Deep Brain Stimulation: Mechanisms of Action

Introduction

Deep brain stimulation (DBS) is a well-established surgical therapy for treatment of the motor symptoms associated with a variety of movement disorders, including essential tremor (ET), Parkinson disease (PD), and dystonia. ^, It is also being explored for the treatment of a number of other neurological and psychiatric disorders, such as epilepsy, depression, dementia, obsessive-compulsive disorder, and the tics associated with Tourette syndrome. ^, In this chapter, we review the current understanding of DBS mechanisms, highlighting several potential factors that play into how DBS elicits its beneficial effects, such as the site of stimulation and the type of disorder being treated.

The potential application of DBS in humans was demonstrated in a seminal study by Hassler and colleagues in 1960. They described a series of patients with tremor who received electrical stimulation via electrodes introduced into the globus pallidus for the purpose of creating therapeutic ablations. They observed that low-frequency stimulation (<25 Hz) exacerbated contralateral tremor, whereas high-frequency stimulation (25–100 Hz) could alleviate or abolish tremor entirely. However, they viewed electrical stimulation primarily as an intraoperative tool to identify specific brain regions in which to make therapeutic lesions and had neither the inclination (as far as we know) nor technical ability to use it as a stand-alone therapeutic tool. Subsequent reports indicated that intraoperative stimulation was effective in mitigating pain and reducing the motor signs of movement disorders, but it was not until the maturation of battery-powered implantable pulse generators in the 1980s and the pioneering work of Alim Louis Benabid and colleagues that the usefulness of chronic DBS therapy was realized. Since then, DBS systems have been implanted in tens of thousands of patients with medication-refractory neurological disorders. ^, Electrical stimulation has largely supplanted ablation as the surgical treatment of choice for many disorders because DBS is reversible and the “dose” can be titrated to maximize therapeutic benefit while minimizing side effects.

The anatomic targets of DBS are embedded within higher level sensorimotor, associative, and limbic networks in which stimulation can have a variety of complex motor and behavioral effects on a patient. The therapeutic mechanisms of action of DBS are not fully understood, but they likely depend on multiple factors, including not just how electrical stimulation affects neural activity in the target nucleus and nearby fiber pathways, but also how brain networks and disease processes compensate and coadapt with the nonphysiologic input provided by stimulation. In this chapter, we review what is known about the therapeutic mechanisms of DBS from electrophysiologic, imaging, neurochemical, and computational modeling studies; we focus primarily on DBS for PD. Specifically, we address (1) how DBS affects neuronal tissue near the site of stimulation; (2) how neurophysiologic changes associated with DBS translate into therapeutic benefit; and (3) in which brain regions stimulation appears to provide the most therapeutic benefit for different disorders. An understanding of the physiologic mechanisms of DBS—although daunting, in view of the complexity of networks involved—is critical as researchers seek to improve the therapy for current indications, develop new DBS technology, and expand it to treat emerging indications.

Neural Responses To Deep Brain Stimulation

When examining the mechanisms underlying the therapeutic effect of DBS, the clinician must consider the types of brain tissue that surround the electrode, which includes neuronal cell bodies and axons of the target nucleus, as well as fiber tracts passing through or near the stimulation target.

