Compound A [CF2 = C(CF3)OCH2F], a degradation product of sevoflurane [(CF3)2CHOCH2F], is a vinyl ether and may be an alkylating agent. Thus it is a potential genotoxin.
The capacity of compound A to produce sister chromatid exchanges was measured in Chinese hamster ovary cells with and without metabolic activation. Concentrations of 11 to 468 ppm compound A were applied for 2 h, the Chinese hamster ovary cells were incubated for a further 34 h in the presence of bromodeoxyuridine, and then colcemid was added to produce arrest in metaphase. Coded slides of cells were examined blindly, and 50 chromosome spreads were counted for each test concentration.
The lowest concentration of compound A applied without metabolic activator (27 ppm) significantly increased (P < 0.001) sister chromatid exchanges, and increasing concentrations of compound A increased the incidence of exchanges. Metabolic activation did not increase the incidence of exchanges.
Compound A increases sister chromatid exchanges at concentrations (27 ppm) obtained in low-flow systems when sevoflurane is used at concentrations approximating minimum alveolar concentration.
Compound A [CF2= C(CF3)OCH2F] is a vinyl ether produced by carbon dioxide absorbents from sevoflurane [(CF3)2CHOCH2F] by extraction of hydrogen fluoride. [1,2] Compound A causes corticomedullary renal injury (necrosis of the outer stripe of the outer medullary layer, otherwise known as corticomedullary junction necrosis) in rats. [3–6] In human volunteers, both glomerular and tubular injury (as defined by albuminuria and glucosuria) result from an 8-h administration of 1.25 minimum alveolar concentration sevoflurane at a fresh gas inflow rate of 2 ml/min.  Such administration produces an average compound A concentration of 41 ppm. An 8-h administration of 1.25 minimum alveolar concentration desflurane at a 2 l/min fresh gas inflow rate produces neither compound A nor renal injury, suggesting that these changes do not result from anesthesia per se, or from the level or duration of anesthesia.  Such results suggest that compound A can produce tissue injury.
Compound A spontaneously degrades and probably associates with blood proteins, the degradation rate having a half-life at 37 degrees Celsius of 1–3 min.  The highly reactive nature of compound A suggests that it may be an alkylating agent. Alkylating agents are electrophilic compounds with an affinity for nucleophilic centers in organic macromolecules. Such agents include several proved or suspected carcinogens.  However, no present clinical or animal data indicate that compound A is a mutagen or carcinogen, and results of tests of the mutagenicity of sevoflurane and compound A have been uniformly negative.  Immediately pertinent to the present report, vinyl ether anesthetics, such as fluroxene and divinyl ether, induce sister chromatid exchanges (SCEs),  as do other vinyl-containing compounds.  It is not known whether fluroxene or divinyl ether is mutagenic or carcinogenic.
These considerations prompted us to try to determine if compound A might be genotoxic. One indicator of genotoxicity, the SCE test, measures the exchange of chromatin material between the sister chromatids that form a single chromosome. 
The in vitro SCE assay is commonly used to screen compounds for possible mutagenicity and carcinogenicity, and SCE results correlate with mutagenic or carcinogenic events. [13,14] However, although the assay can detect in vivo carcinogens (73% may be detected), only approximately 67% of chemicals that are positive in the assay are also positive in an in vivo carcinogenicity assay.  Thus a positive SCE test may be best characterized as indicating genotoxicity rather than demonstrating mutagenicity or carcinogenicity. 
The present report describes the induction of SCEs by compound A concentrations similar to those that may be obtained using sevoflurane in low-flow systems.
Materials and Methods
Our test differed from the usual test for SCE because compound A is breathed rather than ingested as a solid or nonvolatile liquid. Accordingly, we applied compound A as a gas from a premixed cylinder containing 1.3% compound A in air. This gas (15 ml) or dilutions of this gas, always with 8% carbon dioxide, was added to 5 ml media containing the Chinese hamster ovary (CHO) cells in a closed, 20-ml syringe, capped with a sterile three-way stopcock. Two such syringes were prepared for each concentration without metabolic activation. We also studied the effect of metabolic activation in two additional syringes when concentrations of 50–800 ppm (the concentrations injected into the syringes) compound A were studied (four syringes total for such concentrations).
