Insertional effect following electrode implantation: an underreported but important phenomenon

Abstract Deep brain stimulation has revolutionized the treatment of movement disorders and is gaining momentum in the treatment of several other neuropsychiatric disorders. In almost all applications of this therapy, the insertion of electrodes into the target has been shown to induce some degree of clinical improvement prior to stimulation onset. Disregarding this phenomenon, commonly referred to as ‘insertional effect’, can lead to biased results in clinical trials, as patients receiving sham stimulation may still experience some degree of symptom amelioration. Similar to the clinical scenario, an improvement in behavioural performance following electrode implantation has also been reported in preclinical models. From a neurohistopathologic perspective, the insertion of electrodes into the brain causes an initial trauma and inflammatory response, the activation of astrocytes, a focal release of gliotransmitters, the hyperexcitability of neurons in the vicinity of the implants, as well as neuroplastic and circuitry changes at a distance from the target. Taken together, it would appear that electrode insertion is not an inert process, but rather triggers a cascade of biological processes, and, as such, should be considered alongside the active delivery of stimulation as an active part of the deep brain stimulation therapy.


Introduction
2][3][4] After electrodes are implanted, clinicians often wait 2-4 weeks before activating the DBS system.Interestingly, during the initial days or weeks following surgery, patients frequently experience a noticeable clinical improvement prior to the delivery of stimulation.First reported in conditions such as tremor and Parkinson's disease, [5][6][7] this response has also been documented in patients treated for chronic pain, 8 epilepsy 9 and even psychiatric disorders. 10Because the clinical effects of electrode insertion are often similar to those observed after ablative procedures, this phenomenon was originally thought to be caused by a focal 'microlesion'.However, decades of clinical observations and preclinical study have demonstrated that the clinical response associated with the 'insertional effect' may be due to complex biological processes rather than merely tissue disruption.
In preclinical models, the insertion of electrodes has been shown to induce substantial behavioural improvements, often mirroring what is observed in humans.Preclinical models allow for histopathologic examination at various timepoints after insertion, which has led to several explanatory mechanisms being proposed, including inflammatory responses near the electrode site, the activation of astrocytes, a focal activation hyperexcitability of neurons and the development of neuroplastic changes.
This review aims to characterize the insertional effect from four different angles.First, we describe the neuropathological changes that occur following the implantation of brain electrodes and probes.Second, we present some of the behavioural effects and mechanisms responsible for the development of an insertional effect in animal models.Third, we report the clinical consequences of electrode insertion.Finally, we outline how the insertional effect may interfere with the results of clinical trials.

Neurohistopathology of the insertional effect
Most neurohistopathological studies to date have been conducted in naïve rodent brains following the insertion of electrodes and microdialysis probes.The latter is discussed as it may help to characterize neurochemical tissue responses to brain implants.

Initial trauma related to the implant
2][13][14][15][16][17][18] At first, the insertion of an implant focally damages cells, capillaries, the extracellular matrix, and disrupts the blood brain barrier (BBB) (Fig. 1). 15,19,20Injured capillaries may lead to leakage of erythrocytes, activation of platelets 15 and the development of a localized haematoma. 21][24][25] Hours after the initial injury, a transient focal reduction in glucose metabolism can be observed surrounding the electrode, 26 and in areas anatomically connected with the lesioned site. 27][35] Within a few days, an encapsulating sheath begins to form around the electrode. 28,29,36Pericytes, present in capillary walls, undergo morphological changes and contribute to local angiogenesis within the newly forming sheath. 37][39] The abovementioned processes lead to neuronal apoptosis and local cell death along the tract and at the target site (Fig. 1). 19Also injured are fibres en passant, which may project to nearby or distant brain regions. 40Argyrophilic neurons and fine silver deposits denoting degenerating neurites, nerve terminals and fibres may be observed as early as 24 h after probe implantation, becoming increasingly prominent with longer survival times. 40However, not all surrounding neurons perish.3][44] The vitality of neurons surrounding the electrodes is essential for a stable signal transmission and eventual stimulation responses. 45

