Awake intraoperative mapping is the most reliable method to locate eloquent cortex and associated subcortical tissue during brain resection (1). Mapping the speech cortex was originally developed to allow accurate localization of epileptic foci in the dominant hemisphere with minimum risk of subsequent language disturbance (2). The technique is equally useful for safely maximizing the extent of other brain resections near language and motor centers, especially those for low grade gliomas and vascular malformations (3). A frequently used method is IV sedation and analgesia (4,5). In our experience, the major drawback to this practice has been inadequate analgesia during craniotomy and prolonged sedation interfering with brain mapping. The development of short-acting, titratable, rapidly cleared IV anesthetics and narcotics has allowed the evolution of an “asleep-awake-asleep” anesthetic method (6). This technique allows induction of general anesthesia for patient preparation and scalp, skull, and dural opening followed by rapid emergence for intraoperative brain mapping. Once resection is completed, general anesthesia is resumed. Advocates of this technique believe that such an approach causes marked reduction in patient discomfort and stress and also optimal conditions for brain mapping. Despite these advantages, the asleep-awake-asleep technique requires either airway instrumentation or (as modified in our series) spontaneous ventilation with a variable degree of airway control (see below). The purpose of this retrospective review was to define the efficacy and morbidity associated with the asleep-awake-asleep technique in a series of patients requiring intraoperative awake brain mapping.
The methods used in this chart review were presented to and approved by the Duke University IRB to assure maintenance of patient confidentiality through compliance with the Health Insurance Portability and Accountability Act of 1996.
A retrospective chart review was conducted on all patients who underwent awake intraoperative brain mapping at Duke University Medical Center. Patients having tumors, epileptic foci, or vascular malformations in anatomic locations requiring intraoperative functional mapping were included in the analysis. The initial pool of patients was obtained by searching the anesthesia database for the 27-mo period from 6/26/2000 to 11/7/2002. Search criteria included “craniotomy,” “cortical mapping,” “awake testing,” or “monitored anesthesia care.” One-hundred-thirty-three records were identified. At this stage, records were excluded from the analysis if the drug dosing was ambiguous or if the timed events (i.e., time from discontinuation of drugs to wakeup) were not clearly recorded. This yielded 98 procedures which form the basis of this analysis. The features recorded for each of the 98 procedures are shown in Table 1.
Indications for performing brain mapping included: 1) proximity of the lesion to suspected eloquent brain areas based on anatomical landmarks determined by magnetic resonance imaging (7,8), 2) the likelihood of patient cooperation during surgery, 3) the adequacy of sufficient motor and language function to test at the time of surgery, 4) the anticipation of at least a 90% resection of the lesion, and 5) the ability to tolerate postoperative adjuvant therapy (radiation or chemotherapy) if indicated. Awake mapping was not limited to patients at risk for postoperative language deficits but was also considered for all patients with lesions adjacent to or encroaching on the sensorimotor cortex in either hemisphere.
Each patient was examined preoperatively for baseline neurological function including object naming. All patients received intraoperative large-dose steroids (dexamethasone 10–30 mg or methylprednisolone 50–125 mg) and preoperative anticonvulsant therapy (doses not available).
After induction of general anesthesia, a Mayfield three-pin head holder was used to secure the head. Five to 8 mL of 0.25% bupivacaine was injected subcutaneously at each pin site. When operating over the lateral convexity, patients were placed in the lateral decubitus position with the operative side up and the falx parallel to the floor. The head and trunk were adjusted to maintain an anatomically neutral position to prevent neck discomfort and fatigue. A urinary catheter was inserted with intraurethral 1% lidocaine jelly and all potential pressure points were generously padded. Before scalp incision, 0.5% lidocaine and 0.25% bupivacaine mixed in a 1:1 ratio with 1:1000 epinephrine was injected into all layers of the scalp. Skin incisions and bone flaps were tailored according to lesion location. Once exposed, the dura adjacent to the middle meningeal artery was infiltrated with 1% lidocaine to provide a regional dural block. Intravenous infusions were discontinued after dural incision in preparation for awake intraoperative mapping.
