neuropsychiatric effects of subthalamic neurostimulation in Parkinson disease

Transcription

1 reviews neuropsychiatric effects of subthalamic neurostimulation in Parkinson disease Jens Volkmann, Christine Daniels and Karsten Witt Abstract Neurostimulation of the subthalamic nucleus (STN) is an established treatment for motor symptoms in advanced Parkinson disease (PD), although concerns exist regarding the safety of this therapy in terms of cognitive and psychiatric adverse effects. The basal ganglia are considered to be part of distributed cortico-subcortical networks that are involved in the selection, facilitation and inhibition of movements, emotions, behaviors and thoughts. The STN has a central role in these networks, probably providing a global no-go signal. The behavioral and cognitive effects observed following STN high-frequency stimulation (HFS) probably reflect the intrinsic role of this nucleus in nonmotor functional domains. Nevertheless, postoperative behavioral changes are seldom caused by such stimulation alone. PD is a progressive neurodegenerative disorder with motor, cognitive, behavioral and autonomic symptoms. The pattern of neurodegeneration and expression of these symptoms are highly variable across individuals. The preoperative neuropsychiatric state can be further complicated by sensitization phenomena resulting from long-term dopaminergic treatment, which include impulse control disorders, punding, and addictive behaviors (dopamine dysregulation syndrome). Finally, personality traits, the social environment, culture and learned behaviors might be important determinants explaining why behavioral symptoms differ between patients after surgery. Here, we summarize the neuropsychiatric changes observed after STN HFS and try to disentangle their various etiologies. Volkmann, J. et al. Nat. Rev. Neurol. advance online publication 3 August 2010; doi: /nrneurol Introduction Over the past two decades, surgery has become an accepted treatment for movement disorders. the develop ment of functional neurosurgery for Parkinson disease (PD) probably represents the second most important advance in the treatment of this condition after the introduction of levodopa. in patients with PD who exhibit either motor complications arising from long-term levodopa therapy, or severe tremor, deep brain stimula tion (DBs) of the subthalamic nucleus (stn) is now considered to be a routine and evidencebased therapeutic option. indeed, in two large clinical trials, stn neurostimulation was found to be superior to best medical therapy alone for the treatment of motor complications in patients with advanced PD. 1,2 notably, while DBs was associated with marked improvements in quality of life measures throughout the 6 month treatment periods in these studies, medication was only able to maintain a low level of daily functioning in these patients, all of whom displayed severe disabilities. the beneficial effects of stn high-frequency stimulation (HFs) on the motor symptoms of PD are now well established; however, the effects of such therapy on cognition and behavior remain controversial. numerous case reports and cohort studies have described various Competing interests J. Volkmann declares an association with the following company: Medtronic. See the article online for full details of the relationship. The other authors declare no competing interests. behavioral changes following stn HFs including the onset of mania, depression, suicidal ideation, apathy and impulsivity as well as negative effects of this treatment on global cognitive abilities, attention, executive functions, and memory. 3 5 By contrast, larger controlled studies employing global cognitive function testing or neuropsychiatric assessments have failed to demonstrate that this procedure elicits such detrimental effects. 1,2,6 these conflicting views are not, however, mutually exclusive. Behavioral effects caused by stn HFs might not show up in controlled studies that are designed to detect consistent differences in mean scores across groups, as such effects occur with a low frequency, are highly variable, and can be partially opposing in nature. On the basis of our clinical experience, we have little doubt that some patients, despite effective management of motor symptoms with stn HFs, can experience psychiatric adverse effects following this treatment. several questions relating to PD, stn HFs and nonmotor adverse effects remain unanswered. For example, can the increased vulnerability of some patients to cognitive and behavioral adverse effects following stn HFs be explained by their past psychiatric history, the nature of their PD, or their individual response to dopaminergic treatment? Furthermore, does stimulation of basal ganglia structures itself contribute to some or all of these neuropsychiatric sequelae and, if so, which symptoms are specifically affected and what are the underlying mechanisms? recent clinical and experimental studies Department of Neurology, Universitätsklinikum Schleswig-Holstein, Campus Kiel, Arnold- Heller-Strasse 3, Haus 41, Kiel, Germany (J. Volkmann, K. Witt). Department of Psychiatry, Universitätsklinikum Würzburg, Füchsleinstrasse 15, Würzburg, Germany (C. Daniels). Correspondence to: J. Volkmann j.volkmann@ neurologie.uni-kiel.de nature reviews neurology advance OnLine PuBLiCatiOn 1

2 Key points The basal ganglia modulate the activity of distinct frontal cortical areas such as motor and premotor cortices, dorsolateral prefrontal cortex and anterior cingulate cortex via parallel corticobasal ganglia thalamocortical loops Clinical findings support the view that the basal ganglia are involved in the selection, facilitation and inhibition of movements, emotions, behaviors and thoughts The subthalamic nucleus (STN) has a central regulatory role in cortico-subcortical networks involving the basal ganglia, providing a global no-go signal Dopaminergic medication and high-frequency stimulation of the STN act synergistically on the various basal ganglia loops Apathy and depression in Parkinson disease (PD) might reflect hypodopaminergic states of the limbic and associative corticobasal ganglia thalamocortical loops Sensitization and dopaminergic or electrical overstimulation of the limbic and associative corticobasal ganglia thalamocortical loops probably underlie the development of impulsivity and mania in PD examining the role of the basal ganglia in nonmotor behaviors have begun to address these questions. stn neurostimulation is a complex therapy, the success of which depends on the selection of appropriate candidates for surgery, optimal placement of the stimulating electrode, and proper postoperative adjustment of stimulation parameters and antiparkinsonian medication. therapeutic failures or adverse effects can be associated with inappropriate management of any one of these factors. 