Intraoperative Testing During the Mapping of the Language Cortex

Intracranial lesions, particularly in the language-eloquent areas of the brain, can affect one's speaking ability. Despite advances in surgery, the excision of these lesions can be challenging. Intraoperative neurophysiological monitoring (IONM) during awake craniotomies can help identify language-eloquent areas and minimize postoperative impairments. Preoperative language testing is performed to establish a baseline before intraoperative language testing. This involves subjecting patients to predetermined tasks in the operating room to evaluate their phonological, semantic, and syntactic capabilities. The current state and future directions of intraoperative language testing procedures are discussed in this paper. The most common intraoperative tasks are counting and picture naming. However, some experts recommend utilizing more nuanced tasks that involve regions affected by infrequently occurring tumor patterns. Low-frequency bipolar Penfield stimulation is optimal for language mapping. Exception cases are discussed where awake craniotomies are not feasible. When dealing with multilingual patients, the patient's age of learning and skill level can be accounted for in terms of making informed task choices and mapping techniques to avoid any damage to language areas.


Introduction
Language-eloquent tissue comprises a wide range of regions of interest, which are Broca's and Wernicke's areas. Broca's area is anatomically located in the inferior frontal gyrus (IFG) and is responsible for the motor production aspect of speech; pathological insults to this region can cause patients to be unable to generate speech but understand what is being communicated to them. On the other hand, Wernicke's area is anatomically located near the auditory cortex in the superior temporal gyrus and directly impacts language comprehension. Patients with this damaged region can still speak, although it is generally incomprehensible [1]. According to Gonen et al.'s study, Broca's region has three different neural pathways encoding phonology, semantics, and syntax relevant for testing purposes [2]. Verbal working memory is involved in the dorsal stream, while the ventral lexical-semantic network involves converting phonological sounds into meaning [3]. The aforementioned response changes are a sign of language ataxia due to damage to language areas, which includes tumors, lesions, arteriovenous malformations, aneurysms, and epilepsy foci.
The Mayo Clinic defines a brain lesion as "an abnormality seen on a brain imaging test," and they are different from normal tissue when seen on functional magnetic resonance imaging (fMRI) [4]. These lesions can result from trauma or even be acquired during surgery, such as fibrosis or scarring. Conversely, tumors are more organic and varied compared to lesions, which can be local or metastatic. Brain tumors can be classified into primary and secondary categories: primary tumors are most commonly intracranial in origin, while secondary tumors originate extracranially (breast, lung, gastrointestinal tract, and melanoma). There are two categories of tumors: benign and malignant. Malignant tumors are cancerous tumors (or tumors that invade other sites) and grow uncontrollably while potentially spreading into unrelated body areas. Malignant tumors tend to require careful treatment because of the way that they spread through the bloodstream and are more likely than benign tumors to reoccur. Glioma is a common malignant tumor. Several types include astrocytoma, oligodendroglioma, and glioblastoma, which are aggressive tumors mentioned often in the literature. There is also medulloblastoma, which is the most common malignant brain tumor in children [5]. Benign tumors stay in their primary location, so the treatment is more aggressive because of their distinct borders, and once removed, they are unlikely to return [6]. The most common benign tumor is meningioma which, while generally benign, can potentially become cancerous. While they are seen in all adults, they are slightly more common in females than in males [5].
Surgical resections of these lesions can be risky, especially due to the eloquent tissue being placed at risk of damage. Intraoperative neurophysiological monitoring (IONM) decreases said risks during awake craniotomies, specifically protecting the nervous system by providing real-time information. The introduction of IONM in surgery has positively impacted the recovery rates of patients in brain and spinal surgeries; it is said that around 53.3% of patients who had IONM intervention made complete or partial recovery of postoperative deficits by the time of discharge compared to patients requiring reoperation before discharge or experiencing little to no recovery by the time of discharge [7]. Direct cortical stimulation (DCS) is considered the gold standard for cortical mapping during awake craniotomy for delineating the boundaries of malignant tumors and accounting for individual differences between different brains. The previously collected evidence shows that DCS can accurately map the brain areas in question and identify cortical dysfunction [2]. During monitoring, the patient is asked to perform various language-based tasks that are verbal, visual, and auditory while the surgeon stimulates exposed cortical surfaces. These responses are categorized as positive and negative responses. Positive mapping signifies that the response was induced; when it cannot be induced, it is known as negative mapping [1]. Based on the responses observed, surgeons are in a more optimal position to make decisions about whether resections should be gross total or subtotal. The present work is centered around the language testing that takes place intraoperatively, aiming to summarize the recent work and future directions surrounding the process.

