Preoperative Neurocognitive Assessment

Gliomas, especially LGG, are usually revealed by inaugural seizures in young patients who have had a normal life, with no or only a mild neurologic deficit. However, recent extensive neuropsychological examinations have demonstrated that most of patients had cognitive disturbances, especially concerning working memory and executive functions.⁷ This is the reason why a systematic preoperative neurocognitive assessment is now recommended to search the possible neuropsychological deficit not identified by a standard neurologic examination, to adapt the surgical methodology (e.g., functional mapping under local anesthesia) to the results of this assessment, to benefit from a presurgical baseline allowing a comparison with the postsurgical evaluation, and to plan specific functional rehabilitation.

It is nonetheless puzzling to note that these deficits are not more pronounced, despite the frequent location of LGG in the so-called “eloquent areas.” This can be explained by mechanisms of brain reshaping allowing functional compensation in cases of slow-growing lesions. Indeed, it was shown that cerebral remapping was possible, with a recruitment of perilesional or remote areas within the ipsilesional hemisphere and/or recruitment of contra-hemispheric homologous areas. The recent integration of these concepts into the therapeutic strategy has resulted in dramatic changes in the surgical management of LGG patients, with an increase of surgical indications in eloquent regions classically considered “inoperable”.⁶

Preoperative Functional Neuroimaging: A Necessary Baseline

In this context, advances in functional neuroimaging (FNI), namely functional MRI (fMRI), positron emission tomography and magnetoencephalography, have enabled to perform a non-invasive cortical mapping of the whole brain, and is currently a standard before resection of gliomas. FNI gives an estimation of the location of the eloquent areas (e.g., regions involved in sensorimotor, language, visual, and even higher cognitive function) in relation to the tumor, and provides information with regard to the hemispheric language lateralization. Thus, these methods may be useful for surgical indications, partly depending on the location of the tumor and its relationships with eloquent areas detected by FNI (allowing an estimation of the tumor resectability); surgical planning, namely the selection of the surgical approach and the delineation of the limits of resection; and selection of surgical technique, especially the decision to wake up the patient intraoperatively if the glioma is close to language or cognitive areas—on the basis of the laterality index on FNI in addition to the handedness of the patient provided by the neuropsychological examination.

However, it is worth noting that FNI methods are not yet reliable enough at the individual scale, despite constant improvement efforts, mainly because the results depend on biomathematical models used for reconstruction. Regarding fMRI, correlations with intraoperative electrophysiology demonstrated that the sensitivity of fMRI was currently only around 71% for movement, and from 59% to 100% for language (specificity from 0% to 97%).⁸ Such discrepancies can be explained by a neurovascular decoupling in cases of glioma (blood-oxygen–level dependence response in the vicinity of gliomas does not reflect the neuronal signal as accurately as it does in healthy tissue), by inadequate tasks (not adapted to the location of the glioma and/or to the neurologic status of the patient), or to methodological problems (e.g., selection of the threshold). As a consequence, there is a risk of false negative and then to operate a patient without intraoperative mapping, although the glioma is actually located in crucial areas for the function, but not detected by preoperative FNI. Moreover, an erroneous interpretation of brain reshaping (“pseudoreorganization”) can be made. Finally, these methods are not able to differentiate the structures essential for the function, which should be surgically preserved, from those which can be functionally compensated and so potentially resected without permanent deficit. Thus, there is a double risk: first, failure to select a patient for surgery while the tumor was operable, and second, to stop the resection prematurely with a lower impact on the natural history of the glioma (or both).

The recent development of the diffusion tensor imaging (DTI) has also allowed the identification of the main bundles and their location in relation to the tumor. However, this new method needs to be validated before it can be used routinely for surgical planning, especially due to the fact that results of DTI, as FNI, strongly depend on the biomathematical models used for the fiber tracking. Indeed, comparison of distinct fiber-tracking software tools found different results, showing that neurosurgeons have to be cautious about applying tractography results intraoperatively, especially when dealing with an abnormal or distorted fiber tract anatomy. Furthermore, correlations between DTI and intrasurgical subcortical stimulation demonstrated that, despite good correspondence, DTI is not yet optimal for mapping language tracts in patients. Negative tractography does not rule out persistence of a fiber tract, especially when invaded by a glioma.⁹ Moreover, DTI enables study of the sole anatomy of the subcortical pathways, but not their function.

With the aim of overcoming these pitfalls, one can currently consider performing longitudinal studies based on pre-, intra-, and post-operative mapping rather than relying exclusively on static information based on a unique presurgical functional neuroimaging analysis. Consequently, the additional use of invasive electrophysiological investigations is highly recommended for surgery in eloquent structures.

