Four years after its clinical introduction in the United States and most other countries, the safety profile of sevoflurane, fluoromethyl 2,2,2-trifluoro-1-[trifluoromethyl] ethyl ether, has proven comparable to that of isoflurane, as documented in numerous clinical studies (1–4) and by experience with over 45 million administrations worldwide without an increased incidence of complications. However, a few investigators (5–7) continue to question its potential of causing renal toxicity for two reasons: first, because it reacts with carbon dioxide absorbents to produce the vinyl ether, CH2F-O-C(=CF2)(CF3), known as Compound A, and, second, because it is biotransformed to inorganic fluoride (F−). In this study, the effects of sevoflurane on renal function, as evidenced by postoperative changes in serum creatinine and blood urea nitrogen (BUN) concentrations, were studied in 22 different controlled, clinical trials, which included 3436 surgical patients anesthetized with either sevoflurane or one of three reference anesthetics. These data were retrospectively analyzed and are presented in this report.
Entrance criteria and the protocol were consistent among the 22 clinical trials that compared sevoflurane to isoflurane (17 trials), propofol (3 trials), or enflurane (2 trials). Eligible patients had no history or laboratory evidence of renal disease (serum creatinine <1.5 mg/dL) in all but three trials, in which 161 patients with renal impairment (serum creatinine ≥1.5 mg/dL) were studied. All female patients were tested and determined not to be pregnant or lactating. Patients undergoing preoperative treatment with known hepatic enzyme-inducing agents or drugs that alter anesthetic requirements were excluded from the study. Also excluded were patients with a history of hypersensitivity or an unusual response to other halogenated anesthetics, a history or familial presence of malignant hyperthermia, or a history of alcohol or drug abuse within the previous year. Additional conditions that precluded taking part in one of the 22 trials were previous sevoflurane administration, general anesthesia of any sort within 7 days before the scheduled surgical procedure, and treatment with any investigational drug within the previous 28 days. The study was approved by the institutional human investigation committee at each site, and every subject provided written, informed consent.
Premedication was used in all trials, and the most commonly used was IV midazolam. Anesthesia was induced with IV drugs according to local practice, except in two trials in which 118 patients underwent induction with sevoflurane via a face mask. Neuromuscular blocking drugs were used to facilitate endotracheal intubation and for muscle relaxation during operation; reversal drugs were administered as required. For maintenance of anesthesia, patients were administered either a volatile anesthetic by using a specifically calibrated vaporizer or IV propofol, in both cases in conjunction with a fresh gas mixture containing up to 70% nitrous oxide. Either a nonrebreathing or rebreathing anesthesia circuit was used, the latter with soda lime or baralyme for carbon dioxide (CO2) absorption. Total fresh gas flow (FGF) rates ranged from 1 L/min to 10 L/min. Electrocardiogram, end-tidal CO2 concentration, pulse oximetry, and vital signs were monitored during operation. Clinical laboratory profiles (hematology, chemistry, and urinalysis) were performed before and after operation.
Serum creatinine and BUN were evaluated by using standardized definitions for clinically significant changes from baseline (8,9). Samples generally were obtained in the morning, each day after operation while the patient was hospitalized. A “clinically significant increase” in serum creatinine was defined according to the criteria of Hou et al. (8) as an increase of ≥0.5, ≥1.0, or ≥1.5 mg/dL, for patients with baseline creatinine values ≤1.9, 2.0–4.9, or ≥5.0 mg/dL, respectively. A significant increase in BUN was defined as an increase 25% greater than the upper limit of normal for patients with normal or low baseline values or an increase 25% greater than baseline for patients with preoperative BUN values greater than normal (9).
