Clinical Sevoflurane Metabolism and Disposition: II. The Role of Cytochrome P450 2E1 in Fluoride and Hexafluoroisopropanol Formation

Twenty-two nonsmoking ASA physical status 1 or 2 patients undergoing anesthesia for elective surgery with anticipated duration of 3-5 h were studied. The investigational protocol was approved by the Institutional Human Subjects Committee, and all patients provided written informed consent. Eligibility and exclusion criteria are described in the accompanying article. [4]Most patients were evaluated and enrolled before the day of surgery, randomized to control or disulfiram groups at that time, and admitted to the hospital the morning of surgery. A few subjects were inpatients. Patients were randomized by blocks to receive 500 mg disulfiram orally at bedtime on the evening before surgery (n = 12) or nothing (control, n = 10). Patients in the control group also are described in the accompanying article. [4]Patients in the disulfiram group were provided disulfiram at the time of enrollment and instructed to ingest the drug at 10 PM the evening before surgery.

All patients received a standardized general anesthetic designed to minimize potential drug interactions other than the desired test interaction. An indwelling venous catheter, used for blood sampling, was placed in the arm contralateral to the arm used for intravenous fluid administration. Patients received midazolam (1 mg) intravenously after completion of the disulfiram breath test (see below). Anesthesia was induced with propofol (1-2.5 mg/kg) and fentanyl (50-100 micro gram), and succinylcholine was administered to facilitate tracheal intubation. Immediately after the trachea was intubated, anesthesia was maintained with sevoflurane (2.7% end-tidal concentration) in oxygen (2-5 l/min). Inspired and end-tidal sevoflurane concentrations were monitored continuously (Capnomac, Datex Medical Instrumentation, Tewksbury, MA). The inspired sevoflurane percentage was titrated to maintain the desired end-tidal sevoflurane concentration of 2.7% (1.3 MAC). Sevoflurane MAC (2.05 end-tidal percent [24]) was not adjusted for age. Use of muscle relaxants was avoided when possible. When muscle relaxation was necessary, patients received atracurium. No neuraxial local anesthetics or opioids were used intraoperatively. Sevoflurane 2.7% end-tidal concentration was maintained for 3 h, unless the duration of surgery was less than 3 h. For operations lasting longer than 3 h, the sevoflurane was discontinued after 3 h, total gas flows were increased to 6 l/min, and anesthesia was maintained with propofol, nitrous oxide, and fentanyl. Nitrous oxide was not started until the end-tidal sevoflurane concentration had decreased to less than 0.2%. Inspired and end-tidal sevoflurane concentrations were recorded at 15-min intervals while patients were intubated.

Venous blood samples for determination of blood sevoflurane concentration and plasma fluoride and HFIP concentrations were obtained at the times described previously. [4]Urine for determination of fluoride concentration was obtained before induction and for the following consecutive intervals beginning at the start of sevoflurane anesthesia: 0-12, 12-24, 24-36, 36-48, 48-72, and 72-96 h (or until the time of hospital discharge). Urine was thoroughly mixed, the volume was measured, and an aliquot was frozen at -20 degrees Celsius for later analysis.

Clinical evaluation included intraoperative and recovery room hemodynamics and laboratory evaluation (hematology, clinical chemistry, and urinalysis) performed before the study, baseline before anesthesia, 24 h after sevoflurane, and 96 h after sevoflurane (or before hospital discharge). Clinical evaluations also included three recovery indexes: emergence, command response, and orientation time. Emergence was defined as the time from the end of surgery to eye-opening in response to verbal command. Command response was the time from the end of surgery until the patient squeezed the observer's hand in response to command or demonstrated purposeful movement. Orientation time was recorded when the patient stated their name, birth date, and age. Clinical recovery indexes were judged by an independent observer who was not an investigator.

