Toxic Products Formed by Interactions with Carbon Dioxide Absorbents

Sevoflurane and Compound A

Several decomposition products are formed during the interaction of sevoflurane with carbon dioxide absorbents, and fluoromethyl-2-2-difluoro-1-(trifluoromethyl) vinyl ether (compound A) is the major degradation product detected^{[250] [251] [252]} ( Fig. 8-17 ). The dehydrofluorination of sevoflurane to form compound A is initiated by soda lime abstraction of a proton from the isopropyl group of sevoflurane in a manner that is similar to the base-catalyzed deprotonation of halothane by soda lime to yield difluorobromochloroethylene (F₂ C=CBrCl, BCDFE). Bito and Ikeda ^[242] studied patients (n = 16) exposed to sevoflurane at low fresh gas flows (1 L/min) with soda lime or Baralyme as the carbon dioxide absorbent. Soda lime produced individual maximum compound A levels of 23.6 ± 2.9 ppm, and Baralyme produced levels of 32 ± 2.3 ppm.^[242] It is now well established that Baralyme is associated with higher compound A production than soda lime.

Figure 8-17 Sevoflurane degradation in the presence of a base.

Compound A has been the subject of intense research and debate since the introduction of sevoflurane into clinical practice in the United States in 1995. Morio and colleagues^[243] found that high concentrations of compound A could cause renal injury and death in rats. Other investigators confirmed the findings of Morio and coworkers and showed that kidney injury occurred when levels of compound A reached 25 to 50 ppm or greater.^[253] Necrosis is found primarily in the proximal tubular cells within the outer strip of the medulla and the percentage of injured cells was lower 4 days after exposure than 1 day after exposure, consistent with repair after renal injury.^[254] These and other studies suggest a threshold for renal injury of 150 to 300 ppm-hours of compound A exposure (i.e., 50 ppm of compound A administered for 3 hours).^{[255] [254]} Keller and associates^[254] reported that compound A inspired concentrations of more than 114 ppm were associated with concentration-dependent increases in BUN and creatinine levels and in the urinary NAG/creatinine ratio. Moderately severe histopathologic changes were seen at 202 ppm, and all chemistry and histopathologic changes reversed within 14 days of exposure. This renal toxicity is dose and time dependent; the higher the exposures over time, the greater is the observed injury. During sevoflurane anesthesia, in a closed or semiclosed system at low fresh gas inflows, patients are routinely exposed to compound A. However, the safe concentration for compound A exposure in humans is unknown.

Several studies enrolling surgical patients and human volunteers have been undertaken to investigate the possible association between sevoflurane anesthesia and renal injury due to compound A exposure during sevoflurane anesthesia. In an early study, Bito and Ikeda^[250] exposed 10 patients to sevoflurane for longer than 5 hours under closed-circuit conditions with soda lime as the absorbent. Peak compound A concentrations were 19.5 ± 5.4 ppm, and serum levels of BUN, creatinine, and electrolytes were essentially unchanged. In a follow-up study by the same investigators, 10 patients were exposed to sevoflurane at 1 L/min fresh gas flows for 10 hours. Mean compound A levels were 24.3 ± 2.4 ppm, and routine tests of renal function showed no change from preoperative values.^[256] Frink and colleagues^[257] measured compound A levels in 16 surgical patients during low-flow sevoflurane anesthesia for 3 hours with soda lime (n = 8) or Baralyme (n = 8) as the absorbent. Maximum compound A concentrations averaged 8.16 ± 2.67 ppm with soda lime and 20.28 ± 8.6 ppm with Baralyme. A compound A concentration of 60.8 ppm was found in the breathing circuit of one of eight subjects to whom sevoflurane was administered using Baralyme. In the same

252

In a more comprehensive investigation, Bito and Ikeda^[258] studied 100 surgical patients undergoing resection of head and neck tumors for which surgery was expected to last longer than 10 hours. Patients received sevoflurane (n = 50) or isoflurane (n = 50) anesthesia at low fresh gas inflows (1 L/min).^[258] The mean peak compound A concentration for the sevoflurane group was 24.6 ± 7.2 ppm. Although bilirubin, AST, and ALT levels increased postoperatively in the sevoflurane and isoflurane groups, there were no significant differences in these parameters between these groups. Preoperative and postoperative BUN and creatinine levels remained unchanged. A number of other reports have evaluated renal function after closed-circuit or low-flow sevoflurane anesthesia using BUN and serum creatinine as markers of renal damage,^[259] and none has demonstrated any change in of these markers preoperatively and postoperatively.^{[258] [259] [260]}

