- Email: [email protected]
A gamma band specific role of the subthalamic nucleus in switching during verbal fluency tasks in Parkinson’s disease
Experimental Neurology 232 (2011) 136–142
Contents lists available at SciVerse ScienceDirect
Experimental Neurology j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / yex n r
Regular Article
A gamma band specific role of the subthalamic nucleus in switching during verbal fluency tasks in Parkinson’s disease Anam Anzak a, b, Louise Gaynor a, Mazda Beigi a, Patricia Limousin a, Marwan Hariz a, Ludvic Zrinzo a, Thomas Foltynie a, Peter Brown b, Marjan Jahanshahi a,⁎ a b
Sobell Department of Motor Neuroscience & Movement Disorders, UCL Institute of Neurology & The National Hospital for Neurology and Neurosurgery, WC1N 3BG, UK Department of Clinical Neurology, University of Oxford, West Wing, John Radcliffe Hospital, UK
a r t i c l e
i n f o
Article history: Received 14 March 2011 Revised 29 May 2011 Accepted 7 July 2011 Available online 26 August 2011 Keywords: Deep brain stimulation Executive control Fronto-striatal circuits Parkinson's disease Subthalamic nucleus Verbal fluency
a b s t r a c t Objective: Decline in verbal fluency is the most consistent and persistent cognitive impairment documented after deep brain stimulation of the subthalamic nucleus in Parkinson's disease. The mechanisms of this deficit are unclear. We aimed to identify and characterise verbal fluency related processing within the subthalamic nucleus through analysis of local field potentials. Methods: Local field potentials were recorded from deep brain stimulation electrodes implanted in the subthalamic nuclei of 8 patients (16 sides) with Parkinson's disease, when patients were on medication. Patients performed phonemic and semantic verbal fluency tasks and a control word repetition task to control for the motor output involved in response generation. Results: Significant increases in local field potential Power (p ≤ 0.05) were seen across a broad gamma frequency band (30–95 Hz) during both verbal fluency tasks, after controlling for motor output. Increases in gamma local field potential Power of + 7.5% ± 2.3% (SEM) in the semantic fluency task and + 6.9% ± 2.0% in the phonemic fluency task were derived when averaging across all electrode contact pairs. Gamma changes recorded from contacts lying in the left hemisphere (dominant in verbal fluency) correlated with average number of correct responses generated (r = 0.81 p = 0.015) and measures of ‘switching’ (r = 0.79 p = 0.020) particularly strongly in the semantic fluency task. Interpretation: Frequency specific power changes observed during task performance are consistent with involvement of the subthalamic nucleus in switching during verbal fluency. Antagonism of such task-related activity with high frequency stimulation of this nucleus may explain the impairments reported. © 2011 Elsevier Inc. All rights reserved.
Introduction Deep brain stimulation (DBS) of the subthalamic nucleus (STN) is now established as an effective surgical treatment for advanced Parkinson's disease (PD) (Limousin et al., 1995; Weaver et al., 2009). While DBS of the STN does not affect general cognitive ability in PD (Ardouin et al., 1999; Parsons et al., 2006; Smeding et al., 2006; Witt et al., 2008; Zahodne et al., 2009), an impairment in phonemic and semantic verbal fluency (VF) is the most consistently reported cognitive side effect, as confirmed in a meta-analysis of 612 patients across 28 studies (Parsons et al., 2006). This deficit in VF persists across time and is present when patients are followed up and tested three (Zangaglia et al., 2009), five (Contarino et al., 2007) and eight (Fasano et al., 2010) years after surgery.
Abbreviations: DBS, deep brain stimulation; LFP, local field potentials; PD, Parkinson's disease; STN, subthalamic nucleus; VF, verbal fluency. ⁎ Corresponding author at: Sobell Department of Motor Neuroscience and Movement Disorders, Institute of Neurology, 33 Queen Square, London, WC1N 3BG, UK. E-mail address: [email protected] (M. Jahanshahi). 0014-4886/$ – see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.expneurol.2011.07.010
A number of theories exist regarding the mechanism of action of DBS of the STN. The ‘noisy signal hypothesis’ (Marsden and Obeso, 1994; Brown and Eusebio, 2008) posits that high frequency stimulation can act to suppress or override the excessive synchrony described in the 13–30 Hz (beta) frequency band in patients with PD. It is conceivable that high frequency stimulation could equally interfere with physiologically normal task-related activity in other frequency bands. An association between deficits in VF after DBS of the STN and active contacts lying closer to this nucleus has recently been described (York et al., 2009). In light of the above, we put forward the hypothesis that local processing within the STN may be modulated during performance of VF tasks. To test our hypothesis, we made direct recordings of STN local field potential (LFP) activity in patients with PD who had undergone implantation of this nucleus as a prelude to therapeutic high frequency DBS, while they performed semantic and phonemic versions of the VF Task. Changes in STN LFP activity or Power occur over specific bands (Beta band 13–30 Hz, Gamma band N30 Hz) during motor processing (Kuhn et al., 2004; Williams et al., 2005), and we sought to identify and characterise any comparable effects during
A. Anzak et al. / Experimental Neurology 232 (2011) 136–142
cognitive VF tasks. We reasoned that involvement of the STN in processing related to VF should be manifest as a modulation in the pattern of such LFP activity as a function of task. Methods Sample Eight patients with PD, (age 55.4 ± 4.5 years (SD), 5 males 3 females) gave informed consent to take part in this study, which was approved by the joint ethics committees of the National Hospital for Neurology and Neurosurgery and the Institute of Neurology. Patients underwent DBS for the treatment of advanced idiopathic PD with motor fluctuations and/or dyskinesias. None of the patients were demented or clinically depressed as established through detailed preoperative neuropsychological assessment. The clinical details of each patient are summarised in Table 1. Surgical procedure Implantation of bilateral STN DBS electrodes was performed under local anaesthesia in all eight patients, after overnight withdrawal of their antiparkinsonian medication. The DBS electrodes used were model 3389 (Medtronic Neurological Division, Minneapolis) with four platinum–iridium cylindrical surfaces (1.27 mm diameter and 1.5 mm length) and a centre-to-centre separation of 2 mm. Contact 0 of each electrode was the lowermost, contact 3 being the uppermost. Fast acquisition T2 weighted axial and coronal stereotactic MRI scans were performed using Leksell's Frame (Elekta, Sweden), with contiguous slices of 2 mm thickness, which allowed visualisation of the STN, especially the medial border (Hariz et al., 2003). The centre of the STN, which formed the anatomical target point, was defined as lying at the level of the anterior border of the red nucleus on the axial image showing the largest diameter of the red nucleus (Bejjani et al., 2000). This is the point where contact 1 of each electrode was intended to reach. Calculations of Cartesian co-ordinates of the target point were performed both manually on enlarged MRI film copies and using Framelink software (Medtronic, Minneapolis). During the surgical procedure, intra-operative high frequency test stimulation and clinical evaluation of the patient were carried out, helping to identify the best functional target. Once the optimum anatomical and functional target point for stimulation had been identified, the electrode was advanced 1–2 mm to ‘encompass’ this optimal target point before
137
it was fixed in position with either the Medtronic burr hole cap or the Navigus system (Image Guided Neurologics, FL, USA). The same procedure was repeated on both STN sides. Finally, implanted electrodes were attached to extension cables that were externalised. Immediate post-operative stereotactic MRI was carried out on all patients, with the Leksell frame still in place on the head, to confirm correct positioning of the DBS electrode (Foltynie and Hariz, 2010). Data collection for this study was carried out prior to connection of the electrodes to a battery-operated programmable pulse generator (Kinetra 7428, Medtronic). VF and control tasks Six trials of each of the semantic and phonemic versions of the VF test (Delis-Kaplan Executive Function System, 2001), and control word repetition tasks (controlling for motor output) were performed, in blocks of 60 second duration. In the semantic version, participants produced words belonging to one of the following specific semantic categories: animals, boy's names, fruit, furniture, items of clothing, and girl's names. In the phonemic version, participants were asked to generate words beginning with a particular letter (F, A, S, B, H, R), excluding names of people or places, numbers and grammatical variations of the same word. The control task involved listening to and simultaneously repeating back all the words which had been generated during each 60 second duration VF trial, which had been recorded on tape. The primary measure derived was the number of correct responses generated within each 60 second trial (sum of words generated in each of four 15 second intervals), based on Delis–Kaplan criteria. Incorrect responses constituted a) words not meeting the criteria for that semantic category and, b) repetitions within the same trial. Similarly, in the phonemic fluency condition, a correct response was a word that a) met the criteria of the condition (starting with the designated letter, and not the name of a person, place or number), b) was not a grammatical variant of a word previously generated in that trial, and c) not a repetition within that trial. In cases of ambiguous responses, the specific guidelines for scoring correct responses based on the DKEFS normative sample (Delis-Kaplan Executive Function System, 2001) were consulted. The total number of correct responses was corrected for age using the scaled scores for 16 age groups derived from the DKEFS normative sample. Optimal fluency performance may be achieved by production of clusters of semantically or phonemically related words, and once a sub-category is exhausted, switching to another. Thus two further
Table 1 Summary of patient details. UPDRS = Unified Parkinson's Disease Rating Scale. Improved UPDRS scores in each of the eight patients after switching the DBS stimulator on are consistent with placement of at least one DBS electrode contact in the STN. Patient
Age/yrs & sex
Predominant symptoms of PD
Disease duration/yrs
Pre-op motor UPDRS (OFF/ON Levodopa)
Post-op motor UPDRS (ON stimulation & OFF Levodopa/ON stimulation & ON Levodopa)
Medications (daily dose at time of operation)
1
55 M
Rigidity, bradykinesia
15
54/10
35/8
2
64 F
Freezing, difficulty walking
24
57/23
44/20
3
56 F
Rigidity, freezing, dyskinesia
15
71/18
32/11
4
54 M
Dyskinesias
7
29/10
25/7
5 6
58 F 48 M
Freezing, rigidity, dyskinesia Freezing ‘off’
8 12
36/10 72/16
11/7 25/11
7
54 M
Bradykinesia, gait freezing
8
38/9
21/11
8
54 M
Tremor, gait difficulties.
