Awake craniotomy for glioma resection: Technical aspects and initial results in a single institution.

Introduction. Although variations in the technique of awake craniotomy (AC) have been widely reported, a key member of this interdisciplinary procedure is the healthcare professional performing assessments of neurological function during resection. The expertise of the latter will depend on the neurological function to be tested and on available resources of the institution. This report details our initial experience of an AC service utilizing the expertise of a speech and language therapist (SLT) and an experienced neuro-physiotherapist (NP) to monitor patient function during glioma resection. Methods. Forty-five patients underwent 50 AC procedures for eloquently located gliomas over a 3-year period. Patients with a glioma involving speech or sensorimotor areas were assessed preoperatively by the SLT/NP respectively. The same therapist monitored the patient's neurological function intraoperatively and executed a rehabilitation program tailored to the needs of the patient in the postoperative period. Results. Three patients underwent biopsy only, due to intraoperative seizures precluding intraoperative mapping (2 cases) or speech arrest on stimulation of a small recurrent tumor. The remaining 47 cases were suitable for repetitive neurological assessment “awake” during tumor debulking. One patient with a large sensorimotor tumor developed intraoperative hemiparesis due to outward brain herniation (which recovered postoperatively). Ten patients developed a new or worsened neurological deficit in the initial postoperative period (6 were detected intraoperatively), of which 5 eventually had resolution and returned to baseline function within 2 weeks. Conclusions. In our initial experience based anecdotally on a previous similar “non-awake” caseload, we have found AC with the input of the SLT/NP to be a key component in ensuring optimal functional outcomes for patients with gliomas in eloquently located areas.


Introduction
Awake craniotomy (AC) is becoming a standard procedure for improving the extent of resection of gliomas involving language and sensorimotor regions. 1,2 Th e procedure is associated with improved neurological preservation rates compared with similar cases performed under general anesthesia. 3 Although diff usely infi ltrative gliomas will inevitably relapse close to the margins of tumor resection, maximal safe resection has been shown to extend survival in both low-and high-grade tumors. 4,5 Indeed, the concept of " supra-complete " resection of low-grade gliomas involving " non-eloquent areas " is gaining vogue in an attempt to delay onset of progression to higher-grade tumors. 6 Th e technique of AC has evolved from a large craniotomy to expose " positive " stimulation-induced language and motor responses to smaller tailored craniotomies, which may not expose any positive sites. Th is " negative mapping " strategy allows for minimal cortical exposure, less extensive intraoperative mapping, and a more time-effi cient neurosurgical procedure. 1 Direct electrical stimulation of cortical and subcortical areas within the vicinity of gliomas helps identify functionally important areas that need to be preserved. While the surgeon is stimulating these areas, there is a reliance on the skill of the assessor to detect any subtle changes in function.

Abstract
Introduction . Although variations in the technique of awake craniotomy (AC) have been widely reported, a key member of this interdisciplinary procedure is the healthcare professional performing assessments of neurological function during resection. The expertise of the latter will depend on the neurological function to be tested and on available resources of the institution. This report details our initial experience of an AC service utilizing the expertise of a speech and language therapist (SLT) and an experienced neuro-physiotherapist (NP) to monitor patient function during glioma resection. Methods . Forty-fi ve patients underwent 50 AC procedures for eloquently located gliomas over a 3-year period. Patients with a glioma involving speech or sensorimotor areas were assessed preoperatively by the SLT/NP respectively. The same therapist monitored the patient ' s neurological function intraoperatively and executed a rehabilitation program tailored to the needs of the patient in the postoperative period. Results . Three patients underwent biopsy only, due to intraoperative seizures precluding intraoperative mapping (2 cases) or speech arrest on stimulation of a small recurrent tumor. The remaining 47 cases were suitable for repetitive neurological assessment " awake " during tumor debulking. One patient with a large sensorimotor tumor developed intraoperative hemiparesis due to outward brain herniation (which recovered postoperatively). Ten patients developed a new or worsened neurological defi cit in the initial postoperative period (6 were detected intraoperatively), of which 5 eventually had resolution and returned to baseline function within 2 weeks. Conclusions . In our initial experience based anecdotally on a previous similar " non-awake " caseload, we have found AC with the input of the SLT/NP to be a key component in ensuring Furthermore, having an eff ective " early warning system " will help ensure that no new neurological defi cits develop during resection. While the operative conditions depend heavily on the experience of the surgeon and anesthetist, 7 the personnel responsible for clinical assessment of the patient are key to ensuring optimal functional outcome of the patient.
Although prevention of neurological defi cit is paramount, the concept of " brain plasticity " , which may facilitate improvement in neurological function (particularly in lowgrade gliomas), is gaining credence. 8 Th is " reactional plasticity " of the nervous system following surgery may be due to an induction in compensatory mechanisms that recruit latent neural networks. Th is functional reorganization may also provide the basis for a signifi cant response to a tailored rehabilitation program early in the postoperative period.
Th is paper, based on our preliminary experience of a new AC service, highlights the key contributions an experienced SLT and NP make to ensure good functional outcomes for glioma patients.