7 in addition, during postoperative recovery, a patient must adapt to a reduction in the their level of disability, which can lead to adjustment problems and psychoreactive behavioral changes. 8 interpretation of behavioral changes after surgery also requires consideration of an individual s underlying disease neuro pathology, the development of drug-induced sensitization, environmental and cultural factors, learned behaviors, and personality traits. an understanding of the complex interplay between predispositions, surgical factors and post operative issues would require the evaluation of large patient samples to detect consistent response patterns and risk factors in psychiatric and cognitive domains. such epidemiological studies have not been conducted and will probably not be performed in the near future. a more promising approach involves experiments evaluating the acute effects of stimulation and medication challenges through standardized cognitive or behavioral tasks. as patients serve as their own controls in this setting, the effects of interindividual variation are minimized. this approach allows the study of the neuropsychiatric consequences of focal neuro modulation of the stn or the global impact of dopaminergic treatment on brain function in isolation. several previous reviews have addressed the types and prevalence of neurobehavioral changes that are observed after stn HFs. a meta-analysis of the literature pertaining to these effects would be problematic, owing to multi factorial etiologies, reporting bias, the small sample sizes of most studies, and the often uncontrolled and retrospective nature of the observations. an uncritical application of this method can result in prevalence data that are clinically unrealistic 5 and not supported by the few large prospective studies. 1,2,6 in this article, we will examine the various adverse effects associated with stn HFs, discussing the possible etiology of such outcomes, whenever supported by the current literature, to generate testable hypotheses for future research and to provide practical recommendations for the clinical management of patients with neuropsychiatric problems. these objectives will be achieved through interpretation of the literature rather than by presenting a complete collection of all available data. Anatomy and physiology the stn is a component of basal ganglia circuits and has traditionally been considered as a relay station in the indirect pathway controlling thalamocortical excitability. 9,10 now, the stn is also thought to be an important input nucleus of the basal ganglia. 11 such nuclei receive inputs not only from large parts of the frontal cortex, but also from various thalamic and brainstem structures. the basal ganglia via parallel corticobasal ganglia thalamocortical loops modulate the activity of distinct frontal cortical areas, including the motor and premotor cortices, the dorsolateral pre frontal cortex, the anterior cingulate cortex, and the frontal eye field (Figure 1). 9,12,13 as with other basal ganglia nuclei, the stn can be subdivided into functionally segregated territories (motor, oculomotor, associative and limbic). 14,15 the internal circuitry of the basal ganglia, and their connectivity with multiple functionally segregated frontal cortical areas, places these structures in a unique position to select between competing goals, actions and movements. neurocomputational models and behavioral experiments suggest that the basal ganglia have two different roles in decision-making. 16,17 Phasic dopamine signals in the ventral striatum (so called bursts and dips ) provide reward-associated teaching signals, 18 driving go learning (through D 1 receptors and the direct basal ganglia pathway) to seek reward and no-go learning (through D 2 receptors and the indirect basal ganglia pathway) to avoid harmful actions. thus, levodopa-treated patients with PD have specific deficits in learning from negative outcomes, probably through dopaminergic overstimulation. 19 the stn has a complementary role in decision-making by providing an adaptive hold-your-horses signal in the face of decision conflicts (Figure 2). 16,17 the subthalamic no-go signal inhibits automatic responses to a stimulus and allows additional time, proportional to the level of conflict, for central processing of a goal-directed behavior. thus, according to this concept, stn HFs would eliminate the subthalamic no-go signal and facilitate movements, 11 while also potentially leading to cognitive or affective disinhibition and premature responding in situations of high-conflict decisions, depending on electrode location and the circuit being stimulated. the classic rate model of basal ganglia physiology predicted that the net firing rate of basal ganglia nuclei was largely dependent on striatal dopaminergic tone. 20 this model provided the first coherent patho physiological 2 advance OnLine PuBLiCatiOn

3 a b c Hyperdirect pathway Hyperdirect pathway Hyperdirect pathway Thalamus Thalamus Caudate Thalamus Caudate GPe GPe GPe GPi GPi GPi Putamen Putamen Putamen STN STN Figure 1 The basal ganglia form anatomically and functionally segregated neuronal circuits with thalamic nuclei and frontal cortical areas. a The motor circuit involves the motor and supplementary motor cortices, the posterolateral part of the putamen, the posterolateral GPe and GPi, the dorsolateral STN, and the ventrolateral thalamus. b The associative loop and c the limbic loop connect the prefrontal and cingulate cortices with distinct regions within the basal ganglia and thalamus. in the STN, a functional gradient is found, with a motor representation in the dorsolateral aspect of the nucleus, cognitive associative functions in the intermediate zone, and limbic functions in the ventromedial region. Via a hyperdirect pathway, the STN receives direct projections from the motor, prefrontal and anterior cingulate cortices that can detect and integrate response conflicts. 