Comparison of Remifentanil and Fentanyl in Patients Undergoing Craniotomy for Supratentorial Space-occupying Lesions

The protocol for this randomized double-blinded study was approved by the institutional review boards of Duke University Medical Center, the University of Iowa College of Medicine, and Columbia University College of Physicians and Surgeons. All patients provided written informed consent. The study was conducted in 63 adults, who weighed no more than 75% above ideal body weight, were classified as American Society of Anesthesiologists physical status 2 or 3, and were scheduled for elective excision of a supratentorial space-occupying lesion. Patients with clinically relevant preoperative systemic disease or with a known hypersensitivity to opioids were excluded, as were patients who had received opioids within 48 h before surgery. Patients with a history of substance abuse were also excluded. Female patients were either infertile or had a negative result of a urine pregnancy test within 24 h before surgery.

A standard neurologic examination was performed by a senior neurosurgery resident before operation and on postoperative days 1 and 7. The neurologic examination included assessment of mental status, cranial nerve function, motor and sensory systems, cerebellar function, and gait.

A cranial computed tomography or magnetic resonance imaging scan was obtained within 6 weeks before operation. These scans were examined by a senior neurosurgery resident who determined maximum tumor diameter and magnitude of midline shift. Radiographic mass effect was categorized as none, mild, moderate, or severe.

Each patient had a 12-lead electrocardiogram before operation and on the first postoperative day. Routine electrocardiogram interpretations by the hospital attending cardiologist were recorded.

Patients were randomized to receive either remifentanil hydrochloride (25 micro gram/ml Ultiva; Glaxo Wellcome, Research Triangle Park, NC) or fentanyl (50 micro gram/ml Sublimaze; Janssen, Titusville, NJ; or 50 micro gram/ml fentanyl citrate; Abbott Laboratories, North Chicago, IL) as continuous intravenous infusions during anesthesia. After enrollment, opioid infusion syringes were prepared by the hospital pharmacy according to randomization schedules. Study drug dilutions were performed as appropriate to allow administration of the drugs on a milligrams-per-kilogram-per-minute basis so that the investigators and staff were blinded to the contents of the syringes. Different syringes were prepared for each stage of the procedure (induction, maintenance, and emergence). All investigators were blinded to the opioid assignment until clinical and laboratory assessments were completed in all patients.

Any long-term medications that the patients were taking were allowed on the day of surgery. No other premedicants except ranitidine (150 mg given orally or 50 mg given intravenously) or midazolam (up to 0.07 mg/kg given intravenously) were allowed. Before anesthesia was induced, a catheter was inserted into a peripheral vein and approximately 10 ml/kg lactated Ringer's solution was infused. Standard monitors included a lead II electrocardiogram, finger pulse oximeter, and an automatic noninvasive blood pressure cuff. A catheter was placed in the radial artery to continuously monitor blood pressure, to measure arterial blood gases and pH, and to collect samples for pharmacokinetic analysis of remifentanil and its metabolite GR90291 (see below). A second, contralateral, intravenous catheter was placed to administer study drug.

Preinduction hemodynamic values were recorded. Baseline systolic blood pressure (SBP) was defined as the mean of the two lowest values recorded at screening, on the morning of surgery, and immediately before induction of anesthesia. The term “light anesthesia responses” was defined as SBP responses more than 15 mmHg above baseline, heart rate responses more than 90 beat/min persisting for more than 1 min, diaphoresis, or movement. These criteria were used during maintenance anesthesia to justify an increase in the study drug infusion rate.

Pancuronium (up to 2 mg given intravenously) was administered. Anesthesia was induced with thiopental (4–6 mg/kg given intravenously) followed by pancuronium (0.1 mg/kg). Ventilation via mask was begun with 100% oxygen. Fentanyl (50 micro gram/ml) or remifentanil (25 micro gram/ml) was administered from the induction infusion syringe at the rate of 2 micro gram [center dot] kg sup -1 [center dot] min sup -1 (fentanyl) or 1 micro gram [center dot] kg sup -1 [center dot] min sup -1 (remifentanil). This infusion continued until the investigator determined that an appropriate level of anesthesia for tracheal intubation was achieved (target interval, 5 min) or that a maximum infusion interval of 10 min had elapsed. A supplementary dose of thiopental (50–150 mg) could be given at the time of intubation. After intubation of the trachea, mechanical ventilation was begun. An inspired mixture of nitrous oxide and oxygen (2:1) was administered. Ventilation was adjusted to achieve a predicted partial pressure of carbon dioxide in arterial blood (P^aCO2) of 28 mmHg. Vital signs were recorded every minute from onset of study drug infusion until intubation and at 1, 3, and 5 min after intubation.

