Subcellular Localization of Trifluoroacetylated Liver Proteins in Association with Hepatitis following Isoflurane

A 66-yr-old man with type II diabetes mellitus, chronic obstructive pulmonary disease, hypertension, and coronary artery disease underwent an uneventful carotid endarterectomy during which anesthesia was conducted with the use of oxygen (30–100%), isoflurane (0.4–1.0%, 135 min), fentanyl, and thiopental. Two months later, he presented with chest pain and underwent three-vessel coronary artery bypass graft. His preoperative medications were morphine sulfate, diazepam, scopolamine, cefazolin, and gentamicin. During this surgery, the patient received fentanyl, pancuronium, ϵ-aminocaproic acid, mannitol, and heparin in addition to isoflurane and 100% oxygen. The cardiopulmonary bypass time was 110 min, and the cross-clamp time was 89 min. The patient had a postbypass hemoglobin concentration of 8.2 g/dl and received 400 ml pump blood. There were no intraoperative episodes of hypotension, although approximately 15 min after incision, norepinephrine (45 min) and nitroglycerin (60 min) were administered to the patient. Moreover, his cardiac output was 3.7 (on norepinephrine and nitroglycerin) before bypass and 5.0 after bypass without these medications. In addition, his arterial blood gases before, during, and after cardiopulmonary bypass were all within normal range:p  H 7.39, partial pressure of carbon dioxide (Pco²) 46, partial pressure of oxygen (Po²) 157, HCO³28 (before);p  H 7.41, Pco²44, Po²405, HCO³27 (during);p  H 7.38, Pco²45, Po²346, HCO³26 (after). His immediate postoperative recovery was unremarkable except for a mild fever of 38.1–38.3°C that developed 7 h after the completion of surgery, which was taken as time 0 (fig. 1). He was transferred from intensive care to the surgical floor at 23 h. That night, his temperature fluctuated between 36.8 and 38.0°C, and mild icterus developed. At 45 h, his temperature increased to 38.6°C, and he became more severely icteric. Examination revealed that he had tender hepatomegaly, and he became progressively drowsier until he was able to be aroused only with painful stimuli. His serum aspartate aminotransferase concentration, measured at 38 h, was 3,973 U/l. At 52 h after surgery, liver biochemistry testing revealed alanine aminotransferase of 10,394 U/l; aspartate aminotransferase of 25,515 U/l; total bilirubin of 3.8 mg/dl, and alkaline phosphatase of 55 U/l. The prothrombin time peaked at 20.6 s, and total bilirubin peaked at 10.9 mg/dl. Serologic assays for hepatitis A, B, and C were negative. An ultrasound image showed a normal gallbladder and common bile duct. At 50 h, the patient was transferred back to the intensive care unit, where he had a transient episode of hypotension that was corrected by fluid replacement. From the time of surgery up to that point, his blood pressure had been consistently in the normal range (fig. 1).

On the third day after surgery, a rigid abdomen developed. An exploratory laparotomy, performed to exclude ischemic bowel, revealed a swollen erythematous liver with a firm, rubbery texture. No other pathology was noted in the abdomen. A biopsy of the liver was performed. After the laparotomy, during which fentanyl and cisatracurium besylate were used as anesthetic agents, the transaminases steadily decreased. Four weeks after surgery, the patient was discharged from the hospital with normal liver enzymes and synthetic function.

The patient's medical history was significant for a balloon angioplasty 12 yr earlier and an allergy to intravenous dye and codeine. His only surgery was the recent endarterectomy. His medications at home were gemfibrozil, isosorbide dinitrate, glyburide, aspirin, diltiazem HCl, metoprolol, and oxazepam as needed. The patient was not obese and had no risk factors for hepatitis (history of blood transfusion, sexual promiscuity, alcohol abuse, or intravenous drug use). He had not used alcohol for 15 yr.