11,104 This pathway is a powerful contact to influence basal ganglia outflow. Abbreviations: GPi, internal globus pallidus; GPe, external globus pallidus; STN, subthalamic nucleus. Permission obtained from John Wiley & Sons, inc. Obeso, J. A. et al. Mov. Disord. 23 (Suppl. 3), S548 S559 (2008); STN concepts for hypokinetic and hyperkinetic movement disorders and led to the renaissance of functional neurosurgery in PD. according to the model, akinesia (formerly regarded as a loss-of-function symptom) is associated with abnormal increases in neuronal acti vity in the internal globus pallidus and stn that might be addressed by stereotactic lesioning or inhibitory stimulation. 21 nevertheless, the strict conceptualization of the basal ganglia as a rate-driven, negative-feedback system was incapable of explaining many of the post operative clinical findings in movement disorders, which together have been described as the paradox of basal ganglia surgery. 22 For example, pallidal lesions, which reduce the output tone of the basal ganglia, should not be able to simultaneously treat akinesia and dyskinesia; however, this dual effect is observed clinically. thus, in contrast to the rate model, more-recent pathophysio logical concepts in PD have emphasized additional changes in the pattern ing and synchronization of basal ganglia activity, 9,22 24 the roles of which still need to be explored in nonmotor behaviors. as the extent of neurodegeneration in PD is uneven across the substantia nigra, predominantly affecting the a9 region and, to a lesser extent, the a8 and meso limbic a10 regions, the various corticobasal ganglia thalamocortical loops are differentially affected by a hypodopaminergic striatal state. 25 this finding might explain the variation in motor and nonmotor manifestations across patients. in addition, dopaminergic medication, which is titrated to restore motor function, might simultaneously overstimulate nonmotor striatal areas that only have a low level of denervation. indeed, the overdosing of limbic and orbitofrontal basal ganglia circuits that show minimal dopamine depletion in the early stages of PD has been implicated in the deterioration of some cognitive functions with dopaminergic treatment. 16 Moreover, sensitization resulting from overstimulation has been recognized as an important factor in the develop ment of dopaminergic treatment-induced adverse behavioral effects, including punding, impulse control disorders (icds) and addictive drug seeking (dopamine dysregulation syndrome, or Dss) Cognition Cognitive deficits ranging from mild frontal executive dysfunction to frank dementia constitute part of the nonmotor spectrum of PD symptoms, with the progression of such deficits reflecting the natural course of the disease. 29 the term executive function refers to a psychological process that is involved in handling novel situations outside the realms of more-automatized stimulus-response patterns. such processes include higher-order behaviors such as decision-making, problem-solving, maintaining or shifting attention, and inhibition of habitual responses. in one study, detailed nature reviews neurology advance OnLine PuBLiCatiOn 3

4 Sensory input Pre-SMA R1 R2 R3 M1 motor input DBS Hyperdirect pathway (conflict signal) Striatum R1 R2 R1 R2 Thalamus STN Go No go GPe D1 D2 SNc GPi Modulatory signal Inhibitory signal Excitatory signal Figure 2 Neural network model of the striato-thalamocortical circuit. This neurocomputational model 19 outlines the two complementary functions of the basal ganglia in decision-making. Neuronal units are represented by squares, with increasing box height signifying increasing neuronal activity. Under the influence of phasic dopaminergic signals from the SNc, striatal go units facilitate cortical responses (for example, r1), while no-go units suppress competing cortical response possibilities (for example, r2 or r3). The STN provides a global no-go signal, which is proportional to the degree of response conflict at the cortical level, as indicated by the pre-sma. Through diffuse excitatory projections from the STN to the GPe and GPi, the thalamocortical relay can be halted to allow more time for cognitive processing in high-conflict situations. in this example, conflict is low, because only a single cortical response (r1) is active. r1 is selected by the network and sent to motor output, because the cumulative activity of the go units in the striatum outweighs the no-go activity. DBS of the STN would interfere with normal response slowing and, hence, facilitate impulsive choices. 19 Abbreviations: DBS, deep brain stimulation; GPe, external globus pallidus; GPi, internal globus pallidus; SNc, substantia nigra pars compacta; STN, subthalamic nucleus; pre-sma, presupplementary motor area. neuropsychological testing revealed that mild cognitive disturbances were found in 25% of de novo PD cases, and mainly comprised deficits in executive functioning, memory and psychomotor speed. 30 a follow-up study showed that after a 3 year period of disease progression, 50% of patients had further declines in measures of psycho motor speed and attention, and 9% of patients had developed dementia. 31 Many symptoms of dysexecutive syndrome have been shown to improve with levodopa therapy. 32 such observations indicated that these symptoms are linked to the hypodopaminergic state of the basal ganglia and are not primarily caused by extranigral neurodegeneration. the cognitive response profile of dopaminergic therapy, however, is now thought to be a u-shaped curve, with the optimal dose range occurring at intermediate drug concentrations. Detrimental effects develop at low and high doses, and are caused by understimulation and overstimulation, respectively, of the associative basal ganglia loop. 16,32 Chronic treatment effects as most studies investigating the impact of stn HFs on cognition did not include a control group, the risk of cognitive deterioration resulting from surgery or neuro stimulation could not reliably be disentangled from cognitive decline related to disease progression. Furthermore, most studies only involved small sample sizes and, hence, were inadequately powered to detect even large effects. 33 Following stn HFs, some researchers have reported declines in individual cognitive domains, including verbal memory, psychomotor speed 41 and visuo spatial memory. 41,42 Meanwhile, other investigators have even reported a deterioration in global cognitive function, suggesting that stn HFs might hasten the onset of dementia, particularly in older patients (individuals aged >70 years). 41 nevertheless, improvements in cognitive functioning, including enhancement of both mental flexibility 34 and visuomotor sequencing, have also been described after stn HFs. 43,44 the results of a meta-analysis 4 and two large-scale clinical trials, 2,6 which included randomized control groups of patients treated by best medical therapy, did not support the finding that stn HFs causes global cognitive deterioration. Moreover, a prospective study found that the incidence of dementia over a 3 year period after stn HFs was similar to the incidence of this condition in medically treated patients, and suggested that dementia was secondary to the natural progression of PD rather than being caused by 4 advance OnLine PuBLiCatiOn