It has been argued that BUN and creatinine are not sensitive enough tests for detecting renal damage after compound A exposure.^{[258] [259] [260] [261]} In an early attempt to address this issue, Kumano and colleagues measured urinary excretion of several renal tubular enzymes as measures of renal injury. Renal excretion of alanine aminopeptidase and NAG (i.e., enzymatic markers of renal integrity) increased 7 days after exposure to 7.6 MAC-hours of enflurane, but no increase was observed in patients receiving 9.6 MAC-hours of sevoflurane. Renal excretion of γ-glutamyl transpeptidase, β₂ -microglobulin, BUN, and creatinine were unchanged. ^[262] Bito and colleagues^[263] studied 48 patients undergoing gastrectomy under low-flow (1 L/min) (n = 16) and high-flow (6 to 10 L/min) (n = 16) sevoflurane anesthesia and low-flow isoflurane (1 L/min) (n = 16) anesthesia, looking at BUN, creatinine, creatinine clearance, NAG, and alanine aminopeptidase.^[263] The average inspired compound A levels were 20 ppm in the sevo_low group, and the average duration of exposure to this concentration was 6.11 hours. The only difference between the sevo_low and the sevo_high groups was the degree of compound A formation. Postanesthesia laboratory data showed no significant effects of compound A during sevoflurane anesthesia on any markers of renal function. Postanesthesia BUN and creatinine levels decreased, creatinine clearance increased, and urinary excretion of NAG and alanine aminopeptidase increased in all groups compared with preanesthesia values, but there was no significant differences observed between any of the groups. Kharasch and coworkers^[264] completed a similar study in which 73 surgical patients received sevoflurane or isoflurane anesthesia at a fresh gas flow of 1 L/min. Anesthetic duration was 3.7 or 3.9 hours, and the maximum inspired compound A concentration was 27 ppm (range, 10 to 67 ppm). Areas under the inspired and expired compound A concentration versus time curves (AUC) were 79 ppm-hours (range, 10 to 223 ppm-hours) and 53 ppm-hours (range, 6 to 159 ppm-hours), respectively. There were no significant differences between anesthetic groups in postoperative BUN, creatinine or urinary excretion of protein, or glucose (i.e., markers of proximal tubular resorptive function) or in NAG, proximal tubular α-GST, or distal tubular π-GST (i.e., markers of tubular cell necrosis). The investigators concluded that moderate duration of low-flow sevoflurane anesthesia during which compound A formation occurs was as safe as low-flow isoflurane anesthesia.

Nishiyama and associates^[265] studied the effect of repeat sevoflurane exposure within 30 to 90 days in neurosurgical patients. Ten patients received sevoflurane at a fresh gas flow rate of 6 L/min on two occasions. Serum inorganic fluoride was measured 1 day after surgery and BUN, creatinine, serum and urine β₂ -microglobulin, urine NAG, serum AST, serum ALT, and total bilirubin were measured 7 days postoperatively. Urine β₂ -microglobulin, AST, and ALT all increased to abnormal levels after both anesthetics, with no observed differences between the anesthetics. The second sevoflurane exposure did not change the hepatic or renal metabolism of sevoflurane. These changes were indicative of mild hepatic and renal injury. The duration of anesthesia for the first and second anesthetics was 4.3 ± 0.6 hours and 4.0 ± 0.6 hours, respectively.