Changes in Somatic Activity in the Stimulated Nucleus

Surgical ablation and DBS in the subthalamic nucleus (STN), internal segment of the globus pallidus (GPi), and motor thalamus produce similar effects on motor signs in patients with PD. This observation led to the hypothesis that high-frequency electrical stimulation produces a functional lesion, inhibiting neuronal activity and reducing output from the stimulated nucleus. ^, Indeed, suppression of somatic activity near the site of DBS has been observed via microelectrode recordings in the STN and in the GPi both in humans and in animal models. Explanations for this inhibition include depolarization block as a result of inactivation of sodium channels ^, and increase in potassium current ; presynaptic depression of excitatory afferents ; and activation of inhibitory afferents. ^, According to the hypothesis of driving afferent input with DBS, the observed decrease in activity in STN and GPi is not surprising. The majority of afferents to the STN and GPi modulate the γ-aminobutyric acid (GABA) system (are GABAergic), ^, and one study showed that inhibition of GPi during GPi-DBS is mediated by GABA receptors. Not all neurons near the site of stimulation, however, are inhibited by DBS. A small fraction of STN neurons increase their firing rate during STN-DBS, possibly because of the activation of afferent excitatory cortical projections. Similarly, for GPi there are both GABAergic (from the striatum and external globus pallidus [GPe]) and glutamatergic (from the STN) projections that may all be activated during GPi-DBS, the net result thus arising from the predominant effect on these afferents. In addition to changes in firing rate, stimulation can cause striking changes in the pattern of neuronal activity. Neurons in the stimulated nucleus have been shown to entrain to the stimulus pulse train, firing at fixed latency after the preceding stimulus pulse. Such “entrainment” has been shown in the STN, GPi, and motor thalamus, and it is believed to derive from the repetitive activation of somatic and dendritic membrane-bound ion channels. ^, Although overall firing rates during stimulation may be suppressed, altered firing patterns consisting of multiple excitatory and inhibitory phases after each stimulus pulse have been observed. ^{, , ,}

Axonal Output From the Stimulated Nucleus

Changes in the firing rate and pattern of somatic activity in the stimulated nucleus do not necessarily translate, however, to similar changes in output from the stimulated nucleus. Although many studies have shown suppression of activity near the site of stimulation, various other studies have demonstrated increased neuronal output to recipient nuclei from the targeted region. This dissociation may be explained by the fact that axons have lower thresholds for action potential generation by electrical stimulation than do cell somas. Therefore, action potentials initiated along the axon can occur irrespective of activity in the soma. The hypothesis that DBS decouples somatic and axonal activity is supported by modeling studies that have shown that somatic activity near the stimulated electrode is suppressed by activation of presynaptic inhibitory terminals, whereas efferent axons are activated by stimulation. ^,

Experimental studies support the activation hypothesis (e.g., activation of output from the site of stimulation). Although it is typically not feasible to record axonal activity directly, it is possible to record neuronal activity in nuclei receiving input from the stimulated nucleus. Hashimoto and colleagues showed in monkeys made parkinsonian using the neurotoxin 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) that therapeutic STN-DBS increased neuronal firing rates in the GPe and GPi. Because STN has glutamatergic projections ( Fig. 113.1A ), this result suggests that DBS increases the output from the stimulated nucleus. Hashimoto and colleagues also found that therapeutic STN-DBS at 130 Hz altered the firing pattern of most GPe and GPi cells, producing entrained responses with increased probability of firing at 3 and 6.5 msec after each stimulus pulse; periods in between showed pronounced inhibition, particularly in the GPi ( Fig. 113.2A –D ). In contrast, lower amplitude, nontherapeutic STN-DBS did not significantly alter GPi neuronal activity. The inhibitory phases in GPi responses were demonstrated to be induced by GABAergic projections from the GPe. In a complementary computational modeling study based on the findings with STN neurons in the monkeys in Hashimoto and colleagues’ study, Miocinovic and colleagues showed that stimulation caused action potentials first in the axon; 50% of STN axons were entrained with the stimulus pulses at least 80% of the time. Results of other studies also support the hypothesis that STN-DBS elicits action potentials along STN axons and activates STN target nuclei. ^,