Cortical stimulation mapping was performed using a bipolar electrode with a 5-mm lead separation and a train of biphasic, square wave pulses applied at a frequency of 60 Hz and a single-phase duration of 1 ms (Ojemann cortical stimulator, Radionics, Inc. Burlington, MA; or Model Excel Stimulator, Cadwell Laboratories, Inc., Kennewick, WA). During all cortical electrical stimulation, surface electroencephalography (EEG) was simultaneously monitored using a 1 × 4 strip electrode placed on the brain surface (Ad-Tech Medical Instruments Corp.). Before mapping, the stimulation current was increased incrementally (from 2 to 16 mA) until after-discharge (AD) potentials were observed. Sensorimotor and language mapping then proceeded at a current that generated one to two AD potentials. In cortical regions overlying tumor, the AD threshold was frequently lower, making it necessary to decrease the stimulation current. To increase our confidence that a lack of stimulation-evoked responses indicated non-eloquent cortex, we first established a positive control by localizing primary motor cortex.
During language mapping, the patients engaged in number counting and visual object naming. The exposed cortical surface (including that overlying or infiltrated by tumor) was mapped systematically while patients counted aloud. At each site, electrical stimulation was applied for 2–4 s (2,3). Following the counting task, patients were presented with pictures of objects at a rate of about 1 every 5 s and directed to name the objects beginning with the carrier phrase, “This is a….” Each site was stimulated twice. If a naming error occurred, the area was retested. Essential speech and language sites were identified as those from which stimulation elicited speech arrest or naming errors on three separate trials. Small lettered paper tickets were placed over these sites for reference. Excessive sedation was defined as the inability to perform these tests during cortical mapping.
Subcortical motor mapping to define epileptic foci was performed when resection approached suspected motor fibers or when bipolar cautery stimulated motor responses. When speech arrest, naming errors, weakness, or numbness occurred, resection was suspended, and the surface electroencephalogram was assessed for evidence of seizure activity. If no seizure activity was identified, resection was discontinued. When stimulation elicited seizure activity persisting for more than 10 s as evidenced by surface EEG or behavioral changes, the brain surface was gently irrigated with cooled lactated Ringer’s solution. On completion of resection but before resuming general anesthesia patients were routinely retested for motor and language function.
The conduct of the anesthetic is depicted in Figure 1. The anesthetic technique summarized in this article was not performed as a predefined protocol. However, all patients received bolus and continuous infusions of propofol and remifentanil titrated to achieve general anesthesia. The basic goals of this technique were maintenance of a patent airway and spontaneous ventilation, a rapid and predictable time to recovery, the ability to follow intraoperative verbal commands, and minimal residual drug effect during testing.
The premedication varied from patient to patient and included midazolam, fentanyl, acetaminophen, metoclopromide, and ranitidine. Continuous IV infusions of remifentanil and propofol were started to facilitate insertion of invasive monitoring and bladder catheterization. Depth of anesthesia was assessed by response to these maneuvers and an IV bolus of propofol was given as judged necessary before application of the Mayfield pin head holder. Expired carbon dioxide was continuously monitored in all patients during the asleep phase of the procedure. Expired gas was sampled from nasal cannula, face mask, or nasal trumpet, depending on the mode of oxygen administration. Airway patency was maintained when necessary with the use of nasal trumpets. Using the connector from an endotracheal tube, the trumpet could be connected to the anesthesia circuit to allow gentle positive pressure ventilation when required. The remifentanil and propofol infusion rates were adjusted with the goal of maintaining spontaneous ventilation at a rate of 10 breaths per minute. The infusions were continued until notification by the surgeon that dural opening was about to occur. Infusions were discontinued and the patient was allowed to emerge from anesthesia. After completion of testing, anesthesia was reinduced by restarting the infusion of remifentanil and propofol so as to maintain spontaneous ventilation and absence of motor response to noxious stimuli during wound closure. All physiologic values were automatically recorded from vital signs on an automated anesthesia record keeping system (Saturn Anesthesia Information Management System; Draeger Medical Inc, Telford, PA). The pertinent vital signs (Table 1) were then entered into a deidentified database for analysis.
All data are presented as median (interquartile range) unless otherwise specified. Data were analyzed using JMP statistical software version 3.1.6 (SAS Institute, Inc., Cary, NC). Timed outcome measurements were analyzed using Kaplan-Meier curves. Cox proportional hazards analysis was used to assess the influence of covariates on time-dependent variables. Multiple linear regression was used to evaluate the effect of body mass index (BMI), age, sex, maximum propofol dose, and maximum remifentanil dose on Sao2 or Paco2 while anesthetized.
