Parkinson Disease 1 ERP sources in middle cingulate and precuneus differentiate Parkinson ’ s patients from 2 healthy controls and lingual gyri sources reflect human recombinant EPO effects in a 3 Flanker task 4

17 We used EEG source analysis to identify which cortical areas were involved in the suppression 18 of competing responses on a flanker task and compare the potential efficacy of recombinant19 human erythropoietin (rHuEPO) in the performance of Parkinson’ s Disease patients. 20 The samples were 18 medicated PD patients (9 with rHuEPO and 9 without rHuEPO) and 9 21 age and education-matched healthy controls (HCs) who completed the flanker task with 22 simultaneous EEG recordings. N1, N2 and P3 event-related potential (ERP) components were 23 identified and a low-resolution tomography (LORETA) inverse solution was employed to 24 localize the neural generators. 25 Reaction times and errors were increased for the incongruent flankers for PD patients compared 26 to controls. EEG source analysis identified an effect of rHuEPO on the lingual gyri for the early 27 N1component. N2-related sources in middle cingulate and precuneus were associated with the 28 inhibition of automatic responses evoked by incongruent stimuli differentiating PD and HCs. 29 From our results the rHuEPO seems to mediate an effect on N1 sources in lingual gyri but not 30 on behavioural performance. N2-related sources in middle cingulate and precuneus 31 differentiated PD and HCs. 32


Introduction
The basal ganglia structures particularly the striatum and the subthalamic nucleus are part of 34 the fronto-striatal-subthalamic-pallidal network considered to mediate habitual/automatic and 35 goal-directed inhibition as well as habitual and goal-directed action [1,2]. Thus, these circuits 36 are hypothesized to coordinate the selection and suppression of competing responses. 37 Parkinson's disease (PD), the prototypical basal ganglia disorder, is associated with deficits in 38 inhibitory control on a number of experimental tasks such as the stop signal [3], go no-go 39 reaction times [4], the Stroop and the Hayling Sentence Completion task [3], and the Simon 40 task [5,6]. PD patients also have difficulty in suppressing interference arising from the 41 automatic activation of prepotent responses evoked by incongruent flankers in the Ericksen's 42 Flanker Task [7]. Relative to controls, PD patients show increased reaction times (RTs) and 43 errors on incongruent trials compared to congruent trials (eg. [8,9,10,11]). 44 In PD there is an ongoing search for neuroprotective agents which may slow down the 45 progression of the illness and improve cognitive deficits [12]. The recombinant-human 46 erythropoietin (rHuEPO) is studied with great interest due to its neuroprotector properties in 47 neurologic diseases [13]. The anti-apoptotic, anti-inflammatory and cytoprotective effects of 48 EPO in parkinsonism animal models have been described elsewhere [14,15]. The aim of our 49 study is to use a flanker task to identify if rHuEPO produces beneficial effects on performance 50 of PD patients and locate the neural generators involved in the selection and suppression of 51 competing responses in comparison with healthy controls (HCs). 52 were measured to the nearest milliseconds and errors were recorded. 86

