Preoperative Assessment by Navigated Transcranial Magnetic Stimulation

General

After the introduction of transcranial magnetic stimulation (TMS) into clinical neurology in 1985, several studies quickly demonstrated the potential of the modality to assign individual muscles to their cortical representation. TMS is based on the law of induction, which states that a time-varying magnetic field induces an electrical current in an electrically conductive medium. In TMS, the magnetic field is induced by a strong, rapidly rising current in a coil. The resulting magnetic field decreases exponentially with increasing distance from the coil. The current intensity induced in the body depends directly on the conductivity of the medium. As a result, the primary magnetic field reaches the cortex almost completely undamped, because skin, hair, and bones have a high specific resistance—that is, poor electrical conductivity. This also explains the painlessness of the method, as the TMS pulse has only a very small effect on the tissue between coil and cortex.

Two technical innovations were necessary for useful application within neurosurgery. First, the development of so-called double-ring or figure-eight coils was a prerequisite to enable meaningful mapping. With this type of magnetic coil, a conically configured magnetic field is created at the intersection of two round coils, enabling focused stimulation. Focused stimulation is especially possible when working with low stimulation intensities, so that only the tip of the conical magnetic field stimulates the cortex at suprathreshold intensity. The spatial resolution is in the millimeter range.

Furthermore, the combination of the TMS stimulation and the spatial image information was necessary in order to make the mapping of the cortical representation of the limb muscles visible—that is, to assign it to the respective anatomic localization. This could be guaranteed for the first time at the end of the 1990s. The first study, which focused on the accuracy of navigated transcranial magnetic stimulation (nTMS) compared with the “gold standard” of direct intraoperative electrical cortex stimulation in a larger patient population, was published in 2009. Further improvement of the localization systems and the modeling of the TMS effect on the cortex level (“e-field navigated TMS”; Fig. 8.1 ) enabled a continuous optimization of the accuracy of frameless nTMS. ^, Modern systems enable software-controlled optimization of the coil tilting in order to optimally align the induced electric field with respect to the local cortical anatomy and provide online feedback on the strength (V/m) of the induced electric field at cortex level. By this means, individual differences in the local anatomy and varying distance between coil and cortex can be taken into account.

Motor Function

Mapping of cortical motor representations with an nTMS system achieves highest focality when applying just suprathreshold stimulation intensities. To implement this concept in practice, it is necessary to determine the individual resting motor threshold (RMT) before each measurement, because the RMT varies interindividually and also intraindividually, depending on various internal (e.g., level of alertness) and external (e.g., electrode montage) factors. The subsequent mapping of the relevant peritumoral cortex area is then performed usually at 105% to 110% of the RMT. Owing to the somatotopy of the gyrus, 10% to 20% higher stimulation intensities are required for mapping muscles of the lower extremity than for the small hand muscles. During the mapping, it must also be ensured that the induced current flow is always perpendicular to the nearest sulcus. This is due to the fact that the axons of the pyramidal cells are arranged perpendicular to the gyral surface, and axons are depolarized most easily with a parallel current flow. ^, Another factor that significantly influences the reliability of the investigation is the quality of the electromyography (EMG) signal, and it is important to ensure a sufficiently good EMG quality during the examination, wherein the resting activity is always below the threshold for positive motor evoked potential (MEP) responses (usually 50 mV).

With high stimulation intensities, an electrical current can be induced at a depth of several centimeters, but the focus is lost, so targeted stimulation in the sense of mapping subcortical structures is not possible. The presurgical TMS examination is therefore limited to the cortex. The hodotopic concept of brain function—that is, its organization in dynamic networks—emphasizes the role of the long association fibers in maintaining functional integrity. ^, There is increasing evidence that a loss of connectivity between two cortical nodes has even more serious consequences (i.e., worse prospects for functional recovery) than cortical damage. Accordingly, maintaining subcortical connectivity is at least as important as maintaining cortical integrity to preserve neurological function. Diffusion tensor imaging (DTI) fiber tracking (FT) has found rapid acceptance in recent years in the neurosciences and is increasingly being used in presurgical diagnostics. However, the result of DTI imaging depends largely on the experience of the examiner and the software used. In particular, the selected analysis threshold and the selection of starting points for the DTI algorithm influence the configuration of the resulting fiber networks. By identifying those motor areas at cortical level that are essential for motor functioning with high spatial accuracy, nTMS provides starting points for the DTI algorithm, which will display only functionally irreplaceable and therefore surgically relevant tracts. However, when evaluating DTI images, one needs to keep in mind that DTI is anatomic imaging and not functional imaging. Its interpretation must be based on the knowledge that it has not yet been fully clinically validated and that diffusion is susceptible to confounding tumor effects.

