Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Cortical and Subcortical Brain Mapping
Introduction
The ultimate goal of brain surgery, especially in neuro-oncology, is to maximize the extent of resection (EoR) while preserving or even improving the patient’s quality of life (QoL)—that is, to optimize the oncofunctional balance. , Indeed, maximal safe resection of gliomas, when feasible, is currently the first treatment both in low-grade gliomas (LGGs) and high-grade gliomas. All the recent surgical series that have objectively calculated the EoR on repeated postoperative magnetic resonance imaging (MRI) have supported EoR as a statistically significant predictor of overall survival (OS). In World Health Organization (WHO) grade II gliomas—when a simple biopsy is obtained at diagnosis eventually followed by chemotherapy and/or radiotherapy—the survival is around 6 to 7 years, , whereas the OS is about 14 to 15 years with early surgical resection, in particular when there is no residue but also if the residual tumor volume is less than 10 to 15 mL. In the case of glioblastomas, it has also been found that complete removal of the enhanced part of the tumor (controlled on postsurgical MRI) significantly increased the median survival. , , Moreover, because the tumoral cells diffuse beyond the signal abnormalities on neuroimaging, the principle of removing a “security margin” around the lesion visible on FLAIR-weighted MRI has recently been suggested. In LGGs, this “supracomplete” resection prevented malignant transformation with a long-term follow-up of 11 years (range 8 to 16.5 years). In glioblastoma, the removal of more than 53.21% of the surrounding FLAIR abnormalities resulted in a significantly longer OS compared to the resection of only the contrast-enhanced area. Therefore these improved oncologic results have challenged the dogma of a “watch and see” policy, especially in LGG, as well as the classical concept of achieving a simple tumorectomy in brain neoplasms: quite the opposite, they support the principle of an early and supramarginal surgical resection.
However, such aggressive management can result in a higher rate of neurologic deterioration. Indeed, because supratentorial gliomas are often situated near or within the eloquent structures and because they have an infiltrative (poorly demarcated) nature, for a long time it was believed that the chances of achieving an extensive glioma excision were low while the risk of causing permanent postoperative deficits was high. Indeed, in the old literature, many surgical series have reported a rate of persistent and severe neurologic impairment between 13% and 27.5% after the removal of intra-axial tumors (for a review, see Ref. 16). This may be explained by the fact that although the removal of anatomic landmarks is crucial during brain surgery, that is not sufficient, since there is considerable interindividual variation in anatomic function. This is particularly true in cases of gliomas that can generate functional reorganization, explaining why many patients have no or only mild deficits before surgery, especially in the case of slow-growing tumors such as LGGs. In this setting, to preserve neural function while achieving a more radical removal, an emerging philosophy suggests performing individual functional mapping-guided resection as opposed to image-assisted surgery as classically proposed. This optimization of the benefit-to-risk ratio of surgery, aiming at improving QoL as well as prolonging OS, can be implemented only by studying the distribution of the parallel, delocalized, large-scale cortical-subcortical neural circuits at the individual level by means of intraoperative electrical mapping and real-time cognitive monitoring in awake patients. In this regard, surgical resection is pursued until eloquent structures have been reached—that is, up to functional boundaries—with no margin left around the networks critical for brain function. A perfect comprehension of the hodopical organization of the brain—namely the dynamics of the neural subcircuits mediating specific subfunctions, their interactions, as well as their potentials and limitations of functional compensation—is mandatory to optimize the oncofunctional balance of surgery. In this connectomal view of cerebral processing, breaking with the traditional localization model, mechanisms of neuroplasticity can be efficient only if subcortical connectivity is preserved in order to permit spatial communication and temporal synchronization within and between interconnected distributed subcircuits. In summary, it is crucial to investigate for each patient (1) the individual cortical functional organization, (2) the subcortical connectivity, and (3) the brain’s plastic potential, with the aim of tailoring the resection according not only to oncologic but also to cortical-subcortical functional limits (i.e., to the functional connectome).
