Brain mapping in tumors: Intraoperative or extraoperative?


  • Hugues Duffau

    Corresponding author
    1. Department of Neurosurgery, Gui de Chauliac Hospital, Montpellier, France
    2. Institute of Neuroscience of Montpellier, INSERM U1051, Team “Plasticity of Central Nervous System, Human Stem Cells and Glial Tumors,”, Saint Eloi Hospital, Montpellier, France
    • Address correspondence to Hugues Duffau, Department of Neurosurgery, Hôpital Gui de Chauliac, CHU Montpellier, 80 Avenue Augustin Fliche, 34295 Montpellier, France. E-mail:

    Search for more papers by this author


In nontumoral epilepsy surgery, the main goal for all preoperative investigation is to first determine the epileptogenic zone, and then to analyze its relation to eloquent cortex, in order to control seizures while avoiding adverse postoperative neurologic outcome. To this end, in addition to neuropsychological assessment, functional neuroimaging and scalp electroencephalography, extraoperative recording, and electrical mapping, especially using subdural strip- or grid-electrodes, has been reported extensively. Nonetheless, in tumoral epilepsy surgery, the rationale is different. Indeed, the first aim is rather to maximize the extent of tumor resection while minimizing postsurgical morbidity, in order to increase the median survival as well as to preserve quality of life. As a consequence, as frequently seen in infiltrating tumors such as gliomas, where these lesions not only grow but also migrate along white matter tracts, the resection should be performed according to functional boundaries both at cortical and subcortical levels. With this in mind, extraoperative mapping by strips/grids is often not sufficient in tumoral surgery, since in essence, it allows study of the cortex but cannot map subcortical pathways. Therefore, intraoperative electrostimulation mapping, especially in awake patients, is more appropriate in tumor surgery, because this technique allows real-time detection of areas crucial for cerebral functions—eloquent cortex and fibers—throughout the resection. In summary, rather than choosing one or the other of different mapping techniques, methodology should be adapted to each pathology, that is, extraoperative mapping in nontumoral epilepsy surgery and intraoperative mapping in tumoral surgery.

Rationale in Epilepsy Surgery

In nontumoral epilepsy surgery, the main goal for all preoperative investigations is to accurately define the epileptogenic zone (EZ), in order to maximize postsurgical seizure relief. In addition to neurocognitive examination and functional neuroimaging (positron emission tomography [PET], single-photon emission computed tomography [SPECT], functional MRI [fMRI], magnetoencephalography [MEG], which are beyond the scope of this article), electrophysiologic techniques such as extracranial electroencephalography (EEG) and video-EEG are helpful to determine the extent and distribution of the EZ (Lüders et al., 1993). In complex cases, for example, when noninvasive studies do not sufficiently localize the EZ, if their results are discordant, or if the suspected seizure onset is within or near eloquent cortex, invasive intracranial monitoring can be considered. To this end, placement of subdural strip-electrodes or grid-electrodes, even depth electrodes, is usually performed, in particular in patients with suspected bitemporal epilepsy (Wyllie & Awad, 1993).

Beyond the delineation of the EZ, which is to be removed to control seizures, strip/grid studies also allow bedside extraoperative functional mapping by electrostimulation (Winkler, 2011). Using this method, the patient is comfortable in his or her room for the performance of these tasks: This point is particularly important for children. Moreover, recent advances in the interpretation of the electrophysiologic signal, such as electrocorticographic spectral analysis evaluating the event-related synchronization in specific bands of frequency, have allowed a better understanding of the organization of the functional cortex, and a better study of connectivity, in particular via the recording of “cortico-cortical evoked potentials” (Matsumoto et al., 2004). However, extraoperative electrophysiologic mapping has several disadvantages. The usual grids with 1 cm-spaced electrodes have limited accuracy. It is necessary to perform two surgical procedures, one to implant grids, and another to remove the EZ. In addition, there is a significant risk of infectious complications due to the presence of subdural grids in situ for several days (Onal et al., 2003). Above all, although this method is appropriate for the detection of seizure foci through (video)-EEG monitoring, insofar as the identification of eloquent areas is concerned, it is worth noting that only cortex can be mapped; it provides no information about axonal connectivity and it is therefore not possible to investigate subcortical tracts. Therefore, this technique is not well adapted to neurooncology, because gliomas migrate along white matter bundles (Mandonnet et al., 2006).