Language

Starting with the first language model in the late 19th century, it has become evident that the capacity for language is maintained in complex connections between language-related areas, encompassing highly specialized and less specific areas for language processing in both hemispheres. Knowledge about patients’ individual language networks predominantly stems from intraoperative stimulation mapping. These studies during awake surgery have revealed strong interindividual differences in the cortical sites producing language disturbances. ^, The present models include a frontoparietal “dorsal stream” involved in mapping sound onto articulation-based representations and a “ventral stream” in the temporal lobes, which maps sound onto meaning. In light of the recent paradigm shift from the traditional localizationist view of language function located in specific cortical regions toward a view of parallel, highly dynamic, corticocortical and cortico-subcortical networks supporting speech and language function, TMS, as the only noninvasive methodology allowing for electrical stimulation mapping analogous to direct cortical stimulation (DCS), has received increasing interest as a tool for presurgical language mapping.

From the very first reports on TMS language mapping to recent studies, stimulation frequencies between 4 and 10 Hz have been found most effective. ^, The induced disturbance of language processing can result in a variety of behavioral changes ranging from discrete prolongation of response delays to clear anomias. The clinical experience from awake surgery suggests that object naming is the most effective experimental setup to map the language-related cortical areas, because it is robust, explores different language submodalities, and can be easily introduced into a short task design.

rTMS trains usually have 4-10 Hz, start immediately with the object presentation or delayed for up to 300 ms. The whole pulse train is then applied for 1 to 2 seconds. In modern TMS devices, the electric field is calculated at the stimulation site. This allows avoiding too low stimulation intensity, which should be above 50 V/m. Depending on the patient’s abilities and the region of interest, around 200-300 different stimulation sites are chosen per hemisphere. The baseline session in which the patients needs to name all presented objects properly leads to discarding of all misnamed objects. During stimulation, these recordings of properly named objects can then be directly compared with the patients’ answers during stimulation which clearly reveals the stimulation effects. ^, Because of the distributed and highly individual composition of the language network, the indication for preoperative TMS language mapping has a wide range, going beyond tumors in classical left-hemispheric language areas. ^, Preservation of language function also depends on preventing disconnection of the cortical nodes identified by TMS mapping. Therefore combination of the cortical TMS language mapping with white matter tractography is essential for presurgical planning. Here, different approaches implementing either anatomic and/or functional seed areas for DTI tractography have been proposed. The TMS and tractography data enable one to counsel the patient with respect to the difficult risk-benefit balancing of surgery based on these objective measurements. Because of the easy-to-grasp methodology and transparent nature of TMS mapping, the patient will be able to take part in the shared decision-making process well informed and confident. In addition to providing a map of individual language function, the TMS experiment also prepares the patient for potential awake surgery, because the procedure of stimulus presentation, stimulation, and experience of language impairment is analogous to the events during intraoperative language mapping. Despite all efforts, the occurrence of new language deficits after preparatory TMS and DTI and intraoperative awake language mapping cannot be completely prevented when the goal is also to maximize resection for longer survival.

Safety

TMS mapping of presumed eloquent brain areas has been shown over the years to be extremely safe with minimal side effects. ^, The guideline recommendations for upper limits for number, frequency, intensity, and duration of stimulation refer primarily to the risk of inducing an epileptic seizure. The incidence of TMS-induced clinical seizures for all types of TMS patterns is low with, with rates of 0.01% to 0.1% being reported in the literature. For single-pulse TMS, only anecdotal reports are available for induced epileptic seizures in patients with intracranial disease. Other possible side effects are syncope or pain, the incidences of which are also very low in the literature.

While nTMS was introduced in some departments over 10 years ago, it is still new for many. Since it is not affected by oxygenation changes and proofed a high navigation accuracy as well, it was repeatedly shown that the correlation of nTMS to DCS is superior to other noninvasive techniques, such as FMRI and MEG. nTMS therefore offers the possibility of a standard workflow for the preoperative workup before surgery for eloquent lesions. Despite its active stimulation, no severy adverse events were reported for nTMS in the past 10 years. This mentioned study showed in 733 patients of which half suffered from symptomatic seizures and were operated on in 3 large neuro-oncological centers, that no patients showed any severe side effect except pain during or after stimulation. The very recent 2021 guidelines on brain stimulation analyzed the literature of recent years comprehensively and came to the same conclusion. TMS is therefore widely applicable, including in the pediatric population and in preparation for epilepsy surgery. ^, These have also been put in context of newer data in the most recent guidelines. With respect to the importance and benefit of such data, the potentially minor risks of nTM should be put in context.