The modern practice of awake craniotomy (AC) evolved with the development of sedative agents that facilitated existing local anesthetic techniques.1,2 This allowed the precise anatomic localization of neurologic function in patients who are undergoing supratentorial craniotomy, with most benefit seen predominantly for epilepsy surgery.3 Awake craniotomy with brain mapping has become the gold standard for patients undergoing surgery for tumours near or within eloquent areas of the brain. A wide variety of techniques have been described including the asleep-awake-asleep approach (SAS), monitored anesthesia care, and asleep-awake, awake-awake-awake, and conscious sedation approaches.4 Meng et al. recently succinctly summarized AC as a pre-awake, awake, and post-awake phase, with and without airway intervention.5 The goals include maintaining airway patency, optimizing cerebral perfusion, facilitating real-time brain mapping, minimizing postoperative pain, and allowing rapid recovery and assessment of neurologic function after surgery.6

Figure
figure 1

Correlation between intraoperative neurologic testing (a, b), cortical area exposed at craniotomy (c), and functional magnetic resonance images (d, e)

Full size image

Various anesthesia drugs have been used to achieve these goals, including propofol, remifentanil, fentanyl, and dexmedetomidine.7 The use of dexmedetomidine was first reported for AC surgery in 2001, predominantly as an adjunct to the SAS method, but with an additional benefit of decreased respiratory depression.8,9 While a propofol/remifentanil technique provides satisfactory conscious sedation, recognized drawbacks of this approach include hypoventilation and hypercapnia due to airway obstruction.10 Dexmedetomidine is a lipophilic imidazole derivative that acts as a selective pre- and post-synaptic alpha2 adrenoceptor agonist. It has anxiolytic, sedative, and anesthetic properties. Dexmedetomidine provides sedation that resembles natural sleep without cognitive impairment, making it an excellent anesthetic choice for AC surgery.11 An additional feature of dexmedetomidine making it attractive for AC surgery is its shorter arousal time compared with propofol.12 It has also been hypothesized that dexmedetomidine has a capacity to maintain the cerebral metabolic rate to cerebral blood flow (CBF) coupling. This results in a decrease in CBF, via an alpha2B receptor-mediated vasoconstriction, to match the reduced cerebral metabolic requirements during sedation and anesthesia.13,14 As research continues to elucidate the effect of dexmedetomidine on cancer cells, characterizing the impact of dexmedetomidine on tumour biology will be an important step in establishing the role of dexmedetomidine in neurooncology.15,16

In 2012, we instituted a standardized anesthetic technique of intravenous dexmedetomidine and scalp nerve blocks for patients undergoing AC surgery. This technique has been previously described by our group for select high-risk neurosurgical patients and is now the primary technique used at our institution.17,18 In this present study, we examined whether a dexmedetomidine-based sedation facilitates optimal conditions for AC defined as the avoidance of the need for airway manipulation or conversion to general anesthesia prior to complete brain mapping and brain tumour resection. Secondary outcomes included the establishment of optimal conditions for successful brain mapping and maximal tumour resection during AC.5

Methods

Eligibility and data collection

Following institutional Research Ethics Board approval in November 2016, waiving the need for patient consent, all adults (age ≥ 18 yr) who underwent AC at our institution during the study period from 1 March 2012 to 1 September 2016 were eligible for inclusion. Patients were excluded if their health records were incomplete or if they had a documented refusal of consent to participate in research. The intraoperative records of all patients who underwent AC surgery during the study period were screened for eligibility and, if applicable, data were extracted. Demographic and perioperative data collected included American Society of Anesthesiologists physical status (ASA-PS) classification, age, height, sex, weight, use of preoperative anti-epileptic drugs, tumour characteristics (pathology and location), and amplitude of direct cortical electrical stimulation (DCES), types and doses of anesthetic agents, intraoperative adverse events (airway manipulation or instrumentation, conversion to general anesthesia, nausea, and seizure activity), surgical outcomes (extent of resection and deviation of surgical plan based on preoperative functional magnetic resonance imaging), and select postoperative outcomes (new neurologic deficits, duration of hospital stay).

Anesthetic technique

Standard monitors were applied, intra-arterial blood pressure monitoring was instituted in all cases, and end-tidal carbon dioxide was monitored via a sampling channel integrated into a facemask or nasal cannula (depending on patient preference). A dexmedetomidine loading dose of 1 µg·kg−1 was administered over 15 min and an infusion started at 0.3-0.4 µg·kg−1·hr−1. Scalp nerve blocks were performed using an anatomical approach with 0.375% bupivacaine and epinephrine prior to pin insertion for both perioperative anesthesia and postoperative pain.19,20 Midazolam (0.01-0.05 mg·kg−1) and fentanyl (1 µg·kg−1) were administered before performance of the scalp blocks.