5 neurostimulation. 45 nevertheless, a moderate decrease in verbal fluency and a small but marked deterioration in stroop test performance have been consistently revealed in neuropsychological testing 6 12 months after surgery. 2,6 the clinical relevance of these cognitive changes remains controversial, as patients quality of life does not seem to be affected. 6 the mechanisms underlying the deficits in executive functions after stn HFs are still unresolved and are probably multifactorial in nature. Potential risk factors that could explain such deficits include predisposing factors, surgical issues and postoperative management issues. individuals with impaired attention, advanced age ( 70 years), motor symptoms poorly responsive to levodopa, a high levodopa equivalence dosage and/ or advanced axial motor symptoms (for example, postural instability, gait freezing, dysarthria) before stn HFs have been identified to be predisposed to cognitive decline following such therapy (C. Daniels, unpublished work). 46 similar risk factors have been described for dementia in patients with PD who are not treated by neurostimulation. 47 interestingly, in one study, age but not disease duration was found to correlate with cognitive changes in patients with PD after surgery. 48 this finding supports the concepts that an interaction exists between normal aging and disease progression in PD, and that the pathological process undergoes a nonlinear acceleration towards the end of a patient s life. 49 to date, the roles of surgical trajectory and electrode location in stn HFs in the onset of postoperative cognitive changes have received little attention. typically, the electrodes are inserted through the frontal lobes and frequently traverse the head of the caudate nucleus. Of note, these brain regions are both involved in cognitive processes, particularly verbal fluency. York et al. found preliminary evidence that cognitive and emotional changes observed in patients with PD 6 months after bilateral stn HFs were related to the cortical entry point and the position of the stimulating electrode in the subthalamic area. 50 these findings await confirmation in larger patient populations. when interpretating neuropsychiatric effects that occur after stn HFs, one must consider medication changes, which are mandatory during the postoperative period. On average, the levodopa equivalent dosage is reduced by 50 60% after stn HFs compared with before surgery, although the synergistic effect of stimulation and medication on dyskinesia can require the complete withdrawal of dopaminergic drugs in some patients. 1,48 a substantial reduction of dopaminergic medication after surgery can cause apathy, 38,49 a syndrome of decreased responsiveness to internal and external stimulation. in patients with PD, apathy is frequently associated with cognitive dysfunction. 51 nevertheless, most studies that have investigated the effects of medication changes following DBs have not found any correlation between the magnitude of dopaminergic drug reduction and the degree of cognitive change, 6,35,43,44 suggesting that drug adaptations after surgery are not a major contributing cause to cognitive decline. Acute neurostimulation effects the intrinsic impact of subthalamic neuro stimulation on cognitive functions is best evaluated in studies comparing task performances when stimulation is turned on and off. the results of such studies in accordance with studies evaluating the impact of levodopa treatment on cognition have been variable, and might in part reflect methodo logical differences between the various studies and the evaluation of heterogeneous patient populations. 39,44,52 54 Of note, highly variable results have been obtained in tests evaluating aspects of executive functioning. Pillon et al. reported that stn HFs led to marked improvements in a graphic and motor series sensitive to perseveration, the stroop and the trail making tests, psycho motor latency in simple and choice reaction time tasks, and spatial working me mory. 44 Other researchers, however, have reported impairments in stroop test performance with stimulation, 52,54 but a slight improvement in random number generation (a test to overcome habitual counting tendencies). a relatively consistent finding with stimulation turned on is impairment of response inhibition in conflict situations, as assessed by the stroop test, go no-go tasks or stopsignal paradigms the increases in error rates observed in these tasks can be interpreted as premature responding, and are in line with experimental studies in rodents, which have shown marked rises in premature responding in animals that undergo excitotoxic lesioning of the stn. 56,57 the aforementioned stimulation-induced cognitive changes correlate with the activation pattern of the an terior cingulate cortex, 53,54 a cortical target of stn projections that has been implicated in monitoring of decisions under conflict. 58 stn HFs also alters the reaction time to a stop stimulus intended to inhibit a planned or ongoing motor response. indeed, functional neuro imaging studies have demonstrated that the stn becomes activated during a stop-signal task the exact role of the stn in response inhibition is still a matter of debate, as two studies have reported opposing effects of stn HFs on stop-signal test reaction times. 62,63 nevertheless, both studies documented the involvement of the stn in inhibition control. in addition, a Pet study demonstrated that in the stimulation-on condition, the fronto-orbital cortex a brain region known to be involved in the process of response inhibition showed a reduction in activation. 64 taken together, these studies confirm the neurobiological role of the stn in handling decision conflicts and predict that stn HFs should result in a deterioration in the cognitive tasks that are particularly dependent on an intact basal ganglia no-go signal. Only a few studies have assessed the lateralized effects of neurostimulation. in a stop-signal task, reaction times to the stop signal were markedly longer when the left stn was stimulated than when the right stn was stimulated. 62 Moreover, unilateral left-sided stn HFs led to a slight but notable shift of attention towards the right visual field, 65 but this hemineglect disappeared after switching on right-sided stn HFs as well. together, these findings indicate that the right stn and its cortical projections can influence the attentional system of the nondominant hemisphere. nature reviews neurology advance OnLine PuBLiCatiOn 5

