NEURAL CORRELATES OF IMPULSE CONTROL IN PARKINSONS DISEASE By PETER JUSTIN ROSSI A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2016
2016 Peter Justin Rossi
To Peter and Patricia Rossi
4 ACKNOWLEDGMENTS Although this short note scarcely does justice, I wish to thank those whose couns el and support made this project possible. My first debt is to Dr. Michael Okun, my advisor and teacher, who patiently shepherded my progress and gave a sense of direction to this undertaking. No student could have asked for a more devoted mentor. I am especially grateful to Dr. Aysegul Gunduz who come ntored me during this endeavor. H er contribution of time, energy, and insights were indispensable to the success of this project. I am also grateful to Dr. Habibeh Khoshbouei Dr. Dawn Bowers and Dr. Christo pher Hess for their contributions as members of my supervisory committee. I owe particular gratitude to my laboratory colleagues, Jonathan Shute, Rene Molina, and Enrico Opri, whose companionship and diligent efforts on my behalf were much appreciated. I extend my gratitude to Oscar Castellanos, M.S., and Corinna Peden, B.A., for their indispensable assistance in various stages of the project. Finally, I thank Dr. James Giordano for first inspiring my passion for neuroscience. I extend heartfelt thanks to the generous funding institutions that made possible the research contained in this manuscript. This project was funded by a fellowship from the National Institutes of Health T32 Neuromuscular Plasticity Training Program and the University of Florida Depar tment of Health Outcomes and Policy. Finally, and most importantly, I must thank my parents, to whose care, dedication, and constant attention my studies here and from my earliest education are most indebted. To my father, whose passion for medicine f irst sparked my interest in the subject, and to my mother, who followed my progress i n this undertaking with so much patience and diligence, this thesis is dedicated as a small token of gratitude.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................. 4 LIST OF FIGURES .......................................................................................................... 7 LIST OF ABBREVIATIONS ............................................................................................. 9 ABSTRACT ................................................................................................................... 10 CHAPTER 1 THE STN AND GPI DIFFERENTIALLY ENCODE VALENCE ................................ 12 Introduction ............................................................................................................. 12 Materials and Methods ............................................................................................ 16 Patients ............................................................................................................ 16 Behavioral Task ................................................................................................ 17 Localization of Recording Sites ........................................................................ 18 Data Acquisition ............................................................................................... 18 Recording Protocol ........................................................................................... 19 Behavioral Data Analysis .................................................................................. 19 Electrophysiological Analysis ........................................................................... 20 Recording Site Mapping ................................................................................... 22 Behavioral Results .................................................................................................. 23 Reaction Time .................................................................................................. 23 Response Accuracy ......................................................................................... 23 Electrophysiological Results: Valence Processing in the STN ................................ 24 Responses to Reward Opportunity and Threat of Loss .................................... 24 Responses to Reward and Loss Avoidance ..................................................... 25 Movement related Activity ................................................................................ 26 Errorrelated Activity ......................................................................................... 27 Exclusive Correct Neurons ............................................................................... 28 Neuronal Mapping ............................................................................................ 28 Electrophysiological Results: Valence Processing in the GPi ................................. 28 Responses to Reward Opportunity and Threat of Loss .................................... 29 Responses to Reward and Loss Avoidance ..................................................... 29 Movement related Activity ................................................................................ 30 Errorrelated Activity ......................................................................................... 31 Neuronal Mapping ............................................................................................ 32 Discussion: Comparing STN and GPi ..................................................................... 32 Behavioral Data ................................................................................................ 33 Electrophysiological Data ................................................................................. 35 Neuronal Mapping ............................................................................................ 37 Conclusion ........................................................................................................ 40
6 2 IMPULSIVITY AND DIFFERENCES IN STN NEURON BEHAVIOR ...................... 57 Introduction ............................................................................................................. 57 Materials and Methods ............................................................................................ 59 Patients ............................................................................................................ 59 Assessment of Impulse Control Disorders ........................................................ 60 Results .................................................................................................................... 61 Patient Characteristics ...................................................................................... 61 Subthalamic Nucleus ........................................................................................ 61 Globus Pallidus internus ................................................................................... 62 Discussion: Comparing STN and GPi ..................................................................... 63 3 DEEP BRAIN STIMULATION MODIFIES IMPULSE CONTROL ............................ 69 Introduction ............................................................................................................. 69 Materials and Methods ............................................................................................ 70 Patients ............................................................................................................ 70 Assessment of Impulse Control Disorders ........................................................ 71 Statistical Analysis ............................................................................................ 72 Results .................................................................................................................... 72 ICD Prevalence ................................................................................................ 72 Rate of Change in ICD Status Post DBS .......................................................... 73 Qualitative Changes in Impulse Control Post DBS ........................................... 74 Changes in ICD Status and Dopaminergic Medication ..................................... 74 Other Psychological Changes .......................................................................... 76 Discussion .............................................................................................................. 76 LIST OF REFERENCES ............................................................................................... 86 BIOGRAPHICAL SKETCH ............................................................................................ 96
7 LIST OF FIGURES Figure page 1 1 Behavioral t ask ................................................................................................... 41 1 2 Task p erformance ............................................................................................... 42 1 3 Neuronal and waveform characterization ........................................................... 43 1 4 Valence r esponsiveness in STN n eurons ........................................................... 44 1 5 Excitation and i nhibition at s timulus p resentation in STN n eurons ..................... 45 1 6 Excitation and i nhibition at f eedback presentation in STN n eurons .................... 46 1 7 STN m ovement related a ctivity ........................................................................... 47 1 8 Error sensitive n eurons in the STN ..................................................................... 48 1 9 Positions of r ecorded STN n eurons .................................................................... 49 1 10 Anatomical coordinates of r esponsive p opulations ............................................. 50 1 11 Valence r esponsiveness in GPi n eurons ............................................................ 51 1 12 Excitation and i nhibition at s timulus p resentation in GPi n eurons ....................... 52 1 13 Excitation and i nhibition at f eedback p resentation in GPi n eurons ..................... 53 1 14 GPi m ovement related a ctivity ............................................................................ 54 1 15 Error sensitive n eurons in the GPi ...................................................................... 55 1 16 Positions of r ecorded GPi n eurons ..................................................................... 56 2 1 ICD status impacts reward and loss sensitivity in STN but not GPi n eurons ...... 67 2 2 ICD status impacts r es ponse s pecificty in STN but not GPi n eurons ................. 68 3 1 Comparis on of p re DBS and p ost DBS p revalence of ICDs ............................... 80 3 2 ICD changes after STN and GPi DBS surgery ................................................... 81 3 3 P ost DBS ICD c hanges are not d ependent on d opamine t herapy ...................... 82 3 4 Comparison of i mpulsive and n oni mpulsive p atients ......................................... 83
8 3 5 Changes in impulsivity following GPi and STN DBS ........................................... 84 3 6 C hanges in ICD status in b ilateral versus u nilateral DBS ................................... 85
9 LIST OF ABBREVIATIONS AC BG D2 D3 DAA DDS Anterior commissure Basal ganglia Dopamine receptor D2 Dopamine receptor D3 Dopamine agonist Dopamine dysregulation syndrome GPe Globus pallidus externus GP i Globus pallidus internus ICD LEDD LFP LHb MRI PC Impulse control disorder Levodopa equivalent daily dosage Local field potential Lateral habenula Magnetic resonance imaging Posterior commissure PD PETH QUIP S D SEM Parkinsons disease Peri event time histogram Questionnaire for impulsivity in Parkinsons disease Standard deviation Standard error of the mean STN Subthalamic nucleus
10 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy NEURAL CORRELATES OF IMPULSE CONTROL IN PARKINSONS DISEASE By Peter Justin Rossi August 2016 Chair: Michael S. Okun Major: Medical Sciences Neuroscience A significant portion of Parkinsons disease (PD) patients suffer s fr om i mpulse control disorders (ICDs). A hallmark feature of ICDs is the repetitive pursuit of rewarding behaviors despite negative consequences. The pathophysiology of ICDs is not well understood. Recent evidence suggests that alterations in reward/punishme nt (i.e. valence) processing in two subcortical brain regions the subthalamic nucleus (STN) and the globus pallidus internus (GPi) may contribute to this disorder. The objective of this research was to prospectively study the effect of STN and GPi DBS on i mpulse control, investigate the role of the human STN and GPi in valence processing, and determine if ICDs are associated with aberrations in valence processing in these structures. PD patients ( 1 2 with an ICD and 31 without) undergoing DBS electrode place ment were studied. Patients performed a behavioral task in which their action choices were motivated by the potential for a reward or a loss. T he activity of individual STN or GPi neurons was simultaneously recorded. Neuronal activity was analyzed to deter mine firing rate modulation in response to task conditions. Patients were also
11 evaluated preand post operatively using clinical measures to determine the effects of DBS on impulse control. We found that STN and GPi neurons encode valencerelated information during action control, but the proportion of valenceresponsive neurons was greater in the STN compared to the GPi. In the STN, rewardrelated stimuli mobilized a greater proportion of neurons than loss related stimuli. We also found surprising limbic overlap with the sensorimotor regions in both the STN and GPi ICDs were associated with significantly greater proportions of reward responsive neurons and significantly lower proportions of loss responsive neurons in the STN, but not in the GPi. Of the I CDs present at baseline, 60% resolved post operatively ; 20% of patients experienced new onset ICDs Change in ICD status was not associated with change in dopamine agonist medication These findings support the idea that STN and GPi DBS can directly alter impulse control. C hanges in the STN possibly via functional conversion of neurons in the limbic circuit may underlie impulsive behavior in the PD population.
12 CHAPTER 1 THE STN AND GPI DIFFERENTIALLY ENCODE VALENCE Introduction The common requirement for animals to perform goal directed behaviors for survival suggests that phylogenetically older (i.e. subcortical) brain regions have retained fundamental mechanisms for rewardseeking and loss avoidant behavior1. The study of subcortical involvement in the reward system has been largely focused on the striatum, the nucleus accumbens, and the ventral pallidum2; however, roles for the subthalamic nucleus (STN) and, to a lesser degree, the globus pallidus (GP) have only recently been recognized1. The observation that lesions or high frequency stimulation of the STN or the globus pallidus internus (GPi) in some human patients were associated with changes in affective or goal directed behavior, including depression, suicidal ideation, apathy, mania, and impulsivity3, motivated much of the exploration of the potential inv olvement of these structures in the reward system. Substantial evidence from animal studies over the past decade indicates a role for the STN in processing reward3. It h as been shown that a large subset of STN neurons in the rat encodes the prospect of reward, reward receipt, the magnitude of different possible reward outcomes, and reward prediction error4 6. Darbaky and colleagues showed that neurons in the monkey STN were modulated during movement, just before reward delivery, and during reward delivery7. More recently, EspinosaParrilla and colleagues demonstrated that monkey STN neurons modulated in response to reward during task performance as well as to the expectation of reward delivery when the reward was delayed in time8. Interestingly, they also showed t hat movement related modulations were commonly combined with reward delivery modulations, suggesting a
13 convergence of signals related to the animals movement and its outcome in the same neurons8,9. Until recently, it was widely believed that the STN could be divided into three distinct functional zones: an anteromedial limbic zone, a posterior sensorimotor zone, and an overlapping associative zone betwee n the two10 12. The evidence for this socalle d tripartite hypothesis derived mainly from tracer studies13 14, although limited electrophysiological findings supported this model as well15,16. However, the notion of a tripartite division has recently been called into question17. Comprehensive reviews of the extant evidence suggest that substantial overlap exists between the putative functional zones and that the tripartite model may thus require revision13,17,18. Less is known about the putative limbic function of the GPi, as it has been most intensively studied as a purely motor nucleus. However, a few critical animal studies suggest it has a role in reward processing. T he dorsal pallidum the region corresponding to the globus pallidus externus (GPe) and GPi collectively has been shown to encode reward signals1. Hikosaka and colleagues identified a subpopulation of GPi neurons in the monkey that project to the lateral habenula (LHb) which strongly enco de reward prediction signals19, 20. Arkadir and colleagues studied single neuron activity in the GPe in monkeys while they performed rewardmotivated arm movements21. Although they identified only a small percentage of neurons (3%) that responded exclusively to reward, they found that 41% of neurons comodulated activity in response to both reward and movement. Another single unit studies in monkeys examining rewardmotivated movement specifically in the GPi found 15% (12/82) of neurons modulated with reward delivery22. Recently, functional neuroimaging in humans
14 has shown that the pallidum exhibits greater neural activation in response to visual cues predicting potential gains compared to predictors of neutral outcomes and to actual financi al gains compared neutral outcomes23. Regarding the existence of a limbic zone in GPi, immunohistochemical studies in humans implicates the anterior GPi as a recipient of limbic afferents, although, importantly, the boundaries between putative limbic and motor territories were not sharp but rather were consistent with overlap24. Further evidence of a limbic hot spot in the anterior GPi comes from clinical observations, particularly that DBS of the anterior GPi has been shown to relieve obsessive compulsive symptoms in Tourette syndrome25 and self mutilation behavior in Lesch Nyan syndrome26. The construct of reward is best understood within the broader construct of valence, which is defined as the intrinsic attractiveness or aversiveness of a given stimulus. Stimuli that are rewarding are considered to be positively valenced, while stimuli that are undesirable or aversive are considered to be negatively valenced27. Valence and action control are intimately intertwined in that organisms tend to act or inhibit action to obtain positively valenced outcomes (rewards) and avoid negatively valenced outcomes (punishments or loss). Moreover, a well documented feature of the effect of valence on action control is the socalled Pavlovian bias in which many organisms (including humans) tend to learn better to act to obtain reward (as opposed to avoid punishment) and to inhibit action to avoid punishment (as opposed to obtain reward)28 30. While such interactions between action and valence are behaviorally apparent, their neurobiological substrates are not well understood. For example, the observation that the same GPe neurons can convey information regarding both
15 movement and gain f rom that movement (predicted or attained)21 raises the question of wh ether such comodulation occurs in other BG structures implicated in both movement and reward (i.e. STN and GPi). Moreover, it is unclear if such comodulation might apply to negatively valenced motivations as well. Elucidating the neurobiological bases for these interactions could have implications for treating certain neuropsychiatric disorders, such as impulsivity, where valence perception and action control are believed to be dysfunctional3. The successful pursuit of goal directed motor behavior is also dependent on the detection of errors and the adaptive learning required to avoid future errors31. Error detection involves the perception that one has failed to obtain an expected, positively valenced outcome or has obtained an unexpected negatively valenced outcome. Error detection is thus another concept in which action and valence are intimately intertwined. The error monitoring system is complex, and it is believed to involve both cortical and subcortical structures. The observation of abnormal cortical error related activity in patients with diseases of the basal ganglia (PD, Huntingtons disease, Tourette syndrome, or focal BG lesions) implicates the BG in error monitoring and error associated motor adaptation31. Studies of single cell neural activity in the rodent STN2,6,32 and monkey dorsal pallidum1,21 show distinct error related changes in firing rate. With the exception of a single local field potential ( LFP ) study in parkinsonian humans showing error related changes in thetaband oscillations in GPi31, little is known about error responsiveness in the human STN and GPi. We are thus interested in investigating the neural encoding of valence in human STN and G Pi neurons and assessing potential neural integration of valence and action
16 selection signals within these structures. To this end, we assessed the behavior of individual neurons in the STN or GPi of Parkinsonian humans undergoing DBS lead placement surger y while they performed a cognitive task that orthogonalizes valence and action requirements such that four conditions are tested: act to obtain reward, act to avoid punishment, inhibit action to obtain reward, and inhibit action to avoid punishment (see design pioneered by Guitart Masip et al.29,33 and modified by van Wouwe et al.34). Our working hypothesis was that both STN and GPi neurons would modulate activity in response to 1) predictors of positively and negatively valenced outcomes (opportunity for reward and threat of punishment), 2) movement and inhibition, 3) reward receipt and loss avoidance, and 4) failure to obtain reward/ loss (error feedback). Given the ongoing debate about the location of limbic zones within these structures, we also sought to document the topographical positions of any valenceresponsive neurons identified. Based on the extant data discussed above, we hypothesized that valenceresponsive neurons would be concentrated toward the anterior portion of the GPi and the anteromedial portion of the STN. Materials and Methods Patients Patients with Parkinsons disease undergoing unilateral de novo STN or GPi DBS electrode placem ent surgery were studied. All patients fulfilled the Movement Disorder Society clinical diagnostic criteria for PD35. All participants were nondemented, hearing, and sighted. A total of 53 patients participated and had a mean age of 65.69 years (S D=7.85). The Institutional Review B oard approved the study and all patients provided their informed consent before entering the study.