The majority of patients underwent brain mapping in order to define safe margins for tumor resection (Table 2). There was a slight predominance of male over female patients in our study. Most patients were young to middle aged adults, but the series encompassed children as young as 11 yr and adults as old as 71 yr.
Patients were premedicated with midazolam 1 (1–2) mg. Patients also received fentanyl during the preoperative, intraoperative, and immediate postoperative periods, 100 (50–250) μg. Anesthetic infusion requirements for each patient were summarized by recording maximum and minimum rates of infusion of remifentanil and propofol during the general anesthetic phase before intraoperative wakeup (Table 3).
Respiratory and cardiovascular values are reported in Table 4. Respiratory depression was frequently observed (i.e., Paco2 = 50 [36–69] mm Hg, minimum respiratory rate 0 [0–3] breaths/min, lowest Sao2 95 [92–98]). We recorded at least one 30-s epoch of apnea in 69 of 96 patients. Respiratory rates showed marked variability, and there was a 3% incidence of apparent apnea over the entire course of the drug infusion, as shown in Figure 2. Neither Sao2 nor Paco2 correlated with BMI, age, gender, maximum propofol dose, or maximum remifentanil dose. Hypertensive episodes were generally short and most frequently occurred at Mayfield head pin application and during intraoperative emergence. Maximum systolic blood pressure during the period of pretesting general anesthesia was 150 (139–175) mm Hg. Hypotension was not a significant problem. Minimum systolic blood pressure was 100 (90–120) mm Hg. No patient displayed clinically evident signs of intracranial hemorrhage or ischemia at conclusion of craniotomy.
The durations of each anesthetic infusion, durations of craniotomy before brain mapping, and the time required for emergence to allow brain mapping are shown in Figure 3 and Table 3. The duration of drug infusion, age, sex, weight, maximum propofol infusion rate, maximum remifentanil infusion rate, and diagnosis were not significant determinants of emergence time.
Anesthetic complications are listed in Table 5. Headache was the most frequently noted complication on awakening. A variety of approaches were used to prevent or treat this problem, including preoperative acetaminophen, small-bolus IV doses of fentanyl, or continuation of small-dose remifentanil infusion. Nausea was treated with IV ondansetron. Midazolam was administered for agitation. Applying ice slush irrigation to the surgical field treated seizures. Only two patients were unable to tolerate the awake state intraoperatively and required conversion back to general anesthesia to complete the procedure. No procedure was stopped as a result of cerebral edema. The surgical procedure was completed for all patients.
We found that remifentanil and propofol infusions allowed adequate conditions for intraoperative awake brain functional mapping in 98% of patients. Aside from brief periods of apnea, most patients maintained satisfactory respiratory and cardiac function during anesthetic administration. Intraoperative emergence from anesthesia occurred in a short, predictable period after discontinuation of drug infusions and was not determined by the duration of drug infusion or craniotomy in our series.
Numerous strategies to provide anesthesia for awake craniotomy have been reported (4,6,9–16). The series of awake craniotomy anesthetics published by Archer et al. (4) is the largest to date and serves as a useful benchmark when evaluating a new anesthetic technique for craniotomy requiring awake intraoperative brain mapping. There are many important differences between Archer et al.’s study (4) and the present series. Archer et al. (4) studied patients undergoing craniotomy for excision of seizure foci whereas most of our population required craniotomy for wide local excision of brain tumors. Another difference between these two studies is length of surgery. The mean surgical duration in the present series was 209 minutes versus 578 minutes reported by Archer et al. (4). Thus, complication rates may not be directly comparable. However, in the absence of other evidence, we have qualitatively compared complication rates in the two studies (Table 5). The incidence of intraoperative seizure was more frequent in epilepsy patients. Rates of other complications for which direct comparison can be made appear to be similar. This suggests lack of a meaningful difference in complication rates for the anesthetic techniques. However, randomized comparison between the IV sedation versus asleep-awake-asleep techniques in similar patient populations is necessary to draw this conclusion.