Glial response and reactive gliosis
7][48][49][50][51] Reactive gliosis involves the recruitment of astrocytes, microglia and NG2-expressing glial precursors, 48,52,53 as well as angiogenesis and revascularization of the area surrounding the lesion (Fig. 1). 54Within the first week after the insertion of brain electrodes or probes, a large number of reactive astrocytes, microglia and laminin labelled vessels may be observed around the implant. 557][58][59][60][61][62] This process has been demonstrated in rodents, [46][47][48][49]63   The stress induced by the electrode implantation leads to the formation of several damage-associated molecular patterns (DAMPs) that are responsible for the chemoattraction of classically activated microglia and astrocytes. Thee cells form a protective capsule, commonly called 'glial scar', that leads to local metabolic and functional changes due to the release of pro-inflammatory and pro-resolving mediators.These mediators may influence the maturation of NG2-precursor cells into astrocytes and axonal-growth neurons that will influence the connectivity of the target area.
7][68][69] Following the insertion of electrodes/ probes into the brain, glia cells in the vicinity of the implants are activated and undergo morphological changes (Fig. 1). 19,20eactive astrocytes are characterized by hypertrophic somata and processes, 70 synthesize a higher than normal amount of glial-fibrillary acidic protein (GFAP) and express progenitor markers, such as vimentin, nestin and synemin. 48,71,72When active, a large proportion of astrocytes assume an A1 pro-inflammatory secretion phenotype, losing their ability to prune synapses and regulate glutamatergic clearance. 734][75] Mature astrocytes do not normally divide, but may acquire stem cell properties and generate cells that initiate repair after brain injury. 46icroglia also contribute to the formation of a physical barrier between the device and brain parenchyma. 197][78] Upon electrode insertion, they are attracted to the injured territory, and extend lamellipodia to help ensheath the implants.These cells are stimulated following exposure to pathogen-or endogenassociated molecular patterns, endogenous ligands that act on toll-like-receptors (e.g.heat-shock proteins and fibrinogen), or when there is a lack of immunosuppressive signals. 77ctivated microglia undergo structural changes, assume an amoeboid morphology, 77,79 proliferate, 80 and migrate to the site of injury in an attempt to enclose the pathological process and shield the BBB (Fig. 1). 77,78,81,82In contrast to the resting state, active microglia secrete chemokine attractants (e.g.monocyte chemoattractant proteins and macrophage inflammatory proteins), pro-inflammatory cytokines (e.g.tumour necrosis factor-α, certain interleukins and interferons), transforming growth factors, reactive oxygen and nitrogen species (ROS, RNS) and neurotrophic factors (e.g.brain derived neurotropic factor, nerve growth factor and neurotrophin-3). 15,19,83,84The immune response promoted by microglia is critical to restrict damage and facilitate repair, helping to limit toxicity and eliminate foreign materials and debris.
6][87][88] NG2 cells are primarily the precursors of oligodendrocytes, but can also form other glial and neuronal cell types. 85In addition to their role as glial progenitors, these cells migrate, proliferate and inhibit axonal growth, being an important element in the formation of glial scars. 89,90Both NG2 glia and microglia migrate towards inserted probes (Fig. 1).NG2 glia, however, act in a more protracted manner (within hours), forming synaptic connections with nearby neurons. 91When neurons are injured by the insertion of implants, the differentiation of NG2 glial cells into astrocytes to form the glial scar reduces the number of cells that may differentiate into oligodendrocytes. 91This may yield deficits in the remyelination of injured neurons, contributing to the neuronal apoptosis observed following a surge in focal inflammatory processes in the injured tissue. 90,91

Chronic implants
By 6-12 weeks post-electrode or probe implantation, reactive astrocytes are largely condensed in an increasingly compact sheath (Fig. 2). 55,58The continuous presence of the implant results in a sustained response that maintains the gliotic sheath. 41,51,583][94] Tight junctions between microglia and astrocytes during scar tissue formation may act as a biochemical and mechanical barrier.Should the implant be a microdialysis probe or a neurochemical sensor, this may prevent the diffusion of ions, solutes and chemical signalling molecules secreted from surrounding neurons.Over time, the activity of microglia and the acute foreign body reaction to the implant may fade, despite the presence of the sheath. 957][98] That said, neurons near the gliotic process display increased activity, with reduced inhibitory synaptic currents, a downregulation of glutamine synthetase and reduced gamma-aminobutyric acid transmission. 99102]99 Evidence of altered neuronal function can be found in regions projecting to or receiving projections from the target.4][105] This suggests that the chronic presence of electrodes in the brain parenchyma is not an inert event.
Astrocytes and other glial elements change the normal functioning of nearby neurons, leading to a series of consequences, both focally and at a distance from the target.
While many of the findings above have been studied in rodents, similar observations have been noted in non-human primates, including reactive gliosis, the accumulation of microglia, the development of a glial sheath and some degree of neuronal loss. 65,106,107