17 Behavioral Task All stimuli were presented on a 15inch screen positioned at eye level approximately one meter from the patient. Responses were made with th e thumb contralateral to the brain hemisphere being recorded using a handheld, singlebutton joystick. In the pre operative clinic approximately 24 hours prior to surgery, subjects completed practice sessions (2 blocks) of the actionvalence learning task to associate combinations of action (action, inaction) and valence (reward acquisition, loss avoidance) to a specific stimulus color. The same actionvalence learning task was then presented to the subject intraoperatively during the microelectroderecord ing (MER) phase of the DBS operation. The details of the action valence task are as follows (Fig. 11 ). S ubjects were instructed that they had 1000 milliseconds (ms) to either act (i.e. make a button press) or to withhold action after being presented with a color patch. Button press or expiration of the 1000 ms window was followed by a 250 ms delay period (to facilitate separation of movement and feedback neuronal signals). Because delayed outcome of a choice is known to reduce the subjective value of the r eward, the shortest possible delay duration was chosen while still permitting accurate signal separation. After this delay, feedback was displayed for 500 ms in the center of the color patch indicating that the action decision had led to monetary reward (+ $100), monetary loss ( $100) or no monetary outcome ($0). Monetary rewards were accompanied by a positive contingency audio feedback simulating the jingle of coins. Monetary losses were accompanied by a negative contingency audio feedback simulating an aversive buzzer sound. The feedback and color patch were then extinguished, and an inter trial interval followed for 500ms after which the next trial began. A running total of earnings was presented in the
18 upper center of the screen throughout the task. T he four color patches appeared in pseudorandom order and with equal probability across the 120 trials per block. Thus, each color appeared 30 times within a block of trials. Unbeknownst to the subject, two of the color patches provided outcomes that were either rewarded (+$100) or unrewarded ($0), and the other two colors provided outcomes that were either punished ( $100) or unpunished ($0)34. Two colors were thus a ssociated with reward learning and the two other colors were associated with loss avoidance learning Also unknown to the subject, one color from each set produced the optimal outcome (either gain of reward or avoidance of loss) when the button was pressed (action) but the other color from each set produced the optimal outcome by not pressing the button ( withholdin g action)34. Locali zation of Recording Sites A proprietary stereotaxic surgical software system was used to visualize MRI images, estimate anatomical locations of the STN and GPi, define target location, predict trajectories for microelectrode penetrations, and map results o f microelectrode recordings36. To avoid sampling bias, cells were selected strictly on the basis of signal integrity rather than the presence or absence of sensorimotor activity37. Anti Parkinsonian medications were withheld 24 hours before neural recording, and no sedatives were given prior to or during recordi ngs37. Data Acquisition Extracellular spiking activity of subthalamic and pallidal neurons was recorded using platinum/iridium tipped microelectrodes ( F HC, Inc., Bowdoin, ME) with impedances of 0.5 mounted in a motorized, hydraulic microdrive (FHC, Inc., Bowdoin, ME). The behavioral task was administered by a Dell Optiplex 9020 computer using BCI2000 software38. Neuronal activity was bandpass filtered between 300 Hz 6
19 kHz and sampled at 48 KHz (16 bit resolution; Tucker Davis Technologies, Alachua, FL). The neuronal recording was monitored via computer display and audio speakers. Recording Protocol All recordings were performed using an array of three microelectrodes, separated by 2 mm in an Lshaped configuration, lowered transdurally into th e STN or GPi. For recordings of the STN, the microelectrode array was configured with a central contact, a second co axial contact positioned 2 mm laterally, and a third positioned 2 mm anteriorly. For recordings of the GPi, the microelectrode array was configured with a central contact, a second coaxial contact positioned 2 mm laterally, and a third positioned 2 mm posteriorly. Electrophysiological determination of STN and GPi borders was based on electrode depth along the planned trajectory, spontaneous firing rate and pattern, and kinesthetic responses39. When one (or, less frequently two) neuronal units were isolated by at least two of the recording electrodes in the array and the recording was determined to be stable, the cognitive task was initiated. Neuronal data, electrode location, and task performance data were then collected following a standard protocol of 120 stimulus presentations (30 of each of the 4 conditions) in pseudorandom order to ensure an equal number of trials of each condition. Data collection was stopped only if signal quality deteriorated. Upon completion of the task, the microelectrode was advanced so that other units could be evaluated until the microel ectrode array was determined to no longer be in the region of interest. Behavioral Data Analysis Performance on the behavioral task was assessed via methods used previously by Wagenbreth and colleagues40. Accuracy was defined by the percentage of t rials in which the patient selected the optimal response34. In brief, mean accuracy on the
20 behavioral task was analyzed utilizing twoway repeated measures ANOVA, followed by post hoc Tukey test for multiple comparisons to compare individual conditions. Mean reaction times for correct responses during the Go conditions were compared using an unpaired students t test ; p values <.05 were considered significant. Electrophysiological Analysis Amplitude thresholds for neural spike data were selected, and candidate action potential s were sorted into clusters in principal components space (Spike2, Cambridge Electronic Design, Cambridge, UK). Neurons were considered acceptable for further analysis only if their action potentials were of a consistent shape, had a clear refractory period of at least 2 ms, and could be reliably distinguished from the waveforms of other units and from background noise41. Additionally, analyzed neurons had to be associated with a complete s et of task related data (i.e. 120 stimuli trials). Peri event time histograms (PETHs) were then constructed from the correct trials of each of the 4 conditions, and for the additional conditions where incorrect responses occurred. Stimulus related changes, movement related changes, and outcome related changes in firing rate were defined as a significant deviation from the baseline inter trial interval. B aseline activity was defined as activity occurring in the 500 ms period that preceded cue presentation in each trial. Event related activity was determined in the 500 ms that followed the onset of the visual stimulus (stimulus related), the 500 ms centered on the button press (movement related), the 500 ms centered on trial expiration (inhibition related), and the 500 ms that followed feedback delivery (outcome related). Analyses were based on binned peri event firing rates (50 ms bins). For each event of the task, we generated a PETH centered on that event using Matlab
21 (Mathworks, Cambridge, MA)32. For all trial types, the neuronal responses to stimulus presentation, button press, trial expiration, and feedback delivery were analyzed separately. The responses to the stimulus presentation, button press, and feedback were analyzed separately for correct and incorrect trials. To minimize the contamination of signals by activity related to a previous event, the neural response to each event was analyzed across the 500 ms event based epoch and was compared with the activity over the 500 ms inter trial interval preceding the onset of that trial. The 500 ms baseline interval was chosen to maximize sampling while also preventing event related activity from colliding by using the period between two consecutive events (the feedback delivery from a prior trial and the presentation the stimulus in the subsequent trial). Analyses were performed according to the analysis of Teagarden and Rebec (2007)4 and Breysse et al (2015)2. Briefly, the mean firing rate for each perievent bin was expressed as a z score (zi) based on the followi ng formula: z = Fq SEM baseline the mean firing rate of the intertrial baseline period preceding each event, and SEMbaseline indicating the standard error of the mean of the bas with z activation or inhibition. Finally, for each event, neurons were classified as either similar and specific)32. The firing rate of neurons that responded to an event was compared for each outcome with a t test on the normalized data. Thus, neurons were similar if they
22 responded to one event in a similar manner for both rewards ( t te st, p > 0.05)32. Neurons were specific if they responded to one event for both rewards with a significantly higher response to one reward than the other or if they responded exclusively to one outcome32. The proportions of neuronal subpopulations (e.g., Go/Reward selective vs 2test. The average of the z scores of the population PETH were illustr ated by separating the specificity of neurons for either each reward or correct versus incorrect trials, based on the criteria defined above (three consecutive bins with z were also analyzed based on their response type (activati on or inhibition), and the z scores have been calculated. The percentage of variation for activated and inhibited neuronal populations was calculated by comparing the mean firing rate during baseline period and during the event related period. Recording Si te Mapping We defined the spatial location of the recording sites in STN and GPi in relation to the structures borders, as defined by the surgical planning software. We then transposed these positions onto the appropriate anatomical slices of the Schaltenbrand and Bailey atlas42 via a method previously described by Plaha and colleagues43. Briefly, this method involves defining the recording positions intrastructural vertical location by measuring its distance from the structures dorsal boundary. Anterior posterior and medial lateral positions are defined as a proportion of the lengths of the structure along the appropriate axis, and this spatial location was transposed onto the Schaltenbrand and Bailey atlas by proportional measurements43. Finally, anatomical coordinates were
23 derived for each recording position based on electrode trajectory and recording depth utilizing the surgical planning software. Behavioral Results Reaction Time Figure 1 2 shows the differences in reaction times across the conditions. During the training phase, reaction time for Go to Win was .72s (.02 SEM) for game 1 and .69 s (.01) for game 2. Mean reaction time for the Go to Avoid Loss was .73s (.02) for game 1 and .73s (.02) for game 2. During the intraoperative period, average reaction time for the GO trials decreased. Reaction times for the goto win condition were .51s (.02), .49s (.01), and .48s (.01) for games 1, 2, and 3, respectively. Mean reaction times for the go to avoid loss condition were .55s (.02), .54s (.02), .52s (.01) for games 1, 2, and 3, respectively. Mean reaction times were significantly shorter for the go to win condition during games 2 and 3, (unpaired students t test, p =0.038 and p =0.041, respectively, Fig 12 B). Response Accuracy Response accuracy for the four conditions differed considerably in the training period. The twoway repeated measures ANOVA with the factors action (go/no go) and valence (win/avoid loss) revealed no significant main effects for the factors action (F(1,228)=.49; p =.49) or valence (F(1,228)=.38; p =.54) but did reveal a significant valence x action interaction (F(1,228)=17.00); p <.0001). Thus, for PD patients the choice of action was modulated by the anticipation of the outcome. The post hoc Tukey test for multiple comparisons (alpha=0.05) revealed significant differences between the Go Avoid Loss and No GoAvoid Loss conditions and between both No Go conditions (Fig 1 2C).
24 Response accuracy for the four conditions differed considerably in the intraoperative period. The twoway repeatedmeasures ANOVA with the factors action (Go/No Go) and valence (win/avoid loss) revealed a significant main effect for the factors action (F(1,380)=16.57; p<.0001) and a significant valence x action interaction (F(1,380)=17.54); p<.0001). Thus, the patients performed better in Go trials than in No Go trials, and the choice of action was m odulated by the anticipation of the outcome40. The post hoc Tukey test for multiple comparisons revealed significant differences between the GoWin and No Go to Win conditions, between the Go to Avoid Loss and No Go to Win conditions, and between the No Go to Win and No Go to Avoid Loss conditions (Fig 1 2D). Electrophysiological Results: Valence Processing in the STN A total of 100 STN cells from 20 patients and 100 GPi cells from 30 patients met signal quality criteria. The mean duration of recor dings for these neurons was 224s (range: 204s 232s). STN neurons exhibited an average firing rate of 3622 Hz (meanSD), a finding consistent with the results of other studies44 (Hutchison et al. reported a rate of 3717 Hz for 248 cells in eight patients45, a nd Bejjani et al. reported a rate of 38.924 Hz for approximately 530 cells in 12 patients46). The waveforms of the action potentials were biphasic (Fig 1 3 A). Most neurons recorded where single units, although several multi unit recordings occurred (Fig 13 A). Responses to Reward Opportunity and Threat of Loss Of the STN neurons recorded, 57% (57/100) responded to stimulus presentation. The proportion of neurons responding to reward opportunity was significantly greater than the proportion responding to the threat of loss for both the Go [38% vs. 25%, 2=3.916, p =0.0478, Fig. 1 4E ] and No Go conditions [25% vs. 12%, 2=5.604,
25 p =0.0179, Fig. 1 4F ). These results highlight that STN activity was more responsive to a potential reward than a potential loss, and that this reward sensitivity occurred independent of the action context in which the stimulus was presented. The majority of STN neurons that responded to stimulus presentation were valence specific (i.e. they responded exclusively to either reward opportunity or threat of loss). The degree of specific ity was independent of the action condition. For example, we observed that 80.7% were specific during the Go condition and 85.7% were specific during the No Go condition (Fig 1 4A,B) Interestingly, when the response to reward opportunity was an excitation, there was a 42% increased amplitude of the excitation in the Go condition (Fig 1 5 ), while it reached only a 21% increase in the No Go condition (42% vs. 21%, 2=10.22, p =.0014). Responses to Reward and Loss Avoidance We observed that 66% of STN neurons recorded (66/100) responded to feedback presentation during correct trials. The proportion of neurons responding to obtained reward was significantly greater than the proportion responding to loss avoidance for both the Go [48% vs. 32%, 2=5.33, p =.02, Fi g 1 4G] and No Go conditions [27% vs. 15%, 2=4.34, p =.03, Fig 1 4H). These results highlight that STN activity was more responsive to reward than an avoided loss, and that this reward sensitivity occurred independent of the action context in which the s timulus was presented. A majority of feedback responsive neurons were also valence specific (69.9% for the Go condition, 80.6% for No Go condition), and the ratios were similar to those observed following stimulus presentation. A breakdown of the relative proportions of neuron specificity during feedback is shown in Figure 1 4C,D. The relative changes in
26 firing rate amplitude for excitations and inhibitions following feedback are shown in Figure 16. Movement related Activity Of the neurons recorded, 69% (69/100) responded to button press during correct trials. Proportions of neurons responding in the obtain reward and avoid loss conditions were similar [52% (52/100) and 47% (47/100), respectively (Fig 1 7 A ,B)]. Responsive populations for both conditi ons were also similar in terms of the proportions of activated and inhibited neurons (29 activated neurons/ 23 inhibited for the obtain reward condition; 32 activated neurons/ 15 inhibited for the avoid loss condition). Amplitude changes in responsive neurons were also similar, with respect to both activations (+35% for obtain reward, +32% for avoid loss) and inhibitions ( 20% for obtain reward and 23% for avoid loss). Interestingly, however, we observed evidence contrary to our hypothesis that movement related modulation would occur independent of the movements motivational context. Specifically, a majority of neurons (55%, 38/69) responded to movement for just one of the motivational contexts (obtaining reward or avoiding loss), and just 45% of neurons were similar, responding to movement whether movement was motivated by reward seeking or loss avoidance (Fig 1 7 C). Specifically, 26.1% (18/69) of neurons responded to movement exclusively during the obtain reward scenario while 29% (20/69) r esponded to movement exclusively during the avoid loss scenario. This finding suggests that a sizeable subpopulation of movement responsive neurons exists in the STN whose firing rate is affected by the motivational context of action. Also interestingly, neurons that responded to movement for both motivational contexts were significantly more likely to be movement only neurons
27 Errorr elated Activity Neurons responding to feedback during both correct and incorrect trials (which resulted in either an unexpected monetary loss or a failure to receive an expected reward) were recorded. Consistent with the findings in rats previously described by Lardeaux6,32 and Breysse2, we observed a trend toward higher mean baseline firing rate in correct as compared to incorrect trials ( 37.972.5 vs. 25.644.6 Hz, meanSEM, respectively) a lthough this difference did not reach statistical signific ance (students t test, p >.05). Changes in firing rate for both activating and inhibiting neurons are given in Figure 18 A To accurately determine the proportion of neurons responsive to feedback during incorrect trials, we examined only neurons that were recorded during blocks of the task with at least 20% incorrect responses in one of the conditions (i.e. a premature button press during No Go trials or failure to press the button during Go trials) a total of 52 neurons. Of these neurons, 34.6% (18/52) responded to negative feedback during incorrect trials. This was somewhat higher than the proportion of error responsive neurons observed in the human STN by Bastin et al.47 As shown in Figure 1 8B, most of these neurons (77.7%, 14/18) were exclusively responsive to error, whether monetary loss (50%, 7/14) or failure to obtain reward (50%. 7/14). No neurons were observed to be responsive to both. A small portion of error responsive neurons (22.3%, 4/18) responded to incorrect feedback and some other condition of the task (including stimulus, movement, or correc t feedback). This suggests that error sensitive neurons specially encode responses related to the type of negative feedback experienced.