Compliance with disulfiram ingestion was assessed preoperatively before midazolam administration using a breath test for exhaled carbon disulfide. [25,26]Disulfiram is rapidly metabolized after absorption. After a single dose, major metabolites include carbon disulfide, diethyldithiocarbamate, and diethylamine. [27]Carbon disulfide is detectable in plasma, urine, and exhaled breath, and diethylamine is detected in plasma and urine. Concentrations of these metabolites peak approximately 12 h after disulfiram administration, permitting assessment of compliance 10 h after disulfiram dosing. Patients exhaled into two gas washing bottles connected in series. The first bottle contained phenolphthalein and 15 ml of 75 mM NaOH to trap carbon dioxide, and the second bottle contained 15 ml of modified McKees solution ((w/v) 5% diethylamine, 5% triethanolamine, 0.002% copper (II) acetate, and 10% isooctane in ethanol) to trap carbon disulfide. Patients exhaled until 1.1 mEq of expired carbon dioxide had been collected, indicated by a change in the first solution from pink to colorless. An unequivocal yellow color in the second bottle, due to copper diethyldithiocarbamate, was indicative of disulfiram ingestion and adequate absorption.

Eleven patients taking disulfiram had a positive breath test, and one disulfiram-treated patient had a negative breath test. Subsequent urine and blood tests for disulfiram metabolites were negative in this patient, who had severe diarrhea and, therefore, was excluded from the data analysis. No control patient exhibited a positive disulfiram breath test. Thus, data from 10 control and 11 disulfiram-treated patients were analyzed.

Analytical methods are described in the accompanying article. [4]Briefly, total HFIP (unconjugated HFIP and HFIP-glucuronide) in plasma and urine and blood sevoflurane concentrations were determined by gas chromatography with headspace sampling and flame ionization detection. Plasma and urine fluoride concentrations were measured using a fluoride-selective electrode.

Anesthetic dose was calculated as the product of end-tidal sevoflurane concentration (expressed in MAC, uncorrected for age, where MAC = 2.05%) and time, determined in 2-min intervals until 10 min after incision and every 15 min thereafter, with total exposure expressed in MAC-hours. Sevoflurane dose also was calculated from total pulmonary anesthetic uptake and expressed in millimoles. [4].

Patients' demographic data, recovery indexes, and peak plasma fluoride and HFIP concentrations were analyzed by Student's unpaired t test. Blood sevoflurane concentrations, plasma fluoride, and HFIP concentrations and urine fluoride and HFIP excretion in the two groups were compared by two-way repeated-measures analysis of variance. Net plasma fluoride and HFIP concentrations were obtained by subtracting the pre-anesthetic baseline value from all subsequent values. Results are expressed as the mean plus/minus SE.

Patient demographic data are provided in Table 1. Control and disulfiram-treated groups were similar with respect to age, weight, sex, duration of surgery, and surgical blood loss. Propofol doses were similar in both groups (Table 2). Three methods were used to determine the dose of sevoflurane delivered: end-tidal sevoflurane concentrations, calculated total pulmonary uptake, and blood sevoflurane concentrations (Table 2). All three methods showed that the sevoflurane dose was similar in both groups. Total sevoflurane dose was 3.7 plus/minus 0.1 MAC-h in both control and disulfiram-treated patients, respectively (P > 0.05). Total pulmonary uptake was 88.8 plus/minus 9.1 mmol in control patients and 98.8 plus/minus 8.1 mmol in disulfiram-treated patients, respectively (P > 0.05). Blood sevoflurane concentrations in control and disulfiram-treated patients were not significantly different at any time, and areas under the curves were not significantly different (Figure 1).