General considerations
To investigate the tissue response to the chronic implantation of leads in humans, post-mortem samples from patients treated with DBS have been analysed.Data from such studies, however, need to be considered with caution, as samples are collected from patients with different pathological states, contrasting with preclinical material obtained from naïve animals.Moreover, individuals who undergo post-mortem autopsy have usually been treated with DBS for prolonged periods of time.As a result, disentangling the consequences of electrode implantation and stimulation may not always be possible.

Neuropathology
9][110][111][112][113][114][115][116][117][118][119][120] Studies described the presence of a fibrous gliotic sheath around the electrode site with a thickness that varied from 25 μm to over 150 μm. 108,119119]121,122 Collagenous accumulation has also been reported. 108n contrast to preclinical studies, neuronal loss did not seem to be a prominent feature in human samples. 113,119issue damage surrounding the electrodes has been characterized by vacuolization, myelin loss, axonal spheroids, iron deposits (uncommon) and spongiosis (rare). 1214][125] Most recent pathological studies did not reveal neuronal loss at a distance > 500 μm from the electrode site. 59,126In addition to the material of the implants, differences in neuronal death between human and animal studies may be due to the timing of electrode removal.In preclinical work, electrodes are removed right after death or even in live animals.This may injure local cytoarchitecture, 57 as nearby glial cells may adhere to the implant. 122n pathological human specimens, the post-mortem removal of the implants occurs long after death in patients who had their electrodes implanted for a long period of time.As a result, the surrounding glial sheath is often compact and only a few cells are adherent to the implant. 58o disentangle the neuropathological consequences of electrode implantation and chronic stimulation, a strategy adopted by a few studies was to compare tissue samples surrounding active (i.e.used to deliver chronic stimulation) and pivotal role in the mechanisms of electrode implantation, since anti-inflammatory drugs attenuate its antidepressant-like effect.Classically activated microglia may also be responsible for the clearance of local protein aggregates.On the other hand, glial scar astrocytes are often related to an alternative activation that releases pro-resolving mediators and increases glutamate clearance in an attempt to promote homoeostasis and is associated with the development of hyperexcitability in surviving neurons.Electrode implantation is also responsible for the increased expression of colocalized serotoninergic receptors and p11 (S100A10).This response is influenced by inflammatory mediators and may affect the increased input to efferent areas.Beyond the modulation of previously existing connections, electrode implantation may interfere with the formation of new connections with surrounding areas.Efferent and distant regions are also affected, presenting an increase in tissue volume, accounted by angiogenesis and astrocytes.
non-active stimulation contacts.In general, while both showed similar levels of gliosis, 115,121,127 enlarged axons and axonal spheroids were more often noticed around active contacts. 121Despite these small differences, based on the similar findings noted in the two scenarios, it is likely that the histopathological response to DBS described in recent studies might be largely related to the implants, and not the delivery of current per se.

Preclinical models: behavioural effects of electrode insertion and mechanisms
Despite the vast clinical use of DBS for the treatment of movement disorders, the behavioural consequences of electrode insertion have been mainly documented in preclinical models of psychiatric and cognitive disorders.The most convincing behavioural effects came from studies comparing rodents who either underwent surgery without electrode implantation or received electrode implants with or without stimulation.Another study design in which the insertional effect was appreciated involves the use of serial behavioural measurements before and after surgery (i.e. at baseline, following electrode implantation and after the delivery of stimulation).This design, however, does not allow for the measurement of long-term consequences of electrode insertion, since DBS is often introduced days/weeks after the surgical procedure.
When considering the insertional effect in rodents, one should take into account the ratio between electrode diameter and brain volume.Though electrodes commonly used in humans are ∼5-15-fold thicker than the ones used in rodents, the human brain volume is orders of magnitude larger.This difference is maximal in frontal cortical regions, but is also pronounced in subcortical structures.Despite this limitation, preclinical models have been instrumental to develop mechanistic studies and to support the notion that the insertional effect has a biological substrate.As observed in humans, the behavioural effects of electrode insertion in rodents were often similar to those observed following DBS administration, albeit of lesser magnitude. 128,129