The SCE experiments were conducted using modifications of the protocol of Galloway et al.  Chinese hamster ovary (ATT-CCL61-CHO-K1) cells obtained from American Type Culture Collection (Rockville, MD) were grown in an atmosphere of 8% carbon dioxide at 37 degrees Celsius. Cells were cultured in CHO-S-SFM II (GIBCO, Bethesda, MD), a serum-free medium designed for culturing CHO cells in suspension. The medium was used during exposure of cells to compound A with and without metabolic activation (i.e., with and without the presence of enzymes that degrade the test compound to compounds that are potentially more reactive). The medium used to expose cells to compound A in the presence of metabolic activation consisted of CHO-S-SFM II supplemented with 10%(vol/vol) S9 (Aroclor 1254-induced rat liver homogenate obtained from Molecular Toxicology, Annapolis, MD). Cofactors were added at final concentrations of 2.4 mg/ml nicotinamide adenine dinucleotide phosphate and 4.5 mg/ml sodium isocitrate. The components for metabolic activation were freshly prepared and kept on ice before use.
All procedures to the time of final harvest were performed aseptically. Chinese hamster ovary cells were incubated in 5 ml growth media with 15 ml gas containing various concentrations of compound A (including 0 ppm [control]) with 8% carbon dioxide in air. A positive control (mitomycin C at 0.1 micro gram/ml) also was used. After equilibration at 200 oscillations/min for 2 h, the cultures were expelled into 50-ml tubes, care being taken to expel all of the culture and immediately close the three-way stopcock, trapping the gas at the end of equilibration in the syringe. The media in the culture was changed and 5 micro gram/ml bromodeoxyuridine added. The cultures were incubated at 37 degrees Celsius for 34 h (preliminary studies showed that growth through two metaphases required this duration of incubation). Colcemid was added to produce arrest in metaphase, and after 2 h we prepared chromosome spreads differentially stained by the fluorescence plus Giemsa method. 
Slides for all exposures were coded, and 50-chromosome spreads (25 per culture) showing differential staining were analyzed blindly for SCEs. The frequency of SCE per cell and per chromosome was counted, and means per group with standard errors were determined. We applied a two-tailed t test to determine whether treated cells had a greater incidence of exchanges than did control cells, with P < 0.01 considered significant.
Within an hour after completion of the exposure, the sealed syringes were transported from SRI International to the anesthesia laboratory at University of California, San Francisco, where the contents were analyzed by gas chromatography. As defined by preliminary studies, loss of compound A after the initial equilibration was 5–8% over a period of 4 h. Thus the results of analysis indicate the approximate concentration to which the cultures were exposed.
For compound A analysis we used a Gow-Mac 580 (Bridgewater, NJ) flame ionization detector gas chromatograph equipped with a 1-m-long, 2.16-mm (internal) diameter column containing Poropak Q (Alltech, Deerfield, IL) maintained at 124 degrees Celsius with a 10 ml/min carrier flow of nitrogen. The detector (at 174 degrees Celsius) received 55 ml/min hydrogen and 200 ml/min air. The chromatograph was calibrated before and at intervals during each test with secondary (cylinder) calibration standards. We prepared primary (volumetric) standards to calibrate each secondary standard. The chromatographic output was linear over the range of concentrations tested.
In the absence of metabolic activation, the concentration of compound A measured in each syringe was 55–60% of the concentration injected into that syringe. The ratio of measured to injected concentration did not change as a function of the time from injection to analysis. Duplicate determinations (one for each syringe used for the test at a given concentration) differed by no more than 8% of the mean value. In the presence of metabolic activation, the concentration of compound A was much lower than the concentration injected, was more variable (13–18%), and (as a fraction of the injected concentration) appeared to be smaller if compound A was exposed to metabolic activator for a longer period.
In a preliminary study, 480, 980, 2,200, and 4,400 ppm compound A (no metabolic activator) all significantly increased SCEs. Still larger concentrations (7,100) caused cell death. In subsequent studies, we found that measured concentrations of 27, 57, 118, 228, and 468 ppm without metabolic activator all produced small but significant increases in SCEs, either per cell or per chromosome (Table 1) from control. A comparison of the effects of the smallest and largest doses showed a significant dose-related effect for either SCEs/cell or SCEs/chromosome. For the same applied concentrations (i.e., the concentrations introduced into the syringes), metabolic activation did not increase the capacity of compound A to induce SCEs. As expected, mitomycin C (positive control) produced large increases in SCEs (Table 1).