Disulfiram treatment significantly diminished fluoride production, as assessed by plasma fluoride concentrations and urinary fluoride excretion. Measured plasma fluoride concentrations increased from a baseline of 2.1 plus/minus 0.3 micro Meter before anesthetic induction to a peak concentration of 36.2 plus/minus 3.9 micro Meter (range 23.3-61.5) in control patients (Figure 2). In disulfiram-treated patients, fluoride concentrations increased from 1.7 plus/minus 0.2 micro Meter (baseline) to a peak concentration of 17.0 plus/minus 1.6 micro Meter, which was significantly different from control patients (P < 0.05). Net peak plasma fluoride production, measured as the difference between peak and preanesthetic fluoride concentrations, was 34.0 plus/minus 3.9 micro Meter in control patients but significantly less (15.4 plus/minus 1.5 micro Meter) in disulfiram-treated patients (P < 0.05). Areas under the net plasma fluoride concentration-time curves were decreased to 22% of control patients by disulfiram pretreatment, from 1,110 plus/minus 160 to 241 plus/minus 72 micro Meter *symbol* h. The time to peak plasma fluoride concentration was significantly shortened by disulfiram pretreatment (Figure 2). In contrast to control patients, in whom peak plasma fluoride concentrations occurred 2.0 plus/minus 0.4 h after the end of sevoflurane administration, plasma fluoride concentrations in disulfiram-treated patients peaked almost immediately (0.1 plus/minus 0.1 h) after cessation of sevoflurane (P < 0.05).

Urine fluoride excretion was similarly significantly diminished in disulfiram-treated patients. Urine fluoride excretion by disulfiram-treated patients was significantly less than that by control patients during all four 12-h collection periods (i.e., days 1 and 2) after sevoflurane administration (Figure 3). There was a trend toward similarly diminished urine fluoride excretion in disulfiram-treated patients on days 3 and 4. However, the differences were not statistically significant. Cumulative 96-h fluoride excretion in disulfiram-treated patients was 1,080 plus/minus 210 micro mol, compared to 3,950 plus/minus 560 micro mol in control patients (P < 0.05). The mean decrease in urinary fluoride excretion (73%) in disulfiram-treated patients was similar to the decrease in area under the curve of net plasma fluoride concentration versus time (78%).

Disulfiram treatment also significantly attenuated production of HFIP, the other major metabolite of sevoflurane, as measured by plasma HFIP concentrations and urinary HFIP excretion. HFIP was undetectable in preanesthetic baseline samples in both patient groups. Whereas peak total plasma HFIP concentration was 39.8 plus/minus 2.6 micro Meter in control patients, peak plasma HFIP concentration was decreased to only 14.4 plus/minus 1.1 micro Meter by disulfiram pretreatment (Figure 4). Areas under the plasma HFIP concentration-time curves were decreased to 20% of control patients by disulfiram pretreatment, from 1,370 plus/minus 110 to 268 plus/minus 88 micro Meter *symbol* h. Whereas HFIP concentrations peaked almost immediately (0.6 plus/minus 0.4 h) after the end of sevoflurane anesthesia in disulfiram-treated patients, peak HFIP concentrations occurred 5.6 plus/minus 0.8 h after sevoflurane exposure in control patients.

Urine HFIP excretion also was markedly reduced by disulfiram pretreatment. Urine HFIP excretion (Figure 5) by disulfiram-treated patients was significantly different from that by control patients for the first four 12-h collection periods. There was a trend toward similarly diminished urine HFIP excretion in disulfiram-treated patients on days 3 and 4, but the differences were not statistically significant. Cumulative 96-h HFIP excretion in disulfiram-treated patients was 960 plus/minus 240 micro mol compared to 4,300 plus/minus 540 micro mol in control patients (P < 0.05). The mean decrease in urinary HFIP excretion (78%) in disulfiram-treated patients was similar to the decrease in area under the curve of net plasma fluoride concentration versus time (80%).

Clinical indexes of recovery were not affected by disulfiram pretreatment. Times from the end of surgery to eye-opening (emergence), response to command, and orientation were not different for disulfiram-treated patients compared with control patients (Table 3). Recovery times were longer than those published previously [7,28]because patients received sevoflurane at 1.3 MAC without decrement until the end of surgery, or received propofol after sevoflurane.

Patient pretreatment with a single dose of disulfiram before anesthesia resulted in significant inhibition of sevoflurane metabolism, evidenced by substantial (73-80%) reductions in plasma fluoride and HFIP concentrations and urinary fluoride and HFIP excretion. This effect of disulfiram could not be attributed to differences in sevoflurane dose, which was similar in both groups, as indicated by end-tidal sevoflurane concentrations, total pulmonary sevoflurane uptake, and blood sevoflurane concentrations. Rather, the data demonstrate that differences between groups were due to disulfiram inhibition of sevoflurane metabolism. Disulfiram inhibition of sevoflurane metabolism suggests that cytochrome P450 2E1 is a predominant P450 isoform catalyzing human sevoflurane metabolism in vivo.