Behavioural data
In patients with treatment refractory depression, DBS delivered to the subgenual cingulum has been investigated with promising results. 10,128,130,131To study potential mechanisms and the kinetics of DBS, stimulation has been applied to an analogous region (i.e.33][134][135][136] In models of cognitive degeneration, contrasting results have been reported.A significant improvement in the radial arm water maze test was found in Alzheimer's disease transgenic animals briefly implanted with hippocampal electrodes. 137In contrast, no major differences in memory performance were found between rats that did or did not receive nucleus basalis of Meynert electrode implants in the 192 IgG-saporin model. 138n humans, DBS delivered to the subthalamic nucleus (STN) is considered to be a standard of care treatment for motor symptoms in Parkinson's disease. 139,140To study mechanisms of STN DBS in preclinical models, naïve and parkinsonian rodents implanted with electrodes were tested in different motor and cognitive paradigms.In naïve rodents undergoing a serial reaction time task, electrode insertion increased motor time, while low current DBS reduced premature responses. 141In addition, rats with implanted electrodes had a significant memory impairment in the novel object recognition task. 82As for motor effects, both electrode insertion and stimulation delivered to hemiparkinsonian rodents reduced immobility time on a bar, 142,143 while only the latter treatment improved amphetamine-induced rotations [142][143][144] and motor performance in the rotarod test. 144In summary, STN electrode insertion was found to play a role on impulsivity and memory disturbance, while it was only associated with motor improvement in some tests.In contrast, the insertion of electrodes in the pedunculopontine nucleus led to a mild exacerbation in freezing of gait. 145

Mechanisms
Though the mechanisms responsible for the insertional effect are largely unknown, preclinical studies have considered a few hypotheses.The first is the development of a neuroinflammatory response to the injury produced by the electrodes and a subsequent glial reaction. 82As described above, the continuous presence of brain implants results in a sustained response that maintains a gliotic sheath, which is largely comprised of reactive astrocytes and microglia. 41,51,58The systemic administration of anti-inflammatories, but not opioids, countered the antidepressant-like effects of vmPFC electrode insertion, as well as the increased expression of GFAP, TNF-alpha and COX associated with the implants. 104,133Another proposed mechanism is the activation of astrocytes (Fig. 2).The insertion of brain electrodes in rodents induces an astrocytemediated release of glutamate and adenosine, which modulate the activity of neurons and glial cells in the vicinity of the electrodes. 104,133,146,147In cultured astrocytes, the insertion of electrodes activates calcium signalling and adenosine triphosphate (ATP) release. 148,149lso important for the development of an insertional effect are the physiological responses of neurons in the region of the implants.As described above, the immediate vicinity of the electrodes is characterized by a reduction in nerve fibre density and the number of cell bodies. 45,96,97However, surviving neurons show reduced inhibitory synaptic currents, a downregulation of glutamine synthetase and reduced gamma-aminobutyric acid transmission. 99With the chronic interaction between active astrocytes and neurons, the latter tend to become hyperexcitable, 99 a scenario that may lead to focal and distant changes (Fig. 2).
On a circuitry level, the insertion of electrodes in preclinical models was shown to exert changes at a distance from the target.In rodents, positron emission tomography (PET) studies following vmPFC electrode implantation and DBS showed different metabolic patterns, suggesting the recruitment of alternative networks. 103Mice implanted with vmPFC DBS electrodes presented local hippocampal and thalamic volumetric enlargement compared to animals receiving no electrode implants, 105 an effect attributed to an increase in the size of astrocytes and the diameter of blood vessels. 105In AD transgenic mice, the insertion of hippocampal electrodes was found to increase hippocampal neurogenesis and reduce the number of amyloid plaques. 13711 is a protein expressed in brain regions implicated in the pathophysiology of depression [150][151][152] that co-localizes with 5-HT 1B and other serotonergic receptors. 153,154p11 mRNA is reduced in depression states, 155 and increases following the administration of different classes of drugs and electroconvulsive therapy. 156,157This protein has been associated with the antidepressant-like effects of vmPFC electrode implantation. 133Favouring the involvement of a serotonergic mechanism in the behavioural consequences of electrode insertion, the administration of para-chlorophenyl-alanine methyl ester, a drug that reduces serotonin synthesis, countered the antidepressant-like response of vmPFC implants in rodents. 133n addition, the increase in p11 associated with an insertional effect was mitigated by the administration of antiinflammatory agents, 104 suggesting a potential link between neuroinflammation, p11 expression and serotonergic transmission after electrode insertion.DBS has been shown to reduce inflammatory responses in cultures of astrocytes 143 and in structures at a distance from the target in rodent models of epilepsy, 158 stroke, 159 Alzheimer's disease, 160 traumatic brain injury 161 and parkinsonian states. 142,143Similarly, the insertion of STN electrodes mitigated the increased expression of multiple cytokines observed in the striatum of parkinsonian rats, reduced striatal glutamate release and augmented the relative concentration of GABA. 142n summary, the insertion of electrodes into the brain is associated with processes that go far beyond a simple mechanical effect.The initial inflammatory response and the subsequent gliotic reaction that occurs near the electrode site induce lasting changes in glial cells and neurons, which may present altered patterns of neurotransmitter release and become hyperexcitable.These combined responses may ultimately induce a complex interplay of focal and distant changes that actively lead to behavioural modifications.

