progressive MCI. Neurobiol. Aging 30, 1444–1452. Eckerström, C., Olsson, E., Borga, M., Ekholm, S., Ribbelin, S., Rolstad, S., Starck, G., Edman, a, Wallin, a, Malmgren, H., 2008. Small baseline volume of left hippocampus is associated with subsequent conversion of MCI into dementia: the Göteborg MCI study. J. Neurol. Sci. 272, 48–59. Fernández, A., Gil Gregorio, P., Maestú, F., 2012. Spontaneous electroencephalographic and magneto-encephalographic activity as a marker of Alzheimer's disease and mild cognitive impairment. Rev. Esp. Geriatr. Gerontol. 47 (1), 27–32. Niedermeyer, E., 1999. Electroencephalography: Basic Principles, Clinical Applications, and Related Fields, 4th ed. Williams & Wilkins, Baltimore, MD. Gaál, Z.A., Boha, R., Stam, C.J., Molnár, M., 2010. Age-dependent features of EEG-reactivity —spectral, complexity, and network characteristics. Neurosci. Lett. 479, 79–84. Giannakopoulos, P., Missonnier, P., Kövari, E., Gold, G., Michon, A., 2009. Electrophysiological markers of rapid cognitive decline in mild cognitive impairment. Front. Neurol. Neurosci. 24, 39–46. Greicius, M., 2008. Resting-state functional connectivity in neuropsychiatric disorders. Curr. Opin. Neurol. 21, 424–430. Jeong, J., Gore, J.C., Peterson, B.S., 2001. Mutual information analysis of the EEG in patients with Alzheimer's disease. Clin. Neurophysiol. 112 (5), 827–835. Jiang, Z., Zheng, L., 2006. Inter- and intra-hemispheric EEG coherence in patients with mild cognitive impairment at rest and during working memory task. J. Zhejiang Univ. Sci. B 7, 357–364. Klimesch, W., 1999. EEG alpha and theta oscillations reflect cognitive and memory performance: a review and analysis. Brain research. Brain research reviews 29 (2–3), 169–195.

7

Knyazev, M.G., Carmeli, C., Khadivi, A., Ghika, J., Meuli, R., Frackowiak, R.S., 2012. Evolution of source EEG synchronization in early Alzheimer's disease. Neurobiol. Aging 34 (3), 694–705. Knyazev, G.G., 2012. EEG delta oscillations as a correlate of basic homeostatic and motivational processes. Neurosci. Biobehav. Rev. 36 (1), 677–695. Koenig, T., Prichep, L., Dierks, T., Hubl, D., Wahlund, L.O., John, E.R., Jelic, V., 2005. Decreased EEG synchronization in Alzheimer's disease and mild cognitive impairment. Neurobiol. Aging 26, 165–171. Liu, Z., Zhang, Y., Yan, H., Bai, L., Dai, R., Wei, W., Zhong, C., Xue, T., Wang, H., Feng, Y., You, Y., Zhang, X., Tian, J., 2012. Altered topological patterns of brain networks in mild cognitive impairment and Alzheimer's disease: a resting-state fMRI study. Psychiatry Res. 202, 118–125. Missonnier, P., Deiber, M.-P., Gold, G., Herrmann, F.R., Millet, P., Michon, a, Fazio-Costa, L., Ibañez, V., Giannakopoulos, P., 2007. Working memory load-related electroencephalographic parameters can differentiate progressive from stable mild cognitive impairment. Neuroscience 150, 346–356. Park, Y.-M., Che, H.-J., Im, C.-H., Jung, H.-T., Bae, S.-M., Lee, S.-H., 2008. Decreased EEG synchronization and its correlation with symptom severity in Alzheimer's disease. Neurosci. Res. 62, 112–117. Petersen, R.C., 2004. Mild cognitive impairment as a diagnostic entity. J. Intern. Med. 256, 183–194. Petersen, A., Kurz, R.C., D.R., et al., 2001. Current concepts in mild cognitive impairment. Arch. Neurol. 58, 1985–1992. Pijnenburg, Y.A.L., vd Made, Y., van Cappellen van Walsum, A.M., Knol, D.L., Scheltens, P., Stam, C.J., 2004. EEG synchronization likelihood in mild cognitive impairment and Alzheimer's disease during a working memory task. Clin. Neurophysiol. 115, 1332–1339. Qi, Z., Wu, X., Wang, Z., Zhang, N., Dong, H., Yao, L., Li, K., 2010. Impairment and compensation coexist in amnestic MCI default mode network. NeuroImage 50, 48–55. Raichle, M.E., MacLeod, A.M., Snyder, A.Z., Powers, W.J., Gusnard, D.A., Shulman, G.L., 2001. A default mode of brain function. Proc. Natl. Acad. Sci. U. S. A. 98, 676–682. Rodriguez, E., George, N., Lachaux, J.P., Martinerie, J., Renault, B., Varela, F.J., 1999. Perception' s shadow: long-distance synchronization of human brain activity. Nature 397 (6718), 430–433 (4). Rossini, P.M., Buscema, M., Capriotti, M., Grossi, E., Rodriguez, G., Del Percio, C., Babiloni, C., 2008. Is it possible to automatically distinguish resting EEG data of normal elderly vs. mild cognitive impairment subjects with high degree of accuracy? Clin. Neurophysiol. 119, 1534–1545. Scheltens, P., Twisk, J.W.R., Blesa, R., Scarpini, E., von Arnim, C.A.F., Bongers, A., Harrison, J., Swinkels, S.H.N., Stam, C.J., de Waal, H., Wurtman, R.J., Wieggers, R.L., Vellas, B., Kamphuis, P.J.G.H., 2012. Efficacy of Souvenaid in mild Alzheimer's disease: results from a randomized, controlled trial. J. Alzheimers Dis. 31, 225–236. Scheeringa, R., Petersson, K.M., Kleinschmidt, A., Jensen, O., Bastiaansen, M.C.M., 2012. EEG α power modulation of fMRI resting-state connectivity. Brain Connectivity 2 (5), 254-. Schliebs, R., Arendt, T., 2011. The cholinergic system in aging and neuronal degeneration. Behav. Brain Res. 221, 555–563. Stam, C.J., 2010. Use of magnetoencephalography (MEG) to study functional brain networks in neurodegenerative disorders. J. Neurol. Sci. 289, 128–134. Stam, C.J., Nolte, G., Daffertshofer, A., 2007. Phase lag index: assessment of functional connectivity from multichannel EEG and MEG with diminished bias from common sources. Hum. Brain Mapp. 28, 1178–1193. Stam, C.J., Reijneveld, J.C., 2007. Graph theoretical analysis of complex networks in the brain. Nonlinear Biomed. Phys. 1, 3. Yener, G.G., Güntekin, B., Oniz, a, Başar, E., 2007. Increased frontal phase-locking of eventrelated theta oscillations in Alzheimer patients treated with cholinesterase inhibitors. Int. J. Psychophysiol. 64, 46–52. de Haan, W., van der Flier, W.M., Koene, T., Smits, L.L., Scheltens, P., Stam, C.J., 2012. Disrupted modular brain dynamics reflect cognitive dysfunction in Alzheimer's disease. NeuroImage 59, 3085–3093.