It is difficult to compare the intraoperative wakeup times observed in our population with previous series because most series did not report wakeup times. Bekker et al. (17) reported a single case using dexmedetomidine for awake craniotomy. In this report, a waiting interval of 15 minutes was sufficient to observe changes in the sedation level from a Ramsey score of 6 (asleep with no response) to a Ramsey score of 2 (patient cooperative, oriented, and tranquil). This appears to be substantially longer than the 9-minute median interval to intraoperative wakeup observed in the present study. As both techniques provide sufficient anesthetic conditions for intraoperative awake brain mapping, a randomized comparison between the two strategies seems necessary and appropriate to define an optimal technique.
A major purported benefit of the asleep-awake-asleep technique is patient comfort and lack of awareness during intervals not required for brain mapping. Neither factor was reported in our study or others. Based on our anecdotal experience, we believe that the incidence of awareness was negligible. However, we are unable to comment on the frequency of awareness before intended wakeup when using propofol/remifentanil infusion because of the retrospective design of our analysis. Definition of patient comfort/satisfaction is difficult to assess because most patients are subjected to awake craniotomy only once and thus an appropriate control condition is difficult to propose. However, randomized comparative trials of true awake craniotomy versus the asleep-awake-asleep technique could yield such information. Given the already heavy burden to the patient with brain tumor, we believe that optimization of anesthetic techniques for intraoperative brain mapping should include efforts to address patient perceptions in addition to assessment of major medical complications.
In our study, the effect of infusion duration on wakeup time was not significant. This may have been a result of the fact that the duration of infusion was relatively brief (median infusion duration, 78 minutes). Remifentanil does not accumulate appreciably even after infusions of up to 600 minutes (18). However, it is plausible that longer durations of propofol infusion (i.e., more than 6 hours) might have influenced wake-up time (19). Johnson and Egan (20) reported the use of remifentanil and propofol for awake craniotomy in a case report and performed a computer simulation of effect-site concentrations. Wakeup time was reported at 2 minutes. They used maximum infusion rates of remifentanil and propofol that were approximately half those used in our series. However, their pre-wakeup infusion duration was approximately 3 times the median infusion duration for our series (200 minutes versus 70 minutes). This suggests that remifentanil/propofol infusion may also provide reliable emergence from extended durations of surgery before the requirement for intraoperative wakeup and that infusion rates of propofol could be substantially less than those used in our patients.
Respiratory compromise is possible when using a technique with an unsecured airway. Diminished minute ventilation can be caused by decreased ventilatory drive or loss of airway patency. Despite this, episodes of apnea were brief and infrequent and never required emergency tracheal intubation. Our reporting of apneic epochs (Fig. 2) was extremely sensitive but nonspecific because it included both true 30-second periods of apnea and an unknown number of artifacts resulting from loss of end-tidal CO2 recording. One of the potential weaknesses of our technique was the frequent incidence of hypercarbia. It is not possible to compare the magnitude of respiratory depression between propofol/remifentanil with the IV sedation data reported by Archer et al. (4) because arterial blood gas values were not reported. Although the benefits of hyperventilation during craniotomy have been called into question (21), it is generally accepted that hypercarbia should be avoided to minimize intracranial hypertension and provide optimal operating conditions. We did not measure intracranial pressure or systematically record the surgeons’ satisfaction with operating conditions. However, all patients with space-occupying lesions received furosemide and mannitol immediately after induction of anesthesia. This may explain why none of the 98 patients needed to be awakened prematurely because of the potential effects of hypercapnia on brain bulk.
Brief periods of hypertension were observed in some patients during application of the Mayfield pin head-holder and during intraoperative emergence. In the case of pin head-holder application, this indicates deficiency in assessment of depth of anesthesia or reluctance to provide sufficient anesthesia that might otherwise cause sustained apnea. It seems reasonable to consider use of short-acting vasodilators as prophylactic adjuncts at the time of pin head-holder application in these patients. This strategy may also be beneficial in anticipation of immediate intraoperative wakeup. In both cases, comparison of pharmacologic strategies for improved hemodynamic control during these intervals can be readily assessed with prospective trials.
In conclusion, data were retrospectively collected from 98 patients undergoing craniotomy requiring intraoperative awake functional brain mapping. The combined continuous infusion of remifentanil/propofol sufficient to cause general anesthesia during craniotomy provided satisfactory anesthetic conditions in most patients and allowed a median interval wakeup time of 9 minutes. Airway support other than nasal trumpets was not required but the technique was associated with brief episodes of apnea and transient increases in arterial blood pressure.