28 Exclusive Correct Neurons In contrast to the neurons exclusively responsive to error, we identified a small subset of neurons that responded exclusively to positive feedback during correct trials (8/100, 8%), which we call exclusive correct neurons. This proportion of exclusive correct neurons is close to that observed by Breysse and colleagues in the rat STN (11.7%)2. The majority of these neurons (87.5, 7/8) were specific, with a slight majority for reward receipt as compared to loss avoidance (57% vs 43%); interestingly, this corresponds closely to the ratio Breysse and colleagues documented when rats experienced both positive and negatively valenced outcomes (32% sucrose solution vs. quinine, respectively)2. Also interestingly, exclusive c orrect neurons were responsive to feedback for the most part only during the Go condition (87.5%, 7/8), suggesting that goal directed action (rather than passive or inhibitionrelated feedback) evokes positivereinforcement feedback responses in this subpopulation of STN neurons. Neuronal Mapping The mean position of the recording sites in STN were 10.69 1.2 mm (mean SD) lateral to the AC PC line in the mid sagittal plane, 1.34 1.39 mm posterior to the inter commissural point and 3.8 1.25 mm below t he AC PC plane (Fig 1 9). For the mean position of subpopulations of responsive neurons, see Figure 1 10. Electrophysiological Results: Valence Processing in the GPi A total 100 GPi cells from 30 patients met signal quality criteria. The mean frequency o f GPi neurons was 66.23.4 Hz (meanSE) similar to the mean human GPi frequency of 77.2 6.6 found by Lee and colleagues48 and 72 2.9 by Levy and colleagues49, although mean discharge rates as high as 91.73.0 have been reported50.
29 The waveforms of the action potentials were biphasic (Fig 1 2A). Most neurons recorded where single units, although several multi unit recordings occurred (Fig 1 2A). Responses to Reward Opportunity and Threat of Loss We observed that 46% of GPi ne urons recorded (46/100) responded to stimulus presentation. In contrast to the STN, the proportion of GPi neurons responding to reward opportunity was similar to the proportion responding to the threat of loss for both the Go [18% (18/100) vs. 14%, respect ively; 2=.5952, p =0.4404, Fig. 1 11E ] and No Go conditions [13%, vs. 16%, respectively; 2=.363, p =0.5469, Fig. 1 11F ]. The majority of GPi neurons that responded to stimulus presentation were also valence specific. Similar to the STN, the degree of specificity was independent of the action condition. We observed that 76.5% were specific during the Go condition and 81.4% were specific during the No Go condition. A breakdown of the relative proportions of neuron specificity is shown in Figure 111A,B. The relative changes in firing rate amplitude for excitations and inhibitions following feedback are shown in Figure 112. These results suggest that the GPi is equally responsiveness to potential reward and potential loss. Responses to Reward and Loss Avoidance We ob served that 38% of GPi neurons recorded (38/100) responded to feedback presentation during correct trials. In contrast to the STN, the proportion of GPi neurons responding to reward receipt was similar to the proportion responding to loss avoidance for bot h the Go [26% vs. 25% 2=.0263, p =0.8711, Fig. 1 11G ] and No Go conditions (19% vs. 16%, respectively; 2=.312, p =0.5766, Fig. 1 11H ). The relative changes in firing rate amplitude for excitations and inhibitions following feedback are shown in Figure 113.
30 A majority of feedback responsive neurons were also valence specific (69.9% for the Go condition, 80.6% for No Go condition), and the ratios were similar to those observed following stimulus presentation. A breakdown of the relative proportions of neuron specificity is shown in Figure 1 11C,D Movement related Activity Of the GPI neurons recorded, 49% (49/100) responded to button press during correct trials. Proportions of movement responsive neurons responding in the obtain reward and avoid loss conditions were similar, w ith 34% (34/100) during obtain reward and 32% (32/100) during avoid loss. Actionresponsive populations for both valence conditions were similar in terms of the ratios of activated to inhibited neurons (21 activated neurons/13 inhibited for the obtain reward condition; 16 activated neurons/ 16 inhibited for the avoid loss condition; 2=.9623, p=.3358). Amplitude changes in responsive neurons were also similar, with respect to both activations (+23.69% for obtain reward, +19.52% for avoid loss; Fig 1 1 4 ) and inhibitions ( 14.6% for obtain reward and 15.24% for avoid loss; Fi g 1 1 4 ). Here again, we observed evidence contrary to our hypothesis that movement related modulation would occur independent of the movements motivational context. Specifically, a majority of neurons (65.3%, 32/49) responded to movement for just one of the motivational contexts obtaining reward [34.7% (17/49)] or avoiding loss [30.6% (15/49)], and just 34.7% (17/49) of neurons were similar, responding to movement whether movement was motivated by reward seeking or loss avoidance (Fig 1 1 4 C). This finding suggests that a sizeable subpopulation of movement responsive neurons exists in the GPi whose firing rate is affected by the motivational context of action.
31 Errorr elated Activity GPi neurons responding to feedback during both correct and incorrect tr ials (which resulted in either an unexpected monetary loss or a failure to receive an expected reward) were recorded. In contrast to the STN, we observed a trend toward lower mean baseline firing rate in correct as compared to incorrect trials ( 643.57 vs. 71.859.2 Hz, meanSEM, respectively) although this difference also did not reach statistical significance (students t test, p >.05). Also in comparison to the STN, average amplitude changes in the GPi were not significantly different between neurons res ponsive to correct and incorrect feedback. When the response to error feedback was an excitation, there was on average a +25.53.11% change in amplitude and when the response was an inhibition the average amplitude change was 23.22 % (Fig 1 15 A). To acc urately determine the proportion of neurons responsive to feedback during incorrect trials, we examined only neurons that were recorded during blocks of the task with at least 20% incorrect responses in one of the conditions (i.e. a premature button press during No Go trials or failure to press the button during Go trials) a total of 59 neurons. We observed that 42.4% (25/59) of this population responded to feedback presentation during incorrect trials. As shown in Figure 1 15B, the vast majority of these n eurons (92%, 23/25) were specific exclusively responsive to either monetary loss (44%, 11/25) or failure to obtain reward (48%, 12/25). Just 8% (2/25) of these neurons were responsive to both feedback conditions. Just 24% (6/25) of these neurons responded to incorrect feedback and some other condition of the task (including stimulus, movement, or correct feedback). Thus, the majority of error sensitive neurons (19/25) were conditionspecific. These results are
32 consistent with the findings of Ruiz et al. regarding involvement of the human GPi in error monitoring31. Neuronal Mapping The mean position of the recording sites in GPi were 22.22 1.65 mm lateral to the AC PC line in the mid sagittal plan e, 2.08 2.55 mm posterior to the inter commissural point and 1.06 1.94 mm below the AC PC plane (Fig 1 16). For the mean position of subpopulations of responsive neurons, see Figure 1 10. Discussion: Comparing STN and GPi The present study shows that STN and GPi neurons in PD patients can encode multiple valence conditions, including the opportunity for reward, the threat of loss, reward receipt, and the successful avoidance of an aversive outcome. In accordance with the findings of previous animal s tudies2,6,32, STN neurons are mostly specific to one of the two opti mal outcomes, based on the responses observed at both stimulus presentation and feedback delivery. Here we also show a clear tendency for STN neurons to encode the positively valenced optimal outcome (reward) versus the neutrally valenced optimal outcome ( loss avoidance). It has been theorized that the proportion of mobilized neurons may encode the relative value of one outcome compared to another, with a greater proportion of neurons responding to the socalled preferred outcome2. Assuming rewarded outcomes are preferred to neutral outcomes where losses are avoided ( which seems reasonable given the observation of lower mean reaction times for Goto Win vs. Go to avoid losing and higher mean accuracy for Go to win vs. Go to avoid losing), our findings appear to corroborate this theory. Since the STN and GPi are well known to be involved in motor behavior, it may be argued that responses at stimulus presentation might not be strictly valencerelated2.
33 Our experimental design holding motor behavior constant while valence is modified (and conducting this test in both movement and nonmovement scenarios) should in theory permit dissoc iation between valencerelated and motor related activity. The observation that movement centered and stimulus centered activity can cooccur in the same neuron raises the possibility that valenceand motor related signaling in these neurons temporally converge; that is, it is not clear precisely when valencerelated and movement related activity begin and end and where overlap might occur (if at all). Experimental designs that implement a gating approach to motor response, i.e. intervals between stimulus presentation and the time when a successful motor response can be registered, may permit clearer dissociation between valence and motor activity, but they introduce an artificial component into the evaluation of action control and may obscure the character and temporal dynamics of actionvalence interactions in their natural state. A change in activity after stimulus presentation might thus be associated partly with motor preparation, but the signal is modulated by the motivational context associated with a specific stimulus (clearly shown by our results when motor condition is constant but valence is varied)2. Finally, the possibility of overlap between movement and valencerelated activity in the feedback epoch is greatly diminished by the delay period between completion of action selection (button press or expirat ion of the decision interval) and feedback delivery. Behavioral Data Wagenbreth and colleagues previously studied the performance of Parkinsons patients on a very similar version of this task40. The only difference was a directional component (the stimulus cue instructed subjects to press either a left or right button), which was not feasible in our experiment given our goal of simultaneously studying the
34 neuronal responses in contralateral brain regions. Their objective was to compare the perform ance of 16 PD patients in the task in the STN DBS ON vs. OFF states. They observed that PD patients OFF DBS performed equally well in the NO GO conditions (~100% accuracy) and equally poorly (~80% accuracy) in the GO conditions. These findings varied consi derably from our results. We consistently had patients perform poorly in the No Go condition. Importantly, Wagenbreths experimental design consisted of many more trials than we were able to implement (320 vs. 120 per session, respectively), due to the constraints of the operating room environment. Also, the appropriate response for each condition was made explicit to patients during their training period. Furthermore, Wagenbreth did not report performance change over time across the 320 trials; it is there fore possible that subject fatigue could be an important factor explaining the observed differences. Also the subjects in the Wagenbreth study were ON medication and the subjects in the present study were OFF medication. Interestingly, the relative perfor mance of PD patients in their natural OFF medication state tended to correspond to the performance biases observed in this task (with the directional component) in healthy control populations30, 29. Among healthy controls, No Go to Avoid Loss performance exceeded No Goto Win performance, and Go to Win performance exceeded Go to Avoid Loss performance. Our results thus indicate that the typical Pavlovian actionvalence performance biases are maintained in the PD population. It is important to note that during the training phase, patients remained on their supplementary dopaminergic m edication, while for the intraoperative trials dopaminergic medications had been withdrawn 24 hours prior. Therefore, it is difficult to directly
35 compare performance between these epochs, especially given that motor performance during the intraoperative period could be confounded by the emergence of motor deficits in the OFF medication state. Thus, it is conceivable that motor performance (accuracy and response time) was underestimated in the intraoperative period. Nonetheless, we observed significant improvements in task performance from the second training epoch to the first intraoperative epoch performed approximately 24 hours later. This suggests firstly that the memory of the actionvalence associations was quite strong across subjects and secondly that any motor deficits in the OFF medication state did not significantly impact the subjects ability to invigorate the relatively simple button press movement the task required. Electrophysiological Data This is the first study to record the responses of human STN neurons to stimuli indicating an opportunity for reward or a threat of loss, in both action and inhibition contexts. We show that distinct neuronal populations responded for each motivational context. Interestingly, the population responding to rew ard opportunity was larger than that responding to threat of loss. As discussed above, if obtaining reward is the preferred outcome (as evidenced by behavioral results), then the STN seems to encode more strongly the preferred outcome. This finding is in l ine with previous studies by Lardeaux and colleagues6 who identified preferential encoding in the rat STN. The internal and external globus pallidus corresponds in most animal studies to the dorsal pallidum1, with the ventral pallidum occupying a distinct region strongly associated with representing reward information1,51,52. Although less well studied, the dorsal pallidum has been shown to encode reward signals1. Specifically, Hikosaka and colleagues identified a subpopulation of GPi neurons in the monkey that projec t to the
36 LHb which strongly encode reward prediction signals19, 20. Our results corroborate the notion that the GPi participates in reward signaling, both in terms of the expectation of reward and rewardrelated feedback /reinforcement. Moreover, GPi neurons also encoded the prospect of loss (stimulus) and successful loss avoidance (feedback), suggesting that the GPi can encode other valenced stimuli beyond pure reward. Finally, our finding that a large proportion of pal lidal neurons were comodulated by both valence and action corresponded to the finding of Arkadirs study in the monkey GPe (41% comodulated by reward expectation and movement)21. Our study provides critical evidence from the individual neuron level that the human GPi plays a role in error monitoring. A few studies of individual pallidal neuron behavior in animals have supported this view19,21, but to date human evidence has been sparse. The most relevant human study was performed by Ruiz and colleagues; studying LFPs in the GPi, they demonstrated modulation of pallidal activity prior to erroneous responses and at the time of erroneous response onset31. They furthermore observed that the error related signal was enhanced in the ipsilateral GPi. Our demonstration of a substantial population of neurons that are responsive to error feedback in the GPi corroborates the notion of a role for the G Pi in error monitoring. Furthermore, the fact that the majority of error responsive neurons responded exclusively to error suggests a dedicated error monitoring role for some GPi neurons. The notion of parallel processing of error related signals in the motor and limbic loops of the cortical BG thalamocortical circuitry is not new31 and is supported by the finding that certain GPi neurons in monkeys project to the LHb and encode error signals19.