Insertional effect in clinical practice
The amelioration of symptoms following electrode insertion has been reported in multiple conditions.The presence of an insertional effect is more easily demonstrated in patients with movement disorders and epilepsy, as it can be objectively documented.In pain and psychiatry, immediate postoperative improvements are only subjectively reported (Fig. 3).

Tremor
6][7] Based on the similarities in outcome between electrode insertion and thalamic lesions (i.e.thalamotomy), the clinical improvement observed after the former was also called 'microthalamotomy'.
3][164] In ET, but also in PD, series that specifically studied the clinical effects of electrode insertion suggested that this phenomenon was of clinical relevance in ∼50% of patients, with an average duration of 25 days. 162In ET, the development of an insertional effect seems to indicate an accurate targeting strategy and predict a good postoperative response to DBS. 163 In addition to Vim, electrode implants in the region of the zona incerta and prelemniscal radiation have also been associated with an insertional effect, with patients presenting a substantial improvement and even tremor arrest. 165In patients with ET and PD, the insertion of electrodes in the posterior subthalamic area was found to reduce the amplitude and duration of saccades. 166rom a mechanistic standpoint, the insertion of thalamic electrodes in patients with multiple forms of tremor (essential tremor, Parkinson's disease, multiple sclerosis) decreased coherence between the cerebellar thalamus and electromyography activity at the tremor frequency. 167In patients with ET, the implantation of thalamic DBS electrodes increased the focal release of adenosine, 168,169 corroborating results previously observed in preclinical models.