5
43/5
26/13
Levodopa 1100 mg Ropinirole 18 mg Selegiline 10 mg Amantadine 100 mg Levodopa 200 mg Ropinirole 3 mg Levodopa 250 mg Pramipexole 2.8 mg Levodopa 1000 mg Cabergoline 4 mg Cabergoline 2 mg Levodopa 1300 mg Rasagiline 1 mg Entacapone 800 mg Levodopa 800 mg Selegiline 10 mg Ropinirole 12 mg
A. Anzak et al. / Experimental Neurology 232 (2011) 136–142
measures of VF processing obtained were clustering (the production of words within semantic or phonemic sub-categories) and switching (the ability to shift between clusters). Scoring rules were based on Troyer et al. (1997). Cluster size was calculated starting with the second word in each cluster. Separate mean cluster size was consequently derived for the semantic and phonemic tests. Switching measures were derived as the number of transitions between clusters, including single words, for the phonemic and semantic tests separately.
A
30
left STN contact pair1-2 right STN contact pair1-2
Power/a.u.
138
20
10
LFP recordings
All analyses were performed off-line using Spike2 version 6. Data were interpolated to a common sampling rate of 1024 Hz, and were first visually inspected for large amplitude low frequency movement related artefact, which was subsequently removed. The bipolar nature of our recordings, and use of control task recordings with speech as a baseline, made any induced high frequency vibration-related artefact from the level of the voicebox unlikely in our results. For each patient, the individual trials for each task were collated (using the MSPLICE script in Spike2), thus excluding any rest periods, to form purely ‘VF’ and ‘Control’ condition files. Power spectra were subsequently calculated with 1 Hz resolution using the Fast Fourier transform for blocks of 1024 data points. These were estimated for VF and Control condition files, as well as one Rest file of ~60 second duration for each patient. LFP Power Spectra for each patient were exported to Microsoft Excel for calculation of the Average % Change in Power during the VF tasks from a baseline of the Control task at each frequency bin (Average % Change in Power = (VF − Control/Control) × 100), and subsequent collation across patients. Graphs representing the Average % Change in Power during the processing of each task were plotted (Fig. 1). These parametric tests were performed subsequent to confirmation of a normal distribution at each frequency bin across patients (one-sample Kolmogorov–Smirnov tests, p N 0.05). Repeated-measures analyses of variance (ANOVA) were performed in SPSS, using the General Linear Model procedure. Where necessary Greenhouse–Geisser corrections were applied. Means are presented in the text as Mean ± Standard Error of the Mean (SEM), unless otherwise specified.
0
20
40
60
80
100
80
100
80
100
Frequency/Hz
B
Power/a.u.
LFP analysis and statistics
0
30
20
10
0
0
20
40
60
Frequency/Hz
C 300 250
% Change in Power
LFP recordings were made 3–6 days after surgery, with the patients on medication while performing VF and control word repetition tasks. All recordings started within 1 h of ingestion of the last dose of each patient's usual anti-parkinsonian medication, when patients were subjectively and clinically in the ‘on’ state. LFPs were recorded bipolarly from adjacent contacts of each DBS electrode (0–1, 1–2, 2–3) in all cases. LFPs were band-pass filtered at 1–300 Hz using a D360 amplifier (Digitimer Ltd, Welwyn Garden City, Hertfordshire, UK), sampled at 1000 Hz and recorded through a 1401 A-D converter (Cambridge Electronic Design, Cambridge, UK) onto a computer using Spike2 software (Cambridge Electronic Design).
200 150 100 50 0 -50
0
20
40
60
Frequency/Hz
Fig. 1. Examples of LFP Power Spectra in Patient 8. (A) Phonemic verbal fluency task (B) control word repetition task, derived from differential activity between the left STN contact pair 1–2 (pink traces) and the right STN contact pair 1–2 (black traces), which recorded the highest gamma Powers. (C) Average % change in Power during the phonemic verbal fluency task relative to the control word repetition task baseline. A clear increase in Power is seen across a broad 30–95 Hz band. Note the feature at ~ 80 Hz is peculiar only to this patient in the cohort.
Results LFP Power increases in the gamma frequency band common to both VF Tasks In an initial analysis of LFPs derived as an average of recordings across all contact pairs on each electrode, significant changes in LFP Power reactivity (p ≤ 0.05, one sample T-test from 0% change) were seen across a broad gamma frequency band (30–95 Hz, excluding a 45–55 Hz range potentially contaminated by the notch filter used to
remove European line-noise) in both semantic and phonemic versions of the VF task (Fig. 2). Increases in gamma band LFP Power of 7.5% ± 2.3% (SEM) in the semantic fluency task and 6.9% ± 2.0% in the phonemic fluency task were observed when averaged across all STN electrode contact pairs. When only the STN contact pair affording the highest gamma Power for each subject at rest was used for analysis, maximal increases in Power of 8.2% ± 2.8% in the semantic fluency task, and 10.1% ± 2.7% in the phonemic fluency task, were observed (Fig. 3).
A. Anzak et al. / Experimental Neurology 232 (2011) 136–142
A
significant (F[1.00,15.00] = 14.09, p = 0.002), but the effects of TASK TYPE (F[1.00,15.00] = 1.81, p = 0.199), and the TASK TYPE × CONDITION interaction (F[1.00,15.00] = 0.47, p = 0.503) were not significant, confirming that the gamma band Power increase was similar during the two VF tasks. The results were similar when only data from the contact pair with the highest gamma power was analysed on each side.
% Change in LFP Power
25
15
5 0
Frequency specificity of LFP Power changes 10
20
30
40
50
60
-5
70
80
90
100
Frequency/Hz
-15
B % Change in LFP Power
25
15
5 0
10
20
30
40
-5
50
60
70
80
90
100
Frequency/Hz
-15 Fig. 2. Average % change in LFP Power during verbal fluency tasks from a control task baseline for (A) semantic fluency tasks (n = 16) and (B) phonemic fluency tasks (n = 16). Black filled as opposed to empty data points indicate those frequencies at which changes in LFP Power from baseline were significant, after application of one sample T-tests, from 0% change p ≤ 0.05 (see Methods). The peaks observed at ~ 80 Hz result solely from the LFP recordings of Patient 8, which skewed the distribution across these frequency bins so that one sample T-tests could not be applied. Power changes at ~ 50 Hz are likely to be contaminated by line noise. Increases in Power observed at frequencies below 8 Hz were excluded from further analysis as they may have been contaminated by movement artefact. Note the presence of a 50 Hz notch filter.
Average % Change in Power
Next, we sought to determine whether the pattern of increases in LFP Power observed across the broad gamma band was dependent on the type of VF task. We hypothesised that there should be no such effect if LFP changes were related to processing common to both semantic and phonemic versions of the VF task. An ANOVA was performed with factors TASK TYPE (semantic versus phonemic VF) and CONDITION (VF vs Control) for LFPs averaged in the 30–95 Hz gamma band (excluding 45–55 Hz potentially artefact contaminated range) across all STN contact pairs. The effect of CONDITION was 15 10
All contact pairs Best contact pair
5 0 -5 -10
139
Beta
Gamma
Beta
Gamma
-15 Semantic Fluency Task
Phonemic Fluency Task
Fig. 3. Average % change in LFP Power across the Beta (13–30 Hz) and Gamma (30– 95 Hz) frequency ranges for semantic (n = 16) and phonemic (n = 16) fluency tasks. Values are derived as an average of recordings from (1) all contact pairs and (2) ‘best’ estimates from contact pairs affording the highest gamma Powers at rest. Error bars = SEM.
The above findings pertain to changes in the gamma band. There was however a significant suppression of Power over the beta (13– 30 Hz) band in the semantic fluency task, which was less apparent in the phonemic fluency tasks (Figs. 2 and 3). Changes in the beta and gamma bands were subsequently compared, to establish whether changes in both were significant and whether contrasting reactivity existed. A repeated measures ANOVA was applied to LFP Power derived as an average across recordings from all three contact pairs, with factors TASK TYPE (semantic vs phonemic fluency), CONDITION (VF vs Control) and FREQUENCY (13–30 Hz Beta frequencies vs 30–95 Hz Gamma frequencies). As expected there was a significant effect of CONDITION (F[1.00,15.00] = 5.83, p = 0.029) but no effect of TASK TYPE (F[1.00,15.00] = 1.52, p = 0.237). While a significant effect of FREQUENCY was not seen (F[1.00,15.00] = 1.88, p = 0.191), the TASK × FREQUENCY (F[1.00,15.00] = 4.58, p = 0.049), TASK × CONDITION (F[1.00,15.00] = 17.79, p = 0.001) and TASK × FREQUENCY × CONDITION (F[1.00,15.00] = 8.58, p = 0.010) interactions were all significant, indicating a difference in the profile of changes in beta and gamma bands during VF and Control conditions, between semantic and phonemic tasks. The FREQUENCY × CONDITION interaction (F[1.00,15.00] = 1.98, p = 0.180) was not significant. Subsequent post-hoc analysis with paired T-tests revealed a difference between LFP Powers recorded in the VF and Control conditions of the semantic fluency task (p = 0.007), but not the phonemic task (p = 0.198) over the 13–30 Hz beta frequency range. Association between verbal fluency measures and LFP Power change Age-corrected normative data from the Delis–Kaplan manual was used to obtain scaled scores for the total number of words generated in the VF tasks. The average scaled scores on the 6 trials of each of the semantic and phonemic VF tasks for each of the 8 patients are shown in Fig. 4. The average scaled score across our sample of PD patients for the phonemic fluency task was 13.1 ± 5.5 (SD) and the semantic fluency task was 8.3 ± 4.1 (SD). These values were compared to the preoperative mean scaled scores for a sample of 40 PD patients assessed prior to DBS surgery at the NHNN (mean age 56.9 ± 7.9 (SD), 27 male, 13 female) from which our sample of 8 patients was drawn. Phonemic scores were similar (13.3 ± 4.0 (SD), p = 0.90, two sample T-test), but semantic scores were significantly lower in our study sample as compared to the control PD group (11.2 ± 3.4 (SD), p = 0.03). The difference may be reflective of a post-operative stun effect. In addition, for our 8 subjects, a mean cluster size ratio of 1.7 ± 0.84 (SD) was observed for phonemic fluency tasks and 1.4 ± 1.67 (SD) for the semantic fluency tasks. The mean switching score was 10.5 ± 3.5 (SD) for phonemic fluency and 12.4 ± 6.7 (SD) for the semantic fluency tasks. Following confirmation of a Gaussian distribution in both task scores and change in LFP Power averaged across each frequency band (Kolmogorov–Smirnov test, p N 0.05), Pearson correlation coefficients were calculated. For semantic fluency there was a significant correlation between average number of correct responses scaled score in interval 1 (first 15 s of each trial, during which VF output is at its best) and average gamma Power across contacts (r= 0.85, p = 0.0008). Interestingly average gamma Power in only left-sided contacts, considered to be the dominant hemisphere in VF tasks, also correlated significantly with
140
A. Anzak et al. / Experimental Neurology 232 (2011) 136–142 Semantic Fluency
20
28% Contact Pair 23
16
100
14
90
12 10 8 6 4 2 0 1
2
3
4
5
6
7
8
% of Maximum Power
Age-scaled score
A
Phonemic Fluency
18
80 70 60 50 40 30 20 10
Patient Number
0
both average scaled score of correct words generated in interval 1 (r= 0.93, p = 0.0007), and the average total correct words generated across the 60 second trial (r= 0.81, p = 0.015). There was no such correlation with right sided contacts. For phonemic fluency the correlation between number correct in interval 1 and average gamma Power across contacts was of a sizeable magnitude but did not reach significance (r = 0.62, p = 0.10). Also of interest was a significant correlation between left-sided gamma Power and average switching in the Semantic Fluency Task (r = 0.79, p = 0.020). For the phonemic task, the correlation with the switching measure was of a sizeable magnitude but only approached significance (r = 0.66, p = 0.07). No significant correlations were found between the LFPs and measures of clustering, or changes in the beta band and measures of VF.