Case selection and assessment
Since February 2012, our institution has performed " asleepawake-asleep " craniotomy for gliomas involving eloquent areas (language and/or sensorimotor areas). Case selection was based on mode of presentation and correlation with the patient imaging to predict those likely to benefi t from AC. Patients were initially seen in the pre-assessment clinic (following MDT approval) where the patient is counseled by nursing and medical staff (surgeon and anesthetist) as to the need for an AC and what the procedure entails.
Pre-assessment also aff ords the patient an opportunity to visit the operating theater, and undergo detailed SLT/NP assessment which reassures the patient and allays any anxieties they might have regarding the procedure. Anesthetic involvement is key early in the process of case selection to ensure that there are no anesthetic contraindications to the procedure (e.g., diffi cult airway, morbid obesity). Th e standard volumetric magnetic resonance imaging datasets (T1 ϩ contrast, volumetric DTI for tractography planning, and FLAIR for low-grade tumors) for surgical planning are also performed during this visit, and patients get an opportunity to review their imaging with their surgeon.
Adjustment of patient ' s dexamethasone (for high-grade gliomas) and anti-seizure medication is also undertaken at pre-assessment to reduce the chance of brain swelling and seizure activity intraoperatively. Low-grade glioma patients tend not to be on dexamethasone preoperatively. Following this initial preparation, the patient then undergoes a detailed baseline clinical evaluation by the SLT/NP.

Speech and language therapy preoperative assessment
Th e patient ' s speech and language is assessed using formal tests covering the full spectrum of receptive and expressive language tasks (Table I). Th e SLT and patient also plan a range of speech and language tasks to be used intraoperatively, for example, describing an aspect of their occupation, details of a hobby, or daily routine. Th ese tasks are based on the patient ' s known premorbid ability. Any preoperative " baseline " language defi cit is documented for careful monitoring during the operation, to aid with identifi cation of new errors during stimulation. Th e preoperative assessment also facilitates a rapport to be established with the patient, which lends itself to " free-fl owing " speech with which language function can be closely monitored during tumor resection.

Preoperative physiotherapy assessment
For sensorimotor tumors, the patient undergoes an assessment by an experienced NP to record muscle power throughout all limbs using the Oxford grading scale, and an assessment of muscle tone using the Ashworth scale. Sensation is tested for light touch, " pin-prick " , joint position sense, fi nger -nose and heel -shin coordination, and stereognosis to establish any baseline defi cits. Th e therapist assesses the ability to perform a variety of simple functional tasks that can be replicated in theater to test dexterity and strength, for example, grip strength, the ability to direct a plastic needle through a button hole, dab a swab to the mouth, or use a key to open a padlock.