Patient Selection Based on Clinical Presentation
This research will concentrate on awake craniotomies and language cortical mapping during malignant and benign tumor resections. The inclusion criteria were based on the patient's brain tumor affecting the brain areas involved in language formation/processing [8]. Meanwhile, individuals with recurrent seizures and any form of a pacemaker or deep brain stimulation device are often excluded [8]. Males and females are, by all indications, roughly equally represented in the patients chosen.

Imaging/Investigation: Magnetic Resonance Imaging (MRI) and Diffusion Tensor Imaging (DTI)
Preoperatively, several preparations must be taken to ensure that direct cortical stimulation (DCS) and electrocorticography (ECoG) aid the surgical procedure. The first and most important step in a preoperative setting is establishing baselines for brain activity and performance by answering language-related questions. Preoperative magnetic resonance imaging (MRI), magnetoencephalography (MEG), and diffusion tensor imaging (DTI) are utilized to localize lesions. The MEG provides information regarding long-distance connectivity in language-relevant brain areas [9]. Finally, DTI visualizes the subcortical language-related axonal connections the tumor may have shifted [1]. Raffa and collaborators have shown that navigated transcranial magnetic stimulation (nTMS) can be used to pinpoint brain functional areas relevant to language skills akin to fMRI [8]. However, this practice is not considered standard yet (Figure 1). Functional MRI (fMRI) is utilized to localize brain functional regions related to language production and comprehension [9]. To identify the dominant language hemisphere, functional MRI provides an accurate, noninvasive alternative to the Wada test [1]. It is also crucial to note that these imaging approaches are decisively not intended to be a replacement for DCS. Giussani et al. (2009) showed that fMRI helps preoperatively. Still, studies comparing fMRI with DCS have yielded contradictory results due to inconsistencies in validation methods, differences in suitability for specific types of lesions, the feasibility of tasks that can be administered, and discrepancies in intrinsic features of both methods [10]. The anatomic location of eloquent regions may be altered by long-standing brain masses leading to differences between data from fMRI and awake mapping. Finally, all these imaging modalities require intraoperative language tasks to compare performance during the procedure. These tasks will be expanded upon in the operation section.

Electrodes
Four different types of electrodes are commonly used during the linguistic cortical mapping process. These four, however, can be divided into three categories: monopolar electrodes (direct cortical motor mapping) (Figure 2), bipolar electrodes (language mapping) ( Figure 3), and grid electrodes (electrocorticography) (Figure 4). Bipolar electrodes can be either subdermal needles for recordings [11] or ball tip electrodes for The patient is under sedation, and the head is secured with a three-pin Mayfield holder to the operating table.

Anesthesia
Awake craniotomy uses two types of anesthesia: asleep-awake-asleep or asleep-awake techniques. Asleepawake-asleep is a technique that sedates the patient during the dura opening, wakes the patient for language cortical mapping, and then sedates the patient again after that process is complete. Hyperventilating the patient before dural opening helps control brain swelling in asleep-awake-asleep and asleep-awake [13]. Before dural opening, propofol is combined with an analgesic, such as fentanyl or dexmedetomidine, and administered by total intravenous anesthesia (TIVA) [11]. To stop sedation during the waking stage, the use of propofol is discontinued, while analgesics are continued [11]. Lastly, direct laryngoscopy is used to reintubate the patient to assist in a quick recovery postoperatively. An alternative may be blind nasal intubation, but this is avoided unless it is an emergency. Minors older than 10 years of age may require intermittent intravenous sedation [9].
Conscious sedation is significantly less frequently used and discussed than asleep-awake-asleep since it is more uncomfortable for the patient and has underlying complications that will be listed in the section covering the postoperative phase. Conscious sedation necessitates the administration of droperidol, propofol, or short-acting narcotics for dural opening, followed by analgesic administration until both cortical mapping and resection are complete [13]. Finally, the previously mentioned anesthetics are supplied again following resection until the procedure is completed [1].
Note that inhalants are not mentioned. The main reason is that while inhalants can be used to replace propofol during the opening of the dura, they disrupt motor evoked potential (MEP) thresholds, which can lead to postoperative impairments in the patient. As a result, their use is avoided wherever possible [14].

Direct Cortical Stimulation and Intraoperative Language Tasks
There are two common approaches to DCS/direct cortical electrical stimulation (DCES): the Penfield method and the Taniguchi method. The Penfield method is a low-frequency (50 Hz) stimulation method done via bipolar ball tip/subdermal needle electrodes in which stimulation pulse duration is between 300 and 500 µs. Stimulation is applied between three and five seconds allowing for an interstimulation interval of 5-10 seconds [1, 14].
The Taniguchi method, in contrast, uses monopolar ball tip electrodes to deliver high-frequency (250-500 Hz) stimulation pulses with an approximate 500 µs stimulation pulse duration. With an average pulse duration of 500 µs, stimulation pulses are administered four to five times [1]. Starting at 2.0 mA, the intensity of these stimulations gradually increases until a positive response is obtained or until discharges are detected [1] (Tables 1-2).