Immediately after intubation, the induction infusion syringe was replaced with a maintenance infusion syringe. Fentanyl (8 micro gram/ml) or remifentanil (48 micro gram/ml) was administered at an initial rate of 0.03 micro gram [center dot] kg sup -1 [center dot] min sup -1 (fentanyl) or 0.2 micro gram [center dot] kg sup -1 [center dot] min sup -1 (remifentanil).

Study drug boluses or infusion rate increases were used to treat light anesthetic responses as previously defined. Each bolus consisted of 10 ml of either fentanyl (2 micro gram/kg) or remifentanil (1 micro gram/kg) given intravenously. If three study drug boluses were administered for a particular response and a fourth bolus was required, then the fourth bolus had to be followed by an infusion rate increase. Infusion rate increases were made in increments of 0.016 micro gram [center dot] kg sup -1 [center dot] min sup -1 (fentanyl) or 0.2 micro gram [center dot] kg sup -1 [center dot] min sup -1 (remifentanil). A minimum of 1 min was required between boluses and at least 2 min between infusion rate increases. A maximum of five boluses or a maximum infusion rate of 0.06 micro gram [center dot] kg sup -1 [center dot] min sup -1 (fentanyl) or 0.4 micro gram [center dot] kg sup -1 [center dot] min sup -1 (remifentanil) was allowed. If the maximum number of boluses, infusion rate, or both had been administered, then 0.2% isoflurane was added to the inspired inhalational gas mixture. Thereafter, isoflurane could be increased or decreased in 0.2% increments as needed. Throughout surgery, neuromuscular blockade was maintained, as deemed appropriate for the surgical procedure, by administering intermittent boluses of pancuronium.

A SBP less than 80% of baseline, or heart rate less than 45 beats/min persisting for more than 1 min warranted a decrease in the rate of anesthetic administered. If isoflurane had been given, it was discontinued before decreasing the opioid maintenance infusion rate. The maintenance infusion rate was to be decreased in increments of 0.016 micro gram [center dot] kg sup -1 [center dot] min sup -1 (fentanyl) or 0.2 micro gram [center dot] kg sup -1 [center dot] min sup -1 (remifentanil).

Any adverse hemodynamic event that did not respond to changes in anesthetic regimen could be managed at any time with ephedrine, phenylephrine, atropine, labetolol, or esmolol as appropriate.

For pin head-holder placement only, a supplemental dose of thiopental (50–150 mg) could be given before the stimulus. Hemodynamic values were recorded at time of pin placement and 5 min afterward.

After the first burr hole was drilled, the physician ensured that the head was positioned 15 degrees above the horizontal axis. A Gaeltec ICP monitor (Medical Measurements, Hackensack, NJ) was inserted into the epidural space to measure ICP. This measurement was performed at Duke University and the University of Iowa only and only in patients without previous craniotomy (fentanyl n = 16; remifentanil n = 17). For patients at Columbia University, the cerebral blood flow response to P^aCO2was determined during the same interval, the results of which will be reported elsewhere when available. After recording of a stable mean ICP value, an arterial blood sample was collected to measure P^aCO2. Thereafter mannitol was administered if requested by the neurosurgeon. When opening the dura, the surgeon was asked to assess brain volume using a four-point brain relaxation score (1 = excellent, no swelling; 2 = minimal swelling, but acceptable; 3 = serious swelling, no change in treatment required; 4 = severe swelling requiring intervention). [7] 

At bone flap replacement, the opioid maintenance infusion syringe was replaced with a new unlabeled infusion syringe provided by the pharmacy to be used during scalp closure and dressing placement. Patients in the fentanyl group were infused with saline at the previous maintenance infusion rate for fentanyl. Patients in the remifentanil group continued to receive remifentanil at the previous maintenance infusion rate.