37 The present study shows that STN neurons also respond to er ror and that the majority of these are error specific, a finding consistent with rodent studies reporting that a majority of error responsive STN neurons are also error specific2. The observation that error responsive STN neurons can respond either to overt loss or reward omission, in similar proportions but specifically, suggests that in encoding error STN neurons also encode information about the relative value of the errors outcome. Neuronal Mapping The prevailing view of the functional topography of the STN is division into three zones: an anteromedial limbic zone, a posterior sensorimotor zone, and an overlapping associative zone located between the two10 12. Initially, Lambert and colleagues put forth data consistent with a model of distinct limbic and motor zones with an associative zone representing a transition between the two14. Alkemade and others interpreted this, but Lambert and colleagues later cautioned that their methods would not be able to accurately model graduated architectural features and instead artificially provide anatomically distinct boundaries. Thus, we postulated the existence of a graduated region and expected to find a subtle transition across the anterior and medial axes where limbic processing neurons would become more rare toward the STNs lateral and posterior boundaries. Interestingly, we did not observe such a transition but rather identified limbic processing neurons distributed throughout the structure. Moreover, we observed significant overlap of limbic processing with motor responsive neurons. The findings of our study contribute some valuable evidence to the ongoing controversy about the parcellation of the STN into functional zones14,18. Specifically, we show significant overlap between neurons responding to limbic/affective cues, and, importantly, these neurons were not confined to the anteromedial portion of the STN
38 the region widely held to be the zone of limbic involvement. It is important to note that our study was unable to sample a significant number of neurons in the anteromedial STN, due to the inherent limitations of recording neurons encountered exclusively along pre defined trajectories for therapeutic DBS lead placement. Thus, our findings do not preclude the possibility that the anteromedial portion of the STN is more heavily involved in limbic processing (either by proportion of sampled neurons or magnit ude of modulation). However, our results do suggest that limbic overlap with the sensorimotor region is greater than has been previously reported. Similarly for the GPi, our findings did not corroborate the prevailing notion of a limbic GPi limited to the anterior portion of the nucleus. Instead, we found a wide distribution of neurons that were responsive to valence. Here again, our results do not necessarily preclude a higher concentration of valenceresponsive neurons in anterior GPi given that predefi ned electrode trajectories did not permit significant sampling from this region; however, we conclude that valenceresponsive neurons are more widely distributed in the GPi than has been acknowledged elsewhere. Part of the motivation to map valencerespons ive neurons in these structures is to explain the occurrence of cognitive and behavioral decline following DBS utilizing either target but with the susceptibility to decline apparently greater following STN DBS53 59. It is important to note key anatomical differences between the two structures. The human STN is approximately 140 cubic mm60 and con sists of approximately 250,000 neurons61, while the human GPi is approximately 460 cubic m m62 and the pallidum comprises approximately 700,000 neurons63. In light of 1) this size and neuron density difference, 2) the apparently wide distribution of valenceresponsive neurons in
39 both structures, and 3) the higher concentration of valenceresponsive neurons in the STN, one plausible explanation for the observed clinical differences is that the effective DBS electrical field targeting STN would incorporate (and thereby manipulate) more valenceresponsive neurons than would a comparable field targeting the GPi. Moreover, since it is generally accepted that the volume of a target structure positively influences stereotactic lead placement64, the likelihood of misplaced leads increases in STN targets, which in turn creates the requirement of a spat ially larger field to achieve therapeutic effect, providing yet another reason why STN DBS might incorporate more valence neurons despite wide distribution across the nucleus. Finally, we note that individual neuron recording in vivodespite its decreasing popularity in favor of techniques offering broader spatial resolution such as local field potentials may provide a greater degree of subtlety in elucidating graduated changes across the structure and provide more accurate estimates of overlaps and transit ions in topographic organization. In such cases, the use of larger arrays could be beneficial65. Also, the proposed intentional targeting of perceived limbic zones in these structures to treat psychiatric and behavioral disorders25,66,67 could enable greater sampling of the regions not possible in this cohort of patients bei ng treated for motor symptoms. Indeed, combining single unit neuronal recordings with advanced imaging techniques (eg. 7T MRI68 and tractography) could be of even greater value, as it must be recognized that our method of identifying recorded neuron locations, no matter how rigorous, is s ubject to inherent limitations, including the resolution of imaging used for surgical targeting (3T MRI)68 and the need to superimpose these images onto a twodimensional brain atlas
40 with known inconsistencies36,69,70. The implications of these limitations for mapping neurons in a structure as small as the STN could be significant. Conclusion In conclusion, we report here that STN and GPi neurons encode valencerelated information during action control, with valenceresponsive neurons comprising a greater proportion of neurons in the STN compared to the anatomically larger GPi. Valenceresponsive STN and GPi neurons appear to be evenly distributed throughout both structures. In the STN, rewardrelated stimuli mobilize a greater proportion of neurons than loss related stimuli. These attributes of the STN and GPi suggest that both structures occupy critical positions in decis ion making circuitry, and they help to explain the numerous behavioral complications that have been associated with highfrequency electrical stimulation of these structures, as well as the higher incidence of these complications following manipulation of the anatomically more compact STN. Future work elaborating on valence processing and actionvalence interactions in STN and GPi in broader contexts will be important to the refinement of DBS surgery techniques and the further development of neuromodulatory therapies targeted at psychiatric disorders.
41 Figure 11. Behavioral Task. On each trial, one of four possible color patches indicated the combination of required action (making or withholding button press) and outcome valence (win or loss). An action decision was required following presentation of the color patch. After a brief delay, the outcome was presented: +$100 in green indicated a reward, a red $100 indicated a loss, and a black $0 indicated the absence of a win or a loss. In go to win t rials (purple cue), button press was rewarded; in go to avoid losing trials (blue cue), button press avoided punishment; in nogo to win trials (orange cue) withholding button press was rewarded, and in nogo to avoid losing trials (green cue) withholding button press avoided punishment. A running total score was presented at the top of the screen throughout the task. Each trial was preceded by an interstimulus interval in which a blank screen was presented. Stimuli were presented in pseudorandom order for a total of 120 trials (30 trials of each color/condition).
42 Figure 12. Task Performance. A B) Mean reaction times for all training A) and intraoperative B) Go trials, by consecutive game. Results for Go to Win trials are depicted on the left in blue; Go to Avoid Losing trials are depicted on the right in red. Error bars indicate SEM. *= p <.05, unpaired students T test. Intraoperative games occurred approximately 24 hours after training session. C D) Mean task accuracy for all training C) and intraoperative D) trials, by consecutive game. Error bars indicate SEM. *= p <.05, two way repeated measures ANOVA with post hoc Tukey test for multiple comparisons.
43 Figure 13. Neuronal and waveform characterization. A) Example of different waveforms of some representative neurons recorded in the STN showing spikes (left) and biphasic waveforms (middle and right) from two distinct neurons recorded simultaneously from the same electrode. B) Example of the firing pattern of one STN neuron classified as reward opportu nity specific, showing increased activity to the color patch stimulus signifying Go for Reward. Rasters are centered on the occurrence of the stimulus presentation (time = 0). The stimulus is indicated with a black arrow. The area to the left of the vertic al red lines represents the baseline period (intertrial interval) on which the bins were analyzed [0:500 ms]. The black bins represent the bins significantly different score >1.64. Light grey bins represent the bins not significantly different from baseline, with a Z score <1. 64. This neuron was determined to respond with significant neuronal activation due to the occurrence of 3 or more consecutive bins with Z scores >1.64. Top, raster plot of spike firing on each trial (each row illustrates one trial), with the top row of dot s corresponding to the first trial. Bottom, peri event time histogram showing mean firing rate across all trials, with a bin size of 50 ms. C) Distribution of the 100 STN neurons according to their mean firing rate (Hz).
44 Figure 14. Valence Responsiveness in STN Neurons Proportions of the neuronal population responding in the post stimulus period to reward opportunity, threat of loss, and both conditions during A) Go trials and B) No Go trials. Proportions of the neuronal population responding in the post feedback period to reward, loss avoidance, and both conditions during C) Go trials and D) No Go trials [reward receipt (dark blue), loss avoidance (dark orange), and both (grey)]. Total number = total number of responding neurons out of 100 neurons anal yzed. Comparison of neurons responding to reward opportunity and neurons responding to threat of loss for the E) Go and F) No Go conditions. Comparison of neurons responding to reward obtained and loss avoided for the G) Go and H) No Go conditions. (* = p < 0.05, 2 -test).
45 Figure 15. Excitation and I nhibition at Stimulus P resentation in STN Neurons Average z scores (mean SEM) of the firing activity for STN neurons responding by an activation (red line) or an inhibition (blue line) to the visual stimulus (tim e= 0 ms) in Go to Win A), Go to Avoid Loss B), No Goto Win D), and No Goto Avoid Loss E) conditions. The percentages represent the mean variation of activity after each event for activated (red) and inhibited (blue) neuronal population. The z scores are represented for the period on which the time bins were analyzed ( 250:500ms).
46 Figure 16 Excitation and I nhibition at Feedback P resentation in STN Neurons Average z scores (mean SEM) of the firing activity for STN neurons responding by an activation (red line) or an inhibition (blue line) to the visual stimulus (time= 0 ms) in Go to Win A), Go to Avoid Losing B), No Go to Win D), and No Go to Avoid Losing E) conditions The percentages represent the mean variation of activity after each event for act ivated (red) and inhibited (blue) neuronal population. The z scores are represented for the period on which the time bins were analyzed (final 250 ms of baseline from the intertrial interval and 500ms after stimulus presentation).
47 Figure 1 7 STN M ovement related A ctivity. A B) Excitation and inhibition at button press. Average z scores (mean SEM) of the firing activity for STN neurons responding by an activation (red line) or an inhibition (blue line) at button press (time= 0 ms) in Go to Win A) and G o to Avoid Losing B) conditions. The total number of neurons responding (black), total number of neurons responding by excitation (red), and total number of neurons responding by inhibition (blue) are given with n. The percentages represent the mean variat ion of activity after each event for activated (red) and inhibited (blue) neuronal population. The z scores are represented for the period on which the time bins were analyzed (final 250 ms of baseline from the intertrial interval, and (0:500ms). The far r ight panel C) shows the proportion of the neuronal population [Goto Win (grey), Go to Avoid Loss (red), and similar (green)] responding to movement. Total number = total number of responding neurons out of 100.
48 Figure 18. Error Sensitive Neurons in t he STN. A) Excitation and inhibition at feedback presentation for incorrect trials. Average z scores (mean SEM) of the firing activity for STN neurons responding by an activation (red line) or an inhibition (blue line) to error feedback (time= 0 ms) acro ss all conditions. The total number of neurons responding (black), total number of neurons responding by excitation (red), and total number of neurons responding by inhibition (blue) are given with n. The percentages represent the mean variation of activit y after each event for activated (red) and inhibited (blue) neuronal population. The z scores are represented for the period on which the time bins were analyzed (final 250 ms of baseline from the intertrial interval and (0:500ms). B) Proportions of error responsive neurons that were specific for error (purple) and responsive to error and some other stimulus (green).
49 Figure 19. Positions of Recorded STN Neurons. The spatial location of the microelectrode recording sites in STN as transposed onto the Sch altenbrandBailey atlas. A) The STN from coronal slices F.p 4.0 (drawn in red), F.p 1.5 (drawn in green), and F.a 2.0 (drawn in blue) are superimposed on F.p 3.0 (drawn in black). B) The STN from axial slices H.v 1.5, (drawn in red) H.v 3.5, (drawn in gr een) and H.v 6.0 (drawn in blue) are superimposed on H.v 4.5 (drawn in black). The portion of each slice shown is 10mm by 15 mm. Black dots represent recording positions. C) The STN from sagittal slices S.l 10 (drawn in red), S.l 13.5 (drawn in green), a nd S.1 15 (drawn in blue) are superimposed on S.1 11 (in black). Colored squares represent the centroid of clusters of neurons responsive to feedback in each of the conditions.
50 Figure 110. Anatomical Coordinates of Responsive Populations. The mean anat omical coordinates are given in mm. Lateral coordinates are given in mm lateral to the AC PC line in the mid sagittal plane, posterior coordinates are given in mm posterior to the inter commissural point and inferior points are given in mm below the AC PC plane. SD= standard deviation
51 Figure 1 11. Valence Responsiveness in GPi Neurons. Proportions of the neuronal population responding in the post stimulus period to reward opportunity, threat of loss, and both conditions during A) Go trials and B) No Go t rials. Proportions of the neuronal population responding in the post feedback period to reward, loss avoidance, and both conditions during C) Go trials and D) No Go trials [reward receipt (dark blue), loss avoidance (dark orange), and both (grey)]. Total n umber = total number of responding neurons out of 100 neurons analyzed. Comparison of neurons responding to reward opportunity and neurons responding to threat of loss for the E) Go and F) No Go conditions. Comparison of neurons responding to reward obtain ed and loss avoided for the G) Go and H) No Go conditions. (*= p <0.05, 2 -test).
52 Figure 1 12. Excitation and I nhibition at Stimulus P resentation in GPi Neurons Average z scores (mean SEM) of the firing activity for GPi neurons responding by an activation (red line) or an inhibition (blue line) to the visual stimulus (t ime= 0 ms) in A) Go to Win, B) Go to Avoid Losing, C) No Go to Win, and D) No Go to Avoid Losing conditions. The percentages represent the mean variation of activity after each event for activated (red) and inhibited (blue) neuronal population. The z score s are represented for the period on which the time bins were analyzed (final 250 ms of baseline from the intertrial interval, and (0:500ms).
53 Figure 1 1 3 Excitation and I nhibition at Feedback P resentation in GPi Neurons Average z scores (mean SEM) o f the firing activity for all GPi neurons responding by an activation (red line) or an inhibition (blue line) to the visual stimulus (time= 0 ms) in A) Go to Win B) Go to Avoid Losing, C) No Go to Win, and D) No Go to Avoid Losing conditions. The percentages represent the mean variation of activity after each event for activated (red) and inhibited (blue) neuronal population. The z scores are represented for the period on which the time bins were analyzed (final 250 ms of baseline from the intertrial inter val, and (0:500ms).
54 Figure 1 14. GPi Movement related A ctivity. A B) Excitation and inhibition at button press. Average z scores (mean SEM) of the firing activity for STN neurons responding by an activation (red line) or an inhibition (blue line) at button press (time= 0 ms) in Go to Win A) and Go to Avoid Losing B) conditions. The total number of neurons responding (black), total number of neurons responding by excitation (red), and total number of neurons responding by inhibition (blue) are given wi th n. The percentages represent the mean variation of activity after each event for activated (red) and inhibited (blue) neuronal population. The z scores are represented for the period on which the time bins were analyzed (final 250 ms of baseline from th e intertrial interval,and (0:500ms). The far right panel C) shows the proportion of the neuronal population [Goto Win (grey), Go to Avoid Loss (red), and similar (green)] responding to movement. Total number = total number of responding neurons out of 100.
55 Figure 1 1 5. Error Sensitive Neurons in the GPi. A) Excitation and inhibition at feedback presentation for incorrect trials. Average z scores (mean SEM) of the firing activity for GPi neurons responding by an activation (red line) or an inhibition (blue line) to error feedback (time= 0 ms) across all conditions. The total number of neurons responding (black), total number of neurons responding by excitation (red), and total number of neurons responding by inhibition (blue) are given with n. The perc entages represent the mean variation of activity after each event for activated (red) and inhibited (blue) neuronal population. The z scores are represented for the periods on which the time bins were analyzed (final 250 ms of baseline from the intertrial interval, and 0:500ms). B) Proportions of error responsive neurons that were specific for error (purple) and responsive to error and some other stimulus (green).