The goal of this chapter is to review how the philosophy of cognitive monitoring during resection combined with intraoperative electrostimulation mapping (IESM), both at cortical and subcortical levels, can result in significantly better results of glioma surgery. It can intervene in the natural history of the tumor by delaying its malignant transformation and thus increasing OS, and also by preserving QoL. The fundamental implications of IESM in cognitive neurosciences are also discussed.
Preoperative Neurocognitive Assessment in Glioma Patients: A Necessary Baseline
Diffuse gliomas, especially LGGs, are usually revealed by inaugural seizures in young adults who enjoy a normal life, with no or only slight neurologic impairments. As mentioned, this can be explained by mechanisms of brain reshaping allowing functional compensation in case of slow-growing lesions due to the recruitment of perilesional or remote areas within the ipsilesional hemisphere and/or the recruitment of contrahemispheric homologous areas. However, recent series in which objective neuropsychologic examinations have been performed before treatment have revealed that most glioma patients had cognitive disturbances—especially concerning executive functions, speed and information processing, language, visual perception, working memory, attention, reasoning, learning or semantics as well as emotional and behavioral function—deficits that have long been underestimated. , Even in incidentally discovered LGGs, only two-thirds of patients reported not only subjective complaints at diagnosis (in particular, fatigue and attentional deficits) but also objective disorders of neurocognitive functions (in particular concerning executive functions, working memory, and attentional processes). This means that the standard neurologic examination is too crude to be capable of detecting subtle cognitive alterations. Furthermore, neuropsychologic evaluation is useful to tailor the appropriate individualized therapeutic sequence —for example, to propose upfront chemotherapy rather than surgical resection in very diffuse “gliomatosis-like” gliomas that have already generated severe neurocognitive deterioration or to adapt the surgical methodology on the basis of the presurgical neuropsychologic scores (especially to select the best intrasurgical tests for awake mapping, see later) as well as to elaborate specific programs of postoperative cognitive neurorehabilitation. Moreover, the existence of an abnormal baseline neurocognitive score is a strong predictor of shorter OS in glioma patients.
Of note, these deficits are related to the migration of glioma cells along the white matter fibers. For example, a significant relationship between scores of semantic fluency and the infiltration of the subcortical tracts underlying the ventrolateral connectivity (i.e., the inferior fronto-occipital fascicle [IFOF]) has been demonstrated. In the same spirit, deterioration in mentation (a key function in understanding and performing complex social interactions) was correlated to the degree of glioma invasion of the right frontoparietal connectivity. These findings show that the migration of glioma cells along the subcortical bundles can cause specific cognitive or mood disorders depending on the neural subcircuits invaded by the neoplasm. This is explained by the fact that the limitations of cerebral plastic potential are mainly represented by the axonal pathways, as revealed by a recent atlas of neuroplasticity —conversely to the huge potential of cortical neuroplasticity—making it possible to perform extensive resections in eloquent cortex without inducing permanent impairments. Of note, the recent integration of these concepts into therapeutic strategy has resulted in dramatic changes in the surgical management of glioma patients, with an increase of surgical indications within eloquent regions classically considered as “inoperable” on the condition that the white matter connectivity is not predominantly invaded (see later).
In addition to objective neurocognitive scores, the individual patient’s QoL must be subjectively defined by the patient before surgery based on personal criteria such as job, hobbies, lifestyle, and projects. Therefore “mapping à la carte” must be elaborated in agreement with the needs of the patient, resulting in the selection of the best tasks to perform throughout surgery according to preoperative discussion with the patient and his or her relatives. For instance, mapping different languages and the capacity to voluntarily switch from one language to another will be performed in multilingual patients, particularly in translators; calculation will be tested intraoperatively in mathematicians or schoolteachers; spatial awareness will be mapped in dancers or athletes; working memory can be monitored in managers or businesspeople; bimanual coordination in surgeons or pianists; syntax in writers; cross-modal judgment in lawyers; theory of mind in medical doctors, to preserve their empathy; visual field in taxi drivers; and so forth. In other words, intraoperative neurocognitive monitoring must be personalized according to the QoL of the patient and not according to the lobar location of the tumor. ,
In summary, a systematic and standardized preoperative neuropsychologic assessment is now recommended (1) to search the possible cognitive and mood disorders not detected by a standard neurologic examination, (2) to plan the operative procedure accordingly (e.g., optimal selection of intraoperative tasks for the real-time cognitive monitoring), (3) to benefit from a presurgical baseline allowing a comparison with the postsurgical evaluation, and (4) to plan a personalized functional rehabilitation.