The authors acknowledge the participation of faculty neurosurgeons including Michael M. Haglund, MD, PhD and John H. Sampson, MD, PhD. We also thank members of the neuroanesthesia faculty including Randall P. Brewer, MD, Jeffrey M. Taekman, MD, and Ziaur Rahman, MD. We thank Laraine Tuck for expert secretarial assistance.
1. Vinas FC, Zamorano L, Mueller RA, et al. [15O]-water PET and intraoperative brain mapping: a comparison in the localization of eloquent cortex. Neurol Res 1997;19:601–8.
2. Penfield W, Roberts L. Speech and brain mechanisms. Princeton, NJ: Princeton University Press, 1959.
3. Ojemann G, Ojemann J, Lettich E, Berger M. Cortical language localization in left, dominant hemisphere: an electrical stimulation mapping investigation in 117 patients. J Neurosurg 1989;71:316–26.
4. Archer DP, McKenna JM, Morin L, Ravussin P. Conscious-sedation analgesia during craniotomy for intractable epilepsy: a review of 354 consecutive cases. Can J Anaesth 1988;35:338–44.
5. Keifer J. Anesthesia for craniotomy with intraoperative speech testing. Curr Rev Clin Anesth 2001;21:283–90.
6. Huncke K, Van de Wiele B, Fried I, Rubinstein EH. The asleep-awake-asleep anesthetic technique for intraoperative language mapping. Neurosurgery 1998;42:1312–6.
7. Berger M. Preoperative magnetic resonance imaging localization of the primary motor cortex. Perspect Neurologic Surg 1991;2:23–32.
8. Berger M, Cohen W, Ojemann G. Correlation of motor cortex using intraoperative brain mapping data with preoperative magnetic resonance imaging anatomy. J Neurosurg 1990;72:383–7.
9. Sarang A, Dinsmore J. Anaesthesia for awake craniotomy: evolution of a technique that facilitates awake neurological testing. Br J Anaesth 2003;90:161–5.
10. Berkenstadt H, Ram Z. Monitored anesthesia care in awake craniotomy for brain tumor surgery. Isr Med Assoc J 2001;3:297–300.
11. Soriano SG, Eldredge EA, Wang FK, et al. The effect of propofol on intraoperative electrocorticography and cortical stimulation during awake craniotomies in children. Paediatr Anaesth 2000;10:29–34.
12. Taylor MD, Bernstein M. Awake craniotomy with brain mapping as the routine surgical approach to treating patients with supratentorial intraaxial tumors: a prospective trial of 200 cases. J Neurosurg 1999;90:35–41.
13. Herrick IA, Craen RA, Gelb AW et al. Propofol sedation during awake craniotomy for seizures: electrocorticographic and epileptogenic effects. Anesth Analg 1997;84:1280–4.
14. Sartorius CJ, Wright G. Intraoperative brain mapping in a community setting–technical considerations. Surg Neurol 1997;47:380–8.
15. Gignac E, Manninen PH, Gelb AW. Comparison of fentanyl, sufentanil and alfentanil during awake craniotomy for epilepsy. Can J Anaesth 1993;40:421–4.
16. Anesthetic considerations for craniotomy in awake patients. Int Anesthesiol Clin 1986;24:1–190.
17. Bekker AY, Kaufman B, Samir H, Doyle W. The use of dexmedetomidine infusion for awake craniotomy. Anesth Analg 2001;92:1251–3.
18. Egan TD, Minto CF, Hermann DJ, et al. Remifentanil versus alfentanil: comparative pharmacokinetics and pharmacodynamics in healthy adult male volunteers. [erratum appears in Anesthesiology 1996;85:695]. Anesthesiology 1996;84:821–33.
19. Barr J, Egan TD, Sandoval NF, et al. Propofol dosing regimens for ICU sedation based upon an integrated pharmacokinetic-pharmacodynamic model. Anesthesiology 2001;95:324–33.
20. Johnson K, Egan T. Remifentanil and propofol combination for awake craniotomy: case report with pharmacokinetic simulations. J Neurosurg Anesthesiol 1998;10:25–9.
© 2005 International Anesthesia Research Society
21. Gelb AW, Craen RA, Chan M. Hyperventilation does not improve operating conditions during supratentorial craniotomy. J Neurosurg Anesth 2003;15:376(A56).