Awake craniotomy for tumor resection
Th e " asleep-awake-awake " craniotomy performed is broadly similar to that described by Sarang,Dinsmore,and Murphy et al. 7,9,10 Briefl y, patients are positioned in a head clamp and woken up following bone fl ap removal prior to dural opening. Th e therapist ensures that the patient responds readily to voice with normal tone and normal speech and facial expression, is orientated, and clearly demonstrates receptive and expressive language function, before allowing the surgeon to proceed.
Th e boundaries of the tumor are verifi ed using surgical navigation (BrainLAB) and intraoperative ultrasound (Flex Focus 800, BK Medical Systems) guidance systems. Cortical stimulation with an Osiris Neurostimulator (Inomed) bipolar electrode is performed to identify a corridor of safety into the tumor and adjacent eloquent sites that need to be avoided while tumor removal is performed using the Awake craniotomy for glioma 837 cavitating ultrasonic aspirator. Th e stimulation settings (for cortical and subcortical mapping) used for speech mapping are: a frequency of 50 Hz (continuous mode), " Burst " of 5, and a current of 3 -15 mAmp, while sensorimotor stimulation settings are: a frequency of 1 Hz (continuous mode), " Burst " of 5, and a current of 3 -25 mAmp. Early in our experience, our current setting was typically 7 mAmp, but we have since reduced it to 3 -5 mAmp, to reduce seizure risk. Each approximately 1 ϫ 1 cm area of cortex exposed directly over the tumor is stimulated in a methodical fashion three times, taking care not to retest the same area in succession. An area of positive stimulation is regarded as exhibiting a response (either movement or speech arrest) to confi rm eloquence. Low settings are generally used to minimize risk of inducing seizures. In the event of a seizure, which was usually detected at an early stage by the therapist, the operative fi eld was copiously irrigated with ice-cold saline, allowing resection to continue.

Speech and language assessment intraoperatively
Th e SLT ' s aim is to elicit continuous language function using as many language elements as possible (e.g., monologue, automatic speech, language sequencing tasks, explanation of idioms, and word association; see Table I), while avoiding gaps and hesitation that could be mistaken as errors. As the neurosurgeon stimulates and/or resects, the SLT alerts him to ' new ' speech errors, highlighting his proximity to the eloquent speech area to be avoided. Th is process continues until the neurosurgeon has completed resection, incurred a defi cit, or until patient discomfort becomes problematic.

Intraoperative neuro-physiotherapy assessment
Intraoperatively, there is continual assessment of upper and lower limb movements of all joints, sensory testing, and assessment of functional performance by the NP. In some cases, combined SLT/NP intraoperative monitoring is used in particular for patients with preoperative speech and sensorimotor symptoms, and tumors involving the dominant posterior frontal (pre-motor/supplementary motor) areas.

Postoperative general management
Management of these patients followed standard practice including gradual weaning of dexamethasone in the postoperative period. Typically, patients with high-grade gliomas are weaned to 2 mg of dexamethasone daily (or taken off dexamethasone completely) by the time of oncology review 2 -3 weeks postoperatively. Usually, low-grade glioma patients are not commenced on dexamethasone postoperatively unless they have a neurological defi cit, and weaning takes place after a week on a reasonable dose (e.g., 4 mg twice daily). An early postoperative MRI (with standard sequences, for example, T1, T1 ϩ contrast, T2, FLAIR, DWI) is usually performed within 72 h to document the extent of tumor removal and identify surgeryinduced hemorrhage, edema, or infarction.
Th e extent of tumor resection (EOR) was graded as follows: gross total resection (GTR) indicated complete resection of the enhancing mass in high-grade gliomas, or high-signal lesion in T2/FLAIR in low-grade gliomas; subtotal resection (STR) indicated anything from 90% resection and above, while anything less than 90% was regarded as partial resection (PR).

Postoperative therapy
Th e same SLT/NP involved in preoperative and intraoperative assessment also assesses the patient in the postoperative period to avoid any inter-observer variation in functional assessments. Follow-up and review is planned regardless of whether the patient has a defi cit or not.
Our team has devised a postoperative patient questionnaire suitable even for patients with a high-level language defi cit (Supplementary fi le to be found online at http:// informahealthcare.com/doi/abs/10.3109/02688697.2015. 1054354). It captures the patient ' s experience including pain, discomfort, positioning, information given pre-/intraand postoperatively, and relates to all neurosurgical, anesthetic, speech and language, and physiotherapy aspects of the patient ' s journey.