Limitations of Functional Neuroimaging for Individual Preoperative Mapping
Details regarding the science and usefulness of diffusion tensor imaging (DTI) and functional imaging are discussed elsewhere in this book. Advances in functional neuroimaging techniques—namely, functional magnetic resonance imaging (fMRI) and tractography by DTI—enable noninvasive mapping of the entire brain. Yet fMRI is not reliable enough at the individual level to be used routinely in surgical practice as it does not directly reflect the reality of brain processing but gives only a very indirect approximation of cerebral functions based upon biomathematic reconstructions.
Comparison between fMRI and intraoperative electrostimulation mapping (IESM) found only 71% of positive correlations for motor function. Poor correlations have also been found for language (about 33%), with a sensitivity around 66% (specificity from 0% to 97%). By comparing preoperative language fMRI and IESM, Kuchcinski et al. observed a sensitivity of fMRI at 37.1% and a specificity at 83.4%. In a review on fMRI and language, Giussani et al. concluded that the contradictory results of these studies did not allow consideration of language fMRI as an alternative tool to direct cortical stimulation in tumors located in language areas. Moreover, fMRI cannot differentiate areas that are essential for brain function and that should be preserved during surgery from areas involved in but not critical to a given function—that is, areas that can be surgically removed because functional compensation will occur. In practice, due to the frequent false-positive results of fMRI, there is a risk of failing to select a tumor patient for surgery, resulting in a loss of opportunity from an oncologic perspective. This issue has been described for gliomas located within the so-called Broca area and Wernicke area or the insula: these tumors have nonetheless been excised with a very good recovery, although these sites were wrongly thought to be critical according to the results of the preoperative fMRI. Additionally, because of frequent false-negative fMRIs, the other risk is to remove a glioma involving a structure that ultimately proved to be critical for brain functions but was not identified by the preoperative fMRI, with a loss of opportunity from a functional point of view. Besides activation task-based fMRI, a recent investigation comparing IESM mapping with preoperative resting-state fMRI in 98 LGG patients revealed that 96% ± 11% of sensorimotor stimulation was located within 10 mm sensorimotor independent component analysis maps versus 92% ± 21% for language. Moreover, 3.1% and 15% of excised cortex overlapped sensorimotor and language circuits, respectively. Therefore, even if resting-state fMRI succeeded to some extent in distinguishing eloquent versus resectable sensorimotor and language sites, these original findings revealed a high interindividual variability of mapping accuracy and a rate about 80% in the detection of functional cortical areas—making it clearly insufficient to use resting-state fMRI on its own for presurgical mapping. Further validation studies are needed to increase the reliability of this technique for surgical planning, especially by using network automatic identification and/or subnetwork analysis. In conclusion, fMRI is not ready for prime time in guiding glioma resection.
DTI is a useful adjunct, but when dealing with abnormal or distorted fiber tract anatomy induced by gliomas, the risk of erroneous DTI results is increased because diffusion properties can be affected by the neoplasm. These drawbacks explain why correlation studies between DTI and IESM have found concordance in only 82% of cases. Consequently, negative tractography does not mean that no critical tracts are present within the glioma. In other words, DTI is unable to provide any information about the functionality of a specific fiber: it can give only indirect data concerning its structure. DTI is not currently reliable for surgical indications or planning. Of note, there is no article in the literature demonstrating that the use of fMRI and/or DTI for tumor surgery resulted both in an increase in EoR and a decrease in postoperative morbidity in the same patients —in contrast to IESM (see later).
