The goal of cortical stimulation (CS) is to change the excitability or activity of cortical and related subcortical networks involved in pathophysiological processes. Any neurological or psychiatric disorder can be affected by CS, either by reactivating hypoactive neuronal structures, as first proposed by us (‘whenever SPECT [single photon emission computed tomograhy] shows cortical disactivation, the therapeutic rationale would be trying to stimulate it’) or inhibiting overactive structures (epilepsy, auditory hallucinations, tinnitus), or both, such as in depression and stroke, i.e. by activating one side and simultaneously inhibiting the contralateral one .
Considering the risk, albeit small, of serious intracerebral hemorrhages and mortality attendant to electrode insertion in deep brain stimulation (DBS), it seems surprising that the much more benign procedures involved in CS have not given the latter the edge in the field of brain stimulation. While DBS for movement disorders may confer a superior benefit (although this awaits head-to-head trials for confirmation), CS outdoes DBS for neuropathic pain, stroke rehabilitation, tinnitus, and probably coma rehabilitation and epilepsy. Importantly, Extradural CS and transcranial direct current stimulation (tDCS) have been proven better than placebo stimulation – given the lack of physiologic effects elicited, whereas DBS cannot be evaluated with the same degree of confidence for several applications. Finally, CS has the potential for neuroprotection (by hyperpolarization of neurotoxic currents) and has clear neuroplasticity-promoting effects.
Several reasons can be adduced:
- 1.
DBS is approved for the treatment of central nervous system disorders; the huge marketing efforts from the manufacturers may have ‘swamped’ other experimental procedures. However, approval by regulatory bodies of repetitive transcranial magnetic stimulation (rTMS) for depression in the past few years might help reverse the trend
- 2.
Not many neurosurgeons have experience with invasive cortical stimulation. Even worse, in view of the supposed ‘simplicity’ of such procedures, some surgeons simply rushed in without an adequate competence and came away with negative results
- 3.
A philosophical reason: neurosurgeons are both enamored of their ability to be precise (as required by the small size of DBS targets) and the empowering high-tech glittering technology involved. Contrast this with the relative low-tech simplicity of CS, which does not necessitate stereotactic equipment and allied paraphernalia
- 4.
Results with cortectomy were attempted for pain and motor disorders in years gone by but results were less than compelling
- 5.
The daunting vastness of the cortical mantle and the astonishing structural intricacy thereof: suffice to say that only in 2009 we finally learned the number of neurons in the human brain (86 billion neurons – 16 billion in the cerebral cortex and a mere 85 billion non-neuronal cells, one tenth of previous estimates) . Also, much of our knowledge of cortical microanatomy and corticocortical connections is based on non-human primates.
History
Systematic application of electromedical equipment for therapeutic use started in the 1700s. Although clearly any form of electricity applied to the head also stimulates the cortex (including the discharge from electric fish used to therapeutic effects since 4000 BCE), CS was applied for the first time by Giovanni Aldini (1762–1834), Luigi Galvani’s nephew, at the end of the 1700s and it was his demonstrations (and the sensationalist newspaper reports) in London that spurred Mary Shelley’s highly successful novel ‘Frankenstein, or the modern Prometheus’. Aldini stimulated the cerebral cortex of one hemisphere in criminals sacrificed about an hour earlier and obtained contralateral facial muscular contractions . This finding was not exploited and had to be rediscovered by Fritz and Hitzig in the second half of the 19th century. Despite attempts by others (including John Wesley and Benjamin Franklin), Aldini was the first to develop transcranial direct current brain stimulation by exploiting Alessandro Volta’s bimetallic pile ( Fig. 2.1 ) and apply it to psychiatric patients, in particular depressed ones, by stimulating the shaved and humidified parietal area. Sir Victor Horsley (1888–1903) triggered movements in the extremities of human patients by electrically stimulating the cerebral cortex. Keen (1887–1903) did the same with a rubberized handpiece with two partially isolated end poles fed by a battery. Others followed, in particular Penfield and Boldrey in the 1930s. In the 1890s, Jacques d’Arsonval induced phosphenes in humans when their heads were placed within a strong time-varying magnetic field which stimulated the retina. This was the first magnetic stimulation of the nervous system. In 1985, Barker and colleagues introduced the first TMS apparatus and transcranial direct current stimulation (tDCS) was ‘rediscovered’ at the end of the 1990s (for historical reviews see ).