Cases of humans with renal changes after sevoflurane administration have all occurred under conditions of prolonged sevoflurane exposure. Eger and colleagues ^[266] and Ebert and associates^[267] exposed human volunteers to 8 hours of 1.25 MAC of sevoflurane or desflurane anesthesia at 2 L/min of fresh gas flows and investigated BUN levels, creatinine concentration, and enzymatic markers of urinary function for up to 3 days before or 5 to 7 days after exposure, or both ( Table 8-7 ). Both groups of investigators found no differences in BUN or creatinine levels before and after exposure in either group, but they found significant differences in enzymatic markers of renal integrity (i.e., α-GST and π-GST), albuminuria, and glucosuria in the sevoflurane-exposed group. All renal changes were transient, with normal return of renal function by 4 days after exposure in all but one volunteer, whose renal function was normal 2 weeks after exposure. In Eger's volunteers, rebreathing of sevoflurane produced an average inspired concentration of compound A of 41 ppm, and the total compound A exposure was 328 ppm-hours. In the Ebert study, volunteers were exposed to an average inspired compound A concentration of 27 ppm, and the total exposure was 216 ppm-hours. With the exception or BUN and creatinine, both studies showed significant changes in urinary markers; the Ebert volunteers were lower on the compound A toxicity threshold curve and exhibited less drastic changes in all markers of injury. Goldberg and colleagues^[268] replicated Eger's study obtaining lower average compound A levels (253 ppm-hours) and found similar qualitative renal findings. Higuchi and associates^[269] found increased excretion of urinary enzymes after average compound A concentrations of 215 ppm-hours. In another study, Higuchi and colleagues^[270] found proteinuria in patients (n = 14) given low-flow (1 L/min) sevoflurane anesthesia undergoing orthopedic or dental surgeries but not those (n = 14) given high-flow (6 L/min) sevoflurane or those (n = 14) given low-flow isoflurane. Increased urinary NAG excretion occurred in the low-flow and high-flow sevoflurane groups, whereas levels of BUN, creatinine, and creatinine clearance were unchanged in all groups. In human

253

Several studies have examined the use of sevoflurane in the pediatric population. In one report,^[273] sevoflurane was compared with halothane anesthesia in children who had a mean age of 6 years. With comparable exposures, the sevoflurane group had significantly higher inorganic fluoride levels. There were no significant differences in BUN and creatinine levels, and there have been no reports of fluoride-associated renal toxicity in pediatric patients after sevoflurane anesthesia. Other tests of renal integrity were not studied. In another study, ASA class I and II children between the ages of 3 months and 7 years received 1 MAC of sevoflurane exposure at a fresh gas flow rate of 2 L/min. ^[274] The mean anesthetic exposure was 240 minutes, and fresh carbon dioxide absorbent was used for each patient. The soda lime temperature varied from 23°C to almost 41°C, and the inspired and expired compound A concentrations were 5.4 and 3.7 ppm, respectively. There were no differences in clinical chemistry measures, as reflected by AST, ALT, alkaline phosphatase, bilirubin, BUN, and creatinine levels. The maximum soda lime temperature correlated with the maximum compound A concentrations and the body surface area of the child. There have been no reports of markedly abnormal renal chemistry values for pediatric patients after sevoflurane exposure.

There is some concern about pediatric patients related to the major metabolite of sevoflurane, hexafluoroisopropanol (HFIP) because the levels of this metabolite may become relatively high due to a deficiency in UGTs that conjugate glucuronic acid to HFIP (see Fig. 8-7 ). For example, neonatal rats were found to have high levels of free HFIP for as long as 21 days after birth after sevoflurane exposure.^[275]

It is hypothesized that compound A causes serious injury to kidneys in rats but not humans because of differences in levels of renal cysteine conjugate β-lyase enzymes,^[276] which is thought to catalyze the conversion of compound A into a reactive thionoacyl fluoride metabolite that can acylate kidney proteins and cause toxicity ( Fig. 8-18 ). Renal cytosolic and mitochondrial β-lyase enzyme levels in rats are approximately 20 to 30 times higher than those found in humans.^{[277] [278] [279]} There is evidence to support the β-lyase theory of compound A-induced nephrotoxicity. Renal β-lyase appears to mediate the renal toxicity produced by several fluorinated alkenes that are structurally similar to compound A,^{[280] [281] [282]} and metabolic intermediates of the β-lyase pathway have been identified in the bile and urine of rats and human tissues exposed to compound A.^{[259] [264] [278] [279] [283]} Moreover, the glutathione S-conjugate of compound A is nephrotoxic,^[284] and the administration of aminooxyacetic acid (AOAA), an inhibitor of the β-lyase ( Fig. 8-19 ), protects against the nephrotoxicity of compound A in rats.^[285] However, there is some controversy surrounding the relevance of the β-lyase pathway in compound A-induced renal injury.^{[283] [286] [287]}

Investigators have reported that pretreatment of rats with AOAA and acivicin (AT-125), inhibitors of the β-lyase pathway (see Fig. 8-19 )^[286] did not attenuate the nephrotoxicity produced by compound A, but it was instead associated with a nearly threefold increase in renal injury (19% to 53%). These findings suggest that the β-lyase pathway may be a detoxification pathway. In this regard, a report by Kharasch and colleagues ^[287] showed that pretreatment or rats with AOAA partially protected against compound A-induced diuresis and proteinuria but failed to protect against glucosuria. In another report, AOAA again failed to decrease histologic renal injury, and pretreatment with acivicin significantly increased renal injury.^[286] Using an immunochemical approach, other investigators were unable to detect protein adducts of a thionoacyl fluoride metabolite of compound A in kidneys of rats after compound A treatment, suggesting that the β-lyase pathway of compound A metabolism in rats does not occur or is a minor pathway of metabolism. These findings suggest that another pathway of metabolism of