Rationale in Tumoral Epilepsy Surgery: A Different Challenge

In contrast to nontumoral epilepsy surgery, the first aim in tumoral epilepsy surgery is to maximize the extent of tumor removal in order to increase median survival. This is particularly true in gliomas (primary brain tumors, the most frequent neoplasms of the central nervous system), in which surgical resection significantly impacts on the natural history of the disease—both in low grade and high grade gliomas (De Witt Hamer et al., 2012; Capelle et al., 2013). However, these lesions actually do not represent a tumor mass but rather a diffuse disease of the brain: Indeed, it has been clearly demonstrated by multiple biopsies that they are not only growing but they are also migrating along subcortical fascicles (Mandonnet et al., 2006). This is the reason that, with the goal to optimize the “onco-functional” balance, that is, to maximize extent of resection while preserving quality of life, the resection should be performed according to functional boundaries both at cortical and subcortical levels (Duffau, 2012). The principle is to remove the part of the brain invaded by tumor on the condition that this region is not crucial for cerebral functions—namely, until eloquent structures have been encountered. Consequently, due to major interindividual anatomofunctional variability, especially in low grade gliomas which regularly induce mechanisms of brain plasticity, an accurate and reliable detection of critical cortical areas and white matter bundles is mandatory (Duffau, 2005).

To this end, intraoperative electrostimulation mapping is currently the best adapted technique in neurooncology. Indeed, a meta-analysis based on 20 years of glioma surgery (which investigated >8,000 cases) has recently demonstrated that the use of intrasurgical electrical mapping significantly decreased the rate of permanent neurologic deficit while increasing extent of resection and increasing surgery in so-called eloquent regions (De Witt Hamer et al., 2012). For example, excision of gliomas involving areas classically considered as “inoperable” (such as Broca's area, Wernicke's area, central area, and insula) can be achieved without generating significant postoperative functional worsening (Desmurget et al., 2007). This is made possible due to dynamic redistribution of neural networks within a “hodotopical” framework, challenging the traditional localizationist view of brain processing. Indeed, in pathology, according to this new concept, a topologic mechanism (from the Greek topos = place) refers to dysfunction of the cortex (deficit, hyperfunction of a combination of the two), whereas a hodologic mechanism (from the Greek hodos = road or path) refers to dysfunction related to connecting pathways (disconnection, hyperconnection, or a combination of the two). In other words, it is mandatory to take into account the complex functioning of a large-scale distributed cortico-subcortical network to understand both its physiology as well as the functional consequences of a lesion of this circuit—with possible different deficits depending on the location and the extent of the damage (e.g., purely cortical, or purely subcortical, or both) (De Benedictis & Duffau, 2011).

Intraoperative Electrical Mapping in Glioma Surgery: A Window to “Hodotopy”

In this setting, understanding the individual organization of complex circuits, that is, both cortical epicenters and subcortical connectivity for each patient, is essential to optimize the risk–benefit ratio of glioma surgery (Duffau, 2011). This is a crucial issue, based on the fact that axonal connectivity represents a limitation of brain plastic potential. In other words, damage of the white matter fibers has a high risk of inducing permanent neurologic deficits, whereas a cortical lesion is more likely to be functionally compensated thanks to recruitment of perilesional and/or remote cerebral regions (Ius et al., 2011). Of interest, intraoperative direct electrostimulation is actually the sole technique allowing functional mapping of white matter tracts (in addition to cortical hubs), converse to extraoperative mapping by strip/grid, which in essence is able to investigate only cortex. Furthermore, whereas depth electrodes could arguably provide some functional information concerning subcortical fascicles, a comparison of this technique with intraoperative mapping in awake patient has shown that stereoencephalographic language stimulation mapping was unreliable (Gil Robles et al., 2008). In addition, in voluminous tumors, the implantation of numerous deep electrodes all around the lesion before resective surgery in order to investigate subcortical connectivity underlying multiple networks surrounding the glioma, seems difficult and impractical.

Consequently, despite the rare report that advocates the use of extraoperative mapping before tumor surgery, intraoperative direct electrostimulation is now considered the “gold standard” in oncologic neurosurgery (De Witt Hamer et al., 2012). Beyond the traditional monitoring of motor tracts under general anesthesia, intrasurgical mapping in awake patients also enables identification of pathways subserving complex functions such as somatosensory function, language (phonology, syntax, semantics, pragmatic, multilingualism), spatial cognition, calculation, judgment, executive functions, and even emotional aspects. The original concept of considering glioma surgery as “brain networks surgery” has led not only to a dramatic decrease of permanent neurologic impairment (<2% in recent series using intraoperative cortico-subcortical mapping) but also to neurocognitive improvement following resection of gliomas that are known to disturb wide cerebral circuitry. For example, preoperative and postoperative neuropsychological assessments have demonstrated improvement of higher-order functions such as working memory 3 months following surgery (Duffau, 2012).