In the 1970s, Alberts reported that stimulation at 60 Hz with a 7-contact Delgado cortical plate electrode of an area near the rolandic fissure between motor and sensory sites (SI) could initiate or augment parkinsonian tremor in patients, while Woolsey temporarily alleviated parkinsonian rigidity and tremor in two patients by direct acute intraoperative stimulation in the primary motor cortex (MI). He wrote:
…marked tremor and strong rigidity…The results suggest the possibility that subthreshold electrical stimulation through implanted electrodes might be used to control these symptoms in parkinsonian patients.
However, it was only 10 years later that Tsubokawa’s group in Japan applied extradural motor cortex stimulation for the treatment of central pain and another 10 years passed before the same technique was brought to bear on Parkinson’s disease and then other neural disorders (see historical review ). On the whole, the progress of therapeutic cortical stimulation has been slow and only gained momentum in the first decade of the 21st century.
Anatomical constraints on targeting
The neocortex is a dishomogeneous, ultracomplex, six-layered structure ( Fig. 2.2 ), and is strongly folded: in humans almost two thirds of the neocortex is hidden away in the depth of the sulci. The individual sulci vary in position and course among subjects, but also between the two hemispheres in the same subject, may show one or several interruptions and some may be doubled over a certain part of their trajectory . There are also several cortical hemispheric structural asymmetries . This severely limits the possibility to make overarching generalizations as of targeting.
Cytoarchitectonically, the cortex has been divided into 44 sharply delineated areas by Brodmann a century ago, whose boundaries generally do not coincide with the sulci on the cerebral surface. This areal distribution has been revised by several authors, but the result has added more confusion: anatomical exploration with basic histological stains gives little insight on functional subdivisions. Numerous attempts at defining functionally segregated areas (including electrical stimulation) are on record, with a harsh conflict between localizationists (neo-phrenologists) and anti-localizationists. Based on neuroimaging data, it can be estimated that about 150 juxtaposed structural and potentially functional entities are present in the human neocortex (e.g. areas 9/46 and 44/45 are distinct architectonic entities). Each cortical area has a unique pattern of corticocortical/corticosubcortical connections (connectional and functional fingerprint). Yet, since the neocortical wiring is characterized by a distributed hierarchical network that contains numerous intertwined, cross-talking processing streams, the identification of functionally segregated domains remains a difficult problem. Moreover, most of the human neocortex is occupied by association areas of various kinds and the boundaries between these areas do not closely correspond to those of cytoarchitectonic fields as delineated by Brodmann and others. Additionally, all cortical areas (primary and association) show considerable intersubject variability: this appears to be a general feature of neocortical architectonic areas, a microstructural variation superimposed upon the also considerable macrostructural variation pertaining to the overall size and shape of the hemispheres, as well as the sulcal and gyral pattern. This variability seriously hampers structural–functional correlations. This means that simply transferring ‘hot spots’ in brain imaging studies to a 3D version of Brodmann’s chart incorporated in the stereotaxic atlas of Talairach and Tournoux is apt to lead to erroneous conclusions, since the atlas neglects variability , imposing a serious limit on CS procedures. Spatial normalization procedures are thus necessary.