Specification
Penfield Taniguchi   The same array of language exercises that were completed during the preoperative stage is repeated concurrently during DCS. The discussion will center on these exercises because they are meant to assess language production and comprehension. The most frequent tasks involve counting (such as from one to 50), picture naming, reading, writing, and language syntax [9] ( Figure 6) ( Tables 3-7).        Picture for naming various objects (tree, volleyball, car, and rabbit).

Protocol for the Minors
Patients younger than 10 years of age conventionally warrant a two-stage protocol where the first stage involves subdural grid placement with language testing for the necessary highlighting of cortical areas, and the second stage entails lesion resection on a different date [9]. Minors older than 10 years of age may necessitate additional coaching throughout the process [9].

Positive/Negative Cortical Mapping, After Discharges (ADs), and Seizures
The goal of DCS approaches is to identify areas of the cortex that may result in language disruption upon stimulation (such as speech or comprehension deficits) and subsequently classify such regions as "positive" parts of the cortex while classifying unaffected regions as "negative." These areas are taped until the mapping of the language cortical regions is complete [8].
Finally, ECoG is set up with grid electrodes over the cortex to detect after discharges (ADs) and, more critically, to alert the surgical team to seizure activity. Applying ice saline to the relevant cortical areas prevents seizures [14].

Postoperative phase
After the operation, all that remains is allowing the patients to recover from anesthesia and test all neurological activity related to the regions surrounding the operated area. To achieve that goal, the patient is examined in multiple time frames, including right after the surgery, after 24 hours, after a week, after six months, and after a year [1].

Discussion
Direct cortical electrical stimulation allows the surgical team to carry out cortical mapping to avoid postoperative deficits manifesting within the patient's life following surgery. Specifically, inadvertently caused damage to eloquent areas within language cortical areas can be circumvented and absent, in which the patient is at risk of suffering from multiple types of aphasia along with issues using language processes in tandem with other brain functions such as sensory input and motor commands. Although preoperative language testing systems alone may fulfill the need for a reference point for intraoperative language testing by characterizing the patient's baseline status [15], supplementing with the use of fMRI preoperatively or intraoperative, MRI can provide high-resolution information regarding the patient's language abilities' localization. Yet, this is not realistically feasible for every case due to the remarkable financial cost of accessing and using MRI machines. The traditional low-frequency Penfield method is utilized for DCS for language mapping due to the more appropriately prolonged stimulation duration. In contrast, the Taniguchi method's lower risk of poststimulation intraoperative seizures is more suitable for motor mapping.
Asleep-awake-asleep methods may not always be optimal for every set of circumstances, as the patient's condition in terms of age or the degree of dysfunction found may not be compatible with cortical stimulation; in these cases, asleep language mapping can be applied for regions primarily connected to Broca's area allowing the surgical team to track the integrity of motor pathways related to language use. Wada testing is also an important part of all language-related surgeries as it allows the identification of the language-dominant hemisphere for the patient. To prevent new damage incurred by the patient, ECoG is an indispensable tool as the team can be alerted to any seizure activity. Ice-cold saline can be applied immediately to avoid after discharge-related seizure activity.
While a relatively novel technique, multi-site stimulation (MSS) provides an avenue for accessing languagerelated regions inaccessible by DCS's single-site stimulation alone. Gonen et al. [2] found response errors they could induce in 11 out of 15 patients that did not present upon single-site stimulation. In contrast, notably, two patients did not exhibit any errors at all during single-site stimulation. The authors attribute three possible mechanical reasons for this phenomenon: firstly, the access of a third location structurally connected to the two stimulated. Secondly, the regions are connected through white matter tracts comprising greater networks, each dedicated to a key part of language function. Lastly, the sites themselves serve subfunctions in separate networks, which in turn co-activate in varying configurations to produce broader aspects of language.
Intraoperative language testing protocols have seen standardization to some degree where recent reports identify a hierarchy of tasks in terms of how commonly they are used, with picture naming ranked the highest due to how intuitive the task demands can be for the patient in addition to their universality, ease of administration, and wide applicability given the six categories of errors, with the trade-off that this task may leave some linguistic processing resources untapped. In contrast, language syntax is the most sparingly used of all tasks not only due to the inherently higher degree of involvement needed from the patient but also due to them targeting a specific subset of regions and pathways for mapping, namely, the pars opercularis and pars triangularis of the inferior frontal gyrus and the arcuate fasciculus [9]. Other tasks include counting, word repetition, reading, and writing, where brain regions such as the supramarginal gyrus, posterior/superior/anterior temporal gyri, inferior frontal, inferior parietal, basal temporal, and posterior middle temporal gyrus are involved (Figure 7).