01
12
23
Contact Pair
B 69% Contact Pair 12 100 90
% of Maximum Power
Fig. 4. Age-corrected verbal fluency scaled scores for each patient. Lower scores represent poorer performance. Error bars = Standard Error of Mean. Age scaled scores for patients 4, 7 and 8 were similar for both trials of the phonemic fluency condition, thus error bars could not be derived. While the semantic VF scaled scores of patients 5 and 7 were low as compared to the pre-operative mean scaled scores for a sample of 40 PD patients assessed prior to DBS surgery (see Results), collation of VF scores for our sample did not result in exclusion of any patient on the basis of potentially sub-optimal performance (VF Score b (Mean-1.96*SD), after confirmation of Gaussian Distribution, Kolmogorov–Smirnov Test, p N 0.05).
80 70 60 50 40 30 20 10 0 01
12
23
Contact Pair
C 3% Contact Pair 01 100
Evidence for locally generated activity
% of Maximum Power
Analysis of post-operative imaging confirmed that out of 16 STNs from which the recordings were taken, all contacts (0,1,2 & 3) lay within or touching the STN in 12 cases. In 2 cases, 2 out of the 4 contacts lay within or touching the STN, in 1 case only 1 contact lay touching the STN. In one further case the contacts were found to be positioned outside but close to the STN in the following distribution : contact 0 caudal to the inferior of the STN, contacts 1 and 2 medial to the central segment of the STN, and contact 3 rostral to the superior segment of the STN. Whether LFP oscillations reflected activity in local generators or volume conduction from distant, possibly cortical sources, was investigated. To address this point the maximum gamma Power, during the active task, among the three contact pairs from each electrode was defined as +100% and the amplitude of the two remaining contact pairs normalised to that maximum. The mean gradient of gamma Power at the remaining contact pairs was 58.7% ± 2.9%. The contact pairs with the highest amplitude for each electrode were distributed as follows: 28% contact pair 2–3, 69% contact pair 1–2, 3% contact pair 0–1 (Fig. 5). A similar distribution of contact pairs recording the highest beta Power was found: 19% contact pair 2–3, 75% contact pair 1–2, and 6% contact pair 0–1. The mean gradient of beta Power at the remaining contact pairs was 55.7% ± 4.7%. The electrode positioning strategy was such that the inferior two contacts (0,1) lay closer to the deeper, non-motor, part of the STN, while the superior two contacts (2,3) lay closer to the upper sensorimotor part of the STN and zona
90 80 70 60 50 40 30 20 10 0 01
12
23
Contact Pair Fig. 5. Distribution of maximal gamma Power amongst the three contact pairs of the DBS electrode, expressed as a percentage of individual maximum gamma Power. (A) DBS electrode site with maximum gamma Power at contact pair 2–3. (B) DBS electrode site with maximum gamma Power at contact pair 1–2. (C) DBS electrode site with maximum gamma Power at contact pair 0–1. Each line is representative of recordings from each STN of 8 patients during the Active trials of semantic (n = 16) and phonemic (n = 16) fluency tasks (32 lines).
incerta. Thus our measurement of the greatest change or differential activity in the gamma band lying between contacts 1–2, is in line with previous reports of localisation of gamma Power to the upper STN (Trottenberg et al., 2006).
A. Anzak et al. / Experimental Neurology 232 (2011) 136–142
Discussion The principal finding of this study was a significant increase in STN LFP Power across a broad gamma frequency band (30–95 Hz), during performance of phonemic and semantic VF tasks. Decreases in LFP Power across a beta (13–30 Hz) range were also seen for both VF tasks which was significant for semantic VF. LFP Power change in the STN was significantly associated with measures of performance, particularly the number of words generated and switching. These LFP neural signatures for VF were identified after controlling for motor components of verbal response generation. The results demonstrate, for the first time, that STN LFP Power is modulated during semantic and phonemic VF. Before proceeding to a further discussion of our results, it is useful to consider the reliability of the conclusions drawn from the LFP recordings. First, the bipolar recordings made across adjacent contacts of the DBS electrodes in our study, representing differential activity across contact pairs, meant that the signals analysed were most likely the product of locally generated activity in the STN rather than reflecting volume conduction from synchronous cortical areas. Further evidence for a focal origin of the signals comes from the clear gradient of LFP Power across the contact pairs (Fig. 5), since in volume conduction, one would expect LFP Power to be equally distributed between adjacent contact pairs of the electrode, or present a decrement at contacts more distant from the cortical source (Williams et al., 2003; Doyle et al., 2005). Finally, that LFP oscillations are truly reflective of neuronal activity in the STN has consistently been demonstrated by synchrony between these two variables in both anaesthetized rats (Magill et al., 2004) and alert Parkinsonian patients (Levy et al., 2002). While our eight post-operative patients performed worse in the semantic fluency task than a PD control group assessed prior to surgery, the similarities in STN gamma activity between this task and the better performed phonemic fluency task suggest an intact involvement of the STN in processing related to a feature common to the two tasks. On the other hand, the small but significant desynchronisation in STN beta Power observed in the semantic fluency but not phonemic fluency task (see Fig. 3), may be reflective of a further role of the STN in processing specific to success in semantic VF. One possibility is a difference in task demand for recruitment of executive processes. For example, while both phonemic and semantic VF require effortful retrieval of information from semantic memory, it is likely that due to differences in depth of encoding, retrieval of semantically associated items involves additional retrieval mechanisms during semantic VF (Henry and Crawford, 2004). What common process(es) of VF tasks might the STN thus be involved in? Certainly, the impairment in VF is not a manifestation of a global cognitive decline, as general cognitive ability is widely reported to remain intact after DBS of the STN (Parsons et al., 2006; Zahodne et al., 2009). Evidence for a correlation between left sided gamma Power and switching scores for both the semantic and phonemic tasks does however suggest a potential role of the STN in the process of disengaging from an exhausted semantic or phonemic sub-category and switching to a new sub-category, in order to optimise the number of words generated. This is in agreement with the evidence of decline in switching but not cluster size during VF tasks with high frequency stimulation of the STN (Saint-Cyr and Trepanier, 2000; De Gaspari et al., 2006). Correlations with left sided activity rather than right are also consistent with the linguistic nature of VF tasks and research describing an association between left frontal–subcortical neurons and VF (Baldo et al., 2001), and imaging evidence of activation of the left frontal areas in VF tasks (Frith et al., 1991) and, importantly, left frontal systems in cognitive set switching (Dove et al., 2000). Involvement of the STN in the processing of VF tasks can be placed in the context of a larger cortical network model for willed or intrinsic generation of words, validated with Positron Emission Tomography
141
(PET) in normal subjects. The model (Friston et al., 1991; Frith et al., 1991) posits that successful word generation in VF tasks is achieved through a modulatory influence of the dorsolateral prefrontal cortex (DLPFC) over a distributed word associative network in the superior temporal cortex (STC), to prevent spreading activation in the network and to allow strategic generation of words which meet the VF criteria. For example, during phonemic VF when generating words beginning with ‘F’, having generated the word ‘farm’ the participant has to suppress highly semantically associated words such as ‘cow’ or ‘milk’ and move on to generate other words beginning with the letter ‘F’. PET studies have shown that stimulation of the STN affects just such fronto-temporal networks, during VF in PD (Schroeder et al., 2003; Cilia et al., 2007). Similarly, using 18-fluorodeoxyglucose (FDG) PET, Kalbe et al. (2009) reported that FDG uptake in the left DLPFC, left Broca's area and right anterior cingulate was significantly associated with VF scores after STN DBS surgery. In line with the above network modulation model of VF, the STN could be considered to implement a ‘no go’ signal to prevent premature or inappropriate responses (Frank et al., 2007), drawing parallels to its posited role in motor response inhibition (Baunez et al., 1995; Kuhn et al., 2004; Ray et al., 2009). This is consistent with a centre surround inhibition model (Mink and Thach, 1993) in which the STN could ultimately inhibit potentially inappropriate responses and facilitate selection of the appropriate response (Redgrave et al., 1999). It is plausible that left DLPFC initiated activity may exert an inhibitory influence on inappropriate spreading activation in the left STC word association network via STN-mediated inhibition of thalamic output to temporal cortex, thus allowing only words appropriate to the task to be selected. Indeed, the STN receives direct input from the DLPFC (Nambu et al., 2002) and has indirect connections to the temporal cortex via the GPi and thalamus (Middleton and Strick, 2000). As such, the nature of the response inhibition required in VF tasks may involve a more tonic and internally driven form of inhibition that may be somewhat distinct from the more phasic, trial by trial, and cued suppression or withholding of responses in previous motor paradigms (Kuhn et al., 2004). A related proposal is the possibility that the STN receives and implements a switching signal from the frontal cortex to allow a shift from automatic selection of a highly associated but inappropriate word to the controlled selection of a word that meets the task criteria. Such a role for the STN in concert with the pre-SMA in switching between automatic and controlled processing has been shown in neuronal recordings in primates (Isoda and Hikosaka, 2008) and in a recent imaging study of decision-making in humans, where the STN together with the inferior frontal cortex were engaged when switches from the default or status quo decision were required (Fleming et al., 2010). To conclude, we provide evidence, for the first time, of task-related power changes across gamma and beta frequencies in the STN during VF, indicating involvement of the STN in processing related to these cognitive tasks. Frequency specific involvement of the STN in VF may provide a viable explanation for the deficits reported with DBS, through superimposition of high frequency stimulation on physiologically normal task-related oscillatory activity. Our results further suggest a role for the STN in inhibition of inappropriate responses or switching during semantic and phonemic VF. Finally, as a characteristic of untreated PD is a suppression of gamma activity and an enhancement of beta oscillatory activity in the STN (Dostrovsky and Bergman, 2004), this pattern would antagonise the task-related changes found here and might explain the general impairment in VF seen in PD patients (Donovan et al., 1999; Henry and Crawford, 2004). Acknowledgments We would like to thank the patients for their participation. Funding: MB is supported by a 3 + 1 ESRC/MRC studentship. PB is supported by the MRC and the NIHR Biomedical Research Centre, Oxford. PL, LZ, MH and TF are supported by the Parkinson's Appeal.