56 Figure 1 1 6. Positions of Recorded GPi Neurons. The spatial location of the microelectrode recording sites in GPi as transposed onto the SchaltenbrandBailey atlas. A) The GPi from coronal slices F.a 7.5 (drawn in red), F.a 3.0 (drawn in green), and F.a 2.0 (drawn in blue) are superimposed on F.p 5.0 (drawn in black). B) The GPi from axial slic es H.v 1.5 (drawn in red), H.v 3.5 (drawn in green), and H.v 6.0 (drawn in blue) are superimposed on H.v 4.5 (drawn in black). C) The GPi from sagittal slices S.l 16 (drawn in red), S.l 18.5 (drawn in green), and S.l 21.5 (drawn in blue) are superimpos ed on S.l 20 (drawn in black). The portion of each slice shown is 10mm by 15 mm. Black dots represent recording positions. Colored squares represent the centroid of clusters of neurons responsive to feedback in each of the conditions.
57 CHAPTER 2 IMPULSIVI TY AND DIFFERENCES IN STN NEURON BEHAVIOR Introduction Parkinsons disease (PD) is a neurodegenerative disorder primarily characterized by motor dysfunction, but PD patients can also suffer from behavioral and cognitive deficits. Impulse control disorders (ICDs) encompass a spectrum of behavioral disorders in PD characterized by an inability to resist an inappropriate drive, often of a hedonistic nature and this issue results in repetitive behaviors, that often have harmful consequences71. ICDs ca n take the form of pathological gambling, compulsive buying, compulsive or altered sexuality, binge eating72, or the compulsive pursuit of hobbies or other manual activities. ICDs occur in approximately 15% of the PD population and can negatively impact quality of life for both patients72, 73 and caregivers74,75, 76. Despite growing interest in ICDs and robust clinical research over the past decade, the underlying pathophysiology of ICDs is not well understood. Deep brain stimulation (DBS) therapy of the STN and GPi has been associated with increases in ICD symptoms, though it should be noted that resolution of ICDs has also been reported77. Animal studies corroborate the notion that manipulation of these structures can alter impulse control. STN DBS has been shown to both increase78 and decrease79 premature responding in rats, and both lesions80,81, 5 and electrical stimulation82 of the STN can alter rat motivation to obtain rewards, while electrical stimulation of the GPi has been shown to increase impulsivity in rodents83. Deep brain stimulation (DBS) of the STN in human patients has been associated with increased aut omatic response activation and susceptibility to impulsive responses84.
58 Studies of individual neuron behavior in rodents and nonhuman primates suggest that the subthalamic nucleus (STN)3 and (GPi)1 mediate the processing of information related to reward and punishment. Large subsets of STN neurons in the rat have been shown to encode the prospect of reward2,32,85 and loss2, and simi lar signaling has been demonstrated in the STN of nonhuman primates7 9. Hikosaka and colleagues documented reward signaling in the primate GPi19,86,87, and a recent study of single neurons in the human GPi demonstrated responsiveness to the prospect of reward and punishment88. Numerous behavioral studies have established a link between impulsive behavior and aberrant processing of reward and punishment8992. Impo rtantly, PD patients with an IC D have been shown to exhibit exaggerated sensitivity to reward93 and diminished sensitivity to punishment94, 95. We hypothesized that in PD patients with an ICD, neurons in the STN and GPi would be more responsive to rewardrelated stimuli and less responsive to loss related stimuli. Specifically, we expected to observe these differences both in the proportion of responsive neurons and in the magnitude of response (i.e. firing rate). To test this hypothesis, we studied a cohort of 43 PD patients (32 males, 11 females; 12 with an ICD and 31 without) undergoing DBS electrode placement surgery. ICD diagnosis was established using the Questionnaire for Impulsivity in Parkinsons disease (QUIP) and semi structured clinical interview. Patients performed a behavioral task in which their action choices were motivated by the potential for either a monetary reward or a monetary loss. During task performance, single unit activity was recorded in either the STN or GPi [100 STN neurons (35 from ICD+ patients and 65 from ICD
59 patients) and 100 GPi neurons (36 from ICD+ patients and 64 from ICD patients)]. Neuronal activity was spike sorted and analyzed to determine responsiveness (i.e. significant change in firing rate from baseline in response to a stimulus), and proportions of responsive neurons were compared using chi square statistics and firing rate changes were compared using twoway repeated measures ANOVA and chi square statistics. ICDs were associated with significantly greater responsiveness to reward opportunity and this was evidenced in the proportion of responsive neurons ( p <.01) and in the significantly diminished responsiveness to the threat of loss ( p <.05) in the STN but not the GPi neurons. Importantly, ICD status was not associated with any differences in firi ng rate among responsive neurons in either structure. These findings suggest that changes in the STN possibly via functional conversion of neurons in the limbic circuit may underlie impulsive behavior in the PD population. Materials and Methods Patients C onsecutive patients with Parkinsons disease undergoing de novo STN or GPi DBS electrode placement surgery were studied. All patients fulfilled the UK Brain Bank criteria for PD. All participants were Englishspeaking, nondemented, not hearing impaired, a nd sighted. We enrolled 43 patients in the study, w ith a mean age of 64.8 years (S D=8.9). The Institutional Review Board approved the study and all patients provided informed consent before entering the study. We recorded complete details of anti parkinsonian medication therapy for each patient. The recorded medications were those being taken at the time of assessment. Due to the strong association of dopaminergic agonists with the development of ICDs96,
60 we compared ICD+ and ICD groups in the use of dopaminergic agonists ( pramipexole, ropinirole, pergolide, rotigotine, apomorphine, or bromocriptine). Assessm ent of Impulse Control Disorders Assessment of impulse control disorders involved modified version of the method employed by Papay and colleagues97. Participants completed the Questio nnaire for Impulsive Compulsive Disorders in Parkinsons Disease Rating Scale (QUIP), a self reported and self completed validated measure of impulsivity in PD98. Participants were instructed to answer questions based on behaviors that occurred in the previ ous six months that lasted at least four consecutive weeks99. The ful l version o f the QUIP includes five q uestions for each of the four ICDs most commonly reported in PD (compulsive gambling, buying, sexual behavior, and eating) and three questions about any other activity or hobby. Based on the results of the validation of the QUIP, the following thresholds were used to represent a positive screen for the patient completed instrument: compulsive gambling (any 2 of the 5 items); compulsive sexual behavior (any 1 of the 5 items); compulsive buying (any 1 of the 5 items); compulsive eati ng (any 2 of the 5 items); other behavior (punding and hobbyism) (any 2 of the 3 items)99,100. The QUIP is designed as a screening instrument and as such is sensitive for the detection of ICD but is not highly specific101. To avoid over identifying ICDs, a fter patients completed the QUIP, a medically trained investigator administered a semi structured, diagnostic interview that was based upon published proposed criteria for ICDs in PD102 and utilized the same time frame as in the QUIP. The interviewer was blinded to the patients QUIP responses; however, patients occasionally verbalized their responses contrary to instructions. Self identified primary caregivers who accompanied patients were given the opportunity to contribute during the interview, given the well
61 documented phenomenon that pati ents suffering from ICDs tend to underreport symptoms either due to embarrassment or because they lack objective insight into negative behavior99. Primary caregivers were present and contributed in all but three of the interviews. Participants performed the same behavioral task previously described. Data acquisition, recording protocol, and electrophysiol ogical data analysis were identical. Results Patient C haracteristics The mean preoperative QUIP score for ICD+ patients was 7.4 .98 (SEM) and ICDpatients was .45 .21 (SEM). Dopamine agonist use was 29% for ICD patients and 67% for ICD+ patients ( p =.02, 2 test) ICDs diagnosed included pathological gambling (n=4), hypersexuality (n=10), pathological buying (n=6), binge eating (n=4), and hobbyism (n=3). Other impulsive activities observed in this cohort included illegal drag racing and shoplifting. Subtha lamic N ucleus A total of 35 STN cell s were recorded from five ICD+ patients, and a total of 65 STN cells were recorded from 14 ICD patients. Overall, w e observed that impulsivity was associated with discernible differences in the proportions of responsive STN neurons. ICD patients exhibited heightened sensitivity to reward opportunity. The proportion of STN neurons responding to the opportunity to earn a reward was significantly higher in ICD+ patients compared to ICD patients (68.6% vs. 41.5%, 2 test, p =.009, Fig 2 1 A). ICDs were also associated with significantly higher proportions of reward exclusive neurons neurons responding only to reward opportunity (63% vs. 31.9%, 2 test, p =.009, Fig 2 2 A). ICD patients exhibited diminished sensitivity to threat
62 of loss. The proportion of STN neurons responding to the threat of loss was significantly lower in ICD+ patients compared to ICD patients ( 28.6% vs. 49.2%, 2 test, p =.045, Fig 2 1 A). ICDs were also associated with significantly lower proportions of loss exclusive neurons neurons responding only to threat of loss (11.1% vs. 42.6%, 2 test, p=.005, Figure 22 A). Finally, ICD+ and ICDpatients exhibit similar firing rates in responsive STN neurons The mean percent change in firing rate for responsive neurons was not significantly different between ICD+ and ICD groups for activating or inhibiting neurons, either in reward or loss scenarios (T test with post hoc Holm Sidak correction for multiple comparisons, Figure 2 1C ). Investigation of the effe ct of ICD status on post stimulus changes in firing rate using twoway repeated measures ANOVA also revealed no significant effect of ICD status on firing rate for any of the conditions. Globus P allidus internus A total of 36 GPi cells were recorded from 7 ICD+ patients, and a total of 64 GPi cells were recorded from 17 ICD patients. ICD+ and ICD patients exhibited similar GPi responsiveness to reward opportunity. The proportion of GPi neurons responding to the opportunity to earn a reward was not signifi cantly different in ICD+ patients compared to ICDpatients (27% vs. 34.9%, 2 test, p =.413, Figure 2 1B ). ICDs were not associated with differences in the proportions of reward exclusive neurons (35.7% vs. 38.5%, 2 test, p =.856, Figure 2 2B ). ICD+ and ICDpatients exhibited similar GPi responsiveness to threat of loss. The proportion of GPi neurons responding to the threat of loss was not significantly different in ICD+ patients compared to ICD patients ( 24.3% vs. 39.7, 2 test, p =.117, Figure 2 1 B). ICDs were not associated with differences in the proportions of loss exclusive neurons (28.6% vs. 41%, 2 test, p =.410, Figure 2 2B ). Finally, ICD+ and ICD patients exhibited similar firing rates in responsive GPi neurons. The mean
63 percent change in firing rate for responsive neurons was not significantly different between ICD+ and ICD groups for either activating neurons (increased firing rate) or inhibiting ne urons (decreased firing rate) (t test with post hoc Holm Sidak correction for multiple comparisons, Figure 2 1 D). Investigation of the effect of ICD status on post stimulus changes in firing rate using twoway repeated measures ANOVA also revealed no significant effect of ICD status on firing rate for any of the conditions. Discussion : Comparing STN and GPi The roles of the STN and GPi in human motor function are well accepte d principles of basal ganglia physiology. The roles of these structures in nonmotor functions, however, continue to be intensely studied. Compelling animal evidence suggests that neurons from this region play a role in processing reward and punishment, an d clinical evidence has shown that their manipulation via lesions or electrical stimulation may result in behavioral changes likely related to changed perceptions of reward and loss. The present study offers strong evidence suggesting that human subthalami c neuron activity is different in PD patients with an impulse control disorder as compared to PD patients without an impulse control disorder. Specifically, in PD patients with an ICD, the proportions of STN neurons that responded to reward were significantly greater and the proportions of STN neurons that responded to loss were significantly lower. In contrast, no such differences were observed with respect to GPi neurons. The finding of heightened reward sensitivity and diminished loss sensitivity makes s ense in light of the behavioral findings that PD patients with ICD exaggerate their response to rewards and attenuate their response to punishment94. What is intriguing about our findings is that these differences in reward/punishment sensitivity
64 are attributable to differences in the relative proportions of responsive neurons rather than to differences in the magnitude of firing rate changes. Indeed, we observed no significant differences in the ICD+ and ICDgroups in terms of the magnitude of firing rate changes following rewardor loss related stimuli. In addition, our finding that the relative size of the responsive STN neuron population was stable across ICD and ICD+ groups (with s imilar populations of neurons responsive to both reward and loss but different populations of reward exclusive and loss exclusive neurons) could possibly be interpreted as ICD+ patients converting STN loss neurons into reward neurons though this point cannot be concluded based on the available data. An important remaining question is whether (and how) the observed differences in subthalamic function can be attributed to the use of dopaminergic agonist medications. There is a known association between the use of dopamine agonist drugs and the development of ICDs. A large multi center study recently showed that up to 39% of patients being treated with nonergot derived dopamine agonists (pramipexole, ropinirole, or rotigotine) developed ICDs103. It was therefore not surprising in or data that the use of dopaminergic agonists was significantly higher among ICD+ patients. Dopamine agonists have high affinities for the dopamine D2 and D3 receptors96. Intra STN administration of the D2like receptor agonist quinpirole does not alter firing rate patterns of STN neurons in Parkinsonian monkeys104, and administration of the D1 and D2 agonist apomorphine has been shown to not affect over all STN neuron firing rates in humans49. However, the effects of acute administration of D2 agonists on the proportions of neurons responsive to limbic cues are not known. Given that subjects in the present study were withdrawn from dopamine agonist medications 24 hours prior to
65 study, it i s unlikely that the functional differences observed could be due to the direct effects of medication. Rather, we speculate that chronic use of dopamine agonist medications may (directly or indirectly) induce persistent functional changes in STN neurons. IC D symptoms often resolve following discontinuation of dopamine agonist medications, but the timeline in the resolution of symptoms is not well understood. Still, a plausible alternative explanation might be that preexisting functional differences (possibl y diseaseinduced) predispose certain patients to the development of ICDs following DA exposure and that the observed differences in STN physiology are related to such preexisting differences. Certainly, the present study raises a number of questions that warrant further investigation. There are a few limitations to our study that must be acknowledged. First, patients performed the task during an ongoing operation, which may raise stress levels and potentially cause a relative devaluation of the outcome stimuli88. Performing this task in a more relaxed environment may soon be possible with emerging technology that is enabling chronic post operative physiological recordings. However, even this technology would permit only recordings of populations of neurons via local field potentials, not the behavior of single neurons. A lso, while all patients were withdrawn from dopamine medication 24 hours prior to surgery, the withdrawal might have differentially affected subjects. Finally, it was only possible to sample neurons along the pre planned trajectory for the DBS electrode. T hus, neuronal sampling from putative limbic regions of the STN and GPi (which are deliberately avoided by neurosurgeons) may have been insufficient. Our use of a triple array microelectrode system was intended to partially mitigate this bias. In addition, it should be noted that the idea of
66 distinct limbic regions in the STN has been recently questioned14,18 and the existing evidence for such a region in the GPi is limited. Despite these limitations, our approach was required and facilitated an analysis at the level of individual human neurons. In conclusion, we present findings from human single neurons suggesting that differences in subthalamic nucleus functioning may underlie ICDs in PD. We found that PD patients with ICD exhibited increased proportions of subthalamic neurons responsive to prospective reward and decreased proportions of subthalamic neurons responsive to prospective loss, compared to PD controls without ICD. No such differences were observed in globus pallidus internus neurons. Future work will be needed to assess long ter m changes in STN physiology in light of post stimulation changes in impulsive behavior. A deeper understanding of the pathophysiology of impulse control disorder in this population could potentially lead to improvements in patient and target selection for DBS surgery.