In-patient stay
Over a 3-year period, 50 consecutive AC procedures were performed on 45 patients: 3 patients underwent repeat debulking for recurrence of a high-grade glioma at 9, 11, and 15 months after initial surgery, while low-grade glioma patients underwent repeat debulking 32 and 33 months after initial surgery (Table II). Th e patients were aged between 15 and 68 years (mean of 45 years), comprising 28 male and 17 female. Twenty-nine procedures were performed with the SLT, 18 with the NP, and 3 cases (both low-grade tumors involving the dominant premotor area) with combined SLT/ NP input. Th e duration of the awake phase of the operation ranged from 1 -2.5 h (median 1.5 h). Th e median length of in-patient stay was 6 days (range: 4 -42 days), with all except 3 patients being discharged home within 2 weeks. ice-cold saline irrigation, allowing tumor resection to continue following return to baseline neurological function (typically within 5 -10 min of seizure).

Postoperative neurological function
Of the 6 patients with a new intraoperative defi cit, 5 returned to baseline function within 2 weeks while 1 patient had persistent proprioceptive loss requiring prolonged in-patient rehabilitation. Four patients developed a new neurological defi cit (not noticed intraoperatively) in the initial week post-surgery -3 experienced worsening of their preoperative sensorimotor defi cit while another developed new-onset expressive dysphasia following debulking of a dominant pre-motor tumor. One of the cases with worsened hemiparesis also developed new-onset expressive dysphasia (Fig. 1a -d). All 4 patients had a persistent albeit improved defi cit at 6-week review. In all cases, a rehabilitation program tailored to the needs of each patient was executed by the responsible therapist in the postoperative period.

Early postoperative imaging
Of the 50 cases, 39 underwent early postoperative MR imaging while 8 underwent CT imaging and 3 did not undergo any early postoperative imaging, as only a biopsy was performed (2 patients), and 1 patient was discharged before postoperative imaging could be performed. GTR was achieved in 11 cases (all HGGs), and STR in 13 cases

Tumor location and pathology
Th irty-one patients initially presented with seizures, 5 presented with dysphasia, 6 with sensorimotor symptoms, 2 with headaches, and another with depressed conscious level. Th e most common tumor locations were the left temporal and frontal lobes (18 and 15 cases respectively), most of which underwent speech monitoring. An additional left-handed patient underwent speech monitoring for a right temporal tumor (having presented with speech disturbance). Eight cases were " redo " AC cases, three of which had their original AC elsewhere -one of whom has undergone a subsequent repeat resection " awake " in Belfast (3 in total). Th e pathological diagnosis was glioblastoma in 27 cases (2 of which were GBM recurrences), 9 were cases of WHO grade III gliomas (4 cases of anaplastic astrocytoma, 3 of anaplastic oligodendroglioma, 2 of anaplastic oligoastrocytoma), and 14 cases that were WHO grade II gliomas (4 cases of astrocytoma, 7 of oligodendroglioma, and 3 of oligoastrocytoma).

Initial phase (wakening and initial stimulation)
One patient had a tonic-clonic seizure while emerging from anesthesia, while a second had multiple seizures in quick succession after initial cortical stimulation resulting in a prolonged postictal recovery time, which precluded further mapping and tumor resection (both cases were anaplastic astrocytoma). A third case had consistent speech arrest following stimulation overlying a small nodular recurrence of a GBM. Selective needle sampling of tumor using image-guidance in these 3 cases was performed to obtain a histopathological diagnosis. All 3 patients ' neurological function returned to baseline within 24 h postoperatively.