142
A. Anzak et al. / Experimental Neurology 232 (2011) 136–142
References Ardouin, C., Pillon, B., Peiffer, E., et al., 1999. Bilateral subthalamic or pallidal stimulation for Parkinson's disease affects neither memory nor executive functions: a consecutive series of 62 patients. Ann. Neurol. 46, 217–223. Baldo, J.V., Shimamura, A.P., Delis, D.C., et al., 2001. Verbal and design fluency in patients with frontal lobe lesions. J. Int. Neuropsychol. Soc. 7, 586–596. Baunez, C., Nieoullon, A., Amalric, M., 1995. In a rat model of parkinsonism, lesions of the subthalamic nucleus reverse increases of reaction-time but induce a dramatic premature responding deficit. J. Neurosci. 15, 6531–6541. Bejjani, B.P., Dormont, D., Pidoux, B., et al., 2000. Bilateral subthalamic stimulation for Parkinson's disease by using three-dimensional stereotactic magnetic resonance imaging and electrophysiological guidance. J. Neurosurg. 92, 615–625. Brown, P., Eusebio, A., 2008. Paradoxes of functional neurosurgery: clues from basal ganglia recordings. Mov. Disord. 23, 12–20. Cilia, R., Siri, C., Marotta, G., et al., 2007. Brain networks underlining verbal fluency decline during STN-DBS in Parkinson's disease: an ECD-SPECT study. Parkinsonism Relat. Disord. 13, 290–294. Contarino, M.F., Daniele, A., Sibilia, A.H., et al., 2007. Cognitive outcome 5 years after bilateral chronic stimulation of subthalamic nucleus in patients with Parkinson's disease. J. Neurol. Neurosurg. Psychiatry 78, 248–252. De Gaspari, D., Siri, C., Di Gioia, M., et al., 2006. Clinical correlates and cognitive underpinnings of verbal fluency impairment after chronic subthalamic stimulation in Parkinson's disease. Parkinsonism Relat. Disord. 12, 289–295. Delis, D.C., Kaplan, E., Kramer, J.H., 2001. Delis-Kaplan Executive Function System (D-KEFS). The Psychological Corporation, San Antonio, TX. Donovan, K., Siegert, R., McDowall, J., et al., 1999. Clustering and switching in verbal fluency in Parkinson's disease. N. Z. J. Psychol. 28, 61–66. Dostrovsky, J., Bergman, H., 2004. Oscillatory activity in the basal ganglia — relationship to normal physiology and pathophysiology. Brain 127, 721–722. Dove, A., Pollmann, S., Schubert, T., et al., 2000. Prefrontal cortex activation in task switching: an event-related fMRI study. Brain Res. Cogn. Brain Res. 9, 103–109. Doyle, L.M.F., Kuhn, A.A., Hariz, M., et al., 2005. Levodopa-induced modulation of subthalamic beta oscillations during self-paced movements in patients with Parkinson's disease. Eur. J. Neurosci. 21, 1403–1412. Fasano, A., Romito, L.M., Daniele, A., et al., 2010. Motor and cognitive outcome in patients with Parkinson's disease 8 years after subthalamic implants. Brain 133, 2664–2676. Fleming, S.M., Thomas, C.L., Dolan, R.J., 2010. Overcoming status quo bias in the human brain Proceedings of the national academy of sciences. Proc. Natl. Acad. Sci. U.S.A. 107 (13), 6005–6009. Foltynie, T., Hariz, M.I., 2010. Surgical management of Parkinson's disease. Expert Rev. Neurother. 10, 903–914. Frank, M.J., Samanta, J., Moustafa, A.A., et al., 2007. Hold your horses: impulsivity, deep brain stimulation, and medication in parkinsonism. Science 318, 1309–1312. Friston, K.J., Frith, C.D., Liddle, P.F., Frackowiak, R.S., 1991. Investigating a network model of word generation with positron emission tomography. Proc. Biol. Sci. 244, 101–106. Frith, C.D., Friston, K.J., Liddle, P.F., et al., 1991. A PET study of word finding. Neuropsychologia 29, 1137–1148. Hariz, M.I., Krack, P., Melvill, R., et al., 2003. A quick and universal method for stereotactic visualization of the subthalamic nucleus before and after implantation of deep brain stimulation electrodes. Stereotact. Funct. Neurosurg. 80, 96–101. Henry, J.D., Crawford, J.R., 2004. A meta-analytic review of verbal fluency performance following focal cortical lesions. Neuropsychology 18, 284–295. Isoda, M., Hikosaka, O., 2008. Role for subthalamic nucleus neurons in switching from automatic to controlled eye movement. J. Neurosci. 28, 7209–7218. Kalbe, E., Voges, J., Weber, T., et al., 2009. Frontal FDG-PET activity correlates with cognitive outcome after STN-DBS in Parkinson disease. Neurology 72, 42–49.
Kuhn, A.A., Williams, D., Kupsch, A., et al., 2004. Event-related beta desynchronization in human subthalamic nucleus correlates with motor performance. Brain 127, 735–746. Levy, R., Hutchison, W.D., Lozano, A.M., et al., 2002. Synchronized neuronal discharge in the basal ganglia of parkinsonian patients is limited to oscillatory activity. J. Neurosci. 22, 2855–2861. Limousin, P., Pollak, P., Benazzouz, A., et al., 1995. Bilateral subthalamic nucleus stimulation for severe Parkinsons-disease. Mov. Disord. 10, 672–674. Magill, P.J., Sharott, A., Bevan, M.D., et al., 2004. Synchronous unit activity and local field potentials evoked in the subthalamic nucleus by cortical stimulation. J. Neurophysiol. 92, 700–714. Marsden, C.D., Obeso, J.A., 1994. The functions of the basal ganglia and the paradox of stereotaxic surgery in Parkinson's disease. Brain 117, 877–897. Middleton, F.A., Strick, P.L., 2000. Basal ganglia output and cognition: evidence from anatomical, behavioral, and clinical studies. Brain Cogn. 42, 183–200. Mink, J.W., Thach, W.T., 1993. Basal ganglia intrinsic circuits and their role in behavior. Curr. Opin. Neurobiol. 3, 950–957. Nambu, A., Tokuno, H., Takada, M., 2002. Functional significance of the corticosubthalamo-pallidal ‘hyperdirect’ pathway. Neurosci. Res. 43, 111–117. Parsons, T.D., Rogers, S.A., Braaten, A.J., et al., 2006. Cognitive sequelae of subthalamic nucleus deep brain stimulation in Parkinson's disease: a meta-analysis. Lancet Neurol. 5, 578–588. Ray, N.J., Jenkinson, N., Brittain, J., et al., 2009. The role of the subthalamic nucleus in response inhibition: evidence from deep brain stimulation for Parkinson's disease. Neuropsychologia 47, 2828–2834. Redgrave, P., Prescott, T.J., Gurney, K., 1999. The basal ganglia: a vertebrate solution to the selection problem? Neuroscience 89, 1009–1023. Saint-Cyr, J.A., Trepanier, L.L., 2000. Neuropsychologic assessment of patients for movement disorder surgery. Mov. Disord. 15, 771–783. Schroeder, U., Kuehler, A., Lange, K.W., et al., 2003. Subthalamic nucleus stimulation affects a frontotemporal network: a PET study. Ann. Neurol. 54, 445–450. Smeding, H.M.M., Speelman, J.D., Koning-Haanstra, M., et al., 2006. Neuropsychological effects of bilateral STN stimulation in Parkinson disease — a controlled study. Neurology 66, 1830–1836. Trottenberg, T., Fogelson, N., Kühn, A.A., et al., 2006. Subthalamic gamma activity in patients with Parkinson's disease. Exp. Neurol. 200, 56–65. Troyer, A.K., Moscovitch, M., Winocur, G., 1997. Clustering and switching as two components of verbal fluency: evidence from younger and older healthy adults. Neuropsychology 11, 138–146. Weaver, F.M., Follett, K., Stern, M., et al., 2009. Bilateral deep brain stimulation vs best medical therapy for patients with advanced Parkinson disease a randomized controlled trial. Jama-J. Am. Med. Assoc. 301, 63–73. Williams, D., Kuhn, A., Kupsch, A., et al., 2003. Behavioural cues are associated with modulations of synchronous oscillations in the human subthalamic nucleus. Brain 126, 1975–1985. Williams, D., Kuhn, A., Kupsch, A., et al., 2005. The relationship between oscillatory activity and motor reaction time in the parkinsonian subthalamic nucleus. Eur. J. Neurosci. 21, 249–258. Witt, K., Daniels, C., Reiff, J., et al., 2008. Neuropsychological and psychiatric changes after deep brain stimulation for Parkinson's disease: a randomised, multicentre study. Lancet Neurol. 7, 605–614. York, M.K., Wilde, E.A., Simpson, R., et al., 2009. Relationship between neuropsychological outcome and DBS surgical trajectory and electrode location. J. Neurol. Sci. 287, 159–171. Zahodne, L.B., Okun, M.S., Foote, K.D., et al., 2009. Cognitive declines one year after unilateral deep brain stimulation surgery in Parkinson's disease: a controlled study using reliable change. Clin. Neuropsychol. 23, 385–405. Zangaglia, R., Pacchetti, C., Pasotti, C., et al., 2009. Deep Brain stimulation and cognitive functions in Parkinson's disease: a three-year controlled study. Mov. Disord. 24, 1621–1628.