The danger for young neurosurgeons who use fMRI/DTI routinely in the operating room (incorporated into a neuronavigational system or acquired in real time with intraoperative MRI) is to become dependent on neuroimaging. They will not be able to operate on the brain without intrasurgical neuroimaging on the sole basis of their knowledge of the functional anatomy validated by online feedback provided by cognitive monitoring and IESM in awake patients. In fact, functional neuroimaging must be seen as an educational and a research tool outside the operating room —for example to explore the mechanisms of neuroplasticity in longitudinal fMRI studies , but not as a clinical tool into the operating room.
The Gold Standard: Intraoperative Cognitive Monitoring and Cortical-Subcortical Stimulation Mapping in Awake Patients
Due to major anatomic-functional variability across patients and possibly within the same patient over time because of cerebral reorganization (particularly in LGG), intraoperative functional mapping must be systematically achieved under local anesthesia in all glioma patients—at least those who do not have severe deficits before surgery. Indeed, even though motor mapping can be done under general anesthesia by means of intrasurgical electrophysiologic monitoring using motor evoked potentials, this technique is nonetheless unable to map motor cognition with the same accuracy as in awake patients, especially regarding complex movements or bimanual coordination (see later). IESM is currently the sole technique that enables the direct identification (and then the preservation) of cortical-subcortical networks critical for neural functions.
In practice, bipolar electrode tips spaced 5 mm apart and delivering a biphasic current (pulse frequency 60 Hz, single-pulse phase duration 1 ms) are applied to the brain. The current intensity adapted to each patient is determined by progressively increasing the amplitude in 1-mA increments from a baseline of 1 mA until a functional response is elicited, with 5 mA as the upper limit under local anesthesia with the goal of avoiding seizures. The patient is never informed when the brain is stimulated. Each cortical site (size 5 by 5 mm, due to the spatial resolution of the probe) of the entire cortex exposed by the bone flap is tested three times to determine whether it is functionally crucial by generating disturbances during the three stimulations and with normalization of the function as soon as the stimulation is stopped. At least one picture presentation without stimulation must separate each stimulation, and no site is stimulated twice in succession so as to avoid seizures. However, in cases of stimulation-induced seizures, irrigation with cold Ringer lactate is recommended to abrogate the seizure activity.
In addition, it is crucial to benefit from positive cortical mapping before beginning to remove the tumor with the goal of tailoring the resection according to the distribution of the neural networks in the given patient at the given moment. In series supporting negative mapping, 1.6% to 9% of new permanent neurologic deteriorations have been observed. It is recommended to perform a wider bone flap with the goal of obtaining systematic functional responses prior to surgical excision by exposing at least the ventral premotor cortex (vPMC); this will generate articulatory disturbances in all cases during stimulation, whatever the hemisphere. Moreover, positive mapping might also enable optimization of the EoR, since the resection can be pursued until eloquent areas are encountered (i.e., with no margin around the functional structures); this may increase the rate of supratotal resections. In a consecutive and homogeneous series of 115 LGGs in the left hemisphere, the rate of permanent deficits remained lower than 2% despite the absence of margin around the language sites. Of note, due to the use of low-intensity electrostimulation (mean of about 2.25 mA in the author’s personal experience), electrocorticography is not necessary to obtain positive mapping and to limit the risk of intraoperative epilepsy. This was obtained in about 3% in a recent prospective series with 374 supratentorial brain lesions without any aborted awake procedures. Therefore “minimal invasive neurosurgery” means “minimal morbidity” and not “minimal bone flap size.”
To benefit from extensive and reliable neurocognitive assessment throughout the resection, it is crucial that a speech therapist/neuropsychologist/neurologist be present in the operating room in order to interpret the kinds of disturbances caused by IESM—for example, arrest of movement, involuntary limb/face movement or ocular saccades, speech arrest, anarthria, speech apraxia, phonologic disturbances, semantic paraphasia, anomia, syntactic errors, nonverbal semantic disorders, deficits in visual attention, perseveration, dyscalculia, mentalizing deficit, and so on. Thus IESM is able to identify in real-time the cortical sites essential for function before the beginning of the resection in order to select the best surgical approach and to define the cortical limits of lesion removal ( Fig. 5.1 ).