Tumor resection
Th e remaining 47 cases had a good baseline functional status intraoperatively and underwent debulking, with the exception of 1 patient with a large sensorimotor tumor who developed progressive intraoperative contralateral limb weakness due to outward brain herniation. Th e resultant cortical compression and venous congestion at the bone edge of a small craniotomy likely contributed to the transient hemiparesis which resolved postoperatively -an observation similar to that reported by Khu and Ng. 11 Using the electrical stimulation parameters outlined above, consistently positive cortical/subcortical sites were mapped in 13 cases that were subsequently avoided during resection. Although there were no positively mapped sites identifi ed in the remaining 34 cases, tumor resection was halted in areas associated with a signifi cant decline in neurological function, leading to a cessation of resection in 6 cases. Th ese patients experienced new-onset facial weakness (2 cases), proprioceptive loss (2 patients), hand clumsiness, and dysphasia.
Resection of infi ltrative zones of sensorimotor tumors extending into the internal capsular territory was also avoided to minimize postoperative neurological defi cit (5 cases).
Five cases had evidence of diff usion restriction near the resection cavity, in keeping with ischemia, one of which had clinically signifi cant neurological deterioration likely due to involvement of deep white-matter fi ber tracts (Fig. 1a -d); the remaining cases were asymptomatic. Although there was a mild degree of edema in the tumor resection margins in all cases, this did not cause clinical concern and invariably settled with a weaning course of steroids in the postoperative period.

Discussion
AC with direct electrical stimulation mapping is becoming the " gold standard " procedure for resection of gliomas in eloquent locations. Patients are much more likely to have better functional outcomes than similar patients who undergo general anesthetic for tumor resection. 3 With the increasing evidence that GTR of gliomas results in survival benefit, 4,5 surgeons are pushing the boundaries of resection to the limit of functional tolerance. The input of SLT/ NP during " awake " resection may help guide the surgeon when to stop if they feel that further deficit may permanently harm the patient. Although several larger studies have already highlighted the efficacy and safety of AC in glioma patients, 1 -3,7,10,13 -15 our preliminary experience with SLT/NP input highlights the key role the therapist plays in optimizing neurological function and may serve as a useful guide to other neurosurgical centers considering setting up an AC service.

Comparative analysis with other series
Although we have not performed a direct comparative analysis with similar " matched " cases from our " preawake " era, anecdotally, the frequency and severity of postoperative defi cits are much less following " awake " debulking compared to previous practices. Our own anecdotal experience is comparable with a single institution study by Duff au et al. who found that 17% of LGG patients developed a severe permanent neurological postoperative defi cit with no mapping, compared with 6.5% in a cohort of patients who were mapped functionally. 13 Future studies will be important to formally assess the neurological outcomes and potential survival benefi ts compared with a case-matched control series from the " pre-awake " era in our institution. It will also be important to assess if ongoing refi nements in our technique result in outcomes signifi cantly diff erent from the 10% morbidity of this initial experience.

Length of stay
In comparison with some other AC series, our length of stay (median of 6 days) was relatively long. 14 This perhaps reflects a degree of caution on the part of the authors in setting up a new service, to avoid discharging patients who may have developed a delayed " missed " neurological deficit. Most patients were happy to avail of in-patient therapy while waiting for their pathology report typically given on the day of discharge (5 days postoperatively). With greater experience and recognition of patients who may need additional rehabilitation, the authors hope to reduce the length of stay for uncomplicated patients in future. Indeed, patients with a preoperative deficit tended to develop worsening symptoms in the postoperative period, suggesting that these patients will likely need to remain an in-patient longer than those without a preoperative deficit.

Learning curve
At the outset of our AC program, perhaps eager to avoid any added morbidity from the procedure such as that reported by Gupta et al., 12 tumor debulking was halted at the fi rst hint of a neurological defi cit (some of which may have been patient discomfort-related). Th is is perhaps refl ected in the relatively low GTR rates for both high-and low-grade glioma cases in our series compared to other more experienced institutions. 1 -3,13 -15 It is recognized that patients can be pushed into a mild or moderate neurological defi cit to maximize EOR in the knowledge that they will likely rapidly recover function thereafter. 6,15 With increasing experience, the ability of the therapist to discern a reversible mild intraoperative defi cit amenable to rehabilitation from an irreversible defi cit, has facilitated more radical resection in latter cases. Th e enhanced ability of the therapist to detect stimulation-induced responses has also resulted in a reduction in seizure current leading to reduction in intraoperative seizures (one in the last 15 cases). A greater awareness of more seizure-prone areas, for example, supplementary motor areas, will also infl uence current settings used in future cases. 16 Our use of a patient questionnaire (supplementary fi le) has also been helpful in refi ning our technique. Th e initial slight problems of dry mouth, urinary issues, and positional discomfort have been reduced. Patient satisfaction ratings of the multi-disciplinary team approach were very high (90%). Time spent during preoperative assessment was greatly appreciated and promoted a sense of confi dence in the team.

