**2. Requirements and methodological constraints**

One of the main requirements for brain mapping is to have a valid, repeatable, sensitive tool to stimulate functional areas while administering an appropriate test. Electrical stimulation parameters and tests may be adapted to the individual patient according to several individual variables. This makes the mapping procedure extremely variable and imposes methodological constraints subject to validation in individual cases. The method involves comparing intraoperative against postoperative findings classified as true positive (with or without intervention), true negative, false positive, false negative (Wiedmayer et al., 2004). Task must be easy to perform and robust, capable to impact postoperative clinical state without limiting surgical resection. Clinical feedback is used to validate the overall intraoperative strategy while considering additional outcome parameters suggested by surgical management guidelines: extent of resection; complications; neurological and neuropsychological state; functional state and survival (Chang et al., 2003). Finally, new

Penfield & Rasmussen, 1950). In recent years, intraoperative ECS has been adopted for the identification and preservation of language function and motor pathways. Of note is that while cortical mapping was originally applied to epilepsy surgery where resection is essentially limited to the cortex, its indications were later extended to tumor surgery which involves the white matter. Whether these differences result in different clinical and operative settings is unclear and there exist mixed situations between the two extremes. The pathology that benefits most from AS is low-grade glioma (LGG). LGGs pose a considerable challenge in that they have characteristics of both epilepsy and tumors, with a long history that could influence neurofunctional anatomy in patients presenting normal neurological findings (Duffau et al., 2005; Duffau, 2005b, 2006a, 2006b, 2007). Importantly, tumor surgery and epilepsy surgery differ as to the aims of treatment: minimizing neurological sequelae is only one aspect, which can be tailored to lesion characteristics, as determined by clinical and instrumental studies. Basically, the two pathologies differ in symptoms and impairment. Improvement of preoperative clinical impairment and radical tumor resection are the endpoints for tumor surgery, while improvement of preoperative performance is the end-point in epilepsy treatment (Buckner, 2003; Hamberger et al., 2007). In glioma surgery, the definitive clinical advantages are broader indications for tumor removal, higher rate of radical tumor resection, and lower rate of postoperative impairment (Duffau 2005a,b).

In surgical treatment of cerebral gliomas the goals are to obtain complete tumor removal to the extent the nature of the pathology allows and to accomplish this without injuring normal anatomic structures (Yasargil, 1996a). Although LGGs and high-grade gliomas (HGGs) are distinct in biological features, clinical behavior and outcomes, understanding the effect of surgery remains equally important for both. This is especially true for lesions in areas of eloquence, where the proximity of critical pathways can present a significant challenge to standard operative strategies. The concept of eloquent area is evolving and may be potentially extended to all measurable functions. Thanks to collaborative teamwork in neuroscience and neuro-oncology, current neurosurgical innovations aim to improve our anatomical, physiological, and functional understanding of the surgical region of interest with a view to prevent potential morbidity during resection and improve the patient's

One of the main requirements for brain mapping is to have a valid, repeatable, sensitive tool to stimulate functional areas while administering an appropriate test. Electrical stimulation parameters and tests may be adapted to the individual patient according to several individual variables. This makes the mapping procedure extremely variable and imposes methodological constraints subject to validation in individual cases. The method involves comparing intraoperative against postoperative findings classified as true positive (with or without intervention), true negative, false positive, false negative (Wiedmayer et al., 2004). Task must be easy to perform and robust, capable to impact postoperative clinical state without limiting surgical resection. Clinical feedback is used to validate the overall intraoperative strategy while considering additional outcome parameters suggested by surgical management guidelines: extent of resection; complications; neurological and neuropsychological state; functional state and survival (Chang et al., 2003). Finally, new

**1.2 Aim of brain mapping** 

quality of life (QoL), an essential outcome measure.

**2. Requirements and methodological constraints** 

mapping techniques, like fMRI and DT imaging–based tractography (DTI), should be compared with ECS to determine their sensitivity and specificity.