Raw data from each patient were analyzed rather than the mean data from each of the 22 trials. All statistical analyses were performed by using Statistical Analysis Systems procedures (GLM, GREQ, and CATMOD; SAS Institute, Inc., Cary, NC). All statistical tests were two-tailed and P values ≤0.05 were considered statistically significant. Minimum alveolar anesthetic concentration (MAC) equivalents in 100% oxygen were 2.05% for sevoflurane, 1.15% for isoflurane, and 1.68% for enflurane. Laboratory data were summarized by shift tables and compared by using Fisher’s exact test. In addition, for each laboratory variable, the change from baseline, i.e., from the last preoperative to the first postoperative value, was analyzed by using analysis of variance to compare treatment groups and subgroups. Results of 3 of the 22 clinical trials that included data for 122 patients have already been published (4,10,11).
A total of 3,468 adult patients were enrolled and assigned to receive either sevoflurane (1941 patients) or one of the control drugs (1495 patients): isoflurane (1210 patients), enflurane (111 patients), or propofol (262 patients), as the maintenance anesthetic. However, data from 32 control patients were excluded because they received an agent other than isoflurane, enflurane, or propofol. Patient age ranged from 17 to 93 yr, with the majority, 82%, between 18 and 65 yr. Fifty-four percent of patients were female. Eighty-four percent were white, 10% were black, 1% were Asian, and 5% were other races. Eighty-two percent of patients were ASA physical status I or II, 16% physical status III and 2% physical status IV. Patients commonly underwent musculoskeletal system procedures (e.g., amputations, joint replacement, reductions, fusions), female genital system procedures (e.g., hysterectomy, oophorectomy), or digestive system procedures (e.g., cholecystectomy, bowel excision/resection). Duration of anesthesia ranged from less than 0.5 h to approximately 11 h. Up to 15.1 MAC/h of volatile anesthetic was administered with most (97%) exposed to less than 4 MAC/h. Propofol was administered at rates ranging from 33 μg · kg−1 · min−1 to 264 μg · kg−1 · min−1, with patients receiving total doses of 400 to 8620 mg.
Mean changes from baseline in the first postoperative serum creatinine and BUN measurements generally were similar among patients administered sevoflurane and the control drugs (Table 1). However, in the studies that used isoflurane as the control drug, statistically greater mean decreases (P < 0.05) from baseline in serum creatinine and BUN were observed in the sevoflurane group.
The difference in the mean maximum change in serum creatinine for all patients administered either sevoflurane or the three control drugs approached statistical significance (P = 0.08), with patients administered sevoflurane experiencing smaller changes (Table 2). There was a significant treatment effect when groups were evaluated based on preoperative creatinine levels. Patients treated with sevoflurane experienced smaller increases (P = 0.024) than did patients treated with the comparative drugs. Similarly, patients receiving antibiotics who received sevoflurane had smaller mean maximal increases from baseline (treatment, P = 0.002; subgroup, P = 0.034; treatment/subgroup interaction, P = 0.013) than did those administered comparative drugs. FGF rate influenced the change in serum creatinine, with smaller changes observed at flows >2 L/min than at flows ≤2 L/min. However, there was no treatment effect, i.e., no difference between patients administered sevoflurane and those administered the control drugs. Additionally, treatment effects were not seen, based on the type of anesthesia circuit, the choice of CO2 absorbent, inorganic fluoride level, MAC/h, of anesthesia, duration of anesthesia, use of nitrous oxide, age, or ASA physical status (data not shown). Results were similar for BUN (data not shown).
When all patients were considered, the incidence of “clinically significant increases” in serum creatinine, which takes into account the magnitude of the increase in relation to the baseline value (8), was the same (P = 0.42) among patients administered either sevoflurane or one of the control drugs (Table 3). However, when the baseline creatinine value was ≥1.5 mg/dL, the incidence of clinically significant increase in serum creatinine was higher (P = 0.004) in both treatment groups than when the baseline creatinine value was <1.5 mg/dL. The incidence of clinically significant increases in serum creatinine was not related to type of anesthesia circuit, type of CO2 absorbent, FGF rate, or ASA physical status (data not shown). Results were similar for BUN (data not shown). There was insufficient data to make the other comparisons noted above for the changes in maximal change in serum creatinine.