254

Figure 8-18 Proposed pathway for the metabolic activation of compound A shows formation of an intrahepatic glutathione (GSH) conjugate of compound A that is translocated to the kidney, where γ-glutamyl transpeptidase, cysteinylglycine dipeptidase, and renal cysteine conjugate β-lyase catalyze the formation of a nephrotoxic thiol, which can acylate kidney proteins. (Adapted from Martin JL, Kandel L, Laster MJ, et al: Studies of the mechanism of nephrotoxicity of compound A in rats. J Anesth 11:32–37, 1997.)

Higuchi and associates^[288] studied the effect of lowflow sevoflurane or isoflurane anesthesia (1 L/min) on 17 patients with stable moderate renal failure (serum creatinine > 1.5 mg/dL). They found no difference in BUN or creatinine levels up to 14 days after exposure or in creatinine clearance at 7 days after exposure in either group for patient exposures of more than 130 ppm-hours of compound A.^[288] In a multicenter, randomized, controlled trial, Conzen and colleagues^[289] studied 116 patients with stable renal insufficiency (serum creatinine > 1.5 mg/dL) with low-flow (1 L/min) sevoflurane (n = 59) or isoflurane (n = 57) anesthesia. Total exposure to sevoflurane was 201.3 ± 98 minutes, with an average total compound A exposure of 18.9 ± 7.6 ppm. Isoflurane exposures averaged 213.6 ± 83.4 minutes, and Baralyme was used as the carbon dioxide absorbent. No significant changes above baseline values were found at 24 or 72 hours after exposure for serum creatinine, BUN, creatinine clearance, urine protein, or glucose for either anesthetic.

So what does this all mean, and how does it translate into clinical care of patients? In a rebreathing system with a carbon dioxide absorber in lime (soda lime or Baralyme), patients exposed to sevoflurane will breathe compound A. The typical levels seen under clinical conditions vary and depend upon several factors, the most important being the inspired fresh gas flow. The key factor in determining toxicity from sevoflurane is the total exposure rather than the absolute concentration; exposure is expressed as the product of concentration and time. It appears from the many studies published that the lower threshold for renal changes associated with sevoflurane exposure and compound A is approximately 150 ppm-hours. At a fresh gas inflow of 2 L/min, these levels would be expected to be seen only under conditions of prolonged sevoflurane exposure and are not of concern for most patients undergoing anesthesia and surgery. Under conditions of prolonged sevoflurane exposure in which renal changes have been observed, these changes have been transient. It appears that compound A is of theoretical concern and academic interest, but no significant clinical renal toxicity has been associated with the use of sevoflurane. It is recommended that patients with preexisting renal disease not be exposed to sevoflurane and that sevoflurane be administered according to approved package labeling guidelines.

Halothane and Bromochlorodifluroethane

Carbon dioxide absorbents degrade halothane by a dehydrofluorination reaction to form bromochlorotrifluoroethane (BCDFE, F₂ C=CBrCl). The potential nephrotoxicity of BCDFE was compared with compound A in a study by Eger and colleagues. ^[290] It was found that BCDFE was 80% less reactive than compound A; halothane degraded to BCDFE 20- to 40-fold less than sevoflurane to compound A in a standard circle system; BCDFE was 75% less nephrotoxic to rats than compound A; and compounds that blocked the renal β-lyase pathway did not change or decreased renal injury from BCDFE, whereas the same blockers did not change or increased renal injury from compound A (see Fig. 8-19 ), suggesting that BCDFE and compound A may cause kidney injury by different mechanisms. These studies demonstrate that the chance of BCDFE causing nephrotoxicity when halothane is being used as an anesthetic is negligible.