Before the end of surgery, labetolol, hydralazine, or both could be given as a prophylactic measures against emergence hypertension. At the end of surgery, residual neuromuscular block was reversed. Once reversal was adequate, both nitrous oxide and remifentanil or saline infusions were discontinued and a timer was started. Thereafter, hemodynamic and respiratory responses were evaluated at 1-min intervals for the first 10 min after discontinuation of nitrous oxide and then at 5-min intervals for as long as 60 min.

Anesthetic recovery was assessed by the investigator using an established two-part scoring system. [7]The first part of the scoring system assessed level of consciousness, orientation, ability to follow commands, motor function, and the presence of agitation. The second used a modified Aldrete score. [8]Patients were considered to have normal scores when alert or arousable to quiet voice, oriented, able to follow commands, had motor function unchanged from their preoperative evaluation, were not agitated, and had modified Aldrete scores of 9 or 10. These measurements were recorded until a normal score was obtained or when 60 min had elapsed after discontinuing nitrous oxide. The investigator also assessed other acute emergence characteristics, including time to return of spontaneous respiration and time to extubation after discontinuation of nitrous oxide. Naloxone administration was allowed, at the discretion of the attending anesthesiologist, at any time during the recovery interval.

The quality of emergence was assessed independently by the neurosurgeon and the investigator, using an identical questionnaire that assessed patient comfort, hemodynamic profile, level of consciousness, and overall quality of emergence (defined as better, similar, or worse than that usually observed for this procedure). Postoperative status after 8 h of recovery was also assessed by the patient's primary nurse in the intensive care unit using a similar questionnaire that addressed level of consciousness and overall quality of recovery (defined as better, similar, or worse than that usually observed for this procedure). Patient recall of intraoperative events and mental clarity were assessed by the investigator on the first postoperative day.

All adverse medical events occurring during the study were recorded, including nausea and vomiting, venous irritation, and muscular rigidity. Phenomena not considered to be adverse events included incisional pain, headache judged by the investigator to not be more severe than generally expected, hemodynamic events that responded to change in study drug or concurrent anesthetic regimen, or prolongation of hospitalization resulting from an intraoperative change in surgical procedure.

At two institutions, serial arterial whole-blood samples were collected to determine fentanyl, remifentanil, and GR90291 (the primary remifentanil metabolite) concentrations at baseline (i.e., before induction of anesthesia), every 2 h during maintenance infusion, and 0.5, 1, 2, 4, 8 and 12 h after completion of surgery. Fentanyl whole-blood concentrations were determined by liquid chromatography-mass spectrometry. GR90291 concentrations were determined by gas chromatography. Remifentanil concentrations were determined by either high-performance liquid chromatographic methods using ultraviolet detection (for concentrations > 1 ng/ml) or gas chromatography-high-resolution mass spectrometry-selected ion monitoring (for concentrations < 1 ng/ml). [9]All intraoperative steady-state remifentanil and fentanyl blood concentrations were used to calculate individual clearance values. The terminal elimination half-life of GR90291 was estimated from the postinfusion concentration-time data using log-linear regression.

An a priori power analysis was performed. Thirty patients per treatment group were anticipated to provide 80% power to detect a clinically significant between-group difference of 3 min in time to verbal command recovery at alpha = 0.05. Cox proportional hazard modeling was used to detect differences and to produce risk ratio estimates in recovery time analyses. Logistic regression analysis was used to detect differences in dichotomous response variables. Fisher's Exact test was performed if logistic regression did not converge. Analysis of variance was used for normally distributed continuous response variables. A probability value less than 0.05 was considered significant. Most parametric values are reported as means +/- SD. Systolic blood pressure values are reported as time-weighted means when stated as such. Time-weighted means were calculated as the area under the curve over the time profile for each interval divided by the duration of the measurement by assuming the trapezoidal rule. Nonparametric values are reported as medians with ranges. Statistical analysis was performed using SAS® System computer software (SAS Institute, Inc., Cary, NC).