#### **2.1.1 Patient cooperation and compliance in awake surgery**

Candidates for awake surgery face an unpredictable experience. To date, the choice depends on the patient who will have received a detailed description of the procedure and provided fully informed consent. Although awake craniotomy is generally considered to be well tolerated, complications such as emotional distress and agitation are not uncommon, with loss of control, the need for more sedation and failure of the mapping project. Failure rates due to agitation vary from 2 to 8% but are not systematically reported (Danks et al., 1998; Sahjipaul, 2000; Whittle et al., 2005).

#### **2.1.2 Preoperative clinical assessment**

Together with imaging, symptoms and objective findings will guide the surgical strategy. Disturbances in language-related functions, whether transient or progressive, functional or organic, are more indicative of operative risks than the lesion location itself (Benzagmout et al., 2007; Peraud et al.,2004). The standard assessments for dominance are the Edinburgh handness test, the Wada test and/or fMRI with the verb generation task (Duffau et al., 2003a).

The second step in patient assessment is neurological examination. It can reveal motor impairment (Medical Research Council scale, John, 1984) and disturbances in speech and cognition; however, it cannot provide reliable or sufficient information about the type of dysphasia or specific classification nor recognize mild impairments. This is an important drawback, since the rate of patients with mild-moderate deficits undergoing AS for mapping is quite high (26-55%) (Bello et al., 2007; Sanai et al., 2008; Skirboll et al., 1996).

While there is general consensus that mapping requires that patients present no significant disturbance at intraoperative task testing, some authors have underlined the utility of preoperative assessment, showing how sensitive tasks can maximize testing efficiency. The clinical aim is to recognize preserved functions or subprocesses in order to preserve them intraoperatively (Petrovich Brennan et al., 2007; Pouratian et al., 2003). This research can be pursued through consultation with a group of cognition experts during operative planning to develop personalized tests and tasks for a given patient. Specific functions include: spontaneous speech; language fluency; object naming; written/oral comprehension; reading; dictation; and repetition (baseline for French authors). Added to these are tasks involving writing sentences and words, oral controlled association by phonetic cue and semantic cue, famous face naming, action picture naming, transcoding tasks (Bello et al., 2007; Sanai et al., 2008). Nevertheless, evaluation was limited to the naming task before intraoperative assessment in the majority of cases (Haglund et al., 1994; Hamberger et al., 2005; Ojemann, 1989).

Reviewing the literature, the role of the neuropsychologist in AS is seldom defined in relation to treatment and little attention has been paid to the impact of primary brain tumors on QoL (Buckner et al., 2001; Giovagnoli & Boiardi, 1994; Taphoorn et al., 1992, 2005; Weitzner et al., 1996; Weitzner & Meyers, 1997). Differently from other cancer patients, where the burden of the disease is assessed, in brain tumor patients a decrease in cognitive and emotional functioning may be the result of cerebral disease. Subclinical symptoms, personality changes and mood disturbances may prove to be as burdensome to patients, or

Surgical Treatment of Supratentorial Glioma in Eloquent Areas 299

cannot statistically correlate with cortical stimulation and therefore cannot be reliably

Stimulation depolarizes a very focal area of the cortex which, in turn, evokes certain responses. For example, the 50-60 Hz Penfield technique has long been used to elicit motor responses, documented through direct visual observation of contralateral tonic limb movements in the beginning and since the late 1990s through motor evoked potentials (MEP) recordings (Cedzich et al., 1996, 1998; Kombos et al., 2001; Neuloh & Schramm, 2002; Penfield & Boldrey 1937). Although the mechanism of stimulation effects on language are poorly understood, the principle is based on depolarization of local neurons and passing pathways, inducing local excitation or inhibition, as well as possible diffusion to more distant areas by way of orthodromic or antidromic propagation (Ranck, 1975). With the advent of the bipolar probe and electrocorticography (EcoG) for after-discharge (AD) detection as a measure of electrical spreading, avoidance of local diffusion and more precise mapping have been achieved with an