Parkinson's disease
STN electrode insertion in patients with PD can result in a 17-30% reduction in Unified Parkinson's disease rating scale (UPDRS) scores, [170][171][172][173][174] and reductions in bradykinesia, tremor and rigidity in the order of 14-42%. 171,172Clinical amelioration after electrode implantation in PD is maximal in the first postoperative days but may still be present weeks after the procedure. 175,176Similar to what has been described for tremor, PD patients who present an insertional effect were shown to have a good postoperative response to DBS. 177 Objective improvements associated with electrode insertion have been documented in the finger tapping velocity, pronation-supination, and the boxing with touch tests. 178arely, negative insertional effects have been observed, including transient dyskinesias in 10% of patients. 171lthough these are often mild, severe dyskinesias and even ballistic movements have been reported. 179One of the most common complications of STN DBS surgery is the development of a decrease in verbal fluency. 180This adverse event has also been described following electrode insertion, [181][182][183][184] being attributed to the microlesion of fibre bundles, the cingulum and basal ganglia structures.STN electrode insertion in the postoperative period has also been suggested to alter the recognition of facial emotions, 185 transiently increase the latency for saccades, 186 and to induce rapid eye movement sleep. 187he development of an insertional effect has also been documented following electrode insertion to the other major target in the treatment of PD-the globus pallidus internus (GPi).One series reported that almost 50% of GPi-implanted subjects experienced a positive insertional effect, with some presenting >50% improvements in UPDRS scores and a reduction in postoperatively LDOPA requirement. 188In addition to STN and GPi targets, a substantial improvement in motor symptoms has also been documented following electrode insertion in the prelemniscal radiation, zona incerta (tremor and rigidity), 189 and in the vicinity of the STN (suboptimally placed electrodes). 190 patients undergoing DBS surgery, microelectrode mapping (MER) is a commonly used strategy to refine surgical targeting that has been associated with an insertional effect in some instances. 172,1913][194] Interestingly, positive insertional effects have not been found to correlate with the number of brain penetrations. 172The development of an insertional effect following MER correlated with the normalization of the H reflex. 195 Neuroimaging studies have helped to understand distant effects of electrode insertion.PET studies have shown that the insertion of STN electrodes in PD patients reduces glucose utilization in the putamen, globus pallidus, ventrolateral and mediodorsal thalamic regions, while increasing metabolism in the sensorimotor cortex and cerebellum. 196his pattern was similar, but with lower magnitude changes, than the one observed following STN lesions and DBS. 197,198n fMRI study involving visually triggered finger tapping has also revealed a lower signal in cortical and subcortical regions (thalamus, basal ganglia) following electrode insertion compared to the preoperative period. 173In some of these studies, rigidity and axial symptoms recorded after electrode insertion were inversely correlated to the activation of the putamen and globus pallidus. 173n contrast to the somewhat homogeneous results of functional neuroimaging reports, connectivity analyses have shown different patterns of response following DBS and electrode insertion.While STN stimulation induced changes primarily in cortical structures, electrode insertion was associated with an increased centrality and functional connectivity in the brainstem, 199 as well as a decrease in homogeneity in the default mode network (DMN), prefrontal cortex and the cerebello-thalamocortical circuit. 200Also favouring the notion that electrode insertion modulates basal ganglia circuits, electrophysiological recordings have shown a significant increase in low-frequency fluctuations in the putamen and precentral gyrus, a decrease in these fluctuations in structures of the DMN and the executive control network. 174aken together, these findings suggest that the insertional effect is mediated, at least in part, by widespread circuitry modulation.Though the mechanisms responsible for this phenomenon in patients with PD are still unclear, cortical and subcortical oedema, 173 and the functional or anatomical disruption of white matter bundles in the trajectory of the electrodes have been suggested to play a role. 182

Dystonia
Similar to other movement disorders, substantial clinical and objective kinematic improvements have been described in patients with dystonia implanted with GPi electrodes. 178In a recent study, almost 80% of patients with Meige syndrome had a 49% reduction in Burke Fahn Marsden Dystonia scores following electrode implantation. 201In these patients, a positive correlation was found between the development of an insertional effect and postoperative outcome at 12 months. 201In patients with generalized dystonia, trials that reported an insertional effect described a substantial amelioration in dystonic symptoms (often >50%) following GPi electrode implantation lasting days to weeks. 202,203

Epilepsy
In epilepsy, patients with distinct seizure types and foci have been treated with DBS in different brain targets. 204Anterior thalamic nucleus (ANT) and hippocampus stimulation have been proposed for the treatment of focal seizures, including those originating in the temporal lobe.In contrast, centromedian (CM) and subthalamic DBS have been investigated in patients with generalized seizures. 2046][207][208] In some ANT-DBS studies, the reduction in seizure rate after electrode insertion accounted for most of the benefits observed in the postoperative period, with no further outcome changes being documented after DBS onset. 206As for CM electrode implants, a study reported a ≥50% reduction in seizure rate in up to 70% of patients, 209 an effect that lasted one month following the procedure. 209Results in trials implanting hippocampal electrodes have been inconsistent with seizure reduction being demonstrated in one study, 210 but not in other. 211