In most cognitive tasks, two or more cortical areas are activated and these may be considered as nodal points in the networks underlying the process. At the same time, cortical regions (e.g. prefrontal cortex, posterior parietal cortex) are engaged in a wide variety of cognitive demands. The most parsimonious explanation is that they reflect cognitive processes that are tapped by tasks in different domains. This, unfortunately, makes the selection of cortical targets for psychiatric neuromodulation, for example, problematic. The dorsolateral prefrontal cortex (DLPFC), the approved primary target for the treatment of depression, is mainly a cognitive, not a limbic area (and might provide benefit by restoring cognitive control over affect): it is quite large and actually there may be subareas whose stimulation would result in stronger effects . In the end, functional localization and specialization are important principles, but do not offer a complete or sufficient explanation of cortical organization. Rather, a process should be explained in terms of distributed patterns of changing neural activity in networks of interconnected functionally specialized areas. In other words, cognitive and mental abilities result from the functional integration of the elementary processing operations occurring in a smaller or larger number of functional areas . In practice, given the inter-areal connectedness, it is logical to conclude that whatever nodal point is stimulated will entrain the whole network. This has been cogently shown for Parkinson’s disease . Recently, a rostrocaudal gradient model of frontal lobe function has been elaborated upon, undermining the discrete model of frontal functions compartmentalized to highly demarcated zones . The rostrocaudal axis (BA10 to BA9/46 to BA8 to BA6) forms a coherent functional network with longer connections being unidirectional: this implies that adjacent regions along the rostrocaudal axis are connected to one another, but do not project to more rostral regions beyond those immediately adjacent. This has a great importance when one considers possible targets in psychiatric CS (i.e. BA10 would stand out as a primary focus for CS attempts).
Hemispheric specialization must also be accounted for: the right hemisphere is tasked with processing negative affect (and vice versa for the left one), an important consideration for psychiatric ECS: interestingly, parameter modulation (e.g. changing frequency) may ‘recode’ the target function and obtain the sought-after clinical benefit.
Can cortical stimulation be optimized through modeling?
Recently, attempts to model cortical structure and function to fine-tune cortical stimulation efforts have been attempted, in the tracks of what has been done for spinal cord and deep brain stimulation (see ). In practice, they are of little help to practitioners. Why?
For starters, there is very little evidence in favor of the concepts that:
- 1.
the entire neocortex is composed of radially oriented columnar units or modules
- 2.
all of these entities represent variations on one and the same theme
- 3.
all of these entities essentially have the same structure and
- 4.
they all essentially subserve the same function .
This represents a major hurdle by factoring out cortical homogeneity as a foundation for understanding electric field effects. Add to this the dazzling intricacy of cortical cyto- and myelo-architecture . Also, electrical resistance is four to six times higher in the gray than in the white matter.
1. Cells
The cortex accommodates pyramidal (typical and atypical) – 60–85% of all neocortical neurons – and non-pyramidal cells (15–40%) (PC and NPC), with a total number of neocortical synapses numbering at about 300 000 billion.
- A.
The somata of PC are not under the direct influence of any extrinsic afferent system, but only of local circuit neurons (basket cells) and other NPC. PC somata projecting to particular cortical or subcortical targets are preferentially located in particular cortical layers and sublayers. Corticocortical and callosally projecting fibers arise from both LII–III and infragranular PC. The smaller, more superficially situated PC tend to project to ipsilateral cortical areas situated nearby, whereas the larger, more deeply placed cells to contralateral and to more remote ipsilateral cortical areas. Lamina V PC project subcortically to multiple targets: the smallest and more superficial project to the striatum, the largest and most deeply situated to the spinal cord, the intermediate ones to the remaining sites including the thalamus. The projections to the specific thalamic relay nuclei project exclusively from layer V PC.
Although most cortical neuronal populations projecting to a particular cortical or subcortical target show a distinct laminar specificity, it is not uncommon to find some degree of overlap in the boundaries demarcating different populations of projection neurons. Importantly, the degree of subcortical collateralization of corticofugal fibers is limited. The axons of all typical PC release a number of intracortical collaterals: together they constitute the largest single category of axons in the neocortex . Apart from local collaterals, PC axons may also give rise to one to five long, horizontally disposed branches (6–8 mm). These long-range collaterals do not remain within the cytoarchitectonic area in which their parent soma lies but project to adjacent cortical areas. They give off secondary branches in regularly spaced, perpendicularly oriented clusters (column-like) which contact dendrites of other PC but also non-PC. The collaterals of one PC contact numerous other PC and, conversely, one PC receives the converging input of numerous other PCs. Thus, neocortical γ-aminobutyric acid (GABA) interneurons receive input directly from PC axon collaterals and, in turn, synapse with PC, accounting for PC feed-forward/back inhibition. The branching process of axons allows for easier activation by stimulation in comparison to axons without branching.