Sixty-three patients were enrolled in the study. Twenty patients were studied at both Duke University and the University of Iowa. Twenty-three patients were studied at Columbia University. Equal numbers of patients were given fentanyl or remifentanil at each institution. One patient receiving fentanyl was withdrawn from the study because surgery was aborted after severe hemorrhage occurred during removal of the bone flap and the protocol was abandoned. Values for this patient were included in the analysis through the point of bone flap removal. This complication was judged by the investigator to be unlikely to be related to use of the study drug.

(Table 1) summarizes patient demographics and tumor characteristics. There were no important differences between groups. The most common masses were meningiomas (25%) and gliomas (27%). Seventy-five percent of fentanyl-treated patients and 81% of remifentanil-treated patients had been given steroids in the preoperative period.

Abnormal signs on screening neurologic examination were found in 49% of patients. Cranial nerve and motor and gait disturbances were the most common abnormalities noted. By chance, the frequency of screening neurologic abnormalities was greater in the remifentanil group (65%) than in the fentanyl group (40%). No patient had clinically significant electrocardiogram abnormalities before operation.

Baseline SBP values were similar between treatment groups, although heart rate was less in the patients receiving remifentanil (P = 0.034; see Figure 1). Systolic blood pressure values were similar between groups from the start of study drug infusion until the time of intubation. One patient in the remifentanil group, and none in the fentanyl group, had a SBP less than 80 mmHg during this interval.

Between intubation and 5 min after intubation, SBP (fentanyl = 127 +/- 23 mmHg; remifentanil = 113 +/- 18 mmHg; P = 0.004) and heart rate (P = 0.001) were greater in patients in the fentanyl group. Three patients in the remifentanil and none in the fentanyl group had SBP values less than 80 mmHg during this interval. In contrast, five patients receiving fentanyl and one receiving remifentanil displayed responses suggesting light anesthesia at the time of intubation (P = 0.196; see Table 2). The mean total dose of study opioid administered during the interval from induction to intubation was 13 +/- 3 micro gram/kg for fentanyl and 7 +/- 1 micro gram/kg for remifentanil.

Most patients in both treatment groups had an infusion rate adjustment or received a bolus dose of opioid. Similar numbers of fentanyl-and remifentanil-treated patients experienced light anesthetic responses (Table 2). Despite this, the mean infusion rate of remifentanil remained similar (0.22 micro gram [center dot] kg sup -1 [center dot] min sup -1) to the initial maintenance rate of 0.20 micro gram [center dot] kg sup -1 [center dot] min sup -1. By comparison, the mean infusion rate for fentanyl was increased from the initial infusion rate of 0.03 micro gram [center dot] kg sup -1 [center dot] min sup -1 to 0.05 micro gram [center dot] kg sup -1 [center dot] min sup -1 (66% increase). The last arterial blood sample taken before discontinuation of the maintenance infusion syringe was considered to best reflect steady-state blood concentrations of the two opioids. The concentrations for fentanyl and remifentanil were 3.2 +/- 2.2 ng/ml and 7.4 +/- 5.7 ng/ml, respectively. The median duration of maintenance infusion at this interval was 4 h.

The percentage of patients given isoflurane was similar (fentanyl = 41%, remifentanil = 26%; P = 0.21). Of those receiving isoflurane, patients in the fentanyl group received more (mean dose in fentanyl patients = 0.64 minimum alveolar concentration (MAC)-hours; remifentanil patients = 0.07 MAC-hours; P = 0.04), where 1 MAC-hour = 1.15% isoflurane given for 1 h.

Hypotension (SBP < 80% baseline) or bradycardia (heart rate <45 beats/min) for more than 1 min occurred with a similar frequency in both groups, with hypotension being the most common event. Heart rate was similar during the maintenance period for both groups (Figure 1).

Mean ICP and the range of ICP values were nearly identical in the two groups (Table 3). Five of 16 patients receiving fentanyl (31%) and 3 of 17 patients receiving remifentanil had an absolute ICP more than 20 mmHg. Mean CPP values were also similar. The P^aCO2values were similar between groups when ICP was measured. Brain relaxation was rated as excellent or acceptable in most patients. Severe brain swelling requiring intervention occurred in three patients given fentanyl (9%) and no patients given remifentanil (Table 3).