Pain
Very few reports have described an effect of electrode insertion in chronic pain patients.In one of the few retrospective series addressing this issue, 43% of patients implanted with DBS systems in sensory thalamic regions had a substantial reduction in pain scores (60-100%) following electrode implantation. 8,212Recurrence of pain in these patients occurred within a median of 3 months after surgery. 8Twice as many patients who presented an insertional effect following electrode placement responded to DBS at long-term. 8imilar to Parkinson's disease, transient side-effects have also been reported after electrode insertion in patients with chronic pain.As an example, a patient with facial pain developed stutter that lasted 12 days following the insertion of a thalamic electrode. 213n addition to chronic neuropathic pain, DBS has also been used to treat patients with cluster headache and shortlasting unilateral neuralgiform headache (SUNCT).Brain regions commonly targeted for these applications are the hypothalamus and the ventral tegmental area.A few studies have clearly reported the presence of an insertional effect.In one of the series, two out of 11 patients implanted with hypothalamic electrodes had a substantial clinical improvement 1 month after surgery prior to DBS. 214 After receiving ventral tegmental area implants, six out of 21 patients with cluster headache 215 and 4 of 11 patients with SUNCT 216 had an improvement in the number of attacks after electrode placement that lasted 1-3 months postoperatively.
Another neuromodulation modality under investigation for the treatment of chronic neuropathic pain is motor cortex stimulation.This technique involves the placement of electrodes over motor cortical regions, followed by electrical stimulation delivered through a pulse generator.Despite the promising results observed in open label studies, 212 trials comparing pain scores during active versus sham treatment 217 or high versus low stimulation (longer 'on' versus 'off' stimulation cycling) 218 found no differences in clinical response.In some of these studies, however, an analgesic effect prior to the delivery of stimulation could be clearly observed. 217In a recent report in which the effects of electrode insertion were taken into account, a >50% reduction in pain scores was reported in ∼40% of patients. 219More strikingly, the probability of observing some degree of postoperative improvement while receiving stimulation in patients who developed an insertional effect was close to 100%. 219

Psychiatry
In psychiatry, DBS has been approved for the treatment of obsessive-compulsive disorder (OCD) and is currently under investigation for conditions such as major depression and post-traumatic stress disorder (PTSD).A meta-analysis of trials on DBS for OCD revealed that the insertion of electrodes into the target in the absence of stimulation (sham) reduced Yale-Brown obsessive-compulsive scale scores by 15%. 220In a retrospective chart-review of studies using DBS to treat refractory depression, patients who did and did not receive anti-inflammatory drugs in the postoperative period were separated in groups.Early response during the first week after surgery was similar between groups. 133The antidepressant effects of DBS were only observed in patients who did not receive anti-inflammatory drugs. 133In a recent clinical trial investigating the effects of DBS in patients with PTSD, one individual (out of a total of four) reported that the best improvement she experienced following surgery occurred in the immediate postoperative period. 221

Discussion and perspectives for the design of clinical trials
Though the clinical effects of electrode insertion have been appreciated for decades, it was not until recently that its mechanisms of action and relevance for trial design have been recognized.In humans, an improvement in symptoms following electrode insertion has been documented in almost every DBS application.As described above, this response is maximal within days of the procedure, but may last for weeks and even months.This can be treacherous for the future of the field because, for DBS applications to be approved, a common trial design involves a blinded comparison of active versus sham stimulation (i.e.systems set at zero amplitude or inactive). 222Clinical assessments are then commonly conducted at 3 months or 6 months after surgery, with no short-term evaluations.As the development of an insertional effect is rarely considered, its occurrence is not always detected, but can certainly influence the overall results.Without a proper trial design, it is difficult to discern the consequences of electrode insertion from those associated with stimulation-induced plasticity and a carryover effect from electrode insertion.Ideally, postoperative scores should be serially measured until their values are back to preoperative baseline.At present, improvements in the sham-stimulated group are often attributed to placebo effects, which are certainly a factor to be considered should other factors be discarded.
In theory, should an insertional effect occur, patients in the sham-treated group would have better scores compared to those recorded at baseline and the effects of DBS would be underestimated.For example, even 3 months after surgery, one RCT conducted in patients with PD treated with DBS reported a significantly better quality of life and a modest reduction in motor scores in non-stimulated controls. 223n a randomized control trial comparing active versus sham anterior thalamic nucleus stimulation for epilepsy, the frequency of seizures within the first month of electrode placement was in the order of 20% in both groups compared to baseline. 224In the sham-treated group, a 20-30% decrease in seizure rate was still observed 3 months after surgery.Only 4 months after the procedure differences between groups became significant. 224If the trial had its primary outcome set at 3 months, no differences between active and sham stimulation would have been detected and the study would have likely failed to demonstrate efficacy.
33][134][135][136] In the clinic, SCG DBS trials have shown a 10-30% reduction in validated depression scores in patients treated with sham SCG stimulation compared to baseline. 10,225In a study that initiated SCG DBS soon after surgery, an important reduction in depression scores was noted in 87.5% of patients in the first postoperative week. 226This response subsided over the first postoperative month, with only 37.5% of patients being characterized as responders.Only 6 months after stimulation onset a substantial DBS response was recaptured in most patients.This indicates the possibility of an insertional effect immediately after surgery that might have subsided within the first postoperative month. 226imilar to animal models, short-term differences between active and sham stimulation in randomized trials are often not significant. 10,225Considering that the effect of electrode insertion is so striking in rodents, clinical studies should consider either comparing DBS effects with non-implanted controls, or only finish the sham stimulation timeframe when scores return to baseline levels.