The favorable anesthetic properties of sevoflurane are well documented, and although no case of an adverse renal outcome attributable only to sevoflurane has been reported, its potential to cause renal toxicity is still argued (5–7). At the core of the issue is how to measure renal function in surgical patients. The literature indicates that for care of surgical patients not in the intensive care unit, the change from preoperative levels in serum creatinine and BUN levels is the most practical predictor of postoperative renal dysfunction. For example, Charlson et al. (12) compared serum creatinine with creatinine clearance values in 278 high-risk, hypertensive, and/or diabetic surgical patients undergoing noncardiac operations. They found that a postoperative increase in serum creatinine of greater than 20% above the preoperative value identified most patients whose creatinine clearance decreased by more than 50%. Among those patients who had postoperative increases in serum creatinine that were sustained for ≥48 hours, more than one third still had evidence of renal impairment when they left the hospital. Another study by Abel et al. (13) looked at more than 20 preoperative risk factors in surgical patients to determine which ones correlated best with acute postoperative renal failure. Increased serum creatinine and BUN and advanced age were the only reliable predictors. A review by Novis et al. (14) arrived at the same conclusion. They assessed 28 studies that included 10,685 surgical patients and examined 30 preoperative risk factors. Only increased serum creatinine and BUN, advanced age, and cardiac failure were considered strongly predictive of postoperative acute renal failure.
In the present analysis, in which serum creatinine and BUN were used to evaluate renal function in surgical patients in 22 clinical trials, patients treated with sevoflurane compared favorably with those treated with the control drugs. Initial postanesthetic serum creatinine and BUN values (Table 1) were significantly lower after sevoflurane than after isoflurane, the “gold standard” of inhaled anesthetics. Furthermore, in all of the other comparisons of serum creatinine between sevoflurane and control drugs (Table 2), i.e., when potentially nephrotoxic drugs were concurrently administered, when patients with high baseline serum creatinine values were examined, and when different FGF rates (including those ≤2 L/min) were used, results from patients administered sevoflurane indicate they fared as well or better than did those patients administered control drugs. It should be noted, however, that within the various subcategories, relatively few patients were studied at the extremes, i.e., only 15% of patients at FGF rates ≤2 L/min, 5% of patients with baseline creatinine ≥1.5 mg/dL, and 3% of patients exposed to 4 MAC/h or more. Finally, when data were examined to determine which group of patients experienced a “clinically significant increase” in serum creatinine values after anesthesia and surgery (Table 3), 1.8% of patients administered sevoflurane had such an increase compared with 2.3% of patients administered control anesthetics. Thus, the present analysis supports the position that sevoflurane does not cause adverse renal effects. In that respect, it is in agreement with the results of every other clinical study that has reported the comparison of preanesthetic to postanesthetic serum creatinine and BUN values.
Why, then, has there been a question that sevoflurane may cause adverse renal effects? The concerns have been two fold. Initially, it was the knowledge that it was biotransformed to F−, and thus, it might cause polyuric vasopressin-resistant nephropathy. However, unlike the experience with methoxyflurane, there was no correlation between peak F− levels after sevoflurane in excess of 50 μM and increased postanesthetic serum creatinine or BUN values. Additionally, in a number of studies of volunteers who received sevoflurane and were tested with vasopressin and/or water deprivation, the ability to concentrate urine was normal (5,15,16). The most convincing of these studies was performed by Eger et al. (5) who administered 1.25 MAC sevoflurane to volunteers for eight hours. Peak F− levels were higher than in any other study (mean, 101 ± 21 μM), but nevertheless, subjects concentrated urine normally when administered vasopressin. The most likely explanation for the difference between methoxyflurane and sevoflurane is that the former is biotransformed to F− in the kidneys and the liver, whereas sevoflurane is significantly biotransformed only in the liver.