The median duration of anesthesia was similar between groups (remifentanil = 298 min; fentanyl = 294 min). The mean total dose of fentanyl was 34 micro gram/kg (range, 17–54 micro gram/kg). The mean total dose of remifentanil was 73 micro gram/kg (range, 25–149 micro gram/kg).

(Table 4) summarizes emergence characteristics. The median times to emergence were similar between groups. Twice as many fentanyl-treated patients did not reach a normal recovery score (31% fentanyl, 16% remifentanil; P = 0.15) within 60 min. Seven fentanyl-treated patients received naloxone during this interval. No remifentanil-treated patients required naloxone (P = 0.01). In two fentanyl-treated patients the tracheas were not extubated during the 60-min recovery period due to excessive somnolence.

Systolic blood pressure was greater in the patients in the remifentanil group during the 60-min recovery observation interval (fentanyl = 134 +/- 16 mmHg; remifentanil = 147 +/- 15 mmHg; P = 0.001;Figure 1). As a result, more remifentanil-treated patients were given labetolol (34% fentanyl; 71% remifentanil) or hydralazine (19% fentanyl; 32% remifentanil).

Before discharge from the recovery area, anesthesiologists and neurosurgeons rated the quality of patient emergence. Ratings by anesthesiologists were not different between groups, although there was a trend favoring remifentanil in the overall quality of emergence (P = 0.08). Neurosurgeons rated remifentanil-treated patients as better for patient comfort (P = 0.04), level of consciousness (P = 0.02), and overall quality of emergence (P = 0.01).

The percentage of patients who required analgesics for headache within 8 h after the end of surgery was similar (fentanyl = 50%; remifentanil = 64%; P = 0.22). However, the median time to first analgesic administration occurred earlier in the patients in the remifentanil group (fentanyl = 136 min, remifentanil = 34 min; P = 0.04). Similar numbers of patients experienced nausea and vomiting within 24 h (fentanyl = 57% and 37%; remifentanil = 50% and 30%). No clinically significant changes between screening and postoperative electrocardiogram were noted in either opioid group.

No differences were found for quality of recovery at 8 h after surgery. Mental clarity on the first postoperative day was not different between groups. No remifentanil-treated patients remembered any aspect of the operation. One patient receiving fentanyl reported feeling the last few stitches.

Of the 19 (61%) fentanyl-treated patients who had normal results of screening neurologic examinations, five (26%) were abnormal on the first postoperative day and four (21%) were abnormal on day 7 (Table 5). Eleven (35%) remifentanil-treated patients had normal results of screening neurologic examinations. On the first postoperative day, 2 (18%) of these patients had abnormal results of these examinations, and by day 7 only one (9%) continued to have abnormal results. Most changes between screening and postoperative day 1 occurred in motor testing and cranial nerve function in both treatment groups. Between screening and day 7, most changes occurred in motor testing, gait, and cranial nerve function.

Four serious adverse events occurred. Three of these events were considered to be unrelated to study drug. One fentanyl-treated patient experienced severe hemorrhage while the bone flap was being removed. One fentanyl-treated patient remained unconscious on postoperative day 1. One remifentanil-treated patient experienced significant hypotension and subendocardial infarction in the early recovery period after receiving a bolus dose of labetotol. The remaining event was considered probably related to study drug (emergence delirium occurring in a remifentanil-treated patient).

Individual concentrations of remifentanil and its primary metabolite (GR90291) 30 min after infusion are listed in Table 6. Rapid elimination was evident. The values for GR90291 reflect a much slower half-life. Fentanyl concentrations at 30 min after surgery were 1.5 +/- 0.6 ng/ml.