Contents lists available at SciVerse ScienceDirect
Experimental Neurology j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / yex n r
Regular Article
A gamma band specific role of the subthalamic nucleus in switching during verbal fluency tasks in Parkinson’s disease Anam Anzak a, b, Louise Gaynor a, Mazda Beigi a, Patricia Limousin a, Marwan Hariz a, Ludvic Zrinzo a, Thomas Foltynie a, Peter Brown b, Marjan Jahanshahi a,⁎ a b
Sobell Department of Motor Neuroscience & Movement Disorders, UCL Institute of Neurology & The National Hospital for Neurology and Neurosurgery, WC1N 3BG, UK Department of Clinical Neurology, University of Oxford, West Wing, John Radcliffe Hospital, UK
a r t i c l e
i n f o
Article history: Received 14 March 2011 Revised 29 May 2011 Accepted 7 July 2011 Available online 26 August 2011 Keywords: Deep brain stimulation Executive control Fronto-striatal circuits Parkinson's disease Subthalamic nucleus Verbal fluency
a b s t r a c t Objective: Decline in verbal fluency is the most consistent and persistent cognitive impairment documented after deep brain stimulation of the subthalamic nucleus in Parkinson's disease. The mechanisms of this deficit are unclear. We aimed to identify and characterise verbal fluency related processing within the subthalamic nucleus through analysis of local field potentials. Methods: Local field potentials were recorded from deep brain stimulation electrodes implanted in the subthalamic nuclei of 8 patients (16 sides) with Parkinson's disease, when patients were on medication. Patients performed phonemic and semantic verbal fluency tasks and a control word repetition task to control for the motor output involved in response generation. Results: Significant increases in local field potential Power (p ≤ 0.05) were seen across a broad gamma frequency band (30–95 Hz) during both verbal fluency tasks, after controlling for motor output. Increases in gamma local field potential Power of + 7.5% ± 2.3% (SEM) in the semantic fluency task and + 6.9% ± 2.0% in the phonemic fluency task were derived when averaging across all electrode contact pairs. Gamma changes recorded from contacts lying in the left hemisphere (dominant in verbal fluency) correlated with average number of correct responses generated (r = 0.81 p = 0.015) and measures of ‘switching’ (r = 0.79 p = 0.020) particularly strongly in the semantic fluency task. Interpretation: Frequency specific power changes observed during task performance are consistent with involvement of the subthalamic nucleus in switching during verbal fluency. Antagonism of such task-related activity with high frequency stimulation of this nucleus may explain the impairments reported. © 2011 Elsevier Inc. All rights reserved.
Introduction Deep brain stimulation (DBS) of the subthalamic nucleus (STN) is now established as an effective surgical treatment for advanced Parkinson's disease (PD) (Limousin et al., 1995; Weaver et al., 2009). While DBS of the STN does not affect general cognitive ability in PD (Ardouin et al., 1999; Parsons et al., 2006; Smeding et al., 2006; Witt et al., 2008; Zahodne et al., 2009), an impairment in phonemic and semantic verbal fluency (VF) is the most consistently reported cognitive side effect, as confirmed in a meta-analysis of 612 patients across 28 studies (Parsons et al., 2006). This deficit in VF persists across time and is present when patients are followed up and tested three (Zangaglia et al., 2009), five (Contarino et al., 2007) and eight (Fasano et al., 2010) years after surgery.
Abbreviations: DBS, deep brain stimulation; LFP, local field potentials; PD, Parkinson's disease; STN, subthalamic nucleus; VF, verbal fluency. ⁎ Corresponding author at: Sobell Department of Motor Neuroscience and Movement Disorders, Institute of Neurology, 33 Queen Square, London, WC1N 3BG, UK. E-mail address: [email protected] (M. Jahanshahi). 0014-4886/$ – see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.expneurol.2011.07.010
A number of theories exist regarding the mechanism of action of DBS of the STN. The ‘noisy signal hypothesis’ (Marsden and Obeso, 1994; Brown and Eusebio, 2008) posits that high frequency stimulation can act to suppress or override the excessive synchrony described in the 13–30 Hz (beta) frequency band in patients with PD. It is conceivable that high frequency stimulation could equally interfere with physiologically normal task-related activity in other frequency bands. An association between deficits in VF after DBS of the STN and active contacts lying closer to this nucleus has recently been described (York et al., 2009). In light of the above, we put forward the hypothesis that local processing within the STN may be modulated during performance of VF tasks. To test our hypothesis, we made direct recordings of STN local field potential (LFP) activity in patients with PD who had undergone implantation of this nucleus as a prelude to therapeutic high frequency DBS, while they performed semantic and phonemic versions of the VF Task. Changes in STN LFP activity or Power occur over specific bands (Beta band 13–30 Hz, Gamma band N30 Hz) during motor processing (Kuhn et al., 2004; Williams et al., 2005), and we sought to identify and characterise any comparable effects during
A. Anzak et al. / Experimental Neurology 232 (2011) 136–142
cognitive VF tasks. We reasoned that involvement of the STN in processing related to VF should be manifest as a modulation in the pattern of such LFP activity as a function of task. Methods Sample Eight patients with PD, (age 55.4 ± 4.5 years (SD), 5 males 3 females) gave informed consent to take part in this study, which was approved by the joint ethics committees of the National Hospital for Neurology and Neurosurgery and the Institute of Neurology. Patients underwent DBS for the treatment of advanced idiopathic PD with motor fluctuations and/or dyskinesias. None of the patients were demented or clinically depressed as established through detailed preoperative neuropsychological assessment. The clinical details of each patient are summarised in Table 1. Surgical procedure Implantation of bilateral STN DBS electrodes was performed under local anaesthesia in all eight patients, after overnight withdrawal of their antiparkinsonian medication. The DBS electrodes used were model 3389 (Medtronic Neurological Division, Minneapolis) with four platinum–iridium cylindrical surfaces (1.27 mm diameter and 1.5 mm length) and a centre-to-centre separation of 2 mm. Contact 0 of each electrode was the lowermost, contact 3 being the uppermost. Fast acquisition T2 weighted axial and coronal stereotactic MRI scans were performed using Leksell's Frame (Elekta, Sweden), with contiguous slices of 2 mm thickness, which allowed visualisation of the STN, especially the medial border (Hariz et al., 2003). The centre of the STN, which formed the anatomical target point, was defined as lying at the level of the anterior border of the red nucleus on the axial image showing the largest diameter of the red nucleus (Bejjani et al., 2000). This is the point where contact 1 of each electrode was intended to reach. Calculations of Cartesian co-ordinates of the target point were performed both manually on enlarged MRI film copies and using Framelink software (Medtronic, Minneapolis). During the surgical procedure, intra-operative high frequency test stimulation and clinical evaluation of the patient were carried out, helping to identify the best functional target. Once the optimum anatomical and functional target point for stimulation had been identified, the electrode was advanced 1–2 mm to ‘encompass’ this optimal target point before
137
it was fixed in position with either the Medtronic burr hole cap or the Navigus system (Image Guided Neurologics, FL, USA). The same procedure was repeated on both STN sides. Finally, implanted electrodes were attached to extension cables that were externalised. Immediate post-operative stereotactic MRI was carried out on all patients, with the Leksell frame still in place on the head, to confirm correct positioning of the DBS electrode (Foltynie and Hariz, 2010). Data collection for this study was carried out prior to connection of the electrodes to a battery-operated programmable pulse generator (Kinetra 7428, Medtronic). VF and control tasks Six trials of each of the semantic and phonemic versions of the VF test (Delis-Kaplan Executive Function System, 2001), and control word repetition tasks (controlling for motor output) were performed, in blocks of 60 second duration. In the semantic version, participants produced words belonging to one of the following specific semantic categories: animals, boy's names, fruit, furniture, items of clothing, and girl's names. In the phonemic version, participants were asked to generate words beginning with a particular letter (F, A, S, B, H, R), excluding names of people or places, numbers and grammatical variations of the same word. The control task involved listening to and simultaneously repeating back all the words which had been generated during each 60 second duration VF trial, which had been recorded on tape. The primary measure derived was the number of correct responses generated within each 60 second trial (sum of words generated in each of four 15 second intervals), based on Delis–Kaplan criteria. Incorrect responses constituted a) words not meeting the criteria for that semantic category and, b) repetitions within the same trial. Similarly, in the phonemic fluency condition, a correct response was a word that a) met the criteria of the condition (starting with the designated letter, and not the name of a person, place or number), b) was not a grammatical variant of a word previously generated in that trial, and c) not a repetition within that trial. In cases of ambiguous responses, the specific guidelines for scoring correct responses based on the DKEFS normative sample (Delis-Kaplan Executive Function System, 2001) were consulted. The total number of correct responses was corrected for age using the scaled scores for 16 age groups derived from the DKEFS normative sample. Optimal fluency performance may be achieved by production of clusters of semantically or phonemically related words, and once a sub-category is exhausted, switching to another. Thus two further
Table 1 Summary of patient details. UPDRS = Unified Parkinson's Disease Rating Scale. Improved UPDRS scores in each of the eight patients after switching the DBS stimulator on are consistent with placement of at least one DBS electrode contact in the STN. Patient
Age/yrs & sex
Predominant symptoms of PD
Disease duration/yrs
Pre-op motor UPDRS (OFF/ON Levodopa)
Post-op motor UPDRS (ON stimulation & OFF Levodopa/ON stimulation & ON Levodopa)
Medications (daily dose at time of operation)
1
55 M
Rigidity, bradykinesia
15
54/10
35/8
2
64 F
Freezing, difficulty walking
24
57/23
44/20
3
56 F
Rigidity, freezing, dyskinesia
15
71/18
32/11
4
54 M
Dyskinesias
7
29/10
25/7
5 6
58 F 48 M
Freezing, rigidity, dyskinesia Freezing ‘off’
8 12
36/10 72/16
11/7 25/11
7
54 M
Bradykinesia, gait freezing
8
38/9
21/11
8
54 M
Tremor, gait difficulties.