Electrophysiologic, neurochemical, and imaging data support the activation hypothesis, with the effects in downstream nuclei depending on the nature (excitatory or inhibitory) of the axonal projection. Therapeutic GPe-DBS was shown to cause significant decreases in firing rate and bursting in 76% and 48%, respectively, of recorded STN neurons in parkinsonian monkeys. On the basis of recordings in pallidal receiving areas of the thalamus, Muralidharan and colleagues found that GPi-DBS inhibited 47% of thalamic neurons, which was consistent with orthrodromic activation of GABAergic GPi projections. Similarly, GPi-DBS decreased overall thalamic activity in patients with PD and dystonia, although this was associated with a brief excitatory phase 3.5 to 5 milliseconds after each stimulus pulse, which was consistent with a change in the pattern of thalamic activity during pallidal stimulation. GPi sends GABAergic projections to the pedunculopontine nucleus (PPN), in addition to the motor thalamus, and neuronal activity in both areas has been shown to be suppressed during GPi-DBS in parkinsonian primates. ^,

Results of imaging studies are also consistent with activation of output from the stimulated region. In several studies, researchers have used positron emission tomography (PET) to show that blood flow in GPi increased during STN-DBS. ^, Similarly, in patients receiving DBS in the ventral intermediate nucleus, PET revealed increased blood flow in cortical regions receiving thalamic projections, and functional MRI studies demonstrated increased blood oxygen level–dependent signals in the GPi of patients receiving STN-DBS. ^, Neurochemical experiments also support the hypothesis of increased output from the stimulated nucleus. Microdialysis studies in rats have shown that STN-DBS increased levels of glutamate in the pallidum outflow. ^, Intraoperative microdialysis in human patients revealed that therapeutic STN-DBS was associated with an increase in cyclic guanosine monophosphate concentrations, which was thought to reflect an increase in GPi activity. ^,

Activation of Fiber Tracts

In addition to activating afferent projections to or efferent projections from the stimulated nucleus, DBS also activates fiber tracts passing through or adjacent to the target. Dopaminergic fibers from the substantia nigra pars compacta (SNc), serotoninergic fibers from the dorsal raphe nucleus, and cholinergic fibers from the PPN pass through the GPi en route to the GPe and the putamen. There are reciprocal connections between the STN and GPe; a large proportion of these fibers pass through the GPi. ^, Computational models predict that GPi-DBS activates these fibers, which suggests that the therapeutic mechanisms of DBS may involve multiple pathways, including those projecting from the target nucleus, as well as fibers passing through and those adjacent to (see later) the site of stimulation. ^{, ,}

Fiber pathways adjacent to the site of stimulation have also been demonstrated to be activated during DBS. ^, Consideration of this current spread is especially relevant to DBS in the STN and PPN, which are small nuclei surrounded by several large fiber tracts. ^, The PPN, for instance, is bordered laterally by the medial lemniscus; medially by the superior cerebellar peduncle; rostrally by the posterolateral substantia nigra; rostrodorsally by the retrorubral field; caudally by the pontine cuneiform nucleus, subcuneiform nucleus, and locus caeruleus; and ventrally by the pontine reticular formation. ^, It is therefore not surprising that stimulation currents can and often do exceed the boundaries of the PPN, as revealed by the presence of paresthesias and oscillopsia during PPN-DBS in patients with PD. ^,

More recently attention has been focused on the impact of stimulation on axons projecting to the STN via the “hyperdirect” corticosubthalamic pathway. ^, Several rodent studies have suggested that STN-DBS directly activates motor cortex through antidromic activation of cortical projections, resulting in a disruption of pathologic neuronal activity in the cortex. In humans putative antidromic activation has been evaluated through evoked potential recordings using scalp electroencephalography (EEG) or subdural electrocorticography (ECoG). Although these studies are supportive of the idea that STN-DBS may derive some of its therapeutic efficacy through this hyperdirect pathway, it is unlikely to be the primary mechanism for achieving therapeutic DBS in the basal ganglia for the following reasons: the GPi is not known to receive strong direct projections from M1, therapeutic levels of STN-DBS but not GPi-DBS produce short-latency evoked potentials indicative of antidromic activation of M1, and yet similar clinical effects are achieved with both STN-DBS and GPi-DBS. ^,