This diversity among protocols is not trivial because it obviously impacts on the results of stimulation. Therefore, function localization may vary across studies as a result of different stimulation parameters and mapping strategies. Moreover, mapping strategies appear as one of the main variables that may affect the results of stimulation: direct cortical stimulation using the so-called short train technique (5 to 7 stimuli, 0.5 ms duration, ISI 4.0 ms = 250 Hz, with a train repetition rate of 1 or 2 Hz) or the 50-60 Hz Penfield technique (Penfield & Boldrey, 1937); maximizing stimulation currents at each cortical site to ensure the absence of eloquent function according to AD threshold or keeping stimulation intensity constant, while mapping the entire cortex and setting the threshold just below the lowest current observed to induce AD; monopolar or bipolar stimulators; stimulation parameters in subcortical white matter according to cortical response or at the lowest threshold to evoke a

Ojemann, using the single sample binomial test to check whether a site is essential for language, examined the accuracy of response during naming (Ojmann & Mateer, 1979; Ojeman, 1989). He gave a nonparametric description to determine if a site can be interpreted as essential: "*a site was determined to be related to language function if the chance probability of errors evoked at that site was less than 0.05 […] evoking errors during two of three stimulations at a site often achieved that level of statistical significance…*". This rate was slightly different in

The visual object naming task is easy to apply but is much more difficult to interpret. When classifying the type of errors, the main distinction that needs to be made is between speech arrest, anomia and speech disturbances. Error classification is strongly related to the aim of a study; it can vary from a simple definition of error (every change that occurs during stimulation) to a more articulate definition (Bello et al, 2007; Benzagmout et al, 2007; Duffau

Because of the infiltrating nature of gliomas, it is more than likely that a portion of the mass will occupy or be continuous with functional tissue. However, some evidence

interpreted as evoked by stimulation.

accuracy estimated to be ~ 5 mm (Haglund, et al., 1992, 1993).

response, all these alternatives may be encountered in similar studies.

others author's strategies (Hamberger, et al., 2005; Peraud, et al., 2004).

**2.4 The role of imaging in the integrated surgical strategy** 

**2.2 Electrical stimulation** 

**2.3 Positive sites** 

et al, 1999, 2003b; Duffau, 2006a).

more so, as certain focal neurological deficits (Giovagnoli et al., 2005; Talacchi et al., 2010b). Often, they go unrecognized at self assessment; judgment requires trained experts, like those cited above, with oncological experience (Pahlson et al., 2003; Taphoorn et al., 2004). Tumors in the dominant hemisphere may profoundly affect a patient's cognitive function well beyond language function. Although some deficits are known to be related to tumor localization, in brain tumor patients, especially in those with a LGG, many studies failed to find deficits restricted to a single cognitive domain (Tucha et al., 2000; Yoshi et al., 2008). This makes the assessment test battery crucial for global evaluation and longitudinal study.

#### **2.1.3 Tasks, functions and circuits**

Ojemann attempted to describe a distribution of language functions (Ojemann, 1989, 2003). He reported that naming interferences occur over a wide area of the left lateral cortex, extending beyond the limits of the classical model. He also found substantial variability in individual organization (Ojemann, 1977; Ojemann & Mateer , 1979; Ojeman G.A. et al., 1989; Ojeman, S.G., et al., 2003). Commonly, at least one area was described in the inferior frontal gyrus and one or more areas in the temporo-parietal peri-sylvian cortex. Other studies identified naming locations in specific regions such as the insular lobe, the striatum and opercular region, and the basal temporal language area (Duffau & Fontaine, 2005; Duffau, et al., 2005; Hamberger, et al., 2001; Ilmberger, et al., 2001; Lüders et al, 1988, 1991; Peraud et al 2004). However, even though surgical resection will ordinarily respect positive site margins, patients may still display a postoperative language deficit (Petrovich Brennan et al, 2007).

Single neuron recording provides a very sparse distribution of circuits which can be difficult to study with focal electrical interference (Waydo et al., 2006). Language, whose underlying primary neuronal substrate was termed "module", small areas measuring 1-2 cm2 with well-defined boundaries, represents a fortunate exception to this observation rather than the rule (Ojemann, G.A. et al., 1989). In addition, many authors showed that naming sites are often in close relationship to specific sites for different language functions, verb generation, reading, counting, comprehension, writing, working memory, and calculation, which justify the terminology adopted for designating them (nodes and shell-core) (Haglund et al., 1994; Schwartz et al., 1999; Schäffler et al., 1996).