The major concern today regarding the renal effects of sevoflurane is that it reacts with the strong bases in CO2 absorbents to form the vinyl ether, pentafluoroisopropenyl fluoromethyl ether (compound A). Compound A is a dose-related nephrotoxin in rats (17), but renal toxicity, as defined by an increase in serum creatinine or BUN, has not been reported in surgical patients. Compound A itself is not nephrotoxic [although this point is disputed (6)]; rather, its toxicity most likely is mediated by renal β-lyase which converts Compound A to potentially nephrotoxic cysteine-S-conjugates (18–21). Renal β-lyase activity in rat renal tissue is approximately 8–30 times greater than in the human kidney (21). It is low renal β-lyase activity that most likely protects humans from the potential toxic effects of Compound A. Further support for this concept comes from a recent study of the effects of Compound A plus sevoflurane in cynomolgus monkeys (22), a species with somewhat more renal β-lyase activity than humans but far less than rats (21). Cynomolgus monkeys exposed to 600 ppm/h of Compound A plus sevoflurane (ppm/h = Compound A level times the duration of exposure) experienced no adverse renal effects and only minimal, transient renal toxicity at Compound A exposures of 800 ppm/h (22). In contrast, 12 of 30 rats exposed to 114 ppm of Compound A (without sevoflurane) for three hours developed some degree of renal necrosis, and transient proteinuria was seen after exposure in most rats exposed to 202 ppm of Compound A for three hours (23). Compound A concentrations in patients which result from exposure to sevoflurane are significantly lower than those used in the rat and monkey experiments cited above (17,22,23). Mean maximal inspired Compound A concentrations in surgical patients at 1 L/min FGF rates range from 8 to 24 ppm when soda lime is the CO2 absorbent and from 20 to 32 ppm when baralyme is used (24). Thus, exposures of more than 20 MAC/h at 1 L/min FGF rates would be safe in humans if we extrapolate the results from the study of monkeys to surgical patients.
Eger et al. (5) assert that serum creatinine and BUN are not sensitive enough indicators of abnormal renal function in patients administered sevoflurane. They reported increased urinary excretion of albumin, glucose, α-glutathione-S-transferase and π-glutathione-S-transferase in volunteers administered 1.25 MAC sevoflurane (but not desflurane) for eight hours. Changes in these urinary markers were transient, peaking on Day 3 and returning to normal by Day 7 after anesthesia administration. There were no changes in pre- to postanesthetic serum creatinine or BUN levels in volunteers anesthetized with sevoflurane nor were there differences in creatinine and BUN when compared with the volunteers anesthetized with desflurane. Eger et al. (5) claim these results are indicative of transient glomerular, proximal tubular, and distal tubular injury. Of great importance, however, is that the results of the study by Eger et al. (5) could not be duplicated by Ebert et al. (25) working at two separate institutions. They used a protocol almost identical to that of Eger et al. (5) to administer sevoflurane to volunteers and found no difference in the preanesthetic to postanesthetic values of the urinary markers nor in serum creatinine or BUN values.
If the study of Eger et al. (5) proves to be reproducible, the basic question it poses is: What is the clinical significance of transient postanesthetic increases in the urinary markers they reported if serum creatinine and BUN remain unchanged? During the last decade, there has been an explosion in the number of urinary tests of renal function, with 61 described at a 1993 symposium in Telfs, Austria (26). It was inevitable that some would be used to evaluate the renal effects of sevoflurane. But do they contribute to our understanding of postoperative renal function if they have not been validated and are not in agreement with the “gold standard” in surgical patients, i.e., pre- to postoperative comparison of serum creatinine levels? Baines (26) commented on the indiscriminate use of renal markers in the “Invited Opinion” at the Telfs symposium: “What are we to do with seven different markers for tubular proteinuria and four different markers for lysosomes that were discussed in this symposium on urinalysis? Adding more markers has rarely, if ever, advanced our understanding of renal pathophysiology, modified treatment, or improved disease outcomes.” Baines (26) noted that of the 61 markers discussed at the Telfs symposium, only progressive, permanent albuminuria in type II diabetics has been validated as a predictor of significant medical renal disease. In that case, validation was accomplished by comparison with a “gold standard,” renal biopsy.