The other critical issue after identification of the eloquent cortical areas is the mapping of subcortical connectivity. Indeed, brain lesion studies have taught us that damage to the white matter pathways generates more severe deficits than cortical injury, with only a very low rate of recovery. Therefore the subcortical tracts subserving motor, somatosensory, visuospatial, auditory-vestibular, language, executive, and emotional functions must be detected during lesion removal in order to preserve the anatomic-functional connectivity while optimizing the EoR—namely, to stop the resection only when these deep eloquent pathways are encountered. Of note, since the patient can be tired after 1 to 2 hours of continuous tasks, glioma removal should be started directly into the contact of the functional structures detected using cortical and axonal IESM in order to avoid the waste of time. The aim is to disconnect the part of the brain invaded by the glioma and not to debulk the tumor from inside by coming closer to the critical regions only at the end of the resection, when the patient is less cooperative. Once the infiltrated brain is disconnected up to functional limits provided by individual mapping, it can be resected under general anesthesia, since real-time feedback from the patient is no longer useful.
IESM is highly sensitive for detecting the cortical and axonal eloquent structures; it also provides a unique opportunity to study brain connectivity, since each area responsive to stimulation is in fact an input gate into a large-scale network and not an isolated discrete functional site.
Mapping the Connectome: What We Have Learned from Intraoperative Electrostimulation Mapping
Recent advances in cognitive neuroscience resulted in a paradigmatic shift from a traditional rigid and localizationist view of brain organization (in which one given area corresponds to one specific function, with no recovery if this area is damaged) to a dynamic hodotopical model, in which the central nervous system is organized in parallel and interconnected large-scale cortical-subcortical circuits. In this networking account of brain processes, IESM plays a major role to map the connectome for each patient—that is, to identify the hubs and pathways subserving neural functions with the aim of optimizing the oncofunctional balance of surgery. Here, the goal is to detail the functional anatomy of these subnetworks as well as their interactions.
Sensorimotor Subnetwork
Recent IESM explorations have changed our view of the neural basis underlying movement away from a hierarchical organization and toward complex circuits. Indeed, voluntary movement engages a set of highly sophisticated neurocognitive processes called motor cognition—namely, intention to act, motor planning, motor initiation, action control, and so forth. A basic way to monitor all aspects of voluntary movement during surgery is to ask the patient to do a simple dual motor task involving the upper limb (i.e., to lower the arm and open the hand, then to raise the arm and close the hand (the lower limb may be also tested, or both upper and lower limbs at the same time). In addition to controlling the velocity and the accuracy of the movement throughout surgery, this task makes it possible to map critical structures for motor cognition using IESM. Typically, stimulation of the supplementary motor area (SMA) can lead to motor initiation disorders. Other motor tasks may also be performed to map more specific motor abilities (e.g., to ask the patients to perform coordinated movements with both hands, which is crucial for certain professions such as manual work or performance on a musical instrument). In the same vein, the patient can achieve more complex movements to assess fine-grained motor abilities such as drumming or to perform reflexive praxis (e.g., imitation of meaningless movements) to evaluate movement planning.
IESM has confirmed that the corticospinal tract originates from the primary motor cortex, runs with a somatotopical distribution within the corona radiata (with, from medial to lateral, the pyramidal tract of the lower limb, upper limb, and the face), and then within the posterior limb of the internal capsule (with, from anterior to posterior, the pyramidal tract of the face, upper limb, and lower limb) before reaching the brain stem and spinal cord. Moreover, the existence of an additional network involved in motor control has recently been evidenced: IESM of this circuit evokes arrest or acceleration of movement in awake patients, with no loss of consciousness. The axonal stimulation sites are distributed in a veil-like pattern, anterior to the primary motor fibers, supporting the existence of descending tracts originating from the SMA, premotor cortex, and anterior cingulate, known for negative motor response characteristics, and running to the striatum through the frontostriatal tract. Additionally, these white matter bundles mediating the control of movement are somatotopically organized. In fact, IESM of the fibers from lateral to mesial and from anterior to posterior causes arrest of movement of the face/speech (laterally and anteriorly), upper limbs, and lower limbs (mesially and posteriorly). Further stimulation areas in the anterior arm of the internal capsule have indicated a large-scale motor control network. Bilateral negative motor responses can also be induced by unilateral axonal IESM. Such data support the existence of a bilateral cortical-subcortical circuit connecting the premotor cortices, basal ganglia, and spinal cord involved in the control of bimanual coordination. Finally, an IESM study that investigated the neural underpinnings of eye movement (implicated in the control of the spatial orientation of attention), revealed that the oculomotor tract originating from the frontal eye field could be part of this motor control network.