PC show ample structural diversity: size, laminar position, branching pattern of dendrites, density of spines along apical dendrites, affinity to particular afferent systems, cortical or subcortical target regions, distribution of axon collaterals and patterns of intracortical synaptic output. The somata of PC projecting to a particular target are located in one and the same layer or sublayer and show striking similarities in dendritic morphology, thalamocortical connectivity and distribution of axon collaterals and are in receipt of similar extra and intracortical inputs. Likely, all PC projecting to a particular target are in receipt of similar inputs and have similar functions.
- B.
Non-PC, especially spiny stellate cells, are equally vital. Their axons may descend superficially or to deeper layers and contact PC, whereas their short collateral branches likely contact similar cells. Spiny stellate cells play a crucial role in the radial propagation of the activity fed by thalamocortical afferents into layer IV of primary sensory areas. Local circuit neurons are, with a single exception, GABAergic (inhibitory); 25–30% of these cells also express one or several neuropeptides. There are different subpopulations based on morphology and neurochemistry:
- 1.
stellate neurons (in all layers), including neurogliaform cells in sensory areas
- 2.
chandelier cells (especially layer II) which especially influence corticocortical activity
- 3.
basket cells (large, small and nest), making up about 50% of all inhibitory neocortical interneurons, with the axon giving rise to 4+ horizontal branches and contacting hundreds of PC and tens of other basket cells
- 4.
vertically oriented neurons (bipolar, bitufted, including double bouquet cells, Martinotti cells (all layers except LI)
- 5.
horizontal cells (layers I, or of Cajal, layer VI and NOS).
Interneurons are contacted and contact other interneurons, forming an intricate network which includes electrical coupling, autaptic innervation and specific extrathalamic input. Chandelier cells terminate on the PC axon hillock, basket cells target the somata and proximal dendrites of PC; both classes control output and oscillatory synchronization of groups of PC. Unfortunately, it is not known whether these interneuronal networks extend indefinitely across the neocortex or have distinct boundaries and this makes modeling a desperate enterprise.
- 1.
To sum up, it can safely be said that each particular neocortical area contains a number of networks of interconnected, type specific PC. The number and extent of pyramidal networks present within a given cortical area is unknown. Likely, the various PC belonging to a particular network are in receipt of afferents from cohorts of inhibitory interneurons, each cohort contacting a specific domain of the receptive surface of the PC involved. The inhibitory cells forming these cohorts are all of the same type and are generally reciprocally connected by chemical and electrical synapses. Thalamic inputs selectively contact and strongly excite the interneurons belonging to particular cohorts, while others receive weaker or no thalamic inputs. Each of the various cohorts of inhibitory interneurons impinging on a particular pyramidal network is specifically addressed by one or more of the extrathalamic modulatory systems. Not only the inhibitory input but also the excitatory input to PC belonging to the same network may be specific. Although the degree of separation among pyramidal and interneuronal networks is largely unknown, likely the abundant double bouquet cells with their vertically oriented axonal systems contact PC belonging to different networks and the neurogliaform cells form gap junctions with several other types of inhibitory interneurons.
2. Fibers
Myeloarchitectonically, the myelinated fibers in the cortex show two principal orientations, tangential and radial. Tangential fibers tend to form laminae which, in general, can be readily identified in conjunction with the corresponding layers observed in Nissl preparations. The radially oriented fibers are arranged in bundles (radii) which ascend from and descend to the subcortical white matter. However, the number and distinctness of the tangential fiber layers show considerable local differences in the cortex and the same holds true for the extent to which the radii penetrate into the cortex. Moreover, our knowledge of the fiber connections is almost entirely based on studies in non-human primates (particularly the rhesus macaque) and fiber tracking with diffusion tensor imaging in the human has yet to bear substantially on this problem. This is a major point in CS models.