67 Figure 21 ICD status impacts reward and loss sensitivity in subthalamic neurons but not internal pallidal neurons. Comparison of responsive neuron populations in I CD+ and ICDpatients for A) STN and B) GPi; neurons responsive to stimuli indicating the opportunity for reward (left, green) and neurons responsive to stimuli indicating t he threat of loss (right, red). Comparison of firing rate changes in ICD+ and ICD patients for stimulus responsive C) STN and D) GPi neurons. indicates p< .05, ** indicates p<.01, 2 test. Activated indicates neurons that exhibited significant increases in firing rate following stimulus presentation. Inhibited indicates neurons that exhibited significant decreases in firing rate following stimulus presentati on. Error bars indicate SEM. ICD+ indicates patients diagnosed with an impulse control disorder; ICD indicates patients not diagnosed with an impulse control disorder; STN= subthalamic nucleus; GPi= globus pallidus internus.
68 Figure 22. ICD status imp acts proportions of loss and reward exclusive neurons in STN but not GPi. Stimulus response exclusivity and comodulation proportions for ICD+ (left) and ICD (right) patients in the A) STN and B) GPi Reward exclusive neurons indicate neurons that exhibited significant changes in firing rate only in response to reward stimuli; loss exclusive neurons indicate neurons that exhibited significant changes in firing rate only in response to loss stimuli; both neurons indicate neurons that exhibited significa nt changes in firing rate in response to both reward and loss stimuli.
69 CHAPTER 3 DEEP BRAIN STIMULATION MODIFIES IMPULSE CONTROL Introduction A large portion of Parkinsons disease (PD) patients (~15%) suffer s from impulse control disorders (ICDs)105. ICDs are characterized by an inability to resist an inappropriate drive which is frequently of a hedonistic nature71. Patients pursue certain behavior s repetitively, excessively, and compulsively to the point of interference in major areas of life functioning106. Common ICDs include pathological gambling, compulsive buying, hypersexual ity, binge eating, or hobbyism. Dopamine agonist (DAA) therapy is a major risk factor for the development of ICDs, and withdrawal or reduction of DAA has been shown to improve symptoms107. It has been argued by several expert groups that deep brain stimulation (DBS) could be used to treat ICDs, giv en that motor symptom control via DBS could enable therapeutic DAA reduction108,109. Studies comparing ICD symptoms before and after DBS have had conflicting re sults and suggest a more complicated pictureone in which DBS could potentially exert direct effects on impulse control. STN DBS coupled with a large reduction in dopaminergic medication has been shown to reduce preexisting impulsive behavior110, but onset of new ICDs and worsening of existing ICDs in the post operative period despite reduction in dopaminergic medicati ons have also been documented77,111. The results of retrospective case series and reports have revealed conflicting effects of DBS on impulse control, including improvement/resolution, worsening/new onset, and lack of change. Previous studies have been mostly retrospective111 114 or crosssectional115 in design and have had small cohorts116, while the few prospective
70 studies to date have not investigated the effects of DBS on both STN and GPi targets108,110. In addition, these studies, with one exception77, have focused on the effects of bilateral STN DBS. Finally, it remains unclear from the available evidence whether changes in ICD status in these studies were due to changes in DAA or due to changes elic ited by electrical stimulation. We present the first prospective study in PD examining the effects of DBS on impulse control in both STN and GPi. Consecutive patients (GPi, n=30; STN, n=20) undergoing DBS surgery were evaluated for the presence of an ICD using the Questionnaire for Impulsivity i n Parkinsons Disease (QUIP) and using a structured clinical interview. Patients were assessed for changes between 6 and 12 months post operatively. Complete details of anti Parkinsonian medication for pre and post operative periods were recorded and compared between brain targets. Materials and Methods Patients Consecutive patients with Parkinsons disease undergoing de novo STN or GPi DBS electrode placement surgery were studied. The cohort was nonrandomized; brain targets were selected by an interdi sciplinary team following individual evaluations (neurology, neurosurgery, neuropsychology, physical therapy, occupational therapy, speech therapy). All patients fulfilled the UK Brain Bank criteria for PD. There were 53 patients enrolled in the study, wit h a mean age of 64. 8 years (S D=8 9). The University of Florida Institutional Review Board approved the study and all patients provided informed consent before entering the study. We recorded complete details of anti parkinsonian medication therapy for each patient. The recorded medications were those being taken at the time of preoperative
71 and post operative assessment. Due to the strong association of dopaminergic agonists with the development of ICDs, we compared ICD+ and ICD groups in the use of dopam inergic agonists ( pramipexole, ropinirole, pergolide, rotigotine, apomorphine, or bromocriptine) as well as levodopa equivalent daily dosage (LEDD). We intended to perform readministration of QUIP and clinical interview between 6 and 12 months following D BS implantation. However, follow up for some patients occurred between 5 and 14 months. Mean follow up time for all patients was 9. 8 months post surgery. We separated patients into those with unilateral DBS and those with bilateral DBS and performed a subanalysis comparing these groups. Assessment of Impulse Control Disorders Participants completed the Questionnaire for ImpulsiveCompulsive Disorders in Parkinsons Disease Rating Scale (QUIP), a self reported and self completed validated measure of impulsi vity in PD98. Participants were instructed to answer questions based on behaviors that occurred in the previous 6 months that lasted at least four consecutive weeks99. The QUIP includes 5 questions for each of the 4 ICDs most commonly reported in PD and 3 questions about any other activity or hobby. We utilized established thresholds to represent a positive screen for the patient completed instrument: compulsive gambling (any 2 of the 5 items); compulsive sexual behavior (any 1 of the 5 items); compulsive buying (any 1 of the 5 items); compulsive eating (any 2 of the 5 items); other behavior (punding and hobbyism) (any 1 of the 3 items)99. The QUIP is designed as a screening instrument and as such is sensitive for detecting ICDs but is not highly specific101. To avoid over identifying ICDs, a fter patients completed the QUIP, a medically trained investigator administered a semi structured, diagnostic interview based on published proposed criteria for ICDs in PD102 and utilized
72 the same time frame as in the QUIP99. The interviewer was blinded to the patients QUIP responses; however, patients occasionally verbalized their responses contrary to instructions. Self identified primary c aregivers who accompanied patients were given the opportunity to contribute during the interview, given the well documented phenomenon that patients suffering from ICDs tend to underreport symptoms either due to embarrassment or because they lack objective insight into negative behavior99. Primary caregivers were present and contributed in all but 3 of the interviews. Statistical A nalysis Categorical variables were compared using statistics. For continuous variables, ShapiroWilk tests for normality were performed. Provided that the data were normally distributed, continuous variables were then compared using unpaired t tests for comparisons by brain target and ICD status and us ing paired t tests for comparisons of pre DBS and post DBS outcomes. Statistical significance was based on an alpha of 05. Results Post operative follow up was obtained for 50 patients (30 GPi, 20 STN). Of the patients enrolled preoperatively, 3 were los t to follow up. These were not included in the final analysis. Of the 50 patients analyzed, 37 underwent unilateral DBS placement (23 GPi, 14 STN) and13 underwent bilateral DBS placement (7 GPi, 6 STN). ICD Prevalence Of the 50 patients studied, 20 patien ts (40%) had an ICD at preoperative baseline ( Fig 3 1). The mean preoperative QUIP score for ICD+ patients was 6.411.0 (SEM); the mean post operative QUIP score for ICD+ patients was 3.30.9 (SEM). There were no significant differences in the prevalence of ICDs or any ICD sub type in the post operative period compared to baseline for either GPi or STN (Fig 31) The
73 prevalence of ICDs and all ICD subtypes in the preand post DBS periods for both STN and GPi are shown in Figure 32 Clinical interviews p rovided valuable subjective insight into the impulsivecompulsive behaviors reported in the QUIP. For example, common manifestations of hypersexual behavior in this cohort included inappropriate touching of others, increased masturbation, exhibitionism, and pornography use. Hobbyism and other impulsive activities included model boat building, drag racing, hedge clipping and gardening, vacuuming, fishing, shoplifting, onychotillomania (nail picking), and excoriation (skin picking). Most patients reported pur suing these activities despite legal consequences, potential or actual harm to personal relationships and financial wellbeing, and risks to personal safety. Many patients also reported engaging in enabling behaviors, such as binge eating after family members have gone to sleep, hoarding of food, picking skin in areas not visible to others, and covertly pursuing their impulsive/compulsive activity to avoid observation by others. Rate of Change in ICD S tatus Post DBS A substantial proportion of patients exhibited a conversion in ICD status after DBS (33.3% of GPi patients and 40% of STN patients; Figure 3 2). For the GPi, 23.3% of patients experienced resolution of a preexisting ICD and 6.7% of patients experienced a new onset ICD. For the STN, 25% of patient s experienced resolution of a pre existing ICD and 15% of patients experienced a new onset ICD. The de novo I CDs in the GPi cohort included two cases of binge eating. One patient developed a new onset case of dopamine dysregulation syndrome (DDS) A lthough DDS is not an ICD this was an important observation. The de novo ICDs in the STN cohort included two cases of hypersexuality and one case of hobbyism (model boat building). There were no
74 significant differences between the STN and GPi targets in terms of the proportions of patients who experienced any type of ICD status change ( Fig 3 4 ) However, GPi patients had a significantly higher mean QUIP score at baseline than STN patients (3.9 vs. 1.6, p =.0486, Fig 34 ), suggesting a clinical bias in favor of tar geting the GPi in patients with ICD symptoms. Qualitative Changes in Impulse C ontrol Post DBS A substantial proportion of patients reported meaningful qualitative changes in their ability to control impulsive behavior, even though they did not meet criteri a for a change in diagnosis (Fig 3 5) During clinical interview, patients were asked whether they had experienced improvement, worsening, or no change in ICD symptoms since DBS. For the GPi, 20% of patients reported improvement and 16.7% reported worsening, in addition to those patients whose changes constituted a conversion in ICD status. For the STN, 5% of patients reported improvement and 15% reported worsening. Changes in ICD Status and D opaminergic M edication Due to the well established connection between dopamine medications (in particular DAA) and ICDs, we sought to determine the degree to which observed changes in ICD status could be attributed to DBS versus post operative modifications in medication regimen. Special attention was paid to DAA. To accomplish this, we performed a multi layered analysis. We first tested for a relationship between ICD symptom changes and changes in DAA regimen. We found that patients that changed ICD status were not more likely to have changed their dopamine agonist medication than those who maintained their ICD status ( test p =.1468). In addition, patients who changed ICD status or reported qualitative changes in their ICD symptoms were not more likely to have changed their dopamine agonist medication than those who
75 maintained their ICD status and reported no qualitati ve changes ( test, p =.2098). We then tested for a relationship between QUIP score (as a surrogate measure for ICD severity) and LEDD, and we repeated the analysis using LEDD for DAA only (DAA LEDD). Linear regression analysis revealed no significant corr elation between QUIP score and LEDD for either preDBS or post DBS measures (preDBS R2=.204, post DBS R2=.006) or between QUIP score and DAA LEDD (preDBS R2=.006, post DBS R2=.013). We also compared patients with a change in ICD status and those without a change in status in terms of the mean percent change in dopamine agonist dosage. We found no significant difference in this comparison (unpaired students t test, p =.3716, Fig 3 3 ). Interestingly, preDBS and post DBS LEDD for STN patients were remarkabl y similar. However, separate analysis of unilateral and bilateral procedures revealed a trend toward medication reduction for bilateral STN cases (1222516.3 preDBS vs. 928291 post DBS, p =.252 Fig 3 6 ). The lack of statistical significance reflects both the small sample size of bilateral cases but also the bias but also the published philosophy of the UF center to employ conservative medication reduction strategies wherever possible to minimize adverse mood effects associated with dopamine withdrawal117 119. We note that patients who were ICD+ at baseline had significantly higher preoperative LEDD compared to those who wer e ICDat baseline ( p =.019, Fig 34 ). These differences in LEDD were maintained at the post operative assessment ( p =.008, Fig 3 4 ). Interestingly, such differences were not observed for LEDD based on dopamine agonist only (preoperative DAA LEDD, p =.479, p ost operative DAA LEDD, p =.501; Fig 3 4 ). Importantly, of the patients who experienced a resolution of ICD symptoms, only
76 41.7% (5/12) decreased their dopamine agonist dosage. Of the patients who experienced a de novo ICD, 0% (0/6) increased their dopamine agonist dosage. These findings suggest that while dopamine medications are strongly associated with ICDs in the preDBS PD population, they do not independently account for the changes in ICD symptoms observed following DBS. Other Psychological C hanges Du ring the follow up clinical interview, patients were asked to describe any other changes in mood, cognition, or psychological functioning. One patient reported pathological crying (34 episodes per day) following STN DBS. Another patient reported euphoric episodes following STN DBS, however these issues resolved with changes in stimulation settings. Discussion Our data reveal that ICD symptoms can improve, worsen, emerge, or remain constant following STN and GPi DBS. There was resolution of ICDs in 60% of patients following surgery. This finding is close to the 65% observed in a retrospective study of STN DBS73. In our study, 20% of patients that did not have ICD at baseline developed ICD post operatively. This finding paralleled the 13% rate that has been reported previously73. The combination of ICD resolution and emergence potentially explains the comparable prevalence of ICDs in the preoperative and post operative period. The overall preoperative prevalence of ICDs in our cohort was 40%. This was consider ably higher than the ICD prevalence typically reported for the PD population at large (~15%)105,116. Higher prevalence of ICDs in the population of pati ents undergoing DBS has been described previously116 and may be potentially explained by the bias of patients referred to DBS surgery tending to have higher LEDD, higher DAA LEDD, and
77 longer disease durationall risk factors for the development of IC D73. Moreover, the pre operative prevalence of ICDs appears to be largely dependent on the impulsive behavior scale used120, and rates from 22.5%73 to 50%110 have been reported. The post operative prevalence of ICDs in our cohort was 26%, and this was comparable to the 25.8% post operative preval ence found in a retrospective study of STN DBS73. DBS has been associated with an increased risk of binge eating121. We therefore expected to observe a higher prevalence of binge eating in the post DBS population. Although there were no significant differences in the prevalence of binge eating ICDs in t he preand post DBS periods, we observed considerable subthreshold changes associated with eating impulsivity. Six patients reported worsening or subthreshold onset of eating related symptoms, with only one reporting improvement. While these results appear to corroborate previous findings, it is difficult to draw conclusions about binge eating given its low prevalence in the present cohort. Our finding that differences in dopaminergic medications do not fully explain the changes in ICD symptoms following DBS is important. It corroborates preclinical findings suggestive of STN and GPi involvement in impulse control, and it aligns with reports that post DBS changes in ICD symptoms do not always conform to expectations based on medication changes116. P re clinical studies in dopaminenave rodents revealed changes in impulsive behavior following high frequency electrical stimulation of the STN and GPi78,79,82,83, and STN DBS in Parkinsonian humans increases measures of impulsivity independent of dopaminergic medication status122. The mechanisms underlying post operative changes in impulse control are likely complex, with dopaminergic medications and electrical stimulation simultaneously
78 affecting behavior in ways that are dependent on a multitude of factors, including the type(s) and dosage of preand post surgical medication, the location of the DBS lead, and the stimulation parameters all of which, in clinical practice, tend to be modified in the months following DBS lead implantation to optimize outcomes. Exploring the relative impact of these factors while controlling variables in a limited clinical population remains a challenge. Our study has some limitations. The QUIP cannot capture qualitative changes in ICD symptoms that do not constitute a diagnosis conversion. Indeed, our data regarding qualitative changes not captured by the QUIP suggests the need for a new instrument in the DBS population. Also, there was some heter ogeneity in the time to follow up. All efforts were made to readminister evaluations within the intended 6 to 12 month window which was viewed as clinically feasible, but this was not always possible given the time required to complete the structured interview. Finally, our study had a male bias; however, our sampling method and previous studies123,124 suggest that this bias was reflective of the DBS population at l arge. Finally, our sample includes patients who underwent both unilateral and bilateral DBS. The fact that our center has adopted a conservative approach to medication reduction, particularly following unilateral STN DBS, can be regarded as both a strength and weakness of the study. The similarity in dopamine therapy in the preand post operative period for STN cases did permit a more controlled study of the effects of DBS, which would be skewed if both variables (medication and stimulation) had been modif ied. The STN patient population was thus ideal for demonstration that stimulation and medication effects could be independent. However, the study sample does limit the conclusions that can be drawn from
79 comparisons of STN and GPi targets, and it would be interesting to repeat the investigation in a center with a bias toward aggressive medication reduction post DBS. The results of the study revealed that patients with more severe ICDs, as reflected in the QUIP score, were more likely to be implanted in the GPi. It would be interesting in a future study to match this variable in each group at baseline. Despite these limitations our study has several strengths, including its prospective approach, relatively large sample size, high rate of follow up, and the fa ct that patients were enrolled consecutively from a single center. In conclusion, we show that ICDs may improve, worsen, or emerge in PD patients following DBS surgery of either the STN or the GPi. Our prospective results thus confirm the findings of prev ious retrospective studies. There were no meaningful differences in the STN and GPi targets in terms of post operative changes in ICD symptoms. Importantly, we provide evidence that post DBS changes in impulsivity cannot be attributed solely to changes in dopaminergic medication. Overall, our findings suggest that STN and GPi DBS can evoke positive or negative changes in impulse control. Clinical approaches to DBS target and patient selection should take this into account.