5
43/5
26/13
Levodopa 1100 mg Ropinirole 18 mg Selegiline 10 mg Amantadine 100 mg Levodopa 200 mg Ropinirole 3 mg Levodopa 250 mg Pramipexole 2.8 mg Levodopa 1000 mg Cabergoline 4 mg Cabergoline 2 mg Levodopa 1300 mg Rasagiline 1 mg Entacapone 800 mg Levodopa 800 mg Selegiline 10 mg Ropinirole 12 mg
A. Anzak et al. / Experimental Neurology 232 (2011) 136–142
measures of VF processing obtained were clustering (the production of words within semantic or phonemic sub-categories) and switching (the ability to shift between clusters). Scoring rules were based on Troyer et al. (1997). Cluster size was calculated starting with the second word in each cluster. Separate mean cluster size was consequently derived for the semantic and phonemic tests. Switching measures were derived as the number of transitions between clusters, including single words, for the phonemic and semantic tests separately.
A
30
left STN contact pair1-2 right STN contact pair1-2
Power/a.u.
138
20
10
LFP recordings
All analyses were performed off-line using Spike2 version 6. Data were interpolated to a common sampling rate of 1024 Hz, and were first visually inspected for large amplitude low frequency movement related artefact, which was subsequently removed. The bipolar nature of our recordings, and use of control task recordings with speech as a baseline, made any induced high frequency vibration-related artefact from the level of the voicebox unlikely in our results. For each patient, the individual trials for each task were collated (using the MSPLICE script in Spike2), thus excluding any rest periods, to form purely ‘VF’ and ‘Control’ condition files. Power spectra were subsequently calculated with 1 Hz resolution using the Fast Fourier transform for blocks of 1024 data points. These were estimated for VF and Control condition files, as well as one Rest file of ~60 second duration for each patient. LFP Power Spectra for each patient were exported to Microsoft Excel for calculation of the Average % Change in Power during the VF tasks from a baseline of the Control task at each frequency bin (Average % Change in Power = (VF − Control/Control) × 100), and subsequent collation across patients. Graphs representing the Average % Change in Power during the processing of each task were plotted (Fig. 1). These parametric tests were performed subsequent to confirmation of a normal distribution at each frequency bin across patients (one-sample Kolmogorov–Smirnov tests, p N 0.05). Repeated-measures analyses of variance (ANOVA) were performed in SPSS, using the General Linear Model procedure. Where necessary Greenhouse–Geisser corrections were applied. Means are presented in the text as Mean ± Standard Error of the Mean (SEM), unless otherwise specified.
0
20
40
60
80
100
80
100
80
100
Frequency/Hz
B
Power/a.u.
LFP analysis and statistics
0
30
20
10
0
0
20
40
60
Frequency/Hz
C 300 250
% Change in Power
LFP recordings were made 3–6 days after surgery, with the patients on medication while performing VF and control word repetition tasks. All recordings started within 1 h of ingestion of the last dose of each patient's usual anti-parkinsonian medication, when patients were subjectively and clinically in the ‘on’ state. LFPs were recorded bipolarly from adjacent contacts of each DBS electrode (0–1, 1–2, 2–3) in all cases. LFPs were band-pass filtered at 1–300 Hz using a D360 amplifier (Digitimer Ltd, Welwyn Garden City, Hertfordshire, UK), sampled at 1000 Hz and recorded through a 1401 A-D converter (Cambridge Electronic Design, Cambridge, UK) onto a computer using Spike2 software (Cambridge Electronic Design).
200 150 100 50 0 -50
0
20
40
60
Frequency/Hz
Fig. 1. Examples of LFP Power Spectra in Patient 8. (A) Phonemic verbal fluency task (B) control word repetition task, derived from differential activity between the left STN contact pair 1–2 (pink traces) and the right STN contact pair 1–2 (black traces), which recorded the highest gamma Powers. (C) Average % change in Power during the phonemic verbal fluency task relative to the control word repetition task baseline. A clear increase in Power is seen across a broad 30–95 Hz band. Note the feature at ~ 80 Hz is peculiar only to this patient in the cohort.
Results LFP Power increases in the gamma frequency band common to both VF Tasks In an initial analysis of LFPs derived as an average of recordings across all contact pairs on each electrode, significant changes in LFP Power reactivity (p ≤ 0.05, one sample T-test from 0% change) were seen across a broad gamma frequency band (30–95 Hz, excluding a 45–55 Hz range potentially contaminated by the notch filter used to
remove European line-noise) in both semantic and phonemic versions of the VF task (Fig. 2). Increases in gamma band LFP Power of 7.5% ± 2.3% (SEM) in the semantic fluency task and 6.9% ± 2.0% in the phonemic fluency task were observed when averaged across all STN electrode contact pairs. When only the STN contact pair affording the highest gamma Power for each subject at rest was used for analysis, maximal increases in Power of 8.2% ± 2.8% in the semantic fluency task, and 10.1% ± 2.7% in the phonemic fluency task, were observed (Fig. 3).
A. Anzak et al. / Experimental Neurology 232 (2011) 136–142
A
significant (F[1.00,15.00] = 14.09, p = 0.002), but the effects of TASK TYPE (F[1.00,15.00] = 1.81, p = 0.199), and the TASK TYPE × CONDITION interaction (F[1.00,15.00] = 0.47, p = 0.503) were not significant, confirming that the gamma band Power increase was similar during the two VF tasks. The results were similar when only data from the contact pair with the highest gamma power was analysed on each side.
% Change in LFP Power
25
15
5 0
Frequency specificity of LFP Power changes 10
20
30
40
50
60
-5
70
80
90
100
Frequency/Hz
-15
B % Change in LFP Power
25
15
5 0
10
20
30
40
-5
50
60
70
80
90
100
Frequency/Hz
-15 Fig. 2. Average % change in LFP Power during verbal fluency tasks from a control task baseline for (A) semantic fluency tasks (n = 16) and (B) phonemic fluency tasks (n = 16). Black filled as opposed to empty data points indicate those frequencies at which changes in LFP Power from baseline were significant, after application of one sample T-tests, from 0% change p ≤ 0.05 (see Methods). The peaks observed at ~ 80 Hz result solely from the LFP recordings of Patient 8, which skewed the distribution across these frequency bins so that one sample T-tests could not be applied. Power changes at ~ 50 Hz are likely to be contaminated by line noise. Increases in Power observed at frequencies below 8 Hz were excluded from further analysis as they may have been contaminated by movement artefact. Note the presence of a 50 Hz notch filter.
Average % Change in Power
Next, we sought to determine whether the pattern of increases in LFP Power observed across the broad gamma band was dependent on the type of VF task. We hypothesised that there should be no such effect if LFP changes were related to processing common to both semantic and phonemic versions of the VF task. An ANOVA was performed with factors TASK TYPE (semantic versus phonemic VF) and CONDITION (VF vs Control) for LFPs averaged in the 30–95 Hz gamma band (excluding 45–55 Hz potentially artefact contaminated range) across all STN contact pairs. The effect of CONDITION was 15 10
All contact pairs Best contact pair
5 0 -5 -10
139
Beta
Gamma
Beta
Gamma
-15 Semantic Fluency Task
Phonemic Fluency Task
Fig. 3. Average % change in LFP Power across the Beta (13–30 Hz) and Gamma (30– 95 Hz) frequency ranges for semantic (n = 16) and phonemic (n = 16) fluency tasks. Values are derived as an average of recordings from (1) all contact pairs and (2) ‘best’ estimates from contact pairs affording the highest gamma Powers at rest. Error bars = SEM.
The above findings pertain to changes in the gamma band. There was however a significant suppression of Power over the beta (13– 30 Hz) band in the semantic fluency task, which was less apparent in the phonemic fluency tasks (Figs. 2 and 3). Changes in the beta and gamma bands were subsequently compared, to establish whether changes in both were significant and whether contrasting reactivity existed. A repeated measures ANOVA was applied to LFP Power derived as an average across recordings from all three contact pairs, with factors TASK TYPE (semantic vs phonemic fluency), CONDITION (VF vs Control) and FREQUENCY (13–30 Hz Beta frequencies vs 30–95 Hz Gamma frequencies). As expected there was a significant effect of CONDITION (F[1.00,15.00] = 5.83, p = 0.029) but no effect of TASK TYPE (F[1.00,15.00] = 1.52, p = 0.237). While a significant effect of FREQUENCY was not seen (F[1.00,15.00] = 1.88, p = 0.191), the TASK × FREQUENCY (F[1.00,15.00] = 4.58, p = 0.049), TASK × CONDITION (F[1.00,15.00] = 17.79, p = 0.001) and TASK × FREQUENCY × CONDITION (F[1.00,15.00] = 8.58, p = 0.010) interactions were all significant, indicating a difference in the profile of changes in beta and gamma bands during VF and Control conditions, between semantic and phonemic tasks. The FREQUENCY × CONDITION interaction (F[1.00,15.00] = 1.98, p = 0.180) was not significant. Subsequent post-hoc analysis with paired T-tests revealed a difference between LFP Powers recorded in the VF and Control conditions of the semantic fluency task (p = 0.007), but not the phonemic task (p = 0.198) over the 13–30 Hz beta frequency range. Association between verbal fluency measures and LFP Power change Age-corrected normative data from the Delis–Kaplan manual was used to obtain scaled scores for the total number of words generated in the VF tasks. The average scaled scores on the 6 trials of each of the semantic and phonemic VF tasks for each of the 8 patients are shown in Fig. 4. The average scaled score across our sample of PD patients for the phonemic fluency task was 13.1 ± 5.5 (SD) and the semantic fluency task was 8.3 ± 4.1 (SD). These values were compared to the preoperative mean scaled scores for a sample of 40 PD patients assessed prior to DBS surgery at the NHNN (mean age 56.9 ± 7.9 (SD), 27 male, 13 female) from which our sample of 8 patients was drawn. Phonemic scores were similar (13.3 ± 4.0 (SD), p = 0.90, two sample T-test), but semantic scores were significantly lower in our study sample as compared to the control PD group (11.2 ± 3.4 (SD), p = 0.03). The difference may be reflective of a post-operative stun effect. In addition, for our 8 subjects, a mean cluster size ratio of 1.7 ± 0.84 (SD) was observed for phonemic fluency tasks and 1.4 ± 1.67 (SD) for the semantic fluency tasks. The mean switching score was 10.5 ± 3.5 (SD) for phonemic fluency and 12.4 ± 6.7 (SD) for the semantic fluency tasks. Following confirmation of a Gaussian distribution in both task scores and change in LFP Power averaged across each frequency band (Kolmogorov–Smirnov test, p N 0.05), Pearson correlation coefficients were calculated. For semantic fluency there was a significant correlation between average number of correct responses scaled score in interval 1 (first 15 s of each trial, during which VF output is at its best) and average gamma Power across contacts (r= 0.85, p = 0.0008). Interestingly average gamma Power in only left-sided contacts, considered to be the dominant hemisphere in VF tasks, also correlated significantly with
140
A. Anzak et al. / Experimental Neurology 232 (2011) 136–142 Semantic Fluency
20
28% Contact Pair 23
16
100
14
90
12 10 8 6 4 2 0 1
2
3
4
5
6
7
8
% of Maximum Power
Age-scaled score
A
Phonemic Fluency
18
80 70 60 50 40 30 20 10
Patient Number
0
both average scaled score of correct words generated in interval 1 (r= 0.93, p = 0.0007), and the average total correct words generated across the 60 second trial (r= 0.81, p = 0.015). There was no such correlation with right sided contacts. For phonemic fluency the correlation between number correct in interval 1 and average gamma Power across contacts was of a sizeable magnitude but did not reach significance (r = 0.62, p = 0.10). Also of interest was a significant correlation between left-sided gamma Power and average switching in the Semantic Fluency Task (r = 0.79, p = 0.020). For the phonemic task, the correlation with the switching measure was of a sizeable magnitude but only approached significance (r = 0.66, p = 0.07). No significant correlations were found between the LFPs and measures of clustering, or changes in the beta band and measures of VF.