Specifically, there is a horizontal axonal system contacting the basal dendrites of PC situated at specific levels, but the cortex also contains vast numbers of vertically oriented axonal elements (columnar radial coupling), including thalamocortical and corticocortical association fibers, axons and recurrent collaterals of PC and the vertically elongated axonal systems of some types of cortical local circuit (bipolar) neurons. The latter two classes assemble in highly characteristic radially oriented bundles.
In view of variations in length and in position of their apical dendrites, different PC may receive different samples of lamina-specific extracortical and intracortical afferents and apical dendrites of different PC may exhibit different specific affinities to particular afferent systems. Plus, there are distinct lamina-specific differences in the density of spines along the apical dendrites, lamina-specific side branches on the apical dendrites are present and apical dendritic segments of different PC passing through a particular layer may receive highly different numbers of synapses from the afferents concentrated in that layer. There is also the apical dendritic tuft extending into lamina I to be considered which is contacted by thalamic, monoaminergic, recurrent LII–III PC, ascending deep multipolar/bitufted neuron and horizontal lamina I neuron axons. The afferents from different thalamic nuclei which, after having traversed the cortex, spread in lamina I terminate in different subzones of that layer and the apical dendritic tufts of the pyramids thus receive stratified input from different sources. Extrinsic afferent fibers follow a radial course and most distribute themselves in layered arrays. Different (groups of) thalamic nuclei project in a particular laminar fashion to smaller or larger parts of the neocortex. Importantly, more than 10 different extrathalamic subcortical structures projecting to the neocortex have been identified. The effects of the cholinergic, GABAergic and monoaminergic systems are not generalized excitation or inhibition, but rather region-specific enhancement or diminution of activity in limited neuronal ensembles during certain stages of information processing. Additionally, each particular neocortical area also receives a strong input from other neocortical ipsi- and contralateral areas ending in layers III and IV.
3. Association fibers
Cascades of short association fibers interconnect modality-specific primary with secondary sensory association areas and these latter with multimodal sensory areas located at the borders. They may remain within the gray matter of the cortex or pass through the superficial white matter between neighboring cortical areas as U fibers and are believed to play a starring role in the mechanism of action of CS ( see also in ). Long association systems connect the modality-specific parasensory association cortex and the multimodal areas in the occipital, temporal and parietal lobes with the premotor and prefrontal cortex . Short association fibers interconnect the prefrontal cortex, the premotor area and the motor cortex with the primary somatosensory cortex. Connections from parasensory and multimodal association cortices and prefrontal cortex (PFC) to limbic structures pass via the cingulum to the medial temporal lobe; other fibers originating from parasensory association cortices reach limbic structures via the insula. Most association connections are reciprocal. Connections from the primary sensory areas to their neighboring association areas usually originate from the supragranular layers and terminate in/around layer IV (forward connection). Feedback connections originate in the infragranular layers and terminate in layers I and VI. The laminar analysis of association connections may therefore reveal the direction of information transfer.
In sum, the apical dendritic branches of neocortical PC receive input from various sources, but corticocortical projections constitute by far the largest neocortical input system, making these one of the obvious candidates in the mechanism of action of CS. Thus, it can be safely stated that the neocortex communicates first and foremost with itself . An important consideration: the literature on CS often quotes distant effects (for instance in the case of chronic pain) on limbic areas and brainstem as paramount in the mechanism of action, but these must actually be understood as ‘knock-on’ effects (see a critique of these studies in ).