Intrasurgical Functional Brain Mapping: Toward a Hodotopic View of Brain Processing

Intraoperatively, the integration of multimodal imaging into frameless stereotactic surgery was extensively used in the past decade and referred to as “functional neuronavigation.” However, a randomized trial failed to demonstrate significant impact of navigation on postoperative results.¹⁰ It can be explained by the limitations of the presurgical neuroimaging detailed above, as well as to the high risk of intraoperative brain shift, due to surgical retraction, mass effect, gravity, extent of resection (especially for voluminous tumors), and cerebrospinal fluid leakage. Several technical improvements have been proposed to reduce the effects of this shift, but their reliability has still to be optimized: combination with intraoperative ultrasound, producing real-time imaging; use of mathematical models based on data from ultrasonography or digital images that track cortical displacement; and intraoperative MRI. Nevertheless, their actual value on the improvement of EOR and preservation of quality of life remains to demonstrate. As a consequence, invasive electrophysiologic investigations currently remain the “gold standard” when operating in eloquent brain structures.

First, the technique of somatosensory- and motor-evoked potentials was extensively used in the past decades for intraoperative identification of the central region. However, its reliability regarding the localization of the rolandic sulcus is not optimal, with accurate localization of the central sulcus reported only 91% to 94%. Estimation of the overall sensitivity and negative predictive value of this method is around 79% and 96%, respectively. Moreover, phase reversal recording identifies only the central sulcus itself, but offers no direct information on the particular distribution of motor function on the adjacent exposed cerebral structures. Also, whereas the method of motor evoked potentials was improved, when recording compound muscle action potentials, only the monitored muscles can be controlled, that is, there is an inability to detect and possibly avoid motor deficits in nonmonitored muscles. Next, monitoring of muscle action potentials does not mean monitoring of complex movements and action adapted to the environment, which is nonetheless the ultimate goal for the patient. Above all, intraoperative-evoked potentials cannot currently be used to map language, memory or other higher functions crucial for patients’ quality of life (for a review, see ref.11).

Numerous authors have also promoted the use of extraoperative electrophysiologic recordings and stimulations via the implantation of subdural grids. Using this method, the patient is in optimal conditions, in his or her room, to perform the 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 study of the connectivity, in particular via the recording of “cortico-cortical evoked potential.” However, extraoperative electrophysiologic mapping, typically in grids with 1-cm-spaced electrodes, has limited accuracy. Also, it is necessary to perform two surgical procedures, one to implant grids and a second to remove the lesion. In addition, there is a risk of infectious complications due to the presence of subdural grids over several days. Although this method was extensively advocated in epilepsy surgery, because it also allows detection of the seizure foci, only the cortex can be mapped. It provides no information about the axonal connectivity, that is, mapping of subcortical structures is not possible.

Intrasurgical Cortical and Subcortical Electrostimulation Mapping Methods

Taking into account the advantages and the limits of these different mapping techniques, more and more neurosurgeons advocate the additional use of intrasurgical electrostimulation mapping (IESM), under general or local anesthesia during surgery in eloquent areas.^12,13 Indeed, except for tumors located within the motor structures, the mapping is performed in awake patients. However, as previously mentioned, since movements and action are more complex than single muscle contractions, it is also currently proposed to map the motor function under local anesthesia with active participation of the patient.¹⁴ The principle is to use IESM as a focal and transitory virtual lesion to obtain an individual functional map both at cortical and subcortical levels, and to test if a structure involved by a lesion is still crucial for the function—which is, for instance, observed in 15% to 20% of LGG cases. Stimulation of an essential area generates a transient disruption of the task performed by the patient, and this area should be preserved. An individual cortical mapping is thus obtained before the resection, which can be tailored according to functional boundaries (Fig. 6-1). Practically, a bipolar electrode tip spaced 5 mm apart and delivering a biphasic current (pulse frequency 60 Hz, single-pulse phase duration 1 millisecond) is 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 2 mA until a functional response is elicited, with 6 mA as the upper limit under local anesthesia, and with 16 mA as the upper limit under general anesthesia—with the goal of avoiding the generation of seizures. The patient is never informed when the brain is stimulated. At least one picture presentation without stimulation must separate each stimulation, and no site is stimulated twice in succession to avoid seizures. Each cortical site (size 5 × 5 mm, due to the spatial resolution of the probe) of the entire cortex exposed by the bone flap is tested three times. Indeed, it is admitted nowadays that three trials are sufficient to ensure whether an area is crucial for language, by generating disturbances during its three stimulations, and with normalization of the function as soon as the stimulation is stopped. This limitation of trials and tasks is required by the timing of the surgical procedure, because the patient is awake and can be tired at the end of the resection.

FIGURE 6-1 A, Preoperative language fMRI in a patient with no deficit, showing an LGG that involves the left inferior frontal gyrus (Broca’s area), with an activation immediately in front of the tumor (within the anterior insula). B, Intraoperative views before (left) and after (right) glioma resection, delineated by letter tags. IESM shows a reshaping of the eloquent maps, with a recruitment of perilesional language sites located behind the glioma. There was no crucial site within the left inferior frontal gyrus. Thus, an extensive resection of Broca’s area was possible, by preserving the subcortical connectivity in the depth of the cavity (50, corresponding to the anterior part of the arcuate fasciculus). Stimulation of the anterior insula-generated anomia.⁴⁹C, Postoperative axial FLAIR- (left) and coronal T2-weighted MRI, demonstrating a complete resection of the glioma, in a patient with no deficit.