Conclusions
DBS has become standard of care in the treatment of refractory movement disorders, and is under investigation for the treatment of several neuropsychiatric disorders.In almost all DBS applications, the insertion of electrodes into the target induces some degree of clinical improvement.While this is often more pronounced at short-term, the effect of electrode insertion may last for weeks or even months.Disregarding this phenomenon may lead to potential biases in randomized clinical trials, which may show a symptomatic amelioration in patients with electrodes implanted and an equal degree of improvement following sham or active stimulation.
From a mechanistic standpoint, preclinical work has shown that, in addition to an initial inflammatory response, the placement of electrodes in the brain leads to more protracted responses.These include changes in the physiological interplay between astrocytes and other neural elements, a focal release of gliotransmitters, the presence of hyperexcitable neurons in the vicinity of the electrodes, as well as plasticity and neural changes at a distance from the target.This suggests that, rather than a simple neuroinflammatory response, the placement of brain electrodes leads to changes in neurophysiological substrates and its consequences culminate in a cascade of biological processes.With this in mind, we suggest that the implantation of brain electrodes should be considered to be an active part of the DBS therapy, which includes not only the administration of electrical current but also the placement of electrodes into the desired targets.

Figure 1
Figure 1 Neurochemical mechanisms of electrode insertion in the brain parenchyma.(A) The mechanical insertion of the electrode into the target leads to a disruption of small blood vessels causing local edema, focal bleeding and leakage from the blood brain barrier.Injury to the brain parenchyma may lead to mitochondrial stress and the release of reactive-oxygen species (ROS) that may be responsible for a transient decrease in local oxygenation.(B) This initial disruption may induce a transient decrease in local glucose metabolism, a functional disruption in connectivity between the target region and interconnected structures, as well as neuronal stress.(C) In response to the insult, pericytes and endothelial cells will attempt to counter the acute microenvironment stress by re-establishing the BBB integrity, attenuating the infiltration of peripheral immune cells, and inducing angiogenesis.(D) Resident NG2-precursor cells migrate to the implanted site and differentiate into astrocytes, oligodendrocytes or neurons in the attempt to mitigate the insult response.(E)The stress induced by the electrode implantation leads to the formation of several damage-associated molecular patterns (DAMPs) that are responsible for the chemoattraction of classically activated microglia and astrocytes.These cells form a protective capsule, commonly called 'glial scar', that leads to local metabolic and functional changes due to the release of pro-inflammatory and pro-resolving mediators.These mediators may influence the maturation of NG2-precursor cells into astrocytes and axonal-growth neurons that will influence the connectivity of the target area.

Figure 2
Figure 2 Mechanisms of electrode implantation.Preclinical studies suggest that classically activated microglia and astrocytes may play a

Figure 3
Figure 3 Clinical consequences of the insertional effect.NSAID, non-steroidal anti-inflammatory drugs; LEDD, levodopa equivalent daily dosage; PD, Parkinson's disease; TRD, treatment-resistant depression; OCD, obsessive-compulsive disorder.*Adverse effects of electrode implantation are similar to those of DBS; **controversial results when the hippocampus is targeted.