Epilepsy Outcomes in Glioma Surgery

Seizures in tumoral epilepsy are only a symptom (not a disease, converse to many nontumoral epilepsies), which in most cases lead to the diagnosis of the tumor. Seizures are particularly frequent in slow-growing tumors such as low-grade gliomas. Furthermore, in this specific entity, intractable epilepsy is regularly encountered, with a negative impact on the quality of life even if the low grade glioma does not (yet) induce neurologic symptoms by itself. Thus, another goal of tumoral surgery, beyond the oncologic considerations and the preservation of brain functions, is to control medically resistant epilepsy. This is particularly true for tumors involving specific locations such as the mesiotemporal structures, the insular lobe, or the central area (Smits & Duffau, 2011).

Tumor resection usually allows seizure relief in up to 80% of cases. It has been demonstrated that a greater extent of resection is related to better control of epilepsy (Rudà et al., 2012). Therefore, the question concerning the additional use of extraoperative intracranial monitoring (in combination with intraoperative mapping in a second stage) in patients with intractable epilepsy has to be raised. The goal would be to identify the EZ, possibly outside the tumor zone, and to tailor the resection according to presurgical recordings, that is, by removing the EZ in addition to the tumor resection, in order to improve epilepsy control. Such a strategy could be considered in benign and circumscribed lesions such as dysplasias or glioneuronal tumors, in particular when they are located in the lateral temporal lobe or in the paralimbic system without involvement of the mesiotemporal structures. Presurgical neurophysiologic (interictal and ictal) EEG and video-EEG monitoring may help define the extent of the EZ and enhance surgical planning to optimize seizure control as well as optimize functional outcome (memory)—that is, to include additional hippocampectomy with the tumor resection (Giulioni, 2013).

However, in cases of diffuse neoplasms such as low-grade gliomas, one should keep in mind that a more extensive excision significantly increases overall survival (Capelle et al., 2013). As mentioned, this is why performing resections according to functional boundaries and not to oncologic boundaries (which in essence do not exist in diffuse glioma) has been recently recommended (Duffau, 2012). This concept led to “supra-total” resection, namely, to remove the brain invaded by tumoral cells even beyond the signal abnormality on MRI (until crucial pathways have been detected)—with a significant impact on malignant transformation (Duffau, 2013). Consequently with the principle of “maximal resection” based on intraoperative cortico-subcortical mapping, if preoperative (invasive) EEG monitoring demonstrates the putative EZ in proximity to the glioma as is usually the case, this region would be resected anyway, even without the presurgical invasive electrode information. An exception that would result in no resection of the EZ would be in the case of its location in critical cortex. For example, in patients with paralimbic low-grade gliomas, especially generating intractable seizures, we propose to remove mesiotemporal structures in all cases even if they are not invaded by the tumor, whatever the results of a possible preoperative or intraoperative neurophysiologic assessment (Ghareeb & Duffau, 2012) (Fig. 1). Indeed, it has already been shown that the impact on epilepsy relief with this approach is significantly higher in comparison to a control group of patients with paralimbic low-grade gliomas in whom the hippocampus was not removed. Extraoperative strip/grid monitoring or mapping is usually not necessary in tumoral epilepsy surgery for low-grade gliomas (Brogna et al., 2008).

Figure 1.

(Upper) Preoperative coronal fluid-attenuated inversion recovery (FLAIR)–weighted magnetic resonance imaging (MRI) in a left-handed patient with intractable epilepsy (about 10 partial seizures a day despite four antiepileptic drugs, preventing the patient from working and driving), which shows a right paralimbic diffuse low-grade glioma not involving mesiotemporal structures. (Middle) Intraoperative photograph after surgical resection according to cortical and subcortical functional boundaries in awake patient; mapping of motor, language (articulatory), and visuospatial functions (number tags) has been carried out throughout the resection. A, anterior; P, posterior. (Lower) Postoperative coronal T2-weighted MRI, demonstrating a subtotal removal of the tumor combined with resection of the mesiotemporal structures (not invaded by the glioma). The patient had no neurologic deficit, no adjuvant treatment (no chemotherapy, no radiotherapy). No seizures occurred within the 3 years of follow-up after surgery (antiepileptic drugs were progressively decreased and then stopped); the patient was able to work and drive again.

In summary, pragmatically adapting and strategizing surgical planning and management to each disease situation is the best approach, that is, intraoperative mapping in invasive cancerous tumors such as gliomas, extraoperative mapping in nontumoral epilepsy surgery, and sometimes both in pathologies such as cortical dysplasia.


The author reports no conflict of interest. The author confirms that he has read the Journal's position on issues involved in ethical publication and affirms that this report is consistent with those guidelines.