80 Figure 31. Comparison of PreD BS and Post DBS Prevalence of ICDs Data are represented as numbers of patients. P values in bold indicate statistical significant at 95% confidence (p< 05); p values were calculated using statistics.
81 Figure 32 ICD changes after STN and GPi DBS su rgery. Changes in prevalence of ICDs, ICD subtypes, and DDS after 6 12 months of unilateral or bilateral DBS of GPi or STN in 50 patients A ). ICD diagno sis conversions following GPi B ) and STN C ) DBS. Abbreviations: ICD+= met diagnostic criteria for an imp ulse control disorder, Gmb=gambling, Sxl=hypersexuality, Buy=pathological buying/shopping, Eat=binge eating, Oth=other activity or hobbyism, DDS=dopamine dysregulation syndrome
82 Figure 33 Change in ICD status following DBS is not dependent on dopamine therapy Linear regression plots between A) LEDD and QUIP score and between DA B) LEDD and QUIP score showed no significant correlation between the variables for preDBS or post DBS periods. C) The mean percent change in dopamine agonist dosage was compared for patients who changed ICD status post operatively and those for whom ICD status was maintained. Error bars indicate SEM.
83 Figure 34. Comparison of Impulsive vs. NonImpulsive Patients in STN and GPi. Data are represented as mean standard deviat ion. Statistical comparisons between ICD+ and ICD populations are based on preoperative diagnosis. Bold p values indicate statistical significance at 95% confidence (p<05). P for continuous variables were calculated using unpaired t tests. Abbreviations: Diz. Dur= disease duration, yrs= years, f/u=follow up, UPDRS=Unified Parkinsons Disease Rating Scale, OFF=off medication, ON=on medication, DRS=dementia rating scale, FSIQ=ful l scale intelligence quotient, WMI=Wechsler memory index, PSI=psychological screening inventory, QUIP=questionnaire for impulsivity in Parkinsons disease, LEDD= levodopa equivalent daily dosage, DA= dopamine agonist, DBS= deep brain stimulation, ICD=impul se control disorder, GPi=globus pallidus internus, STN=subthalamic nucleus.
84 Figure 35. Changes in impulsivity following GPi and STN DBS Comparison of changes in ICD symptoms by target, given in number of cases. Resolved and de novo ICDs indicate a ch ange in ICD diagnosis based on quantitative changes in the QUIP score that passed predefined thresholds and corresponding findings during clinical interview. Qualitative improvement and worsening indicate directional changes in ICD symptoms that did not c onstitute a diagnosis conversion based on QUIP score and clinical interview. No ) indicates patients without ICD at baseline that did not convert diagnosis post operatively. Bold p values indicate statistical significance at 95% confidence ( p < 05). P values calculated using test.
85 Figure 36. Impulsivity outcomes and dopamine therapy i n bilateral vs. unilateral DBS. Sub analysis comparing patients who underwent unilateral and bilateral DBS. Bold p values indicate statistical significance at 95% confidence ( p < 05). P values for categorical variables calculated using test, continuous variables calculated using unpaired t test.
86 LIST OF REFERENCES 1. Hikosaka, O., Bromberg Martin, E., Hong, S. & Matsumoto, M. New insights on the subcortical representation of reward. Curr. Opin. Neurobiol. 18, 203 208 (2008). 2. Breysse, E., Pelloux, Y. & Baunez, C. The Good and Bad Differentially Encoded within the Subthalamic Nucleus in Rats. eNeuro 2, (2015). 3. Rossi, P. J., Gunduz, A. & Okun, M. S. The Subthalamic Nucleus, Lim bic Function, and Impulse Control. Neuropsychol. Rev. 25, 398 410 (2015). 4. Teagarden, M. A. & Rebec, G. V. Subthalamic and striatal neurons concurrently process motor, limbic, and associative information in rats performing an operant task. J. Neurophysiol. 97, 2042 2058 (2007). 5. Lardeux, S. & Baunez, C. Alcohol Preference Influences the Subthalamic Nucleus Control on Motivation for Alcohol in Rats. Neuropsychopharmacology 33, 634 642 (2007). 6. Lardeux, S., Paleressompoulle, D., Pernaud, R., Cador, M. & Baunez, C. Different populations of subthalamic neurons encode cocaine vs. sucrose reward and predict future error. J. Neurophysiol. 110, 1497 1510 (2013). 7. Darbaky, Y., Baunez, C., Arecchi, P., Legallet, E. & Apicella, P. Rewardrelated neuronal activity in the subthalamic nucleus of the monkey. Neuroreport 16, 1241 1244 (2005). 8. EspinosaParrilla, J. F., Baunez, C. & Apicella, P. Linking reward processing to behavioral output: motor and motivational integration in the primate subthalamic nucleus. Front. Comput. Neurosci. 7, 175 (2013). 9. EspinosaParrilla, J. F., Baunez, C. & Apicella, P. Modulation of neuronal activity by reward identity in the monkey subthalamic nucleus. Eur. J. Neurosci. 42, 1705 1717 (2015). 10. Lambert, C. et al. Confirmation of functional zones within the human subthalamic nucleus: Patterns of connectivity and subparcellation using diffusion weighted imaging. Neuroimage 60, 83 94 (2012). 11. Karachi, C. et al. The pallidosubthalamic projection: an anatomical substrate for nonmotor functions of the subthalamic nucleus in primates. Mov. Disord. Off. J. Mov. Disord. Soc. 20, 172 180 (2005). 12. Joel, D. & Weiner, I. The connections of the primate subthalamic nucleus: indirect pathways and the openinterconnect ed scheme of basal gangliathalamocortical circuitry. Brain Res. Brain Res. Rev. 23, 62 78 (1997).
87 13. Alkemade, A., Schnitzler, A. & Forstmann, B. U. Topographic organization of the human and nonhuman primate subthalamic nucleus. Brain Struct. Funct. 220, 3075 3086 (2015). 14. Lambert, C. et al. Do we need to revise the tripartite subdivision hypothesis of the human subthalamic nucleus (STN)? Response to Alkemade and Forstmann. NeuroImage 110, 1 2 (2015). 15. Mallet, L. et al. Stimulation of subterrito ries of the subthalamic nucleus reveals its role in the integration of the emotional and motor aspects of behavior. Proc. Natl. Acad. Sci. U. S. A. 104, 10661 10666 (2007). 16. Greenhouse, I. et al. Stimulation at dorsal and ventral electrode contacts tar geted at the subthalamic nucleus has different effects on motor and emotion functions in Parkinsons disease. Neuropsychologia 49, 528 534 (2011). 17. Keuken, M. C. et al. Are there three subdivisions in the primate subthalamic nucleus? Front. Neuroanat. 6, 14 (2012). 18. Alkemade, A. & Forstmann, B. U. Do we need to revise the tripartite subdivision hypothesis of the human subthalamic nucleus (STN)? NeuroImage 95, 326 329 (2014). 19. Hong, S. & Hikosaka, O. Diverse sources of reward value signals in the basal ganglia nuclei transmitted to the lateral habenula in the monkey. Front. Hum. Neurosci. 7, 778 (2013). 20. Matsumoto, M. & Hikosaka, O. Lateral habenula as a source of negative reward signals in dopamine neurons. Nature 447, 1111 1115 (2007). 21. Arkadir, D., Morris, G., Vaadia, E. & Bergman, H. Independent coding of movement direction and reward prediction by single pallidal neurons. J. Neurosci. Off. J. Soc. Neurosci. 24, 10047 10056 (2004). 22. Gdowski, M. J., Miller, L. E., Bastianen, C. A., N enonene, E. K. & Houk, J. C. Signaling Patterns of Globus Pallidus Internal Segment Neurons During Forearm Rotation. Brain Res. 1155, 56 69 (2007). 23. Bischoff Grethe, A. et al. Striatal and Pallidal Activation during Reward Modulated Movement Using a Tr anslational Paradigm. J. Int. Neuropsychol. Soc. JINS 21, 399 411 (2015). 24. Karachi, C. et al. Three dimensional cartography of functional territories in the human striatopallidal complex by using calbindin immunoreactivity. J. Comp. Neurol. 450, 122 13 4 (2002).
88 25. Nair, G., Evans, A., Bear, R. E., Velakoulis, D. & Bittar, R. G. The anteromedial GPi as a new target for deep brain stimulation in obsessive compulsive disorder. J. Clin. Neurosci. Off. J. Neurosurg. Soc. Australas. 21, 815 821 (2014). 26. Cif, L. et al. Anteroventral internal pallidum stimulation improves behavioral disorders in Lesch Nyhan disease. Mov. Disord. 22, 2126 2129 (2007). 27. Lane, R. D., Chua, P. M. & Dolan, R. J. Common effects of emotional valence, arousal and attention on neural activation during visual processing of pictures. Neuropsychologia 37, 989 997 (1999). 28. Chowdhury, R., Guitart Masip, M., Lambert, C., Dolan, R. J. & Dzel, E. Structural integrity of the substantia nigra and subthalamic nucleus predicts flexibi lity of instrumental learning in older age individuals. Neurobiol. Aging 34, 2261 2270 (2013). 29. GuitartMasip, M. et al. Go and nogo learning in reward and punishment: interactions between affect and effect. NeuroImage 62, 154 166 (2012). 30. GuitartMasip, M., Duzel, E., Dolan, R. & Dayan, P. Action versus valence in decision making. Trends Cogn. Sci. 18, 194 202 (2014). 31. Ruiz, M. H. et al. Involvement of Human Internal Globus Pallidus in the Early Modulation of Cortical Error Related Activity. C ereb. Cortex 24, 1502 1517 (2014). 32. Lardeux, S., Pernaud, R., Paleressompoulle, D. & Baunez, C. Beyond the reward pathway: coding reward magnitude and error in the rat subthalamic nucleus. J. Neurophysiol. 102, 2526 2537 (2009). 33. GuitartMasip, M. et al. Action controls dopaminergic enhancement of reward representations. Proc. Natl. Acad. Sci. 109, 7511 7516 (2012). 34. van Wouwe, N. C. et al. Easy to learn, hard to suppress: The impact of learned stimulus outcome associations on subsequent action control. Brain Cogn. 101, 17 34 (2015). 35. Postuma, R. B. et al. MDS clinical diagnostic criteria for Parkinsons disease: MDS PD Clinical Diagnostic Criteria. Mov. Disord. 30, 1591 1601 (2015). 36. Sudhyadhom, A., Okun, M. S., Foote, K. D., Rahman, M. & Bova, F. J. A Threedimensional Deformable Brain Atlas for DBS Targeting. I. Methodology for Atlas Creation and Artifact Reduction. Open Neuroimaging J. 6, 92 98 (2012). 37. Sarma, S. V. et al. The effects of cues on neurons in the basal ganglia in Park insons disease. Front. Integr. Neurosci. 6, (2012).
89 38. Schalk, G., McFarland, D. J., Hinterberger, T., Birbaumer, N. & Wolpaw, J. R. BCI2000: a general purpose braincomputer interface (BCI) system. IEEE Trans. Biomed. Eng. 51, 1034 1043 (2004). 39. To leikis, J. R. et al. Effect of intraoperative subthalamic nucleus DBS on human single unit activity in the ipsilateral and contralateral subthalamic nucleus. J. Neurosurg. 116, 1134 1143 (2012). 40. Wagenbreth, C. et al. Deep brain stimulation of the subt halamic nucleus modulates reward processing and action selection in Parkinson patients. J. Neurol. 262, 1541 1547 (2015). 41. Zimnik, A. J., Nora, G. J., Desmurget, M. & Turner, R. S. Movement related discharge in the macaque globus pallidus during highf requency stimulation of the subthalamic nucleus. J. Neurosci. Off. J. Soc. Neurosci. 35, 3978 3989 (2015). 42. G Schaltenbrand & Bailey, P. Introduction to stereotaxis with an atlas of the human brain. III, (Georg Thieme Verlag, 1959). 43. Plaha, P., Ben Shlomo, Y., Patel, N. K. & Gill, S. S. Stimulation of the caudal zona incerta is superior to stimulation of the subthalamic nucleus in improving contralateral parkinsonism. Brain 129, 1732 1747 (2006). 44. Sterio, D. et al. Neurophysiological refinement of subthalamic nucleus targeting. Neurosurgery 50, 58 67; discussion 67 69 (2002). 45. Hutchison, W. D. et al. Neurophysiological identification of the subthalamic nucleus in surgery for Parkinsons disease. Ann. Neurol. 44, 622 628 (1998). 46. Bejjani, B. P. et al. Bilateral subthalamic stimulation for Parkinsons disease by using threedimensional stereotactic magnetic resonance imaging and electrophysiological guidance. J. Neurosurg. 92, 615 625 (2000). 47. Bastin, J. et al. Inhibitory control and err or monitoring by human subthalamic neurons. Transl. Psychiatry 4, e439 (2014). 48. Lee, J. I. et al. Internal Pallidal Neuronal Activity During Mild Drug Related Dyskinesias in Parkinsons Disease: Decreased Firing Rates and Altered Firing Patterns. J. Ne urophysiol. 97, 2627 2641 (2007). 49. Levy, R. et al. Effects of Apomorphine on Subthalamic Nucleus and Globus Pallidus Internus Neurons in Patients With Parkinsons Disease. J. Neurophysiol. 86, 249 260 (2001). 50. Tang, J. K. H. et al. Neuronal firing rates and patterns in the globus pallidus internus of patients with cervical dystonia differ from those with Parkinsons disease. J. Neurophysiol. 98, 720 729 (2007).