Accordingly, the characteristics of the underlying neural circuits differ among functions and are not yet well understood anatomically. This differences may influence the mapping modality, recording the neuronal activity that "participates" in a widely distributed function with some regional differences. This is the case of intraoperative ECS in the right hemisphere for mapping spatial functions where "positive sites" could be removed while monitoring the corresponding partial deficit during tumor resection (Bartolomeo et al., 2007; Gharabaghi et al, 2006; Thiebaut de Schotten et al., 2005).

#### **2.1.4 Item selection**

As a standard procedure, an intraoperative task has to be previously verified on each patient before being performed, and those items chosen that patients were able to name during the preoperative testing phase. Items that patients were unable to name at preoperative assessment are deleted (Hamberger et al., 2005; Ojemann, 1989; Roux et al., 2003b). What is not well specified is the cut-off number of items a patient can misname when included in an AS protocol (Roux et al., 2003a; Lubrano et al., 2004). According to Little et al. (Little, et al., 2004), a preoperative object naming error rate greater than 25% cannot statistically correlate with cortical stimulation and therefore cannot be reliably interpreted as evoked by stimulation.

## **2.2 Electrical stimulation**

298 Advances in the Biology, Imaging and Therapies for Glioblastoma

more so, as certain focal neurological deficits (Giovagnoli et al., 2005; Talacchi et al., 2010b). Often, they go unrecognized at self assessment; judgment requires trained experts, like those cited above, with oncological experience (Pahlson et al., 2003; Taphoorn et al., 2004). Tumors in the dominant hemisphere may profoundly affect a patient's cognitive function well beyond language function. Although some deficits are known to be related to tumor localization, in brain tumor patients, especially in those with a LGG, many studies failed to find deficits restricted to a single cognitive domain (Tucha et al., 2000; Yoshi et al., 2008). This makes the assessment test battery crucial for global evaluation and longitudinal study.

Ojemann attempted to describe a distribution of language functions (Ojemann, 1989, 2003). He reported that naming interferences occur over a wide area of the left lateral cortex, extending beyond the limits of the classical model. He also found substantial variability in individual organization (Ojemann, 1977; Ojemann & Mateer , 1979; Ojeman G.A. et al., 1989; Ojeman, S.G., et al., 2003). Commonly, at least one area was described in the inferior frontal gyrus and one or more areas in the temporo-parietal peri-sylvian cortex. Other studies identified naming locations in specific regions such as the insular lobe, the striatum and opercular region, and the basal temporal language area (Duffau & Fontaine, 2005; Duffau, et al., 2005; Hamberger, et al., 2001; Ilmberger, et al., 2001; Lüders et al, 1988, 1991; Peraud et al 2004). However, even though surgical resection will ordinarily respect positive site margins, patients may still display a postoperative language deficit (Petrovich Brennan et al, 2007). Single neuron recording provides a very sparse distribution of circuits which can be difficult to study with focal electrical interference (Waydo et al., 2006). Language, whose underlying primary neuronal substrate was termed "module", small areas measuring 1-2 cm2 with well-defined boundaries, represents a fortunate exception to this observation rather than the rule (Ojemann, G.A. et al., 1989). In addition, many authors showed that naming sites are often in close relationship to specific sites for different language functions, verb generation, reading, counting, comprehension, writing, working memory, and calculation, which justify the terminology adopted for designating them (nodes and shell-core) (Haglund et al., 1994;

Accordingly, the characteristics of the underlying neural circuits differ among functions and are not yet well understood anatomically. This differences may influence the mapping modality, recording the neuronal activity that "participates" in a widely distributed function with some regional differences. This is the case of intraoperative ECS in the right hemisphere for mapping spatial functions where "positive sites" could be removed while monitoring the corresponding partial deficit during tumor resection (Bartolomeo et al., 2007;

As a standard procedure, an intraoperative task has to be previously verified on each patient before being performed, and those items chosen that patients were able to name during the preoperative testing phase. Items that patients were unable to name at preoperative assessment are deleted (Hamberger et al., 2005; Ojemann, 1989; Roux et al., 2003b). What is not well specified is the cut-off number of items a patient can misname when included in an AS protocol (Roux et al., 2003a; Lubrano et al., 2004). According to Little et al. (Little, et al., 2004), a preoperative object naming error rate greater than 25%

**2.1.3 Tasks, functions and circuits** 

Schwartz et al., 1999; Schäffler et al., 1996).