The patient was positioned awake and a Mayfield frame was used in all cases with supplemental lidocaine applied to the pin sites prior to insertion. The incision line was infiltrated with 2% lidocaine and additional infiltration of the muscle and topical anesthesia of the dura were performed by the surgeon as required throughout the case. The dexmedetomidine infusion rate was titrated to achieve a Ramsey sedation score between 2 and 4 until the dura was opened.21 Bolus doses of dexmedetomidine (0.05 µg·kg−1-0.1 µg·kg−1) were permitted throughout the case and infusion rates were increased as necessary to achieve required levels of sedation. Fentanyl for analgesia during bone flap and dural opening were provided at the discretion of the anesthesiologist; if required, a low-dose remifentanil infusion was permitted for supplemental analgesia. All infusions were discontinued at dural opening and restarted following neurophysiologic testing on a case-by-case basis. Hydromorphone, for postoperative analgesia, was administered following completion of the “awake” phase.

Neurologic assessment

Patient assessment involved pre- and postoperative neurophysiologic testing using a battery of tests for speech and cognition, including phenomic word fluency using the letters F, A, and S, semantic fluency tests (animal naming),22 the line bisection test,23 trail making tests A and B,24 the Weschler Memory Scale-Logical Stories (Anna Thompson story)25 judgement of line orientation,26 digit span, digit symbol, Rey-Osterreith Complex Figure,27 Barthel Index,28 Hospital Anxiety and Depression Scale,29 and Patient Assessment of Own Functioning.30

All patients underwent intraoperative speech and sensorimotor testing, which was dictated by the location of the tumour and of the cortical areas exposed by the craniotomy. Speech testing included number counting, naming, and word-generation tasks. Mapping was performed using bipolar Direct Cortical Electrical Stimulation [DCES with the OCS2 Ojemann Cortical Stimulator (Integra Life Sciences; Plainsborough, NJ, USA)]. Stimulation was performed with 500-msec pulses at 60 Hz with a starting amplitude of 2 mA and peak amplitude of 8mA31 (Figure).

Primary and secondary outcomes

The primary outcome was the incidence of failure of the AC anesthetic technique, defined by the need to convert to general anesthesia with a secured airway prior to (or during) brain mapping and brain tumour resection. Secondary outcomes were the incidence of: 1) optimal conditions for successful brain mapping and maximal tumour resection, 2) interventions to restore airway patency or rescue the airway, 3) significant (> 20% from baseline) hemodynamic instability, 4) nausea and vomiting, 5) new-onset neurologic deficits, and 6) seizure activity.

Statistical analysis

Data distribution was tested for normality using the Shapiro-Wilk test. Continuous data are presented as mean [standard deviation (SD)] when normally distributed or median (interquartile range [IQR]) when not normally distributed; categorical data were presented as proportions. The 95% confidence intervals are reported where appropriate. Fisher’s exact test was used to test differences between groups. P < 0.05 was considered statistically significant. All statistical analyses were performed using STATA 14 (STATACorp LP, TX, USA).

Results

We identified 56 patients who underwent AC surgery during the study time period. One patient chart was excluded because of an incomplete health record. Fifty-five patients were included in the cohort, including ten patients described previously by Garavaglia et al.17 Demographic and intraoperative characteristics are summarized in Table 1. Most patients were male (55%), ASA-PS ≥ III (99%), and had preoperative seizure activity (64%) requiring anti-epileptic medication. The median [IQR] patient postoperative hospital stay was 2 [1-3] days.

Table 1 Demographic and intraoperative characteristics for all (n = 55) patients
Full size table

Primary outcomes

All patients included in the study underwent successful AC with DCES for motor or speech function. There was no failure of the approach requiring conversion to general anesthesia with a secured airway. No intraoperative interventions were required to restore airway patency.

Secondary outcomes

The preoperative functional MRI-based surgical plan was often modified using intraoperative information acquired from DCES. Modifications to this plan fell into three categories: 1) identification of the corridor for tumour resection [17/55 (31%)], 2) definition of the limit of resection [31/55 (56%)], and 3) identification of the focus of positive stimulation distant from the corridor or site of tumour resection [6/55 (11%)]. In patients in the third category, our practice was to keep patients awake throughout the phase of tumour resection, often with serial episodes of cortical or subcortical stimulation during physiologic testing. Overall, 17/55 (31%) of patients remained awake throughout the process of tumour resection. The remainder underwent some degree or duration of deep sedation following the brain mapping.