01
12
23
Contact Pair
B 69% Contact Pair 12 100 90
% of Maximum Power
Fig. 4. Age-corrected verbal fluency scaled scores for each patient. Lower scores represent poorer performance. Error bars = Standard Error of Mean. Age scaled scores for patients 4, 7 and 8 were similar for both trials of the phonemic fluency condition, thus error bars could not be derived. While the semantic VF scaled scores of patients 5 and 7 were low as compared to the pre-operative mean scaled scores for a sample of 40 PD patients assessed prior to DBS surgery (see Results), collation of VF scores for our sample did not result in exclusion of any patient on the basis of potentially sub-optimal performance (VF Score b (Mean-1.96*SD), after confirmation of Gaussian Distribution, Kolmogorov–Smirnov Test, p N 0.05).
80 70 60 50 40 30 20 10 0 01
12
23
Contact Pair
C 3% Contact Pair 01 100
Evidence for locally generated activity
% of Maximum Power
Analysis of post-operative imaging confirmed that out of 16 STNs from which the recordings were taken, all contacts (0,1,2 & 3) lay within or touching the STN in 12 cases. In 2 cases, 2 out of the 4 contacts lay within or touching the STN, in 1 case only 1 contact lay touching the STN. In one further case the contacts were found to be positioned outside but close to the STN in the following distribution : contact 0 caudal to the inferior of the STN, contacts 1 and 2 medial to the central segment of the STN, and contact 3 rostral to the superior segment of the STN. Whether LFP oscillations reflected activity in local generators or volume conduction from distant, possibly cortical sources, was investigated. To address this point the maximum gamma Power, during the active task, among the three contact pairs from each electrode was defined as +100% and the amplitude of the two remaining contact pairs normalised to that maximum. The mean gradient of gamma Power at the remaining contact pairs was 58.7% ± 2.9%. The contact pairs with the highest amplitude for each electrode were distributed as follows: 28% contact pair 2–3, 69% contact pair 1–2, 3% contact pair 0–1 (Fig. 5). A similar distribution of contact pairs recording the highest beta Power was found: 19% contact pair 2–3, 75% contact pair 1–2, and 6% contact pair 0–1. The mean gradient of beta Power at the remaining contact pairs was 55.7% ± 4.7%. The electrode positioning strategy was such that the inferior two contacts (0,1) lay closer to the deeper, non-motor, part of the STN, while the superior two contacts (2,3) lay closer to the upper sensorimotor part of the STN and zona
90 80 70 60 50 40 30 20 10 0 01
12
23
Contact Pair Fig. 5. Distribution of maximal gamma Power amongst the three contact pairs of the DBS electrode, expressed as a percentage of individual maximum gamma Power. (A) DBS electrode site with maximum gamma Power at contact pair 2–3. (B) DBS electrode site with maximum gamma Power at contact pair 1–2. (C) DBS electrode site with maximum gamma Power at contact pair 0–1. Each line is representative of recordings from each STN of 8 patients during the Active trials of semantic (n = 16) and phonemic (n = 16) fluency tasks (32 lines).
incerta. Thus our measurement of the greatest change or differential activity in the gamma band lying between contacts 1–2, is in line with previous reports of localisation of gamma Power to the upper STN (Trottenberg et al., 2006).
A. Anzak et al. / Experimental Neurology 232 (2011) 136–142
Discussion The principal finding of this study was a significant increase in STN LFP Power across a broad gamma frequency band (30–95 Hz), during performance of phonemic and semantic VF tasks. Decreases in LFP Power across a beta (13–30 Hz) range were also seen for both VF tasks which was significant for semantic VF. LFP Power change in the STN was significantly associated with measures of performance, particularly the number of words generated and switching. These LFP neural signatures for VF were identified after controlling for motor components of verbal response generation. The results demonstrate, for the first time, that STN LFP Power is modulated during semantic and phonemic VF. Before proceeding to a further discussion of our results, it is useful to consider the reliability of the conclusions drawn from the LFP recordings. First, the bipolar recordings made across adjacent contacts of the DBS electrodes in our study, representing differential activity across contact pairs, meant that the signals analysed were most likely the product of locally generated activity in the STN rather than reflecting volume conduction from synchronous cortical areas. Further evidence for a focal origin of the signals comes from the clear gradient of LFP Power across the contact pairs (Fig. 5), since in volume conduction, one would expect LFP Power to be equally distributed between adjacent contact pairs of the electrode, or present a decrement at contacts more distant from the cortical source (Williams et al., 2003; Doyle et al., 2005). Finally, that LFP oscillations are truly reflective of neuronal activity in the STN has consistently been demonstrated by synchrony between these two variables in both anaesthetized rats (Magill et al., 2004) and alert Parkinsonian patients (Levy et al., 2002). While our eight post-operative patients performed worse in the semantic fluency task than a PD control group assessed prior to surgery, the similarities in STN gamma activity between this task and the better performed phonemic fluency task suggest an intact involvement of the STN in processing related to a feature common to the two tasks. On the other hand, the small but significant desynchronisation in STN beta Power observed in the semantic fluency but not phonemic fluency task (see Fig. 3), may be reflective of a further role of the STN in processing specific to success in semantic VF. One possibility is a difference in task demand for recruitment of executive processes. For example, while both phonemic and semantic VF require effortful retrieval of information from semantic memory, it is likely that due to differences in depth of encoding, retrieval of semantically associated items involves additional retrieval mechanisms during semantic VF (Henry and Crawford, 2004). What common process(es) of VF tasks might the STN thus be involved in? Certainly, the impairment in VF is not a manifestation of a global cognitive decline, as general cognitive ability is widely reported to remain intact after DBS of the STN (Parsons et al., 2006; Zahodne et al., 2009). Evidence for a correlation between left sided gamma Power and switching scores for both the semantic and phonemic tasks does however suggest a potential role of the STN in the process of disengaging from an exhausted semantic or phonemic sub-category and switching to a new sub-category, in order to optimise the number of words generated. This is in agreement with the evidence of decline in switching but not cluster size during VF tasks with high frequency stimulation of the STN (Saint-Cyr and Trepanier, 2000; De Gaspari et al., 2006). Correlations with left sided activity rather than right are also consistent with the linguistic nature of VF tasks and research describing an association between left frontal–subcortical neurons and VF (Baldo et al., 2001), and imaging evidence of activation of the left frontal areas in VF tasks (Frith et al., 1991) and, importantly, left frontal systems in cognitive set switching (Dove et al., 2000). Involvement of the STN in the processing of VF tasks can be placed in the context of a larger cortical network model for willed or intrinsic generation of words, validated with Positron Emission Tomography
141
(PET) in normal subjects. The model (Friston et al., 1991; Frith et al., 1991) posits that successful word generation in VF tasks is achieved through a modulatory influence of the dorsolateral prefrontal cortex (DLPFC) over a distributed word associative network in the superior temporal cortex (STC), to prevent spreading activation in the network and to allow strategic generation of words which meet the VF criteria. For example, during phonemic VF when generating words beginning with ‘F’, having generated the word ‘farm’ the participant has to suppress highly semantically associated words such as ‘cow’ or ‘milk’ and move on to generate other words beginning with the letter ‘F’. PET studies have shown that stimulation of the STN affects just such fronto-temporal networks, during VF in PD (Schroeder et al., 2003; Cilia et al., 2007). Similarly, using 18-fluorodeoxyglucose (FDG) PET, Kalbe et al. (2009) reported that FDG uptake in the left DLPFC, left Broca's area and right anterior cingulate was significantly associated with VF scores after STN DBS surgery. In line with the above network modulation model of VF, the STN could be considered to implement a ‘no go’ signal to prevent premature or inappropriate responses (Frank et al., 2007), drawing parallels to its posited role in motor response inhibition (Baunez et al., 1995; Kuhn et al., 2004; Ray et al., 2009). This is consistent with a centre surround inhibition model (Mink and Thach, 1993) in which the STN could ultimately inhibit potentially inappropriate responses and facilitate selection of the appropriate response (Redgrave et al., 1999). It is plausible that left DLPFC initiated activity may exert an inhibitory influence on inappropriate spreading activation in the left STC word association network via STN-mediated inhibition of thalamic output to temporal cortex, thus allowing only words appropriate to the task to be selected. Indeed, the STN receives direct input from the DLPFC (Nambu et al., 2002) and has indirect connections to the temporal cortex via the GPi and thalamus (Middleton and Strick, 2000). As such, the nature of the response inhibition required in VF tasks may involve a more tonic and internally driven form of inhibition that may be somewhat distinct from the more phasic, trial by trial, and cued suppression or withholding of responses in previous motor paradigms (Kuhn et al., 2004). A related proposal is the possibility that the STN receives and implements a switching signal from the frontal cortex to allow a shift from automatic selection of a highly associated but inappropriate word to the controlled selection of a word that meets the task criteria. Such a role for the STN in concert with the pre-SMA in switching between automatic and controlled processing has been shown in neuronal recordings in primates (Isoda and Hikosaka, 2008) and in a recent imaging study of decision-making in humans, where the STN together with the inferior frontal cortex were engaged when switches from the default or status quo decision were required (Fleming et al., 2010). To conclude, we provide evidence, for the first time, of task-related power changes across gamma and beta frequencies in the STN during VF, indicating involvement of the STN in processing related to these cognitive tasks. Frequency specific involvement of the STN in VF may provide a viable explanation for the deficits reported with DBS, through superimposition of high frequency stimulation on physiologically normal task-related oscillatory activity. Our results further suggest a role for the STN in inhibition of inappropriate responses or switching during semantic and phonemic VF. Finally, as a characteristic of untreated PD is a suppression of gamma activity and an enhancement of beta oscillatory activity in the STN (Dostrovsky and Bergman, 2004), this pattern would antagonise the task-related changes found here and might explain the general impairment in VF seen in PD patients (Donovan et al., 1999; Henry and Crawford, 2004). Acknowledgments We would like to thank the patients for their participation. Funding: MB is supported by a 3 + 1 ESRC/MRC studentship. PB is supported by the MRC and the NIHR Biomedical Research Centre, Oxford. PL, LZ, MH and TF are supported by the Parkinson's Appeal.