The internal capsule runs lateral to the STN, and although it has been hypothesized that some activation of the pyramidal tract fibers may be beneficial to reduce rigidity, excessive activation of the internal capsule by STN-DBS can become problematic because it evokes strong contralateral muscle contractions and paresthesias and can worsen bradykinesia. ^, Results of animal experiments suggest that STN-DBS could influence dopaminergic activity through activation of nigrostriatal and pallidonigral fibers. Although PET studies in patients with advanced PD did not provide evidence of increased striatal dopamine during therapeutic STN-DBS, ^, other animal studies have demonstrated increased levels in the striatum. ^, STN-DBS may also activate pallidothalamic fibers running dorsal to the STN, ^, and it has been suggested that modulation of cerebellothalamic fibers adjacent to the STN underlie the improvements in tremor with STN-DBS. ^, Xu and colleagues showed that neuronal activity in both pallidal and cerebellar receiving areas of the thalamus were modulated by STN-DBS, which is further evidence in support of the hypothesis that activation of adjacent fiber pathways plays a role in mediating the therapeutic benefit of STN-DBS.

Effects of Deep Brain Stimulation on Distal Brain Regions

The neurophysiologic effects of DBS extend beyond the neuronal populations and fibers of passage that are directly modulated by the stimulation and are propagated throughout the entire basal ganglia–thalamocortical network. For example, motor cortex firing patterns were found to be modulated by STN-DBS ^{, ,} and GPi-DBS. ^, This could be a result of stimulation-induced changes in neuronal activity in basal ganglia–thalamocortical pathways, activation of hyperdirect cortical-subthalamic projections, or antidromic activation of fibers within the internal capsule. ^{, , ,} PET studies have shown STN-DBS and GPi-DBS to have broad effects on primary and nonprimary motor areas of the cortex that depend on the stimulation target. Limousin and colleagues found greater motor task–related changes in regional cerebral blood flow in supplementary motor areas, the cingulate cortex, and the dorsolateral prefrontal cortex during STN-DBS than during GPi-DBS, although this difference may have been attributable to suboptimal GPi lead placement, given that GPi-DBS patients did not show the same degree of motor benefit as those who received STN-DBS. Other studies showed that STN-DBS reduced regional cerebral blood flow in motor and prefontal cortical areas. GPi-DBS has been demonstrated to decrease activity in the sensorimotor cortex and to increase activity in premotor cortex and the cerebellum. Regional cerebral blood flow in the sensorimotor cortex was found to be nonlinearly correlated with stimulation intensity during DBS in the ventral intermediate nucleus for ET ; this suggests recruitment of different fiber tracts with increasing voltage and current spread, which results in mixed inhibitory and excitatory effects in sensorimotor cortex. It is evident that DBS for PD can exert a widespread effect, producing changes not just in nuclei within the area being stimulated but also throughout distributed brain networks and that these changes may be dynamic, varying with stimulation parameters and target location.

Therapeutic Mechanisms of Deep Brain Stimulation

How do the observed changes in neuronal activity during DBS translate into therapeutic improvement for the patient? In the case of DBS for PD, stimulation at multiple nodes in the circuit (STN, GPi, GPe) can similarly improve motor symptoms; is there a single therapeutic mechanism of action, or is it different depending on stimulation site? Although the therapeutic effect of DBS for the STN and GPi is akin to that of surgically ablating a region of brain tissue, one could argue that DBS creates a functional lesion in the target site that prevents pathologic signals to be relayed to the rest of the motor circuit. In GPe, however, the effect of lesions exacerbates parkinsonian motor signs and blunts the response to antiparkinsonian medication, whereas DBS in the GPe improves motor signs, suggesting that the effect of DBS is not merely related to removing pathologic activity but rather depends on how it changes activity in the network. Although the specific effects that occur throughout the network may vary based on the target site and stimulation parameters, there would seemingly be some common thread or unifying hypothesis that can explain how these changes improve the disease state. Experimental and computational studies have been addressing aspects these questions, but an explanation for the therapeutic mechanisms of DBS that is consistently supported by experimental evidence remains elusive.