Intraoperative medications administered are described in Table 2. Mean (SD) time from commencement of dexmedetomidine infusion to first cessation for testing was 100 (33) min. The mean (SD) cumulative dose of dexmedetomidine administered prior to initiating neurophysiologic testing was 1.80 (0.76) µg·kg−1. Fentanyl and midazolam were predominantly administered prior to pinning with supplemental fentanyl administered at dural opening as needed. Hydromorphone loading for postoperative analgesia took place following completion of testing in the majority of cases. In 7/55 (13%) of patients, hydromorphone was administered prior to testing in a dose range 0.002-0.13 mg·kg−1. Mean (SD) intraoperative hydromorphone dose administered in all patients was 1.2 (0.6) mg.

Table 2 Mean (SD) doses of anesthetic and analgesic medications administered
Full size table

No significant changes in blood pressure were observed related to dexmedetomidine use. One patient experienced an episode of bradycardia associated with hypotension at dural opening, which responded to bolus glycopyrolate (0.2 mg). At the time of the episode, the infusion dose of dexmedetomidine was 1 µg·kg−1·hr−1. There were no nausea and vomiting during the procedure. Only one patient required low-dose supplemental remifentanil infusion at a rate of 0.005-0.02 µg·kg−1·min−1 for analgesia during the initial period up to dural opening. Emergence agitation or delirium was not evident in any patient and no patient required general anesthesia after the initial pre-awake phase.

Tumour characteristics and surgical outcomes are summarized in Table 3. There were no episodes of generalized seizure. Focal seizures induced by DCS occurred in nine (16%) patients, of which eight (15%) occurred in patients with pre-existing preoperative seizure activity on anti-epileptic medication. All occurred at the time of cortical stimulation. Seizures resolved with direct administration of cold saline alone in seven (13%) patients; two patients required midazolam (0.5 mg) and phenytoin after mapping to control focal seizure activity. None of the events required premature termination of cortical mapping and no seizure resulted in airway intervention or conversion to general anesthesia. There was no association between preoperative antiepileptic use and intraoperative seizures (P > 0.99) or between intraoperative phenytoin use and intraoperative seizures (P = 0.69). Characteristics of seizure activity are summarized in Table 4. Gross total resection of metastatic lesions, the area of enhancement in high-grade glioma cases, or the area of FLAIR signal abnormality in low-grade gliomas cases was achieved in 32 (56%) patients. Three patients (5%) developed new postoperative neurologic deficits following surgery, one of which was associated with a postoperative hemorrhage within the tumour cavity.

Table 3 Tumour characteristics and surgical outcomes for all (n = 55) patients
Full size table
Table 4 Incidence of intraoperative seizure activity, classified by anticonvulsant use and tumour pathology
Full size table

Discussion

We describe our experience of using a predominantly dexmedetomidine-based anesthetic technique in conjunction with scalp blocks to facilitate AC surgery. This technique shows that dexmedetomidine can be used as the primary anxiolytic and sedative drug for AC surgery up to four hours in duration with no adverse respiratory events and providing optimal neurosurgical conditions. In our cohort, using dexmedetomidine anesthesia, no patient required unplanned airway intervention or conversion to a general anesthesia.

Our reported outcomes are in keeping with the relatively low conversion rates to GA utilizing other anesthetic protocols. Previously published anesthesia failure rates during AC surgery vary, but typically 2-6% of patients require some form of airway manipulation during AC.32 The reasons for airway interventions generally include poor patient selection, an inadequate anesthetic regimen, and intraoperative stimulation-induced seizures. The choice of anesthetic technique influences perioperative adverse events. Goettel et al. found no respiratory events when dexmedetomidine/propofol was used for rescue, but 20% with propofol/remifentanil.33 Dilmen et al. more recently reported moderate-to-severe intraoperative desaturations in almost 20% of cases using a dexmedetomidine-based technique, but with unquantified amounts of remifentanil or propofol suplementation.34 Our study adds to the evidence base, suggesting that use of dexmedetomidine without propofol can be safely performed with no adverse respiratory events, no conversion to general anesthesia, and successful intraoperative mapping throughout. Lobo et al. have suggested that for surgeries with an expected duration exceeding four hours, the SAS technique is more appropriate as patients can cooperate better if their awake phase is preceded by an asleep phase. Concern has also been expressed regarding potential accumulation of dexmedetomidine and delayed return of function in prolonged procedures.35 The median duration of surgery in our study was less than four hours, and there were no deleterious effects on intraoperative mapping or delayed return of function observed.