142
A. Anzak et al. / Experimental Neurology 232 (2011) 136–142
References Ardouin, C., Pillon, B., Peiffer, E., et al., 1999. Bilateral subthalamic or pallidal stimulation for Parkinson's disease affects neither memory nor executive functions: a consecutive series of 62 patients. Ann. Neurol. 46, 217–223. Baldo, J.V., Shimamura, A.P., Delis, D.C., et al., 2001. Verbal and design fluency in patients with frontal lobe lesions. J. Int. Neuropsychol. Soc. 7, 586–596. Baunez, C., Nieoullon, A., Amalric, M., 1995. In a rat model of parkinsonism, lesions of the subthalamic nucleus reverse increases of reaction-time but induce a dramatic premature responding deficit. J. Neurosci. 15, 6531–6541. Bejjani, B.P., Dormont, D., Pidoux, B., et al., 2000. Bilateral subthalamic stimulation for Parkinson's disease by using three-dimensional stereotactic magnetic resonance imaging and electrophysiological guidance. J. Neurosurg. 92, 615–625. Brown, P., Eusebio, A., 2008. Paradoxes of functional neurosurgery: clues from basal ganglia recordings. Mov. Disord. 23, 12–20. Cilia, R., Siri, C., Marotta, G., et al., 2007. Brain networks underlining verbal fluency decline during STN-DBS in Parkinson's disease: an ECD-SPECT study. Parkinsonism Relat. Disord. 13, 290–294. Contarino, M.F., Daniele, A., Sibilia, A.H., et al., 2007. Cognitive outcome 5 years after bilateral chronic stimulation of subthalamic nucleus in patients with Parkinson's disease. J. Neurol. Neurosurg. Psychiatry 78, 248–252. De Gaspari, D., Siri, C., Di Gioia, M., et al., 2006. Clinical correlates and cognitive underpinnings of verbal fluency impairment after chronic subthalamic stimulation in Parkinson's disease. Parkinsonism Relat. Disord. 12, 289–295. Delis, D.C., Kaplan, E., Kramer, J.H., 2001. Delis-Kaplan Executive Function System (D-KEFS). The Psychological Corporation, San Antonio, TX. Donovan, K., Siegert, R., McDowall, J., et al., 1999. Clustering and switching in verbal fluency in Parkinson's disease. N. Z. J. Psychol. 28, 61–66. Dostrovsky, J., Bergman, H., 2004. Oscillatory activity in the basal ganglia — relationship to normal physiology and pathophysiology. Brain 127, 721–722. Dove, A., Pollmann, S., Schubert, T., et al., 2000. Prefrontal cortex activation in task switching: an event-related fMRI study. Brain Res. Cogn. Brain Res. 9, 103–109. Doyle, L.M.F., Kuhn, A.A., Hariz, M., et al., 2005. Levodopa-induced modulation of subthalamic beta oscillations during self-paced movements in patients with Parkinson's disease. Eur. J. Neurosci. 21, 1403–1412. Fasano, A., Romito, L.M., Daniele, A., et al., 2010. Motor and cognitive outcome in patients with Parkinson's disease 8 years after subthalamic implants. Brain 133, 2664–2676. Fleming, S.M., Thomas, C.L., Dolan, R.J., 2010. Overcoming status quo bias in the human brain Proceedings of the national academy of sciences. Proc. Natl. Acad. Sci. U.S.A. 107 (13), 6005–6009. Foltynie, T., Hariz, M.I., 2010. Surgical management of Parkinson's disease. Expert Rev. Neurother. 10, 903–914. Frank, M.J., Samanta, J., Moustafa, A.A., et al., 2007. Hold your horses: impulsivity, deep brain stimulation, and medication in parkinsonism. Science 318, 1309–1312. Friston, K.J., Frith, C.D., Liddle, P.F., Frackowiak, R.S., 1991. Investigating a network model of word generation with positron emission tomography. Proc. Biol. Sci. 244, 101–106. Frith, C.D., Friston, K.J., Liddle, P.F., et al., 1991. A PET study of word finding. Neuropsychologia 29, 1137–1148. Hariz, M.I., Krack, P., Melvill, R., et al., 2003. A quick and universal method for stereotactic visualization of the subthalamic nucleus before and after implantation of deep brain stimulation electrodes. Stereotact. Funct. Neurosurg. 80, 96–101. Henry, J.D., Crawford, J.R., 2004. A meta-analytic review of verbal fluency performance following focal cortical lesions. Neuropsychology 18, 284–295. Isoda, M., Hikosaka, O., 2008. Role for subthalamic nucleus neurons in switching from automatic to controlled eye movement. J. Neurosci. 28, 7209–7218. Kalbe, E., Voges, J., Weber, T., et al., 2009. Frontal FDG-PET activity correlates with cognitive outcome after STN-DBS in Parkinson disease. Neurology 72, 42–49.
Kuhn, A.A., Williams, D., Kupsch, A., et al., 2004. Event-related beta desynchronization in human subthalamic nucleus correlates with motor performance. Brain 127, 735–746. Levy, R., Hutchison, W.D., Lozano, A.M., et al., 2002. Synchronized neuronal discharge in the basal ganglia of parkinsonian patients is limited to oscillatory activity. J. Neurosci. 22, 2855–2861. Limousin, P., Pollak, P., Benazzouz, A., et al., 1995. Bilateral subthalamic nucleus stimulation for severe Parkinsons-disease. Mov. Disord. 10, 672–674. Magill, P.J., Sharott, A., Bevan, M.D., et al., 2004. Synchronous unit activity and local field potentials evoked in the subthalamic nucleus by cortical stimulation. J. Neurophysiol. 92, 700–714. Marsden, C.D., Obeso, J.A., 1994. The functions of the basal ganglia and the paradox of stereotaxic surgery in Parkinson's disease. Brain 117, 877–897. Middleton, F.A., Strick, P.L., 2000. Basal ganglia output and cognition: evidence from anatomical, behavioral, and clinical studies. Brain Cogn. 42, 183–200. Mink, J.W., Thach, W.T., 1993. Basal ganglia intrinsic circuits and their role in behavior. Curr. Opin. Neurobiol. 3, 950–957. Nambu, A., Tokuno, H., Takada, M., 2002. Functional significance of the corticosubthalamo-pallidal ‘hyperdirect’ pathway. Neurosci. Res. 43, 111–117. Parsons, T.D., Rogers, S.A., Braaten, A.J., et al., 2006. Cognitive sequelae of subthalamic nucleus deep brain stimulation in Parkinson's disease: a meta-analysis. Lancet Neurol. 5, 578–588. Ray, N.J., Jenkinson, N., Brittain, J., et al., 2009. The role of the subthalamic nucleus in response inhibition: evidence from deep brain stimulation for Parkinson's disease. Neuropsychologia 47, 2828–2834. Redgrave, P., Prescott, T.J., Gurney, K., 1999. The basal ganglia: a vertebrate solution to the selection problem? Neuroscience 89, 1009–1023. Saint-Cyr, J.A., Trepanier, L.L., 2000. Neuropsychologic assessment of patients for movement disorder surgery. Mov. Disord. 15, 771–783. Schroeder, U., Kuehler, A., Lange, K.W., et al., 2003. Subthalamic nucleus stimulation affects a frontotemporal network: a PET study. Ann. Neurol. 54, 445–450. Smeding, H.M.M., Speelman, J.D., Koning-Haanstra, M., et al., 2006. Neuropsychological effects of bilateral STN stimulation in Parkinson disease — a controlled study. Neurology 66, 1830–1836. Trottenberg, T., Fogelson, N., Kühn, A.A., et al., 2006. Subthalamic gamma activity in patients with Parkinson's disease. Exp. Neurol. 200, 56–65. Troyer, A.K., Moscovitch, M., Winocur, G., 1997. Clustering and switching as two components of verbal fluency: evidence from younger and older healthy adults. Neuropsychology 11, 138–146. Weaver, F.M., Follett, K., Stern, M., et al., 2009. Bilateral deep brain stimulation vs best medical therapy for patients with advanced Parkinson disease a randomized controlled trial. Jama-J. Am. Med. Assoc. 301, 63–73. Williams, D., Kuhn, A., Kupsch, A., et al., 2003. Behavioural cues are associated with modulations of synchronous oscillations in the human subthalamic nucleus. Brain 126, 1975–1985. Williams, D., Kuhn, A., Kupsch, A., et al., 2005. The relationship between oscillatory activity and motor reaction time in the parkinsonian subthalamic nucleus. Eur. J. Neurosci. 21, 249–258. Witt, K., Daniels, C., Reiff, J., et al., 2008. Neuropsychological and psychiatric changes after deep brain stimulation for Parkinson's disease: a randomised, multicentre study. Lancet Neurol. 7, 605–614. York, M.K., Wilde, E.A., Simpson, R., et al., 2009. Relationship between neuropsychological outcome and DBS surgical trajectory and electrode location. J. Neurol. Sci. 287, 159–171. Zahodne, L.B., Okun, M.S., Foote, K.D., et al., 2009. Cognitive declines one year after unilateral deep brain stimulation surgery in Parkinson's disease: a controlled study using reliable change. Clin. Neuropsychol. 23, 385–405. Zangaglia, R., Pacchetti, C., Pasotti, C., et al., 2009. Deep Brain stimulation and cognitive functions in Parkinson's disease: a three-year controlled study. Mov. Disord. 24, 1621–1628.