255

Figure 8-19 Crucial steps in the β-lyase pathway: First, conjugation of compound A with glutathione in the liver is mediated by glutathione S-transferase; this conjugate is then transported to the kidney. Second, the glutathione S-conjugate is converted to a cysteine S-conjugate, a process mediated by γ-glutamyl transpeptidase. Third, the cysteine S-conjugate of compound A is then converted to a reactive thiol, mediated by renal cysteine conjugate β-lyase; DL-buthionine-(S,R)-sulfoximine (BSO) depletes endogenous stores of glutathione in the body, acivicin (AT-125) inhibits the activity of γ-glutamyl transpeptidase, and aminooxyacetic acid (AOAA) inhibits the activity of β-lyase. Compound A, its conjugates, and the reactive thiol have all been postulated to cause the renal necrosis that can occur in rats after the administration of compound A and the glucosuria and enzymuria that can result in rats and humans from compound A administration.

Carbon Monoxide

Carbon monoxide (CO) is an odorless, colorless gas that is poisonous because it displaces oxygen from hemoglobin in the blood, leading to carboxyhemoglobin formation. Its affinity for binding hemoglobin is 250 times greater than that of oxygen. All humans have some CO bound to hemoglobin in their blood; on average, this is 1%, for nonsmokers and as much as 10% for those who smoke. High levels of CO can produce neuropsychiatric problems in patients, and when levels of carboxyhemoglobin reach 50%, death can occur. CO poisonings have been reported during clinical anesthesia. ^{[291] [292] [293] [294] [295] [296]} The difficulty in monitoring CO levels clinically arises from the fact that pulse oximeters do not distinguish between carboxyhemoglobin and oxyhemoglobin.

All inhaled anesthetics produce some CO as a result of their interaction with strong bases in relatively dry carbon dioxide (CO₂ ) absorbents.^[297] Several factors influence the level of CO production, including the choice of anesthetic agent, the inspired anesthetic concentration, and the type, temperature, and degree of dryness of the CO₂ absorbent.^[298] Of the inhaled anesthetics, desflurane produces the most CO, followed in descending order by enflurane and isoflurane, with negligible amounts from sevoflurane and halothane ^{[298] [299]} ( Fig. 8-20 ). Intraoperative formation of CO has been reported with

Figure 8-20 Anesthetic degradation to carbon monoxide at equi-minimum alveolar concentrations (1.5 MAC) concentrations by (A) barium hydroxide lime and (B) soda lime. Each data point is the mean ± SD (n = 3). The single data point at 300 minutes reflects undetectable carbon monoxide from sevoflurane, methoxyflurane, or halothane. (Adapted from Baxter PJ, Garton K, Kharasch ED: Mechanistic aspects of carbon monoxide formation from volatile anesthetics. Anesthesiology 89:929–941, 1998.)

256

Figure 8-21 Proposed mechanism of carbon monoxide formation from difluoromethyl-ethyl ether anesthetics. The backbone structures for isoflurane (X = CI) and desflurane (X = F) are shown, along with a putative mechanism for the concomitant formation of trifluoromethane. Water in line 3 may also react as OH^- (Adapted from Baxter PJ, Garton K, Kharasch ED: Mechanistic aspects of carbon monoxide formation from volatile anesthetics. Anesthesiology 89:929–941, 1998.)

Amsorb contains neither NaOH nor KOH and does not degrade inhaled anesthetics to CO or degrade sevoflurane to compound A.^[301] Fully hydrated or rehydrated absorbents do not degrade inhaled anesthetics to CO. Soda lime contains 15% water by weight, and only after this level of hydration has been decreased to about 1.4% are appreciable amounts of CO formed. Baralyme contains 13% water by weight, and the threshold of hydration before CO is formed is 4.8%. However, even at these threshold levels, very little CO is produced. Production of large amounts of CO requires complete or nearly complete desiccation of the absorbent, a process that requires high flows of dry gas (10 L) for prolonged periods (1 to 2 days). It has been reported that certain patients had high levels of CO, a situation that was particularly prevalent in the first case on Mondays.^[291] This appeared to be related to the practice of flushing the anesthesia machine with high fresh gas flows over the weekend, which dried out the CO₂ absorbent. Higher temperatures are also associated with increased CO production.

In general, CO toxicity has little clinical significance, regardless of the agent used, as long as simple guidelines to minimize or eliminate CO are observed. These include the use of fresh absorbent, the use of soda lime instead of Baralyme, and avoidance of techniques that dehydrate the CO₂ absorbent in the anesthetic circuit. Low fresh gas flows are more economical and limit absorbent desiccation. As a last resort, the absorbent can be rehydrated by adding approximately 1 cup of water (230 mL) per 1.2 kg of absorbent (i.e., standard canister).