The role of the therapist in prevention of new defi cits
Although many AC centers use a neuropsychologist to perform preoperative cognitive assessments and to assess language intraoperatively, our department has limited access to neuropsychology services. Instead, we have " on-site " access to SLT, which provides the perioperative services outlined above. However, the authors recognize that this specialist role is undertaken in some centers by neuropsychology, which may result in variation between centers with regard to assessments and pre-and postoperative care.
Although language can be tested in many ways during AC, for example, naming and reading tasks, 17 we have found that continuous speech production allows for the full spectrum of language function to be tested during ongoing resection (as opposed to " on-off " resection synchronous with testing) and may facilitate detection of language errors more readily than a single language function task such as object naming.

Conclusion
Th e multidisciplinary approach of our AC service incorporating the expertise of SLT/NP helps to preserve neurological function and is key to delivery of a tailored program of postoperative rehabilitation to ensure optimal patient outcomes. Such a multidisciplinary approach may serve as a helpful guide for other neurosurgical centers considering setting up their own AC service.

Declarations of interest:
GQ has been paid consultation fees by BK Medical Services. Otherwise, the authors have no confl ict of interest or fi nancial support to derive from this work.
Indeed, in comparing our SLT experience with that of NP, continuous clinical assessment is much easier with speech cases than sensorimotor cases where multi-modality testing (e.g., light touch, pin-prick, proprioception, and motor) is relatively more challenging to perform and monitor. In such cases, an " on-off " resection approach is particularly required at the tumor margin to forewarn the NP of the higher risk of incurred defi cit. In preparation for such cases, we now routinely have a preoperative case conference to review the imaging, in particular overlap of tumor and relevant fi ber tracts, to help focus the intraoperative therapy assessments.
In many cases, the baseline preoperative assessment by the SLT/NP helped detect deficits that would have been easily missed by the untrained observer. These subtle deficits can then be specifically monitored intraoperatively to ensure that no further worsening of neurological function occurs. It is also in the interest of the relevant therapist to ensure that no new deficits are encountered, as this minimizes the burden of rehabilitation postoperatively and helps minimize delays in subsequent radio/chemotherapy for patients with high-grade tumors. Functional preservation is also an important contributing factor to health-related quality-of-life indices -a key component of any treatment algorithm for glioma patients, given the limited life expectancy of many of these patients. 18

The role of the therapist in rehabilitation
Th e importance of rehabilitation of brain tumor patients is gaining increasing recognition following the publication of the latest version of Manual for Cancer services: Brain and CNS measures, May 2014. 19 Having the SLT and NP as key members of the AC team signifi cantly lends itself to meeting these criteria and standards. Patients will in essence have a " key rehabilitation worker " who oversees their rehabilitation needs from the outset to discharge, with a transition to community therapists to ensure that maximal neurological recovery is achieved. Although patients may improve due to brain " plasticity " , 6 active input from the SLT/NP is also likely to yield signifi cant neurological recovery. Indeed, the benefi ts of early and intense rehabilitation in the setting of brain tumor patient management is increasingly being recognized 20,21 and is also shown to be more costeff ective. 22 The role of postoperative imaging in predicting length of follow-up required Our fi ndings on early postoperative imaging are interesting, although future validation of these fi ndings with a larger patient cohort will be required. Th e predictive value of the extent of ischemia on diff usion-weighted imaging and the need for a more prolonged rehabilitation program will also need to be studied in greater detail. However, based on our preliminary experience, the presence of signifi cant vasogenic edema and/or peritumoral ischemia in eloquent areas may predict the need for more intense and prolonged postoperative rehabilitation involving community services.
Awake craniotomy for glioma 841