We have shown previously that disulfiram is an effective inhibitor of human P450 2E1 activity in vivo. [22]A single oral dose of disulfiram (500 mg) administered 10 h before ingestion of chlorzoxazone, used as a specific noninvasive probe of hepatic P450 2E1 activity, [29]significantly diminished P450 2E1 activity in vivo, evidenced by an 85% decrease in chlorzoxazone 6-hydroxylation. Single-dose disulfiram administered the evening before surgery also significantly inhibited enflurane defluorination in patients receiving 3.9-4.1 MAC-h enflurane, as assessed by changes in plasma fluoride concentrations and urinary fluoride excretion. [23]Peak plasma fluoride concentrations were 24.3 plus/minus 3.8 micro Meter in untreated patients, whereas disulfiram treatment abolished the rise in plasma fluoride concentration. Fluoride excretion in disulfiram-treated patients was 62 plus/minus 10 and 61 plus/minus 12 micro mol on days 1 and 2, respectively, compared to 1,094 plus/minus 185 and 1,196 plus/minus 223 micro mol, respectively, in control patients.

Although the effectiveness of P450 2E1 inhibition in vivo by single-dose disulfiram has been established [22]and the selectivity of disulfiram toward P450 2E1 has been shown in vitro, [30]the absolute specificity of single-dose disulfiram toward P450 2E1 in vivo has not been demonstrated. Diethyldithiocarbamate, the active inhibitory metabolite of disulfiram, can inhibit in vitro human liver microsomal coumarin 7-hydroxylation, which is catalyzed by P450 2A6. [31]The influence, however, of disulfiram on human coumarin 7-hydroxylation in vivo has not been established. Furthermore, P450 2E1 is a major component of the P450s in the liver, whereas P450 2A6 represents less than 1% of total hepatic P450. [32]Thus, disulfiram-inhibitable metabolism is attributed predominantly but not necessarily exclusively to P450 2E1. Disulfiram inhibition of sevoflurane metabolism in vivo strongly supports a role for P450 2E1 but does not exclude the possible participation of other P450 isoforms.

Disulfiram inhibition of P450 2E1 activity diminished but did not abolish the metabolism of sevoflurane. Disulfiram-treated patients showed increases in plasma fluoride and HFIP concentrations compared with baseline values, and fluoride and HFIP were excreted in urine. The fluoride production observed in disulfiram-treated patients cannot be attributed primarily to alternative pathways of sevoflurane metabolism, further metabolism of HFIP to fluoride, or fluoride production from sources other than sevoflurane (i.e., sevoflurane compound A). [33,34]This is because fluoride and HFIP production were comparable in disulfiram-treated patients, HFIP is not metabolized to fluoride by human liver microsomes,** HFIP elimination was not altered significantly by disulfiram pretreatment (HFIP clearance was 52.6 plus/minus 6.1 and 71.0 plus/minus 8.9 ml/min in control and disulfiram-treated patients, respectively), and HFIP is not an expected metabolite of compound A.

The degree of disulfiram inhibition of sevoflurane metabolism (73-80%) is consistent with that of chlorzoxazone hydroxylation (85%) [22]but contrasts with the complete inhibition of enflurane metabolism observed previously. [23]The reason for the difference between clinical enflurane and sevoflurane metabolism is not immediately apparent, particularly because in vitro microsomal metabolism of both agents is apparently similar. [21]One possible explanation might be subtotal inhibition of P450 2E1 activity by disulfiram. This incomplete inhibition would be more manifest at the higher intrinsic rate of sevoflurane, compared with enflurane, metabolism. [21,35]Alternatively, P450 isoforms other than P450 2E1 that are not inhibited by disulfiram may contribute to sevoflurane but not enflurane metabolism. Nevertheless, it appears that P450 2E1 is a predominant P450 isoform catalyzing human sevoflurane metabolism in vivo.