Transient albuminuria in surgical patients, as seen in patients administered isoflurane and sevoflurane (2,3), regardless of the quantity, is different from permanent, progressive albuminuria in patients with diabetes. The fact that changes in urinary markers in surgical patients and in volunteers anesthetized with sevoflurane revert to normal within days and are not associated with postoperative changes in serum creatinine or BUN suggests that they are not indicative of significant renal structural or functional changes. Indeed, it is likely that similar changes in urinary analytes have occurred in surgical patients for many years and have not been recognized. This is suggested by the report of Kharasch et al. (2) who found that the incidence of proteinuria was approximately 30% in postoperative surgical patients administered either isoflurane or sevoflurane. Similar results were reported by Bito et al. (3) and by Obata et al. (27) after low-flow isoflurane and sevoflurane administration, and by Ebert and Arain (28) after low-flow desflurane and sevoflurane administration. To the contrary, a recent paper by Higuchi et al. (7) reported transient albuminuria in patients anesthetized with sevoflurane at low FGF rates, but not after isoflurane. Also, Goldberg et al. (29), in an uncontrolled study of volunteers anesthetized with sevoflurane, observed albuminuria and enzymuria when the dose of Compound A exceeded 240 ppm/h. In none of these studies were there postoperative changes in serum creatinine or in BUN.
In summary, based on the present analysis of serum creatinine and BUN data from 22 well controlled, clinical trials that involved 3436 patients, sevoflurane administration was not associated with evidence of renal toxicity. This was true regardless of the FGF rate at which sevoflurane was administered, when baseline serum creatinine and BUN levels were greater than normal, and when potentially nephrotoxic antibiotics were concurrently administered. The results of this analysis of serum creatinine and BUN are in agreement with all similar measurements of serum creatinine and BUN reported in the literature. Given that sevoflurane has been used world-wide in more than 45 million patients without any increase in reports of renal abnormalities, our results provide additional evidence that serum creatinine and BUN are appropriate indicators for evaluating renal function in postoperative surgical patients.
1. Campbell C, Andreen M, Battito MF, et al. A phase III multicenter open-label randomized comparative study evaluating the effect of sevoflurane versus isoflurane on the maintenance of anesthesia in adult ASA class I, II, and III inpatients. J Clin Anesth 1996; 8:557–63.
2. Kharasch ED, Frink EJ Jr, Zagar R, et al. Assessment of low-flow sevoflurane and isoflurane effects on renal function using sensitive markers of tubular toxicity. Anesthesiology 1997; 86:1238–53.
3. Bito H, Ikeuchi Y, Ikeda K. Effects of low-flow sevoflurane anesthesia on renal function: comparison with high-flow sevoflurane anesthesia and low-flow isoflurane anesthesia. Anesthesiology 1997; 86:1231–7.
4. Conzen PF, Nuscheler M, Melotte A, et al. Renal function and serum fluoride concentrations in patients with stable renal insufficiency after anesthesia with sevoflurane or enflurane. Anesth Analg 1995; 81:569–75.
5. Eger EI, Koblin DD, Bowland T, et al. Nephrotoxicity of sevoflurane versus desflurane. Anesth Analg 1997; 84:160–8.
6. Martin JL, Laster MJ, Kandel L, et al. Metabolism of compound A by renal cysteine-S-conjugate β-lyase is not the mechanism of compound A-induced renal injury in the rat. Anesth Analg 1996; 82:770–4.
7. Higuchi H, Sumita S, Wada H, et al. Effects of sevoflurane and isoflurane on renal function and on possible markers of nephrotoxicity. Anesthesiology 1998; 88:307–22.
8. Hou SH, Bushinsky DA, Wish JB, et al. Hospital-acquired renal insufficiency: a prospective study. Am J Med 1983; 74:243–8.
9. Berndt WO. Use of renal function tests in the evaluation of nephrotoxic effects. In: Hook JB, ed. Toxicology of the kidney. New York: Raven, 1981: 1–29.
10. Frink Jr EJ, Ghantous HN, Malan TP, et al. Plasma inorganic fluoride with sevoflurane anesthesia: correlation with indices of hepatic and renal function. Anesth Analg 1992; 74:231–5.