Posterior thalamocortical somatosensory pathways and their somatotopy have also been studied by IESM, which evokes dysesthesias or tingling in awake patients. , It is worth noting that axonal stimulation of the white matter behind the central sulcus can also cause disorders in movement control, likely explained by a transient inhibition of U fibers within the central region. , In addition, in patients who had interference with movement during subcortical IESM, pathways that elicited inhibition or acceleration were located immediately posterior to the thalamocortical somatosensory fibers. Thus, a thalamoparietal connection distinct from somatosensory pathways is probable. Based on these original IESM findings, the existence of a wide frontothalamoparietal sensorimotor subnetwork has been proposed. In other words, these stimulation-based findings gave evidence that the circuit subserving motor control is not centered on the frontal lobe but extends to other regions because it includes frontal and parietal white matter pathways as well as projection fibers with inhibitory and excitatory features.
Subnetwork Mediating Visuospatial Attention
The optic tract originates from the lateral geniculate body in three bundles. The anterior bundle, the so-called Meyer loop, curves anterolaterally above and in front of the temporal horn and then loops backward along the inferolateral wall of the atrium. The middle bundle runs laterally around and turns posteriorly along the lateral wall of the atrium and the occipital horn. The posterior bundle runs directly backward, again along the lateral wall of the atrium and occipital horn. In patients with lateral homonymous hemianopia, driving is prohibited in many countries, and several activities such as reading become difficult. Therefore, visual pathways also have to be mapped in awake patients. A protocol was recently proposed in which, with the vision being fixed at the center of the screen, patients are asked to name successively two pictures disposed in the two opposite quadrants, knowing that it is essential to preserve the inferior quadrant (the superior being compensable in daily life). Whereas IESM of visual pathways usually elicits a range of phenomena subjectively described by the patient (either “inhibitory phenomena,” such as blurred vision or impression of shadow, or “excitatory phenomena,” such as phosphenes of visual illusions), the described task makes it possible to benefit from a more objective confirmation of the transient visual disorders evoked by electrostimulation: in fact, the patient is not able to name the picture presented in the inferior quadrant contralateral to the tumor. The amplitude of visual saccades or the possible increase of naming response time in the visual field under scrutiny must also be taken into consideration. The extent of the visual field must regularly be checked manually.
The inferolateral occipitotemporal cortex plays a major role in object recognition: injury of this system can result in visual agnosia (i.e., a disabling inability to recognize object with the visual modality). A simple way to map such high-order visual processes during awake surgery is to ask the patient to engage in a picture-naming task. When IESM causes disorders of object recognition, the patient usually commits a non–semantically related “visual paraphasia.” Furthermore, axonal IESM of the inferior longitudinal fasciculus (ILF) may elicit contralateral visual hemiagnosia, supporting the existence of an occipitotemporal pathway connecting occipital visual input to higher-level processing in temporal lobe structures, in particular the fusiform gyrus and the anterior temporal lobe. These stimulation findings support a crucial role for the ILF in visual recognition, with specialization of this bundle for visuospatial processing in the right hemisphere and language processing in the left hemisphere—inducing alexia or lexical retrieval difficulties when stimulated. It is worth noting that bilateral disconnection of ILF can provoke prosopagnosia.