Important characteristics of an opioid used during elective supratentorial craniotomy include emergence and recovery profiles and effects on CPP. [7,10–17]Remifentanil may be useful in neuroanesthesia because of its ultra-short duration of actions [2–4,18]allowing a more predictable emergence and recovery. With respect to the effects of remifentanil on ICP and CPP, there is only limited information for humans. In one study, patients undergoing elective supratentorial craniotomy were anesthetized with isoflurane and nitrous oxide. Remifentanil (0.5 micro gram/kg or 1 micro gram/kg) was administered as a single bolus infusion after the first burr hole placement. [6]Intracranial pressure and mean arterial pressure changes from baseline were determined in the absence of surgical stimulation. No changes in ICP were observed. Cerebral perfusion pressure was significantly reduced as a result of the effect of remifentanil on mean arterial pressure and occurred in a dose-dependent manner. Effects of remifentanil on mean arterial pressure and CPP were comparable to effects observed for alfentanil (10 micro gram/ kg or 20 micro gram/kg). The results demonstrated sufficient similarity between cerebrovascular effects of remifentanil and existing opioids so as to allow execution of a full-scale trial of remifentanil in a population of patients having neurosurgery.

In the current trial, we chose to compare a remifentanil-nitrous oxide combination with a fentanyl-nitrous oxide combination, each with low-dose isoflurane rescue. The fentanyl-nitrous oxide regimen is used commonly for neurosurgical procedures, and extensive experience with fentanyl exists when studying opioid effects in patients with supratentorial space-occupying lesions. [7,11] 

Before execution of this protocol, we conducted an open-label pilot study consisting of 30 patients receiving remifentanil (ten patients at each institution) in which the full protocol defined in this study was executed. [19]The induction dose of remifentanil was 1 micro gram [center dot] kg sup -1 [center dot] min sup -1 followed by 0.4 micro gram [center dot] kg sup -1 [center dot] min sup -1 according to the manufacturer's recommendations. Of the 30 patients, none required isoflurane during the procedure, indicating that the maintenance dose of remifentanil was likely to yield an opioid effect greater than that seen for fentanyl in previous work using this model. [7,11]Accordingly, the starting remifentanil maintenance infusion rate used in the current study was reduced by one half (to 0.2 micro gram [center dot] kg sup -1 [center dot] min sup -1) from that used during the pilot work.

Measurement of blood opioid concentrations indicates that equianalgesic states were derived from the opioid infusions during the maintenance phase. Because remifentanil is rapidly hydrolyzed, it is more practical to measure whole-blood concentrations. A remifentanil maintenance concentration of 7.4 ng/ml observed in our study has been reported to yield a isoflurane MAC value of approximately 0.3%. [20]The mean whole-blood fentanyl concentration of 3.2 ng/ml observed in our patients must be converted to plasma concentration for comparison. Using a fentanyl whole-blood:plasma concentration ratio of 0.965 (i.e., fentanyl plasma concentration of 3.4 ng/ml), [21]isoflurane MAC would be predicted to be approximately 0.4%[22]; that is, nearly identical to that for remifentanil. Despite this, significantly more isoflurane was used for the fentanyl-treated patients. We speculate that the relatively faster onset of remifentanil occurring after either bolus or increased infusion rate resulted in sufficient attenuation of light anesthesia responses to limit the administration of isoflurane.

The timing and method of ICP measurement in this study had to be acceptable to anesthesiologists, neurosurgeons, and patients. Intracranial pressure and CPP were measured only at the time of first burr hole placement. Thus we cannot comment on cerebral hemodynamic alterations during induction and intubation. Nevertheless, we consider our ICP measurement to be representative of steady-state ICP values between intubation and aural opening. All ICP measurements were made with an epidural pressure transducer device. Although intraventricular and subdural measurements are believed to be the most accurate methods for ICP measurement, measurement with the Gaeltec monitor has compared well in clinical trials. [23] 

Minimizing or preventing brain swelling during neurosurgical procedures is crucial in providing neurosurgeons with ideal operating conditions. Anesthetic techniques that accentuate brain swelling could make operating conditions more difficult and could contribute to neurological injury by increasing the requirement for surgical retraction. Intracranial pressure and CPP values were similar between the remifentanil and fentanyl treatment groups at the time that the first burr hole was placed. Three patients receiving fentanyl and five receiving remifentanil had ICP values greater than 20 mmHg. However, most patients in both treatment groups were considered by the neurosurgeon to have excellent brain relaxation (i.e., no swelling) or minimal swelling at the time of bone flap removal and dural opening. No patient receiving remifentanil required intervention for severe brain swelling.