6 Mood euphoria, hypomania and mania the reported incidences of mood changes following stn HFs must be interpreted with caution because of methodological problems similar to those outlined for cognitive outcomes. Postoperative euphoria and/or hypomania have been described in 4 15% of patients, and usually occur within the first 3 months after surgery. 3 Manic psychosis has been less frequently documented than euphoria and hypomania, occurring in % of patients. 66 Hypomanic or manic states typically develop in close association with the initiation of HFs; thus, such states seem to be a direct consequence of neuromodulation affecting the limbic basal ganglia pathways. this assertion is supported by the fact that electrode contacts causing hypomania are frequently located within the anteromedial limbic territory of the stn. 67,69,70 HFs within this region has been shown to cause activation changes in mood-related thalamic and cortical regions, whereas stimulation of a more-dorsolateral contact has been demonstrated to improve movements and activate a motor network in patients with hypomania. 67 the hypothesis that mood is directly affected by stimulation of the stn has been challenged in a study involving fiber tracking by diffusion tensor imaging. the results of this investigation suggested that stimulationinduced acute hypomania could be caused by inadvertent axonal activation in the medial forebrain bundle, which runs in close proximity to the anteromedial stn. 71 the medial forebrain bundle is an important pathway of the mesolimbic dopamine system and belongs to the reward circuitry, which has been implicated in affective disorders, addictive behavior, and learning. 72 regardless of whether postoperative hypomania results from an intrinsic or extrinsic stimulation effect, neurosurgeons should avoid targeting the anteromedial stn so as to minimize the risk of inducing this condition. in patients who already have implanted electrodes, stimulation of a more-proximal contact of the quadrupolar electrode often provides better control of motor symptoms and carries less risk of affective disinhibition than does a distal contact. the conversion of hypomania to a manic psychosis is gradual and often triggered by an unbalanced and synergistic action of dopaminergic medication and neurostimulation in the postoperative adjustment period. Both treatments have an acute mood-elevating effect in experi mental settings, which is additive. 73 rapid withdrawal of dopaminergic medication and sometimes the addition of mood stabilizing drugs or neuro leptics typically lead to a rapid remission of postoperative mania. 28 in accordance with dyskinesia, which can be exacerbated in the immediate postoperative period by subthalamic stimulation but become habituated in the long term, 74 hypomania or mania are almost exclusively observed during the first few weeks of stimulation. in the long term, the stimulation threshold for induction of these phenomena gradually increases and, eventually, the contact that initially caused the adverse effect can be used for chronic stimulation. Depression studies utilizing standardized diagnostic criteria to examine mood symptoms have reported that depression develops in 20 25% of patients with PD following subthalamic surgery, and that such symptoms typically develop within the first 2 months of undergoing this procedure. 75,76 interestingly, the two prospective random ized trials of stn HFs incorporating a medically treated control group found small but beneficial effects on measures of depression in the surgically treated group after 6 months, although individual patients in both (surgically and medically treated) groups experienced clinically relevant depression. 1,2 these findings indicate that postoperative depression is neither a consistent finding across patients nor stable in expression with time. as discussed above, risk factors for postoperative depression include rapid or excessive withdrawal of dopa minergic medication, a previous history of depression, and problems adjusting to a reduction in disability level. the effect of psychological factors on this risk must not be under estimated, as the results of other life-changing surgeries (for example, organ transplantation or epilepsy surgery) have shown that depression is among the most common postoperative adverse effects and can often occur in the face of otherwise successful treatment. 8 the prevalence of clinically relevant depressive symptoms in patients with PD is estimated to be 30 40%. 77 this high disease-related risk must be dissociated from risk associated with subthalamic surgery. Castelli et al. found no differences in mood or anxiety in patients with PD following 3 years of treatment with either HFs stn or best medical management. 78 this finding is in line with a 5 year follow-up study, in which only one of 49 patients with PD who underwent stn HFs developed clinically relevant depression at the end of the observational period. 79 another prospective study evaluated mood and psychosocial functioning in such patients for 3 years after stn HFs and also found no marked changes in these states over the study period. 80 in many patients, postoperative depression can be explained by an amplification or reactivation of a preoperative disorder. 76 thus, careful preoperative psychiatric assessments are essential. 81 Medication changes in the postoperative period probably also affect postsurgical depression, as reinstitution or dose increases of dopa minergic drugs often alleviate this condition. Few systematic studies have been conducted on the differential psychotropic effects of levodopa and subthalamic stimulation. we found an improvement in depressive symptoms in patients with PD after an acute challenge with either stimulation or levodopa. 82 However, only levodopa was capable of restoring a normal hedonic tone. thus, psychiatric phenomena associated with the reduction of dopaminergic medication might not be fully compensated for by stn HFs in the early postoperative period, and might contribute to the postoperative risk of anhedonia or depression. whether the more-pulsatile nature of levodopa stimulation (as opposed to the tonic effect of electrical stimulation) on the mesolimbic dopaminergic system is of importance 6 advance OnLine PuBLiCatiOn