Unilateral spatial neglect is characterized by the failure to explore and allocate attention in the space contralateral to the injured hemisphere (especially the right) and has a major impact on QoL—for example, regarding the ability to drive. A classical test to evaluate spatial neglect during surgery is the bisection line task. In a touch-screen environment, the patient is asked to show the true center of an 18-cm line. If IESM elicits a significant rightward deviation (typically 7 mm or slightly more), the brain structure is considered as eloquent for visuospatial cognition. This task is helpful in mapping the posterior parietal cortex, including the inferior and superior parietal lobules and to a lesser extent the posterior temporal cortex. , In addition, IESM of layer II of the superior longitudinal fasciculus (SLF), which connects the superior and inferior parietal lobules with the posterior middle frontal gyrus, also generates rightward deviation during a line-bisection task. These data suggest that parietal-frontal communication, thought to maintain information exchange between the ventral and the dorsal attention systems, is necessary for symmetric processing of the visual scene. Thus spatial awareness depends not only on the cortical areas of the temporoparietal junction but also on a larger parietofrontal network communicating via the right SLF. In this regard, a recent IESM investigation evidenced that the right inferior frontooccipital fasciculus (IFOF)—a brain-wide white matter tract connecting the occipital, parietal, and temporal lobes and the prefrontal cortex —also plays a role in spatial cognition, since its stimulation may cause a transient hemineglect. In summary, the use of a simple line-bisection task makes it possible to map and spare critical subnetworks for spatial attention. Indeed, the effectiveness of the IESM was demonstrated by a recent study reporting that, whereas about half of the patients with a glioma experienced a transient neglect in the immediate postoperative period, none presented a permanent spatial neglect.
Finally, stimulation of another subcircuit in the right SLF can generate a central vestibular syndrome with vertigo due to a transitory disruption of the vestibular inputs assembled in the temporoparietal areas and the prefrontal cortex. This finding demonstrates the role of the SLF in the network coordinating body posture and spatial orientation.
Language Subnetwork
To avoid aphasia, the naming task remains the gold standard during awake surgery because it is very sensitive to all levels of language processing. IESM may evoke different kinds of disturbances: speech arrest, dysarthria (disturbance of motor programming), anomia (disturbance of lexical retrieval), phonologic paraphasia (disturbance of phonologic encoding: production of a word with phonologic deviations such as phelephant for elephant ), semantic paraphasia (disturbance of semantic processing, production of a word semantically related to the target word, such as cow for horse ) or perseveration (disturbance of inhibitory control mechanisms, repetition of a prior word in front of a new picture). , Beyond classic language-related cortical areas, the naming task enables the surgeon to map the main associative connectivities—the arcuate fascicle (AF) for phonologic processes, the lateral SLF for articulatory processes, the IFOF for semantic control processes (see later), and the ILF for lexical retrieval—as well as certain intralobar tracts, such as the frontal aslant tract (involved in speech initiation and control). ,
Based on IESM findings, a dual-stream model for visual language processing has been proposed with a ventral route dedicated to mapping visual information to meaning (semantics) and a dorsal stream dedicated to mapping visual information to articulation through visuophonologic conversion. In comparison with other neurocognitive or computational models (e.g., by Hickock and Poeppel), this model has the great advantage of integrating anatomic constraints, especially the white matter connectivity information.
The first step of this model is object recognition. As mentioned, IESM of the ILF, especially in the right hemisphere, typically elicits (hemi-)visual paraphasia, showing that the patient is no longer able to correctly recognize the picture. These problems of visual cognition are usually caused by IESM of the posterior part of the ILF, which interconnects the primary visual cortex with the visual word form area (VWFA) located in the fusiform gyrus. Of note, stimulation (or surgical resection) of the same white matter fibers in the left hemisphere, especially those interconnecting the primary visual cortex and the VWFA, leads to pure alexia. Disruption of the VWFA per se generates different forms of alexia depending the site of stimulation (i.e., posterior, dorsal, or anterodorsal).