Awake surgery offers the best approach to optimal tumour removal while minimizing the risk of permanent postoperative neurologic deficits, with both improved quality of life and prolonged survival.36 In a number of our cases, the operative plan was modified based on unexpected findings during stimulation of eloquent brain regions within the planned area of tumour resection. In these cases, tumour resection was minimized to optimize intact neurologic function. This finding strongly suggests a benefit to continued patient assessment during tumour resection. The low rate (5%) reported here of new postoperative neurologic deficits following surgery, despite aggressive tumour resection, also compares favorably with other studies. A recent systematic review found a 7% incidence of new focal deficit after AC but 23% after general anesthesia.37 Similarly, Honorato-Cia et al. found significantly lower perioperative neurologic deficits using a predominantly dexmedetomidine-based approach (in a dose range of 0.2-1.4 µg·kg−1·min−1) compared with other techniques.38 Dilmen et al. also reported reduced postoperative neurologic deficits using conscious sedation compared with SAS.34

While other studies have previously shown dexmedetomidine to be effective for AC surgery, the use of supplementary propofol—by either infusion or bolused—was most often used.39

The use of midazolam in AC is controversial, as it can be associated with emergence agitation and delirium when transitioning to the awake phase.6 This was not noted in our cohort where 52 (95%) patients received midazolam perioperatively. A mean (SD) dose of 2 mg (1) was administered prior to neurophysiologic testing. In animal studies, midazolam has been shown to interact synergistically with dexmedetomidine resulting in a dose-sparing effect.40 The ability to use lower doses of dexmedetomidine to achieve sedation may have some benefits in reducing the side effects of dexmedetomidine and did not interfere with neurocognitive testing in this cohort. A recent study by Suero Molina et al. comparing dexmedetomidine and remifentanil to an SAS technique for AC found that dexmedetomidine did result in better quality and reliable neurologic testing upon cessation.41

One of the main criticisms of using dexmedetomidine is the potential for blunting of the carbon dioxide (CO2) response curve,42 and some anesthesiologists advocate for advanced airways for the pre-awake phase to minimize this risk. This is an important consideration for those patients with a “tight brain” and increased surgical difficulty. Despite this, hypercapnia, due to either obstruction or hypoventilation, was not found in this cohort. Indeed, the mean (SD) maximal partial pressure of carbon dioxide was very near physiologic levels of 40 mmHg,8 suggesting that the central physiologic responsiveness to CO2 was maintained. In addition, favorable brain tension was reported in the vast majority of cases in this study. Previous reports of dose-related cardiovascular effects have described episodes of bradycardia of up to 27% during AC surgery,43 but within this cohort we found no clinically significant hypotension and only one episode of bradycardia with a surgical etiology requiring treatment.

Surgical failure during AC is defined as incomplete intraoperative awake monitoring of brain function during tumour resection, typically reported at approximately 2%.44 Nonetheless, even with a continuously awake (i.e., awake-awake-awake) approach, there is still an inherent failure rate (approximately 2%) due to seizure activity.45 Seizures have historically been the major cause of aborted AC surgery with reported incidences in the range of 3-22%. Nevertheless, phenytoin use is also associated with communication and AC failure, with no evidence of reducing perioperative seizure activity.46 The seizure rate in our cohort was comparable to previous studies, and all seizures were triggered by cortical stimulation with most terminated with the application of cold saline.

Our study has some limitations. The retrospective nature of this study introduces several risks of bias and missing data. The surgical team provided information on alterations to the surgical plan following retrospective assessment of patient charts, which may have introduced recall bias. In addition, total doses of local anesthetics used by the surgical team were inconsistently recorded and could not be reported. In addition, formal pain and discomfort scores were not recorded for all patients. As AC surgery is relatively uncommon, there were insufficient data available prior to 1 March 2012 to make a comparison with non-dexmedetomidine anesthesia in our convenience sample.

Conclusions

Awake craniotomy for tumour resection using a dexmedetomidine-based anesthetic and scalp blocks resulted in no airway complications or conversion to general anesthesia. The anxiolytic and analgesic properties of dexmedetomidine enabled patients to remain awake and be surveilled during tumour resection. Further evaluations of clinical outcomes associated with this approach could include the duration of disease-free survival, time to tumour recurrence, and overall quality of life.