7 in the evolution of postoperative mood disorders remains speculative. 23 unlike mania, depressive states have rarely been associated with stimulation itself. in a single case reported by tommasi and colleagues, 83 acute transient depression could be induced by HFs through several contacts of the quadrupolar electrodes. these contacts covered a large subthalamic volume including the zona incerta, stn and substantia nigra. the stimulation-induced mood changes subsequently adapted with time, indicating that the microtraumatic effect of the implantation procedure contributed to the appearance of these changes. the diffuse nature of such lesions makes an anatomical interpretation of this observation difficult. in the famous case of acute stimulation-induced depression reported by Bejjani and colleagues, 84 the electrode contact inducing the emotional adverse effect was located ventrally to the stn within the substantia nigra. the notion that stimulation of ventral contacts might impair mood was further corroborated by the COMPare trial, 85 in which 22 patients with PD rated themselves on average less happy, less energetic and more confused when stimulated ventrally (below the optimal motor target) than when stimulated by a moredistal contact within the sensorimotor region of the stn. no patients in this trial, however, exhibited acute depressive symptoms. taken together, the few published cases of acute depression during HFs probably represent rare effects of stimulation affecting a still-undetermined region outside the stn. ventral subthalamic stimulation might reproduce a mild depressive syndrome through limited current spread into this region. studies evaluating acute mood changes with stimulation inside the stn through contacts providing optimal motor control have consistently reported acute antidepressive and mood-elevating effects in patients. 73,86 to summarize, dopaminergic withdrawal phenomena and problems of psychosocial readjustment probably both interacting with premorbid vulnerability factors are the most common causes of depression or dysphoria during the first few months after subthalamic surgery. increases in the levodopa dose or treatment involving a dopamine agonist with mood-elevating properties (for example, pramipexole) are often sufficient to control such symptoms. 3 Psychotherapy and anti depressants might be used in patients who do not respond to levodopa or dopamine agonists. in all cases, patients should be carefully evaluated for suicidal ideation, with prevention strategies being applied in patients deemed at risk of suicide. Anxiety anxiety (up to the level of panic attacks) is a frequent nonmotor symptom during off periods in medically treated PD patients. 87 in a controlled study of patients with PD, individuals who underwent stn HFs had markedly lower Beck anxiety inventory scores after 6 months of therapy than did individuals receiving best medical treatment. 6 Caution is warranted in the interpretation of this finding, however, as the Beck anxiety inventory includes several items with a strong somatic connection (such as an inability to relax and tremor of the hands) that improve considerably after DBs. nevertheless, reductions in anxiety following subthalamic neurostimulation have been reported in studies using other scales and measures of state anxiety during stimulation challenges. 35,73 whether reductions in anxiety occur secondarily to improvements in motor fluctuations or are a genuine nonmotor effect of stn HFs remains unknown. an intrinsic role for the stn in fear processing has been suggested by two studies that demonstrated specific deficits in recognizing fearful facial expressions or film sequences during stn HFs. 88,89 Behavior a wide spectrum of abnormal behaviors has been described in PD. Patients with this disease who have not received any treatment exhibit apathy as well as deficits in reward processing or novelty seeking. 90 these symptoms are reversible by dopamine agonist treatment. 91 Prolonged dopaminergic treatment alone, however, can be associated with behavioral adverse effects, such as punding, icds or addictive medication use (dopamine dysregulation syndrome or DDs). Punding was first described in the 1970s as an adverse effect of chronic amphetamine abuse, 92 and is defined as an intense fascination with excessive, repetitive and nongoal-oriented behaviors involving acts that can be simple (cleaning, sorting or ordering of objects) or complex (painting, creative writing, computer use or repairing). Manifestations of punding differ across individuals with PD, being often associated with previously learned behaviors. 26 Patients normally have insight into the disruptive and unproductive nature of their behavior; however, forcible attempts by family members or caregivers to interrupt punding often leads to irritability and dysphoria in the affected individual. 93 icds refer to patients inability to control urges or impulses that lead to compulsive performance of pleasurable activities. the icds reported in patients with PD include pathological gambling, hypersexuality, binge eating and compulsive shopping. icds have a proven association with dopamine agonist use, 94 although punding is more frequently seen in patients who have developed DDs, which is characterized by compulsive use of medication, particularly short-acting levodopa formulations. 93 the pathophysiology of the behavioral addictions observed in patients with PD remains to be determined, but a recent survey has emphasized the similarities with drug-induced sensitization phenomena leading to dyskinesia in the motor system. 26 impulse control disorders Cases of acutely altered behaviors induced by stn HFs are rare and isolated. in a case series described by romito and colleagues, 95 5 of 30 patients treated by subthalamic neurostimulation reported development of hypersexuality. three of these patients had coexisting mania, while one had a history of hypomania and another had a history of depression. the hypersexuality exhibited by the patients occurred despite postoperative drug nature reviews neurology advance OnLine PuBLiCatiOn 7