There are also differences in laminar electrophysiology. For instance, there exists a major difference between sensory-evoked and spontaneous activity in primary sensory cortical regions, namely the site of initiation (layer IV but also upper layer VI versus layer V). Layer V neurons are intrinsically more depolarized than layers II–III, on average being about 10 mV closer to action potential threshold (i.e. more excitable). In addition, layer V neurons are strongly synaptically coupled to other nearby layer V neurons in a highly recurrent excitatory microcircuit (making spontaneous waves of excitation more easily spread). Lamina V neurons have relatively weak connections to layers II–III. Both evoked and spontaneous activities have a relatively limited horizontal spread in superficial layers (i.e. more localized coding) and a more extended propagation in deep layers. But this applies only to action potentials: subthreshold activity propagates widely in superficial layers. How is information encoded? Layer V pyramidal cells fire at a higher rate during both spontaneous and evoked activity (dense firing or population code), whereas lamina II–III pyramidal neurons overall fire at low rates during both types of activity (sparse firing or cell-specific temporal code). There appear to be some neurons in each layer that are orders of magnitude more active than other nearby neurons; perhaps the less active neurons provide a reserve pool to become active at the appropriate moment . How these can all be accommodated inside a model seems a daunting task with current tools.
In the end, this discussion highlights the extreme aspecificity of current cortical stimulation paradigms, since stimulation tends to affect the cortex across the board. A first step would be complexity analysis with closed-loop stimulation devices (e.g. the NeuroPace device for epilepsy control), but it is moot that this alone may circumvent the amazing intricacy of cellular architecture . Does cortical stimulation affect differentially positioned cells in the same way? Does a homogeneous wave of excitation create intracortical conflicts (e.g. two self-effacing inhibitions)? Should dendrites, soma, axon hillocks, nodes, internodes and unmyelinated terminals, all having different electrical properties, be stimulated differentially? This is way beyond current technology. When it comes to details, the only currently feasible approach is to consider the cortex a sort of black box, from which a net effect is sought through trial and error.
MI as a paradigm of cortical stimulation
The primary motor cortex (MI) has been the first target of CS endeavors, especially for chronic pain and control of movement disorders . Understanding it may help bring out general principles which can then be applied to other areas and disorders. The upshot can be anticipated: MI is less straightforward than previously thought.
MI is far from the passive servant of higher motor structures. It performs a complex integration of multiple influences, originating in both cerebral hemispheres, in a role as the ultimate gate-keeper that is carefully and differentially tuned to generate well-defined motor behaviors . The discharge pattern of individual MI neurons conveys a bewildering diversity of information. Thus, some neurons receive strong sensory input, whereas others do not. Some neurons respond to contralateral, ipsilateral or bilateral movements; some neurons even reflect sensory signals used to guide action . Many pyramidal tract neurons respond with a wide range of peripheral inputs (visuo-audio-vestibular) .
MI has two subdivisions. A rostral region lacks monosynaptic cortico-motoneuronal cells (evolutionarily old MI) – descending commands are mediated through spinal circuitry, and a caudal region (evolutionarily new MI) with monosynaptic cortico-motoneuronal cells which have direct access to motoneurons in the ventral horn essential for highly skilled movements . Neurons in the rostral portion of MI may be more related to kinematic variables, such as velocity and movement direction, than more caudally placed cells .
MI is partially sensory due to the coexistence within the same neurons of motor and sensory properties. In particular, MI and SI hand cortices overlap and are not divided in a simple manner by the central sulcus and sensory responses are elicitable well outside the classically accepted anatomical borders (see references in ). In functional magnetic resonance imaging (fMRI) studies, the motor hand area may extend to (50% of cases), or be located exclusively, in SI (20% of cases), even during the simplest motor tasks . Apart from intrinsic responses, MI and SI are so tightly interconnected by short corticocortical U-fibers that arborize over a considerable rostrocaudal distance in MI to make them almost a unique structure . SI is a major source of somatosensory input to MI and MI is strongly modulated by sensory flow (and vice versa) . Clearly, uniformly targeting MI in ECS efforts for chronic pain and Parkinson’s disease may be misplaced: SI could be another potential target. Also, BA44 (found 2 cm anterior to MI tongue area) has direct fast conducting corticospinal projections with a role in voluntary hand movements , confirming the haziness of MI borders.