11. Malan TP. Sevoflurane and renal function. Anesth Analg 1995; 81:S39–45.
12. Charlson ME, MacKenzie CR, Gold JP, Shires T. Postoperative changes in serum creatinine: when do they occur and how much is important? Ann Surg 1989; 209:328–33.
13. Abel RM, Buckley MJ, Austen WG, et al. Etiology, incidence and prognosis of renal failure following cardiac operations. J Thoracic Cardiovasc Surg 1976; 71:323–33.
14. Novis BK, Roizen MF, Aronson S, Thisted RA. Association of preoperative risk factors with postoperative acute renal failure. Anesth Analg 1994; 78:143–9.
15. Munday IT, Stoddart PA, Jones RM, et al. Serum fluoride concentration and urine osmolality after enflurane and sevoflurane anesthesia in male volunteers. Anesth Analg 1995; 81:353–9.
16. Frink Jr EJ, Malan TP, Isner RJ, et al. Renal concentrating function with prolonged sevoflurane or enflurane anesthesia in volunteers. Anesthesiology 1994; 80:1019–25.
17. Gonsowski C, Laster M, Eger E, et al. Toxicity of compound A in rats: effect of increasing duration of administration. Anesthesiology 1994; 80:566–73.
18. Iyer RA, Baggs RB, Anders MW. Nephrotoxicity of the glutathione and cysteine s-conjugates of the sevoflurane degradation product 2-(fluoromethoxy)-1,1,3,3,3-pentafluoro-1-propene (compound) in male Fischer 344 rats. J Pharmacol Exp Ther 1997; 283:1544–51.
19. Kharasch ED, Thorning D, Garton K, et al. Role of renal cysteine conjugate beta-lyase in the mechanism of compound A nephrotoxicity in rats. Anesthesiology 1997; 86:160–71.
20. Dekant W, Vamvakas S, Anders MW. Formation and fate of nephrotoxic and cytoxic glutathione s-conjugates: cysteine conjugate B-lyase pathway. Adv Pharmacol 1994; 27:115–62.
21. Iyer RA, Anders MW. Cysteine conjugate beta-lyase-dependent biotransformation of the cysteine s-conjugates of the sevoflurane degradation product compound A in human, nonhuman primate, and rat kidney cytosol and mitochondria. Anesthesiology 1996; 85:1454–61.
22. Mazze RI, Friedman M, Delgado-Herrera L, et al. Renal toxicity of compound A plus sevoflurane in non-human primates [abstract]. Anesthesiology 1998; 89:A490.
23. Keller KA, Callan C, Prokocimer P, et al. Inhalation toxicity study of a haloalkene degradant of sevoflurane, compound A (PIFE), in Sprague-Dawley rats. Anesthesiology 1995; 83:1220–32.
24. Bito H, Ideda K. Long-duration, low-flow sevoflurane anesthesia using two carbon dioxide absorbents. Anesthesiology 1994; 81:340–5.
25. Ebert TJ, Frink Jr EJ, Kharasch ED. Absence of biochemical evidence for renal and hepatic dysfunction after 8 hours of 1.25 minimum alveolar concentration of sevoflurane anesthesia in volunteers. Anesthesiology 1998; 88:601–10.
26. Baines AD. Strategies and criteria for developing new urinalysis tests. Kidney Int 1994; 46(Suppl 47):S137–41.
27. Obata R, Bito H, Goroku M, Sato S. The effects of prolonged low-flow sevoflurane anesthesia on renal and hepatic function [abstract]. Anesthesiology 1999; 91:A406.
28. Ebert TJ, Arain SR. Renal effects of low-flow anesthesia with desflurane and sevoflurane in patients [abstract]. Anesthesiology 1999; 91:A404.
29. Goldberg ME, Cantillo J, Gratz I, et al. Doses of compound A, not sevoflurane, determines changes in the biochemical markers of renal injury in healthy volunteers. Anesth Analg 1999; 88:437–45.