8 cessation, and was largely transient in nature, although sometimes this state lasted for several months. we described the case of an artist who, following surgery, not only changed the frequency at which he painted, but also altered the content of his art to include sexual imagery. 96 this patient underwent DBs in the dorsal stn and, postoperatively, received a marked reduction in drug dosage. His change in artistic expression resolved eventually. One possible factor that pre disposed this patient to hypersexuality might have been his pre operative history of drug-related mood fluctuations. smeding and colleagues 97 reported covert gambling lasting for 3 years after surgery in a man without any previous history of an icd. the patient s symptoms stopped within days after withdrawal of the dopamine agonist and alteration of the stimulation parameters. this case emphasizes the importance of actively seeking historical evidence of icds before and after surgery. Patients are frequently ashamed of their behaviors and do not volunteer such histories readily, particularly during presurgical evaluations for fear of not qualifying for surgery. 76 to date, the limited evidence available suggests that icds developing after stn HFs are transient phenomena during the postoperative adjustment period, and eventually come to a halt when medication is markedly reduced and a stable level of stimulation is maintained. if these assertions are correct, the question arises as to whether stn HFs could be recommended for treating patients with PD whose pre-existing icds are difficult to manage medically or psychologically. ardouin and colleagues reported on a series of seven patients with PD and pathological gambling who received surgery to treat relevant motor fluctuations. 98 these patients showed delayed cessation (one average 18 months after treatment) of their pathological behavior after a combina tion of undergoing chronic stn HFs and receiving a marked reduction in their dopaminergic medication dose. 3 years after surgery, however, three of the patients developed transient depression, which acted as a trigger to gamble again, although no pleasure was experienced by the indivi duals during this action, and the urge to gamble eventually resolved. the decisions of two of the other patients to start gambling postoperatively were associated with transient mania. in another series of patients with pre-existing icds and DDs, the effect of stn HFs on behavioral outcomes was mixed, with a large proportion of these patients showing no improvement. 99 Poor outcomes were consistently associated with no reduction or an increase in medication dose after surgery. typically, medication doses were not reduced, either because the motor outcome would be suboptimal or because the patients resisted drug withdrawal. the concept of desensitizing the limbic and associative basal ganglia loops to excessive dopaminergic stimulation by means of stn HFs needs to be further examined in clinical trials; however, the preliminary data discussed above suggest that patients with icds have a high risk of developing postoperative psychiatric problems and will need careful interdisciplinary neuropsychiatric management. the combination of DDs and icds might be particularly problematic, because craving and addictive behavior could interfere with the intended postoperative drug withdrawal. suicide a higher than expected frequency of suicide has been reported among patients undergoing stn HFs for advanced PD. 3,100 a multicenter survey of 55 movement disorder and surgical centers worldwide reported completed suicides in 0.45% of surgically treated patients (24 of 5,311 individuals). 101 the rate of attempted suicides in such patients was reported to be 0.90%. in the first postoperative year, suicide rates in surgically treated patients were fold higher than the global suicide rate reported by the who, 101 although the former declined and approached the rate of suicide in the general population level during the third post operative year. Postoperative depression, being single, and a previous history of icds were found to be independent risk factors for suicide in these patients (together, these factors accounted for 51% of the postoperative suicide risk). 101 Of note, another study revealed that suicide attempts can occur in surgically treated patients with PD who have no previous or present psychopathology, and under circumstances that underline the impulsive nature of this behavior. 102 the increase in suicide rate in surgically treated PD is even more remarkable when one considers that medically treated PD is associated with an up to a 10-fold lower suicide rate than occurs in the general population, 103 despite the high prevalence of depression (the most important risk factor for suicide in the general population) in this patient group. thus, a tempting presumption is that postoperative suicide in PD is part of the spectrum of impulsive behaviors associated with treatments that interfere with the normal inhibitory role of the stn in decision-making. 19,104 nevertheless, we emphasize that suicidal outcomes are a product of multiple factors, and that such a single neurobiological factor would account for a small amount of the observed variance in risk. Patients should be informed about the increased risk of suicide after DBs and encouraged to be open about predisposing psychiatric factors during the presurgical evaluation. in patients at risk of suicide, a short followup time and careful assessment for suicidality should be implemented. Apathy apathy is the most frequent long-term psychiatric adverse effect of subthalamic neurostimulation. Clinically, apathy is defined as a decrease in motivation and interest that cannot be ascribed to emotional or cognitive disorders, or to an impairment of consciousness. in one study, the proportion of patients with PD who exhibited apathy was documented to be 8.7% before surgery and 24.6% in the third postoperative year. 105 in individuals with PD, apathy is more frequently associated with a frontal dysexecutive syndrome or dementia 106 than with depression, indicating that this symptom is at least in part attributable to 8 advance OnLine PuBLiCatiOn