The language network is mediated by two main pathways that work in parallel while interacting: the phonologic dorsal stream and the semantic ventral stream. A double dissociation between phonemic and semantic errors has been found during axonal IESM, showing that the two processes are performed in a synergetic manner. The dorsal pathway is underpinned by the SLF and consists of two subparts. The deep part is the classic arcuate fascicle (AF) that connects the posterior temporal structures (mostly the middle and inferior gyri) to the inferior frontal gyrus (mainly the pars opercularis). IESM of this subpart elicits conduction aphasia, or a combination of phonemic paraphasia (supporting a role for this subpart in phonologic processing) and repetition disorders, but without semantic paraphasia. Interestingly, the posterior cortical origin of this tract corresponds to the VWFA, a functional hub involved both in phonologic processing dedicated to visual material and in semantics. The superficial portion of the dorsal route is underlain by the lateral part of the SLF (also called the SLF III), stimulation of which generates anarthria (articulatory disorders). This lateral operculo-opercular component of the SLF plays a pivotal role in articulation, by connecting the junction between the posterior part of the superior temporal gyrus (which receives feedback information from somatosensory and auditory areas) and the supramarginal gyrus, with the vPMC (which receives afferents bringing phonologic–phonetic information to be translated into articulatory motor programs). During word repetition, this loop also enables the conversion of auditory input, which is processed in the verbal working memory system, into phonologic and articulatory representations within the vPMC. This observation is in line with recent data obtained from cortical IESM (probabilistic atlas based on 771 stimulation sites), which demonstrated that the Broca area is not the speech output region, thereby challenging the classic theories on language.
The ventral pathway is divided into a direct tract (the IFOF) and an indirect pathway formed by the anterior ILF and the uncinate fascicle (UF). These fascicles relay information to one another in the temporal pole. The IFOF connects the posterior occipital lobe and the fusiform gyrus to anterior corticofrontal areas, including the inferior frontal gyrus and dorsolateral prefrontal cortex. , IESM of this white matter connectivity revealed a major role in verbal semantics because its stimulation reproducibly elicited semantic paraphasias. Nonetheless, recent studies have shown that this bundle is also implicated in nonverbal semantic cognition in the left as well as the right hemisphere, demonstrating that the nonverbal semantic system is distributed bilaterally. The indirect ventral semantic pathway has a relay at the level of the temporal pole, which represents a semantic “hub” that enables plurimodal integration of the multiple signals emanating from the unimodal systems. This indirect ventral stream includes the anterior part of the ILF, which connects the fusiform gyrus with the temporal pole, and information is then relayed by the UF, which links the temporal pole with the pars orbitalis of the inferior frontal gyrus. IESM of this indirect pathway does not cause semantic paraphasia but may nonetheless elicit nonverbal semantic disturbances. This ventral semantic pathway seems to contribute to the repetition of real words or pseudowords and may also participate in proper name retrieval. The difference between UF and IFOF stimulation during picture naming could reflect their distinct frontal terminations. Because picture naming requires minimal levels of semantic control, this might explain the negligible effects of UF stimulation on this process. Naming at a more specific level (including naming of people) might be somewhat more executively demanding; thus UF stimulation does have an effect. Of note, the middle longitudinal fascicle, which connects the angular gyrus with the superior temporal gyrus up to the temporal pole, may also be part of the ventral semantic route. Yet subcortical IESM of this fascicle failed to generate any naming disturbances, so its exact functional role in the language network is still unclear.
Beyond picture naming, IESM showed that syntactic processing was underpinned by delocalized cortical regions (left inferior frontal gyrus and posterior middle temporal gyrus) connected by a subpart of the left SLF. Axonal IESM of the SLF can induce specific problems involving grammatical gender, thus disrupting one subfunction without interfering with the others. Indeed, this subnetwork interacts with but is independent of the subnetwork involved in naming, as demonstrated by double dissociation between disruption of syntactic and naming processes during IESM. These findings support parallel rather than serial theory, calling into question the concept of the “lemma”—an abstract lexical representation of a word before its phonologic properties are assigned.
You're Reading a Preview
Become a Clinical Tree membership for Full access and enjoy Unlimited articles
If you are a member. Log in here