Evidence shows a rough body-centered map of MI that matches the traditional motor homunculus. This map extends to nearby premotor areas. Yet, rather than discrete regions of MI controlling different parts of the arm, control of each part is mediated by an extensive territory that overlaps with the territories controlling other parts . Whereas the prior view suggested that stimulation of different regions of MI should elicit movement of different body parts, it is now clear that stimulation can elicit movement of a given body part from a broad region, i.e. MI has a broadly overlapping mosaic of points where stimulation elicits movements of different body parts. Any given MI neuron may influence the motoneuron pools of several muscles (not just one). Selective stimulation of different regions in MI can produce the same movement, due to intra-MI dense bi-directional projections of up to 1 cm. Limb joints are represented in the cortex more than once, but with different contiguity (shoulder to wrist, shoulder to elbow) . Rather than simply controlling different body parts, MI directs a host of body parts to assume complex postures. The map appears to be organized not just according to muscle groups, but to the positions in space where the movements conclude . Two dissociable systems for motor control (one for the execution of small precise movements – especially distal muscles – and another for postural stabilization – especially proximal muscles) coexist in MI, with the representation of distal and proximal muscles substantially intermingled within the MI arm representation. Depending on duration of stimuli applied on MI, simple or complex movements can be elicited. This clearly proves the difficulty of modeling even such an apparently known cortical area.
The picture gets even more complex. In one out of five patients, there are variations in the organization of MI, i.e. mosaicism (overlapping of functional areas), variability (inverted disposition of MI functional areas) or both : for instance, the sensory hand area may be found between 1 and 7 cm from the sylvian sulcus and leg sensation can be found within 3 cm of the sylvian fissure. These findings suggest that individual neurons over the postcentral gyrus responding to a specific stimulus may appear to be arranged randomly rather than grouped together. There is significant intermixing of sensory neurons that respond to different sensory modalities and similar results apply to MI . Moreover, the local mosaic-like topography (somatotopy) of individual distal arm representations is highly idiosyncratic, with wide variability among subjects . Finally, somatotopic differences not only exist between subjects, but also between hemispheres in the single case. In Parkinson’s disease (PD) specifically, map shifts are found in the majority of the patients, both in untreated early cases and treated cases of long duration, with a correlation between inter-side differences in the severity of PD symptoms and inter-hemispheric map displacement .
The left and right hemispheres are specialized for controlling different features of movement. In reaching movements, the non-dominant arm appears better adapted for achieving accurate final positions and the dominant arm for specifying initial trajectory features (e.g. movement direction and peak acceleration) . Also, the area of hand representation is greater in the dominant (left) than in the non-dominant hemisphere, with greater dispersion of elementary movement representations and more profuse horizontal connections between them, thus leading to more dexterous behavior of the dominant hand . Stronger beta rebound after right median nerve stimulation is observed in the left compared with the right hemisphere . This suggests that left MI ECS may be expected to have different effects.
In sum, MI is not just classical Brodmann’s area 4: more anterior and posterior areas must be investigated. Premotor cortex BA6 lies on the crown of the precentral gyrus, thus needing less energy for activation, while MI is mostly within the central sulcus. SI is another option for both pain and Parkinson’s disease.
Mechanism of action and parameters considerations
- 1.
Neural changes during stimulation include excitation, inhibition ( Fig. 2.3 ), oscillatory changes in corticosubcortical loops and intracortical layers and neuroplastic changes. Despite several authors suggesting an exclusive subcortical action of CS, neuroimaging and electrophysiological data confirm that the primary locus of action is the cortex itself (see discussion in ). This applies to both extradural and non-invasive CS. For instance, the analgesic effects of rTMS of both MI and DLPFC do not depend on the activation of descending inhibitory systems . CS renormalizes a disrupted intracortical function (disinhibition, as demonstrated in the setting of central pain with GABAergic–propofol challenge : by acting on small inhibitory axons (probably Golgi-II cells with long axons) and, via U-fibers, modulates nearby areas, specifically SI in pain patients. At the same time, disrupted oscillatory patterns between cortex and thalamus (e.g. central pain) or basal ganglia (e.g. Parkinson’s disease) are shifted towards more normal patterns , also by way of antidromic effects . On the other hand, the Neuropace apparatus appears to be purely cortical when delivered through cortical paddles.