**2.1.4 Item selection** 

Gharabaghi et al, 2006; Thiebaut de Schotten et al., 2005).

Stimulation depolarizes a very focal area of the cortex which, in turn, evokes certain responses. For example, the 50-60 Hz Penfield technique has long been used to elicit motor responses, documented through direct visual observation of contralateral tonic limb movements in the beginning and since the late 1990s through motor evoked potentials (MEP) recordings (Cedzich et al., 1996, 1998; Kombos et al., 2001; Neuloh & Schramm, 2002; Penfield & Boldrey 1937). Although the mechanism of stimulation effects on language are poorly understood, the principle is based on depolarization of local neurons and passing pathways, inducing local excitation or inhibition, as well as possible diffusion to more distant areas by way of orthodromic or antidromic propagation (Ranck, 1975). With the advent of the bipolar probe and electrocorticography (EcoG) for after-discharge (AD) detection as a measure of electrical spreading, avoidance of local diffusion and more precise mapping have been achieved with an accuracy estimated to be ~ 5 mm (Haglund, et al., 1992, 1993).

This diversity among protocols is not trivial because it obviously impacts on the results of stimulation. Therefore, function localization may vary across studies as a result of different stimulation parameters and mapping strategies. Moreover, mapping strategies appear as one of the main variables that may affect the results of stimulation: direct cortical stimulation using the so-called short train technique (5 to 7 stimuli, 0.5 ms duration, ISI 4.0 ms = 250 Hz, with a train repetition rate of 1 or 2 Hz) or the 50-60 Hz Penfield technique (Penfield & Boldrey, 1937); maximizing stimulation currents at each cortical site to ensure the absence of eloquent function according to AD threshold or keeping stimulation intensity constant, while mapping the entire cortex and setting the threshold just below the lowest current observed to induce AD; monopolar or bipolar stimulators; stimulation parameters in subcortical white matter according to cortical response or at the lowest threshold to evoke a response, all these alternatives may be encountered in similar studies.

#### **2.3 Positive sites**

Ojemann, using the single sample binomial test to check whether a site is essential for language, examined the accuracy of response during naming (Ojmann & Mateer, 1979; Ojeman, 1989). He gave a nonparametric description to determine if a site can be interpreted as essential: "*a site was determined to be related to language function if the chance probability of errors evoked at that site was less than 0.05 […] evoking errors during two of three stimulations at a site often achieved that level of statistical significance…*". This rate was slightly different in others author's strategies (Hamberger, et al., 2005; Peraud, et al., 2004).

The visual object naming task is easy to apply but is much more difficult to interpret. When classifying the type of errors, the main distinction that needs to be made is between speech arrest, anomia and speech disturbances. Error classification is strongly related to the aim of a study; it can vary from a simple definition of error (every change that occurs during stimulation) to a more articulate definition (Bello et al, 2007; Benzagmout et al, 2007; Duffau et al, 1999, 2003b; Duffau, 2006a).

#### **2.4 The role of imaging in the integrated surgical strategy**

Because of the infiltrating nature of gliomas, it is more than likely that a portion of the mass will occupy or be continuous with functional tissue. However, some evidence

Surgical Treatment of Supratentorial Glioma in Eloquent Areas 301

AS is a challenge for a working team, since it implies substantial modifications in the professional behavior of all physicians involved: for the surgeon working in an uncomfortable situation; the anesthesiologist monitoring the patient continuously; the neuroradiologist awaiting intraoperative confirmation of interpretation of findings; the neuropsychologist making real-time evaluation of scans. AS involves a complex scenario: integration of different types of knowledge, organization of a heterogeneous team, cooperation in different settings (operating room, ward, out-patient clinic), surgical and research protocols to be adopted, technical adjustments to make the research comparable.