Our results show that, like other vinyl ethers such as fluroxene, divinyl ether, or ethyl vinyl ether,  compound A can damage DNA as evidenced by increases in SCE. However, in contrast to these vinyl ethers, compound A produces this damage at low concentrations (27 ppm), and does so in the absence of metabolic activator. Fluroxene, divinyl ether, and ethyl vinyl ether were tested at respective concentrations of 11,470 ppm, 7,060 ppm, and 13,380 ppm (1 h exposure) in the presence of metabolic activator. In the absence of metabolic activator, a 24-h exposure to fluroxene did not increase SCE (the other vinyl ethers were not tested in the absence of metabolic activator). Thus compound A is far more potent than these anesthetic vinyl ethers in their respective capacities to increase SCEs.
The finding of induction of SCE by compound A, without metabolic activation, supports other data suggesting that compound A may be a direct alkylating agent. Compound A degrades in various blood proteins, including albumin, pure hemoglobin, erythrocytes, and plasma, precluding the determination of solubility in bloody.  In contrast, fluroxene is stable in blood, permitting the determination of blood/gas partition coefficients.  The induction of SCE by compound A suggests that it may be able to bind to DNA as well as proteins. However, SCE-inducing agents may not necessarily interact with DNA. Chemicals that alter DNA replication or repair may also result in elevated SCE frequency.  Further study will be required to determine whether compound A increases SCE frequency by binding to DNA or by a less direct mechanism.
The lowest compound A concentration producing significant increases in SCE was 27 ppm. We did not test lower concentrations and thus did not define the threshold for producing significant increases in SCEs. The value of 27 ppm approximates the inspired concentrations (partial pressures) that may be obtained in clinical practice at lower inflow rates of sevoflurane.  However, alveolar and tissue partial pressures of compound A will be lower than inspired partial pressures. For example, we have found (E.I. Eger, unpublished data) that the alveolar concentration of compound A is 60–80% of the inspired concentration. Because of destruction of compound A in blood, the tissue partial pressures will be lower still. Thus the partial pressure of compound A in tissues may be less than the threshold partial pressure that produces increases in SCEs.
The implications of our findings are limited by several factors, some of which have been expressed in the preceding paragraphs. We emphasize that although the increases in SCEs produced by compound A were statistically significant, they were small, particularly compared with our positive control (Table 1). Indeed, the increases in SCEs produced by both compound A (present study) and anesthetic vinyl ethers  are considerably less than the increases produced by many other compounds. [13,14] No clinical data directly suggest that sevoflurane or compound A has a link to mutagenicity or carcinogenicity, and no test of mutagenicity to date has shown that compound A or sevoflurane is a mutagen. As indicated earlier, SCE tests correlate imperfectly with known mutagens, providing both false-positive and false-negative results. [15,16] Standard bacterial tests (Ames tests with Salmonella typhimurium and Escherichia coli) of mutagenicity do not produce positive results with compound A, even at concentrations enormously greater than any that might be imposed clinically.  We have confirmed these findings (unpublished data). Compound A also does not produce chromosome aberrations in a Chinese hamster lung fibroblast cell line, nor does it affect micronuclei in mice.  However, neither these nor other studies have determined whether compound A induces gene mutations in mammalian cells.
Also related to the issue of relevance, although in vitro tests of genotoxicity (such as the SCE test) are commonly used to screen compounds for mutagenic or carcinogenic potential, the predictive capacity of such a test is imperfect, and several authorities question the utility of the SCE as indicative of either mutagenicity or carcinogenicity. [22,23] For example, a comprehensive evaluation of four tests (Salmonella mutagenesis, mutations in mouse lymphoma cells, SCEs in CHOs, and chromosome aberrations in CHOs) found an overall concordance with rodent carcinogenicity tests of 60%.  The SCE test was the most sensitive test (73% of carcinogens detected) but had a false-positive rate of 55%. The Salmonella test had a lower sensitivity (45%) and a lower false-positive rate (14%). Thus the implications of our finding that Compound A induces SCEs at concentrations found in clinical practice remain unclear. They suggest a need for further investigations, specifically the study of the mutagenicity of compound A in mammalian cell models.
The authors thank Winifred von Ehrenburg for editorial suggestions and Ohmeda's Pharmaceutical Products Division for providing compound A.