FIGURE 7: Language areas of the brain.
Brain image with the consolidation of the areas involved with language.
Source: Kabir SS, Rinesmith C, Vilches C, Chakarvarty S, Jahangiri FR: Language mapping of the brain. Introduction to neurophysiology. Jahangiri FR (ed): Kendall Hunt Publishing Company, Dubuque, IA; 2023. 310 [12] Some of the more complex and rarely used tasks involve judging semantically incorrect sentences from correct counterparts, naming the odd phonological word/semantic concept missing, verb generation, nonsensical word generation, and rapid syllable generation/repetition [16]. These tasks represent specific targeting or isolation of functions from one another, such as planning and coordination of the motor elements of speech, separating phonological generation processes from those that involve navigating semantic concepts or for the generation of words that belong to a given category, allowing the surgical team a means to deal with cases involving irregular tumor profiles. A necessary distinction must be drawn between common tasks such as picture naming and counting and rarer tasks such as semantic association, where the former is easier to perform and track data from. Still, the latter has higher precision. It could be more suitable for the patient's assessment needs to be based on their imaging data for an approach tailored for maximal safe resections in irregular cases [17]. Therefore, these rarer high-complexity tasks allow the surgical team to preserve the function of areas more indirectly tied to language, preventing the loss of nuanced abilities utilized in conversation or abstract mental processes [16].
Regarding directions for the near future, several challenges must be tackled. Firstly, further delineation of "naming" for objects versus verbs or using visual versus auditory stimuli for descriptions is under investigation and yet to see consensus [9]. Secondly, inter-rater variability for preoperative, intraoperative, and postoperative language testing can still introduce noise sources in the data [18]. Thirdly, the patient's potential inability to answer questions due to elements such as anxiety, inattention, immaturity, and/or confusion remains a concern as the field eventually seeks to fine-tune methods differentiating such instances from the intended inability to answer questions due to stimulation [19]. Furthermore, as language regions overlap with regions involved in other cognitive functions, such as verbal working memory tapping into diverse regions such as the IFG along with the hippocampal complex or the angular gyrus with linguistic functions overlapping with calculation abilities, some authors argue in favor of expanding testing methods to accommodate tasks tapping into other cognitive abilities such as attention, executive function, nondominant hemispheric tasks such as emotion recognition, and those that recruit broader syntactic and semantic abilities [16], and it appears plausible that this could lead to further clarity in distinguishing between "critical" and "participating" brain regions [10]. Finally, the role of speech and language therapists is part of an ongoing discourse regarding capacities and functions that support the team or their training regimens [20]. It is highly recommended that only experienced team members should be involved in language cortical mapping procedures.
The population of multilingual speakers is one manifestation of the field's need for both standardization and customizing assessment processes for the individual patient. Optimally, the patient should present with no deficits in any of their spoken languages postoperatively, yet this presents a daunting challenge as most systems currently in place are geared toward monolingual individuals. There is no established method of accounting for the differences in properties between languages referred to as "distance" as it is necessary for language mapping's purposes [12]. Although an understanding of how the brain regulates mutual versus separate elements of languages is still at a relatively rudimentary stage and semantic meanings of words are thought to be represented in the brain independently from language, an idea that has gained some traction is that properties tied to specific languages tap into different cognitive resources for multilingual speakers. The age of acquisition and the degree of proficiency with which every language is spoken are considered useful metrics in this regard as a starting point.

Conclusions
Multimodal intraoperative cortical mapping protocols have demonstrated positive outcomes allowing for optimized resection in a cost-efficient manner. We strongly recommend that the choice of intraoperative language tasks should be based on stringent preoperative clinical language testing and imaging. The other factors that should be considered may include the patient's age, number of languages known, age of acquisition, and degree of proficiency. Language mapping must be utilized by experienced teams in brain surgeries for lesions involving the language areas to minimize the incidence of postoperative neurological deficits.

Additional Information Disclosures
Human subjects: All authors have confirmed that this study did not involve human participants or tissue. Animal subjects: All authors have confirmed that this study did not involve animal subjects or tissue.

Conflicts of interest:
In compliance with the ICMJE uniform disclosure form, all authors declare the following: Payment/services info: All authors have declared that no financial support was received from any organization for the submitted work. Financial relationships: All authors have declared that they have no financial relationships at present or within the previous three years with any organizations that might have an interest in the submitted work. Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work.