Awake surgery is performed as follows. Continuous sedation (Sarang, et al.,2003) was achieved with rapidly acting agents and infiltrative anaesthesia of the scalp. Airway management remains a concern due to the risk of aspiration or oversedation with oxygen saturation <90%. Patients breathe spontaneously during awake surgery. Propofol, fentanyl, remifentanyl and midazolam are commonly used agents. Volatile anaesthetics should not be given because they interfere with electroencephalographic (EEG) recording and cause a dose-dependent distortion on the EEG, with vasodilatation resulting in increased intracranial pressure (ICP) (Himmelseher, et al., 2001). While propofol can also alter the EEG (Herrick, et al., 1997), intravenous drugs are preferable because of their rapid onset and easily manageable duration of action which causes no nausea or vomiting. The sedation level is very important since oversedation results in an uncooperative patient and medical problems (i.e., respiratory depression), whereas undersedation makes the patient uncomfortable and restless. For this purpose, the Modified Observer's Assessment of

The feasibility and efficacy of AS have been studied in comparison with general anesthesia (GA). The absolute anesthesiologic exclusion criteria for AS are obstructive sleep apnea and

 Duration of surgery: according to Gupta, the mean procedure time was shorter in the GA group than in the AS group (182 min vs. 196 min; p <0,05) (Gupta, et al., 2007). A similar duration was found by Keifer and Taylor (Keifer et al., 2005; Taylor & Bernstein, 1999). Bello reported longer durations: mean 5 h 45 min, longest 6 h 45 min; mean awake time 1 h 45 min (Bello et al., 2007). In Whittle the mean awake duration was 62

 Intraoperative medical complications are classified as *anesthetic* (inadequate or excessive sedation, pain, nausea, vomiting); *respiratory* (oxygen saturation <90%, increased CO2, hypoventilation <8/min, airway obstruction); *hemodynamic* (hyper- o hypotension, tachy- or bradycardia); *neurological* (convulsions, brain swelling, new neurological deficit) (Costello et al., 2005; Keifer et al., 2005; Sarang, & Dinsmore, 2003). In a review of the literature, Skucas demonstrated how hyper- and hypotension could be frequent in AS (11 and 56%, respectively) (Skucas, et al., 2006). However, in their study on 332 patients, they observed that airway problems are not so frequent: only 2% of patients developed hypoxemia and only 1.8% required intubation or positioning of respiratory devices. Respiratory issues could arise more frequently in obese patients or those with asthma or chronic obstructive pulmonary disease (COPD). As concerns intractable seizures in unconscious patients, the study reported that seizures occurred in only 3%

difficult intubation (Picht et al., 2006). Parameters for comparing GA versus AS:

Alertness/Sedation Scale (Bauerle, et al., 2004) was used.

min (range 10-105 min) (Whittle et al., 2005).

**3. Intraoperative setting** 

**3.1 Anesthesiology** 

supports the concept that resection should ideally go beyond the gross tumor margin apparent on preoperative imaging. Therefore, it is not only patients with tumors located within the frontal operculum who may benefit from intraoperative language mapping, but also those with lesions in proximity to this region because of the significant variability in this region's anatomical and functional organization (Edeling et al., 1989; Quiñones-Hinojosa et al., 2003).

Possible causes of damage are: trajectory in subcortical tumors; abnormal anatomy in recurrent tumors; distorted anatomy due to the tumor; tumor infiltrating the functioning brain in LGGs; irregular tumors; and tumor periphery in HGGs. All are known to be crucial factors for surgical outcome, and knowledge of the structural characteristics of eloquent areas may help the surgeon to avoid clinical consequences. Aims may be categorized as linked to: 1) orientation, which is usually not histology-dependent (trajectory, abnormal anatomy, distorted anatomy); and 2) removal, usually used for specific gliomas (low-grade, irregular margins, periphery).