Patients appeared more deeply anesthetized with remifentanil compared with fentanyl during the induction and intubation intervals. In the fentanyl group, heart rate and SBP were greater and more patients displayed light anesthetic responses during and immediately after intubation. In contrast, remifentanil-treated patients were more likely to have reduced SBP (< 80 mmHg) after intubation. The hemodynamics of both treatment groups were similar during the rest of the study, except during emergence and early recovery. In the initial recovery phase, patients receiving remifentanil had greater SBP and required antihypertensive drugs to be given more frequently. This hyperdynamic response was probably related to inadequate analgesia occurring in the remifentanil group after discontinuation of the infusion. This is supported by the absence of measurable blood concentrations of remifentanil 30 min after completion of surgery (as opposed to 1.5 +/- 0.6 ng/ml for fentanyl). Remifentanil-treated patients required analgesic administration much earlier during recovery than did patients receiving fentanyl. Thus analgesic administration before emergence, or early in the recovery interval, may be important in preventing the hyperdynamic responses we found in the remifentanil-treated patients.

Emergence and recovery from approximately 5 h of anesthesia were generally rapid in both treatment groups. Of those patients who reached a normal recovery score in the first 60 min, the median time to normal recovery was 10 min for both treatment groups. This, however, does not represent the nine (29%) fentanyl-treated and five (16%) remifentanil-treated patients who did not achieve normal recovery during that interval. In contrast, only 10% of the patients in the study by Todd et al. [7]who received a fentanyl/nitrous oxide anesthetic failed to reach a normal recovery score at 60 min; in that study, no patients required naloxone. In our study, seven (23%) of the fentanyl-treated patients required naloxone for excessive sedation compared with none of the remifentanil-treated patients. The prolonged recovery and need for narcotic reversal may be partially explained by the total fentanyl dose administered to patients in our study. Patients in the study by Todd et al. [7]were given a mean total fentanyl dose of 26 micro gram/kg for procedures with a mean duration of 338 min, compared with 34 micro gram/kg for procedures lasting a mean of 294 min in our study.

There did not appear to be a difference in new neurological changes between treatment groups. However, we observed a large variance in results of baseline neurological examinations and neurological outcomes 7 days after operation for both treatment groups. A substantially larger patient population probably would be required to detect small differences for the two opioids with respect to neurological outcome should such differences exist. Side effects were similar, with nausea and vomiting occurring frequently for both opioids. One patient treated with remifentanil displayed a brief but severe episode of emergence delirium. This episode may have been caused by sudden postoperative incisional pain after rapid elimination of remifentanil from the blood. In the remaining patients given remifentanil, postoperative pain was not this severe. However, most patients receiving remifentanil anesthesia should benefit from prophylactic analgesic administration before discontinuation of the infusion.

It is unclear why the quality of emergence and recovery from anesthesia was rated differently between anesthesiologists and neurosurgeons. Neurosurgeons rated remifentanil patients significantly better than usual in terms of patient comfort, level of consciousness, and overall quality of emergence. These differences were not noted by anesthesiologists. Perhaps perceptions of recovery and emergence in patients having neurosurgery were influenced by the observer's immediate clinical goals.

The purpose of conducting pharmacokinetic investigations in this study was to confirm that the pharmacokinetics of remifentanil and GR90291 are the same after prolonged administration as that seen previously in shorter surgical procedures. The clearance of remifentanil and the mean half-life of GR90291 are consistent with previous studies. [2,3] 

In summary, patients undergoing supratentorial craniotomy for space-occupying lesions were given either remifentanil or fentanyl as primary opioids during their anesthetic care. During anesthesia induction, remifentanil had a rapid onset of effect and effectively attenuated the hemodynamic response to intubation. During anesthesia maintenance, ICP and CPP were similar for the two opioids. Depth of anesthesia during this interval was easier to manage with remifentanil and allowed a reduced requirement for isoflurane. The incidence of naloxone use was less in patients given remifentanil. At the same time, emergence hypertension and surgical pain were more frequently observed in the remifentanil group. The frequently of adverse events was similar between groups. Remifentanil appears appropriate for use during elective supratentorial craniotomy. Further studies are needed to devise techniques that will improve emergence and recovery analgesia in these patients and allow extension of our findings to other populations of patients having neurosurgery.