9 an advanced stage of disease characterized by widespread neurodegeneration. 106 Other authors have assigned apathy to the spectrum of hypodopaminergic nonmotor symptoms of PD, because in some patients this symptom responds to dopaminergic treatment. 107,108 Disease progression might explain the increasing incidence of apathy seen in long-term follow-ups after DBs surgery. a marked deterioration in apathy scores has, however, been noted as early as 3 6 months postoperatively, 109 indicating that apathy is associated with either neurostimulation itself or medication withdrawal in the immediate postoperative period. a Pet study, which mapped areas of notable metabolic change before and after surgery in patients with apathy after stn HFs, found marked correlations between changes in apathy scores and glucose metabolism in the right prefrontal cortex, the right posterior cingulate cortex and the left middle frontal gyrus. 110 no correlations existed between the extent of drug reductions, apathy scores and metabolic changes. these preliminary data indicate that subthalamic neurostimulation might in advertently modulate a frontal motivational network that is connected to the limbic and associative territories of the stn, thereby causing apathy. this etiological concept is still controversial, however, as a study comparing apathy in the stimulation-on and stimulation-off states found an improvement in self-reported apathy with stn HFs. 111 Moreover, acute neurostimulation and levodopa challenges have been previously reported to have similar drive-enhancing effects, although the magnitude of the psychotropic effect was slightly larger for the latter. 73 as a pragmatic treatment recommendation, one should try to increase levodopa or dopamine agonist dosages when apathy (frequently in association with anhedonia and depression) occurs within the first few post operative months after a marked reduction of dopaminergic drugs. apathy emerging during long-term follow-ups in the context of cognitive decline might, however, be less responsive to drug treatment, and confusional states could result from excessive dopaminergic stimulation. Behavioral strate gies, such as implementation of a structured daily routine, or an increase in psycho logical stimula tion by caregivers, might be attempted under these circumstances. Conclusions neuropsychiatric symptoms account for the majority of adverse effects that occur following DBs of the stn in patients with PD. Of note, however, such symptoms are typically transient and mild if managed appropriately. in patients most suitable for surgery, the motor improvement induced by the surgery outlasts any adverse effects and leads to substantial improvements in quality of life. 1,2 the basal ganglia are now regarded as part of a complex system of interconnected motor, limbic and associative pathways that are involved in the selection, facilitation and inhibition of movements, emotions, behaviors and thoughts. in these networks, the stn has a central regulatory role, providing a global no-go signal. apathy and depression in PD might reflect hypodopaminergic states Impaired reward-associated learning Reduced psychomotor speed Depression Apathy Compulsion Symptom severity Akinesia Too low Optimal dose range (medication or stimulation) of the limbic and associative loops that can be corrected by medication in the initial stages of the disease. icds and mania are symptoms at the opposite end of the spectrum of behavioral effects, and are probably caused by sensitization and dopaminergic overstimulation of these brain structures (Figure 3). in this model, the behavioral symptoms of PD reflect the pathophysiology of nonmotor basal ganglia loops, just as akinesia and dyskinesia reflect the abnormal activity of motor loops. 26 the spectrum and quality of postoperative neuropsychiatric problems do not differ from those seen in medically treated patients. neuropsychiatric symptoms typically occur within the first few postoperative months during adaptation to neurostimulation and medication. the motor (dorso lateral) territory of the stn is the target of functional surgery; however, variability in electrode placement and the small volume of the nucleus could account for diffusion of current to nonmotor territories. the focal effect of neurostimulation in contrast to the global effect of dopaminergic medication could explain why im balances between the activities of the motor, limbic and associative basal ganglia loops might occur in the adjustment period, in which both medication and stimulation are titrated to provide optimal motor symptom control. thus, in some patients, the limbic stn territory might receive current diffusion, which causes hypo mania, whereas in other patients, medication withdrawal and stimulation confined to the motor territory might cause a relative hypo dopaminergic state of the limbic loop, leading to depression. an individual s pattern of striatal dopaminergic denervation, drug-induced sensitization phenomena and premorbid personality traits or experiences could shape the risk and expression of cognitive, emotional and behavioral Impaired failure-associated learning Premature responding Mania Attention-deficit hyperactivity disorder Impulsivity Too high Dose Dyskinesia Motor Associative Limbic Figure 3 Behavioral effects of dopaminergic medication or high-frequency stimulation. The motor, associative and limbic segments of the subthalamic nucleus each have their own dose response curves, which have opposing pathological symptoms at high and low doses. Dopaminergic denervation in Parkinson disease is uneven across the various functional basal ganglia circuits; thus, the dose response curves corresponding to different segments of the subthalamic nucleus might not necessarily be aligned. An electrode that is not correctly located in the motor territory of the nucleus might cause currents to spread to other circuits and, hence, lead to additional functional imbalances in these circuits. in the postoperative period, stimulation and/or medication are primarily titrated to achieve optimal motor symptom control, leaving the other circuits at risk of overdosing or underdosing. nature reviews neurology advance OnLine PuBLiCatiOn 9