90 51. Smith, K. S. & Berridge, K. C. Opioid Limbic Circuit for Reward: Interaction between Hedonic Hotspots of Nucleus Accumbens and Ventral Pallidum. J. Neurosci. 27, 1594 1605 (2007). 52. Tindell, A. J., Smith, K. S., Pecia, S., Berridge, K. C. & Aldridge, J. W. Ventral Pallidum Firing Codes Hedonic Reward: When a Bad Taste Turns Good. J. N europhysiol. 96, 2399 2409 (2006). 53. Zahodne, L. B. et al. Greater improvement in quality of life following unilateral deep brain stimulation surgery in the globus pallidus as compared to the subthalamic nucleus. J. Neurol. 256, 1321 1329 (2009). 54. V idenovic, A. & Metman, L. V. Deep brain stimulation for Parkinsons disease: prevalence of adverse events and need for standardized reporting. Mov. Disord. Off. J. Mov. Disord. Soc. 23, 343 349 (2008). 55. Fukaya, C. & Yamamoto, T. Deep brain stimulation for Parkinsons disease: recent trends and future direction. Neurol. Med. Chir. (Tokyo) 55, 422 431 (2015). 56. Zangaglia, R. et al. Deep brain stimulation and cognitive functions in Parkinsons disease: A threeyear controlled study. Mov. Disord. Off. J. Mov. Disord. Soc. 24, 1621 1628 (2009). 57. Witt, K. et al. Neuropsychological and psychiatric changes after deep brain stimulation for Parkinsons disease: a randomised, multicentre study. Lancet Neurol. 7, 605 614 (2008). 58. Witt, K. et al. Deep brai n stimulation of the subthalamic nucleus improves cognitive flexibility but impairs response inhibition in Parkinson disease. Arch. Neurol. 61, 697 700 (2004). 59. Hershey, T. et al. Stimulation of STN impairs aspects of cognitive control in PD. Neurology 62, 1110 1114 (2004). 60. During, M. J., Kaplitt, M. G., Stern, M. B. & Eidelberg, D. Subthalamic GAD gene transfer in Parkinson disease patients who are candidates for deep brain stimulation. Hum. Gene Ther. 12, 1589 1591 (2001). 61. Lvesque, J. C. & Parent, A. GABAergic interneurons in human subthalamic nucleus. Mov. Disord. Off. J. Mov. Disord. Soc. 20, 574 584 (2005). 62. Tarsy, D., Vitek, J. L., Starr, P. & Okun, M. Deep Brain Stimulation in Neurological and Psychiatric Disorders (Springer Scienc e & Business Media, 2008). 63. Purves, D. et al. Projections from the Basal Ganglia to Other Brain Regions. (2001).
91 64. Vasques, X. et al. Prognostic value of globus pallidus internus volume in primary dystonia treated by deep brain stimulation. J. Neurosurg. 110, 220 228 (2008). 65. Patil, P. G. M. D., Carmena, J. M., Nicolelis, M. A. L. M. D. & Turner, D. A. Ensemble Recordings Of Human Subcortical Neurons as a Source Of Motor Control Signals For a BrainMachine Interface. Neurosurg. July 2004 55, 27 3 8 (2004). 66. Cleary, D. R., Ozpinar, A., Raslan, A. M. & Ko, A. L. Deep brain stimulation for psychiatric disorders: where we are now. Neurosurg. Focus 38, E2 (2015). 67. Mallet, L. et al. Subthalamic nucleus stimulation in severe obsessivecompulsive d isorder. N. Engl. J. Med. 359, 2121 2134 (2008). 68. Vu, A. T. et al. High resolution whole brain diffusion imaging at 7 T for the Human Connectome Project. NeuroImage 122, 318 331 (2015). 69. Nowinski, W. L., Liu, J. & Thirunavuukarasuu, A. Quantificati on and visualization of the threedimensional inconsistency of the subthalamic nucleus in the SchaltenbrandWahren brain atlas. Stereotact. Funct. Neurosurg. 84, 46 55 (2006). 70. Nowinski, W. L., Liu, J. & Thirunavuukarasuu, A. Quantification and visuali zation of the threedimensional inconsistency of the subthalamic nucleus in the SchaltenbrandWahren brain atlas. Stereotact. Funct. Neurosurg. 84, 46 55 (2006). 71. Bugalho, P. & OliveiraMaia, A. J. Impulse control disorders in Parkinsons disease: cros sroads between neurology, psychiatry and neuroscience. Behav. Neurol. 27, 547 557 (2013). 72. Phu, A. L. et al. Effect of impulse control disorders on disability and quality of life in Parkinsons disease patients. J. Clin. Neurosci. Off. J. Neurosurg. Soc. Australas. 21, 63 66 (2014). 73. Kim, Y. E. et al. Impulse control and related behaviors after bilateral subthalamic stimulation in patients with Parkinsons disease. J. Clin. Neurosci. 20, 964 969 (2013). 74. Kadastik Eerme, L., Rosenthal, M., Paju, T., Muldmaa, M. & Taba, P. Healthrelated quality of life in Parkinsons disease: a cross sectional study focusing on nonmotor symptoms. Health Qual. Life Outcomes 13, (2015). 75. Ondo, W. G. & Lai, D. Predictors of impulsivity and reward seeking behavior with dopamine agonists. Parkinsonism Relat. Disord. 14, 28 32 (2008). 76. Leroi, I. et al. Carer burden in apathy and impulse control disorders in Parkinsons disease. Int. J. Geriatr. Psychiatry 27, 160 166 (2012).
92 77. Moum, S. J. et al. Effects of ST N and GPi deep brain stimulation on impulse control disorders and dopamine dysregulation syndrome. PloS One 7, e29768 (2012). 78. Aleksandrova, L. R. et al. Deep brain stimulation of the subthalamic nucleus increases premature responding in a rat gambling task. Behav. Brain Res. 245, 76 82 (2013). 79. Desbonnet, L. et al. Premature responding following bilateral stimulation of the rat subthalamic nucleus is amplitude and frequency dependent. Brain Res. 1008, 198 204 (2004). 80. Baunez, C., Amalric, M. & Robbins, T. W. Enhanced FoodRelated Motivation after Bilateral Lesions of the Subthalamic Nucleus. J. Neurosci. 22, 562 568 (2002). 81. Baunez, C., Dias, C., Cador, M. & Amalric, M. The subthalamic nucleus exerts opposite control on cocaine and natural rewards. Nat. Neurosci. 8, 484 489 (2005). 82. Rouaud, T. et al. Reducing the desire for cocaine with subthalamic nucleus deep brain stimulation. Proc. Natl. Acad. Sci. U. S. A. 107, 1196 1200 (2010). 83. Summerson, S. R., Aazhang, B. & Kemere, C. T. Ch aracterizing Motor and Cognitive Effects Associated With Deep Brain Stimulation in the GPi of Hemi Parkinsonian Rats. IEEE Trans. Neural Syst. Rehabil. Eng. 22, 1218 1227 (2014). 84. Plessow, F., Fischer, R., Volkmann, J. & Schubert, T. Subthalamic deep brain stimulation restores automatic response activation and increases susceptibility to impulsive behavior in patients with Parkinsons disease. Brain Cogn. 87, 16 21 (2014). 85. Lardeux, S., Paleressompoulle, D., Pernaud, R., Cador, M. & Baunez, C. Different populations of subthalamic neurons encode cocaine vs. sucrose reward and predict future error. J. Neurophysiol. 110, 1497 1510 (2013). 86. Tachibana, Y. & Hikosaka, O. The primate ventral pallidum encodes expected reward value and regulates motor act ion. Neuron 76, 826 837 (2012). 87. Smith, K. S., Tindell, A. J., Aldridge, J. W. & Berridge, K. C. Ventral pallidum roles in reward and motivation. Behav. Brain Res. 196, 155 167 (2009). 88. Howell, N. A. et al. Preliminary evidence for human globus pal lidus pars interna neurons signaling reward and sensory stimuli. Neuroscience doi:10.1016/j.neuroscience.2016.04.020 89. Porat, O., Hassin Baer, S., Cohen, O. S., Markus, A. & Tomer, R. Asymmetric dopamine loss differentially affects effort to maximize gain or minimize loss. Cortex J. Devoted Study Nerv. Syst. Behav. 51, 82 91 (2014).
93 90. Dong, G., Hu, Y. & Lin, X. Reward/punishment sensitivities among internet addicts: Implications for their addictive behaviors. Prog. Neuropsychopharmacol. Biol. Psychiat ry 46, 139 145 (2013). 91. Potts, G. F., Bloom, E. L., Evans, D. E. & Drobes, D. J. Neural reward and punishment sensitivity in cigarette smokers. Drug Alcohol Depend. 144, 245 253 (2014). 92. Gaher, R. M., Hahn, A. M., Shishido, H., Simons, J. S. & Gast er, S. Associations between sensitivity to punishment, sensitivity to reward, and gambling. Addict. Behav. 42, 180 184 (2015). 93. Frank, M. J., Samanta, J., Moustafa, A. A. & Sherman, S. J. Hold your horses: impulsivity, deep brain stimulation, and medic ation in parkinsonism. Science 318, 1309 1312 (2007). 94. Piray, P. et al. Impulse control disorders in Parkinsons disease are associated with dysfunction in stimulus valuation but not action valuation. J. Neurosci. Off. J. Soc. Neurosci. 34, 7814 7824 ( 2014). 95. Balasubramani, P. P., Chakravarthy, V. S., Ali, M., Ravindran, B. & Moustafa, A. A. Identifying the Basal Ganglia Network Model Markers for MedicationInduced Impulsivity in Parkinsons Disease Patients. PLoS ONE 10, (2015). 96. Seeman, P. Par kinsons disease treatment may cause impulsecontrol disorder via dopamine D3 receptors. Synap. N. Y. N 69, 183 189 (2015). 97. Papay, K. et al. Patient versus informant reporting of ICD symptoms in Parkinsons disease using the QUIP: Validity and variabi lity. Parkinsonism Relat. Disord. 17, 153 155 (2011). 98. Weintraub, D. et al. Validation of the questionnaire for impulsivecompulsive disorders in Parkinsons disease. Mov. Disord. Off. J. Mov. Disord. Soc. 24, 1461 1467 (2009). 99. Papay, K. et al. Pa tient versus informant reporting of ICD symptoms in Parkinsons disease using the QUIP: validity and variability. Parkinsonism Relat. Disord. 17, 153 155 (2011). 100. Weintraub, D. et al. Dopamine Agonist Use is Associated with Impulse Control Disorders in Parkinsons Disease. Arch. Neurol. 63, 969 973 (2006). 101. Weintraub, D., Papay, K. & Siderowf, A. Screening for impulse control symptoms in patients with de novo Parkinson disease. Neurology 80, 176 180 (2013). 102. Grant, J. E., Steinberg, M. A., Kim, S. W., Rounsaville, B. J. & Potenza, M. N. Preliminary validity and reliability testing of a structured clinical interview for pathological gambling. Psychiatry Res. 128, 79 88 (2004).
94 103. Garcia Ruiz, P. J. et al. Impulse control disorder in patients wit h Parkinsons disease under dopamine agonist therapy: a multicentre study. J. Neurol. Neurosurg. Psychiatry 85, 840 844 (2014). 104. Galvan, A. et al. Localization and function of dopamine receptors in the subthalamic nucleus of normal and parkinsonian monkeys. J. Neurophysiol. 112, 467 479 (2014). 105. Weintraub, D. et al. Impulse control disorders in Parkinson disease: a cross sectional study of 3090 patients. Arch. Neurol. 67, 589 595 (2010). 106. Weintraub, D., David, A. S., Evans, A. H., Grant, J. E. & Stacy, M. Clinical spectrum of impulse control disorders in Parkinsons disease. Mov. Disord. Off. J. Mov. Disord. Soc. 30, 121 127 (2015). 107. Garcia Ruiz, P. J. et al. Impulse control disorder in patients with Parkinsons disease under dopamine agonist therapy: a multicentre study. J. Neurol. Neurosurg. Psychiatry 85, 840 844 (2014). 108. Amami, P. et al. Impulse control behaviours in patients with Parkinsons disease after subthalamic deep brain stimulation: de novo cases and 3year follow up. J. Neurol. Neurosurg. Psychiatry 86, 562 564 (2015). 109. Okun, M. S. & Weintraub, D. Should impulse control disorders and dopamine dysregulation syndrome be indications for deep brain stimulation and intestinal levodopa? Mov. Disord. Off. J. Mov. Disord. Soc. 28, 1915 1919 (2013). 110. Lhomme, E. et al. Subthalamic stimulation in Parkinsons disease: restoring the balance of motivated behaviours. Brain 135, 1463 1477 (2012). 111. Kim, Y. E. et al. Impulse control and related behaviors after bilateral subthalamic stimulation in patients with Parkinsons disease. J. Clin. Neurosci. Off. J. Neurosurg. Soc. Australas. 20, 964 969 (2013). 112. Broen, M., Duits, A., Visser Vandewalle, V., Temel, Y. & Winogrodzka, A. Impulse control and related disorders in Parkinsons d isease patients treated with bilateral subthalamic nucleus stimulation: A review. Parkinsonism Relat. Disord. 17, 413 417 (2011). 113. Demetriades, P., Rickards, H. & Cavanna, A. E. Impulse Control Disorders Following Deep Brain Stimulation of the Subthala mic Nucleus in Parkinsons Disease: Clinical Aspects. Park. Dis. 2011, (2011). 114. Voon, V., Kubu, C., Krack, P., Houeto, J. L. & Trster, A. I. Deep brain stimulation: neuropsychological and neuropsychiatric issues. Mov. Disord. Off. J. Mov. Disord. Soc. 21 Suppl 14, S305 327 (2006).
95 115. Hlbig, T. D. et al. Subthalamic deep brain stimulation and impulse control in Parkinsons disease. Eur. J. Neurol. 16, 493 497 (2009). 116. Gee, L. et al. The Influence of Bilateral Subthalamic Nucleus Deep Brain Stimul ation on Impulsivity and Prepulse Inhibition in Parkinsons Disease Patients. Stereotact. Funct. Neurosurg. 93, 265 270 (2015). 117. Okun, M. S. et al. Acute and Chronic Mood and Apathy Outcomes from a Randomized Study of Unilateral STN and GPi DBS. PLoS O NE 9, (2014). 118. Nirenberg, M. J. Dopamine agonist withdrawal syndrome and nonmotor symptoms after Parkinsons disease surgery. Brain 133, e155 e155 (2010). 119. Thobois, S. et al. Non motor dopamine withdrawal syndrome after surgery for Parkinsons dis ease: predictors and underlying mesolimbic denervation. Brain 133, 1111 1127 (2010). 120. Samuel, M. et al. Management of impulse control disorders in Parkinsons disease: Controversies and future approaches. Mov. Disord. Off. J. Mov. Disord. Soc. 30, 150 159 (2015). 121. Zahodne, L. B. et al. Binge eating in Parkinsons disease: prevalence, correlates and the contribution of deep brain stimulation. J. Neuropsychiatry Clin. Neurosci. 23, 56 62 (2011). 122. Frank, M. J. Hold your horses: a dynamic computatio nal role for the subthalamic nucleus in decision making. Neural Netw. Off. J. Int. Neural Netw. Soc. 19, 1120 1136 (2006). 123. Setiawan, M. et al. Referrals for movement disorder surgery: under representation of females and reasons for refusal. Can. J. Ne urol. Sci. J. Can. Sci. Neurol. 33, 53 57 (2006). 124. Hamberg, K. & Hariz, G.M. The decisionmaking process leading to deep brain stimulation in men and women with parkinsons disease an interview study. BMC Neurol. 14, 89 (2014).
96 BIOGRAPHICAL SKETCH P. Justin Rossi is an M.D./ Ph.D. candidate in the Department of Neuroscience, University of Florida College of Medicine. He was an associate fellow at the Potomac Institute for Policy Studies in Arlington, Virginia. He holds a bachelors degree in history from Harvard University, where he was awarded the Thomas Temple Hoopes Prize for his senior thesis on piracy in the Mediterranean.