Interestingly, recent series show that the surgical procedure can be simplified by avoiding the use of intraoperative electrocorticography despite an equivalent reliability of electrical mapping and without increasing the rate of seizures.¹² However, in cases of stimulation-induced seizures, the use of cold Ringer’s lactate is recommended to abrogate the seizure activity. In addition, some authors emphasized the value of “negative mapping” (no identification of eloquent sites) in the setting of a tailored cortical exposure.¹³ Although such recommendation is acceptable for high-grade gliomas, since the surgical goal is mainly to remove the enhanced part of the tumor, a negative mapping can be dangerous in surgery of diffuse LGG, especially in nonexpert hands. Indeed, due to the fact that LGG is poorly delineated, the limit of the resection will be essentially guided according to functional criteria. Because negative mapping can be due to false negative for methodologic reasons, it does not guarantee the absence of eloquent sites. In the experience reported by Sanai et al., all four patients with permanent postoperative deficits had no positive sites detected prior to their resections.¹² Therefore, other authors continue to promote a wider bone flap, in order to obtain a systematic positive mapping before performing the resection.^11,12 Moreover, a positive mapping might also allow an optimization of the EOR, since the resection can be pursued until eloquent areas are encountered, that is, with no margin around the functional structures. A recent study demonstrated that in a consecutive and homogeneous series of 115 LGG in the left dominant hemisphere, the rate of permanent deficit remained lower than 2% despite the absence of margin around the language sites.¹²

IESM allows the mapping of motor function (possibly under general anesthesia, by inducing involuntary motor response, but also in awake patient by eliciting a disturbance of the movement), somatosensory function (by eliciting dysesthesias described by the patient himself intraoperatively), visual function (by eliciting phosphenes and/or visual field deficit described by the patient), auditivo-vestibular function (by inducing vertigo), language (spontaneous speech, counting, object naming, comprehension, writing, reading, bilingualism, switching from one language to another), and also the mapping of higher-order functions such as calculation, memory, spatial cognition, cross-modal judgment or even emotional processing, by generating transient disturbances if the electrical stimulation is applied at the level of a functional “epicenter.”¹⁴ It is crucial that a speech therapist/neuropsychologist/neurologist be present in the operative room, in order to interpret accurately the kind of disorders induced by IESM, for instance speech arrest, anarthria, speech apraxia, phonological disturbances, semantic paraphasia, perseveration, anomia, dysculia, and so on. Thus, IESM is able to identify in real-time the cortical sites essential for the function before the beginning of the resection, in order to both select the best surgical approach and to define the cortical limits of the lesion removal.

Another major issue is the use of subcortical mapping throughout the resection, in addition to the cortical mapping before the lesion removal.^11,12 Brain lesion studies have taught that damage of the white matter pathways generated more severe deficit than cortical injury. Therefore, the subcortical tracts subserving motor, somatosensory, visual, auditivovestibular, language, and cognitive functions must be detected during the lesion removal, in order to preserve anatomofunctional connectivity while optimizing the EOR, that is, to pursue the resection until eloquent pathways are detected. Interestingly, according to the same principle as that described at the cortical level, IESM can also identify eloquent subcortical structures. It allows the study of anatomofunctional connectivity by directly and regularly stimulating the white matter tracts and deep gray nuclei throughout the resection, and by eliciting functional response when in contact with deep crucial areas (Fig. 6-1). Furthermore, IESM enables a better understanding of the brain connectivity, showing that dynamic cerebral processing is underlain by parallel distributed and interactive networks, the so-called “hodology.”¹⁵ This connectionist view also opens the door to the concept of cerebral plasticity, crucial in LGG surgery.

One of the major advantages of IESM for brain mapping in adult patients is that it intrinsically does not cause any false negatives—if the methodology is rigorously applied as detailed previously. Indeed, IESM is highly sensitive for detecting the cortical and axonal eloquent structures, and 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, rather than an isolated discrete functional site. IESM, however, also has a limitation, as its specificity is suboptimal. Indeed, IESM may lead to interpretation that a structure is crucial, due to the induction of a transient functional response when stimulated, whereas this effect is caused by backward spreading of the electrostimulation along the network to an essential area, and/or the stimulated region can be functionally compensated thanks to long-term brain plasticity mechanisms. In brief, although IESM is still the gold standard for brain mapping, due to the risk of “false positives,” its combination with new methods such as perioperative FNI and biomathematical modeling is now mandatory, to clearly differentiate those networks that are actually indispensable to function from those that can be compensated.¹⁶

IESM: New Insights into Dynamic Functional Organization of the Brain

Complementary methods of functional mapping are used to better understand the pathophysiology of functional areas, and thus to improve surgical planning in various eloquent regions.