Materials and Methods
The physical characteristics of the stimuli were black letters on a white frame with an h = 1.5 87 cms and L= 7 cms, under 6 0 a visual angle. The distance of the participant to the computer 88 monitor was 60 cms. Each stimulus was presented at the center of the screen and kept for 190 89 msec., followed by an interstimulus interval of 1735 msec. A training block of 40 stimuli was 90 designed to ensure task instructions were understood. 91 92 EEG: 93 The Electroencephalogram (EEG) was continuously recorded at a sampling rate of 512 Hz 94 from 64 electrodes located at standard positions of the International 10/20 System using a Brain 95 Vision system (Brain Products https://www.brainproducts.com/products_by_apps.php?aid=5 96 [20]. The electro-oculogram (EOG, horizontal and vertical) was recorded from electrodes 97 placed 1 cm to the left and right of the external canthi, and from an electrode beneath the right 98 eye. The ears were used as on-line reference and the front as earth. 99 Data were filtered using 1-30 Hz and a notch filter to eliminate the 60Hz powerline artefact. 100 All data were referenced using an average reference to all the channels. The baseline was 101 corrected between -400 to -200 msec. Epochs with electric activity exceeding baseline activity 102 by 100 µV were considered as artefacts and were automatically rejected from further 103 processing (15% of epochs related to hits and 11% of the epochs related to errors In order to localize the generators of the ERP components, a lead field was constructed for each 115 participant to calculate the inverse solution at the three selected latencies using LORETA (Low 116 Resolution Tomography) (http://www.uzh.ch/keyinst/loreta) [22]. The significant specific 117 source effects in each latency were independently confirmed by means of permutation methods 118 [23]. The tomographic inverse solution was plotted using an average brain, (volume constraints) 119 with the coordinates of the AAL (Automated Anatomical Labelling of Activations) 116 120 structures atlas of the Montreal Neurological Institute (MNI) [24]. 121 Statistical analysis. The General Linear Model and a priori contrasts were used for statistical 122 analysis, with Group (PD with rHuEPO vs PD without rHuEPO) as the between group factor 123 and the experimental condition (incongruent versus congruent) as the within-subject repeated 124 measures factor. The three windows for analysis were: 100-180, 180-300 and 300-450 msec. 125 This was also applied for the neural sources using voxel-based analysis for the individual 126 source matrices. For the second objective, we analyzed the difference between the two groups 127 of PD vs Healthy controls in the same way. The resulting F statistic was corrected twice. First, 128 using Bonferroni corrected according to the total number of points in the analysis window (700 129 milliseconds) and divided by α=0.05. The second correction was using FDR (false positives 130 (FDR: false discovered rate) for a q-value=0.01, that is, controlling a 1% of the expected value 131 [25]. Analysis was completed with STATISTICA 7.0 and the software (NEEST) from 132 Neuronic http://www.neuronicsa.com/ 133

Behaviour: 135
All groups, PD with or without rHuEPO and HCs had longer RTs and made more errors on the 136 incongruent than the congruent trials (Table 1 and 2). However, when comparing PD patients 137 with or without rHuEPO, there were no significant differences in performance between the task 138 conditions (Table 1). 139

143
When comparing all PD patients and HCs (

Source analysis of the ERP differences between PD patients with and without rHuEPO: 163
The comparison between patients showed that PD group with rHuEPO showed a larger N1

179
Discussion 180 Behavioural results 181 PD patients showed significantly increased reaction times and a higher number of errors to 182 the incongruent stimuli during the performance of the flanker task in comparison to age and 183 education matched HCs. These higher error rates in PD than controls are consistent with the 184 proposal that the basal ganglia together with the anterior cingulate [26] participate in the 185 monitoring of incongruence and error monitoring [27,28] which may be impaired in PD due 186 to the dopamine deficiency (for a recent revision how the progressive dopamine deficiency 187 reduces striatal cholinergic interneuron activity see [29]). But we did not find the expected 188 beneficial effect of rHuEPO on the behavioural performance (RT and accuracy) in PD 189 patients who received the drug in comparison with the others. Nonetheless, the differences 190 between groups of patients were found in the electrophysiological results. For example in the  191  N1 component.  192  This component reflects selective attention, linked to the basic characteristics of a stimulus,  193 and also to the recognition of a specific visual pattern [30]. In terms of spatial localization, the 194 N1 amplitude is greater in occipital regions as well as in discrimination tasks The second component N2 has been found in several studies of incongruence using the Flanker 203 task and its latency was unaltered in medicated PD patients (for a review see [33]). In our study, 204 we did not find any differences between experimental conditions or in the N2 latency, but the 205 HCs had significantly higher N2 amplitudes than PD patients.

242
The authors declare that there is no conflict of interest regarding the publication of this paper.