P450 2E1 participation in human sevoflurane metabolism in vivo mirrors the role of P450 2E1 in human liver microsomal sevoflurane defluorination in vitro. [21]These data provide validation for human microsomal sevoflurane metabolism in vitro as a model for sevoflurane biotransformation in vivo.

In addition to providing in vitro-in vivo correlations of metabolism, the current identification of P450 2E1 participation in human sevoflurane metabolism provides a mechanistic basis for several clinical observations regarding sevoflurane metabolism. [21]For example, Higuchi et al. reported that sevoflurane defluorination was significantly greater in obese than nonobese patients, [11]whereas Frink et al. found no relationship between obesity and sevoflurane metabolism. [12]Higuchi et al. suggested that the high incidence of hepatic fatty infiltration in their obese patients may account for the observed difference. Identification of the role of P450 2E1 in clinical sevoflurane metabolism corroborates this contention. Hepatic P450 2E1 content and P450 2E1-dependent anesthetic metabolism are substantially higher in livers with fatty infiltration compared to normal livers.*** The individual heterogeneity in sevoflurane metabolism observed currently and previously [1,5-10]can be explained partially by the population variability in hepatic P450 2E1 activity. Several investigations have found a broad distribution in hepatic P450 2E1 activity exclusive of the effect of inducers or inhibitors. [21,36-38].

In addition to the known effects of P450 2E1 inducers such as obesity and fatty liver infiltration, other P450 2E1 inducers, including isoniazid, chronic ethanol consumption, untreated diabetes, and prolonged fasting, would be predicted to likewise stimulate clinical sevoflurane metabolism. [39]In contrast, other classic P450 inducers, including phenobarbital and phenytoin, which do not increase P450 2E1 activity, would be predicted to have minimal effect on clinical sevoflurane metabolism. Phenobarbital induction had no effect on sevoflurane metabolism as assessed by urinary fluoride excretion. [40].

Inhibition of P450 2E1 activity and sevoflurane metabolism would not be of expected clinical consequence. We observed that clinical indexes of recovery were not influenced by the rate or extent of sevoflurane metabolism. This is consistent with the small extent of sevoflurane metabolism [1,4,41]and demonstrates that metabolism does not play a significant role in terminating the clinical effect of sevoflurane. Furthermore, no alternate pathways of sevoflurane metabolism have been identified toward which sevoflurane might be "switched" if the primary P450 2E1-dependent pathway was inactive.

Thus, even in patients with minimal or no hepatic metabolic capacity, such as those with diminished enzyme activity or with intrinsic liver disease, recovery from sevoflurane anesthesia should not be affected significantly. This has been confirmed clinically, whereby sevoflurane recovery was similar in healthy patients and those with hepatic disease.****

Clinical consequences of P450 2E1 induction of sevoflurane metabolism and increased metabolite formation have not been fully characterized. Obese patients demonstrated significantly greater sevoflurane metabolism than normal patients, with 11 of 15 obese patients exhibiting peak serum fluoride concentrations greater than 50 micro Meter, but there were no abnormalities of renal function. [11]Other investigations have similarly shown no link between sevoflurane-dependent elevations in plasma fluoride concentration and renal dysfunction. [10,42].

In summary, we have shown that P450 2E1 is a predominant cytochrome P450 isoform responsible for clinical sevoflurane metabolism in humans.

The authors thank Douglas Hankins, for technical assistance with the disulfiram breath test. They also thank the various anesthesiology residents and nurses who participated in this investigation.

* Masaki E, Kondou T, Kobayashi K: Relationship between serum F- and cytochrome P-450 after sevoflurane anesthesia in ethanol treated rats. ANESTHESIOLOGY 73:A396, 1990.

** Kharasch ED: Unpublished data, 1995.

*** Thummel KE: Unpublished observations. 1994.

**** Frink E Jr: Personal communication. 1994.