A wedge biopsy of the liver was obtained 72 h after surgery and was processed for frozen and paraffin sections and for electron microscopy. The liver architecture and the portal tracts were normal, and there was no increase in fibrosis (fig. 2). Hepatocytes in zone 3 exhibited severe macrovesicular steatosis, a finding that is consistent with another case report of isoflurane hepatitis. 7Coagulative necrosis affected all of zones 3 and 2. These areas were sharply demarcated and separated from the viable-appearing hepatocytes of zone 1 (figs. 2A and B). A few scattered neutrophils were present in some of the necrotic areas. Central veins and most sinusoids were empty of erythrocytes, but there was erythrocyte extravasation into the space of Disse in zone 3 (fig. 2C). In the portal tracts, there were early signs of ductular proliferation. Bilirubinostasis was evident in a few ducts and ductules, and occasionally in hepatocytes and sinusoids in zone 1 (fig. 2D). Glycogen was present in hepatocytes in zone 1 but was absent from zones 2 and 3 (fig. 2E). Indicative of the acute nature of the process, a normal reticulin pattern was retained throughout the lobule (fig. 2F).

Electron microscopic evaluation showed that the less severely affected cells in zone 1 exhibited dilatation and proliferation of smooth endoplasmic reticulum, swelling of mitochondria, microvesicular steatosis, abundant lipolysosomes, and increased peroxisomes. In more severely affected cells, which appeared dark in electron micrographs, there were marked nuclear changes, with irregularity of the nuclear outline, peripheral clumping of chromatin, and enlargement and condensation of nucleolar elements (results not shown).

Trifluoroacetyl-modified proteins were detected in liver sections using a well-described procedure. 9Briefly, slides were incubated with affinity-purified rabbit anti-trifluoroacetyl immunoglobulin G (IgG; 2 μg/ml, 1 h) in 1.5% (vol/vol) normal goat serum, followed by biotinylated, affinity-purified goat anti-rabbit secondary antibody (Vector Laboratories, Inc., Burlingame, CA). TFAMPs were visualized with a peroxidase reagent (Vector Laboratories, Inc.) and developed with a diaminobenzidine peroxidase substrate kit (Vector Laboratories, Inc.). Infiltrating cells in the liver biopsy were immunohistochemically characterized using an automated immunostainer. In short, an antigen retrieval method was used with a standard pressure cooker. CD4⁺lymphocytes were detected with 1:20 monoclonal mouse anti-human clone IF6 (Novocastra, Newcastle upon Tyne, United Kingdom); CD8⁺lymphocytes with 1:20 mouse anti-human IgG1, clone C8/144B, (DAKO Corp., Carpinteria, CA); B lymphocytes with 1:200 mouse anti-human IgG2a κ, clone L-26 (DAKO Corp.); and macrophages with 1:50 monoclonal mouse anti-human CD68 IgG1, clone KP-1 (DAKO Corp.). The secondary antibody was a biotinylated anti-rabbit/mouse globulin supplied by Ventana ES (Tucson, AZ). Diaminobenzidine was used as the chromogen throughout. Incubation times were according to the manufacturer's recommendations.

Trifluoroacetyl-modified proteins were detected in intact and necrotic hepatocytes in zones 1 and 2 adjacent to zones of steatosis (figs. 3A and B). No adducts were detected when purified normal rabbit IgG (2 μg/ml) was used in place of the anti-trifluoroacetyl IgG (fig. 3C) or when the trifluoroacetyl hapten was removed from the tissue proteins with 1 m monoethanolamine before the addition of anti-trifluoroacetyl IgG (results not shown). 10Immunochemically identified CD4⁺and CD8⁺T cells and B lymphocytes in zones 1 and 2 adjacent to the zones of steatosis were few in number, whereas macrophages found in the same regions did not seem to differ in number from those found in normal liver tissues (results not shown).

Tissue from the patient's liver biopsy and from the normal liver biopsy of an individual not recently exposed to volatile anesthetics were removed from the paraffin blocks, deparaffinated in xylene, placed in absolute ethanol, and embedded in LR White (SPI, West Chester, PA). Ultrathin sections were mounted on 150-mesh uncoated nickel grids. Grids were floated on blocking solution (20 min, phosphate-buffered saline, 0.1% Tween 20, 0.5% cold water fish gelatin; Ted Pella, Inc., Redding, CA) and incubated with anti-trifluoroacetyl IgG (2 μg/ml, 1 h), or with normal rabbit IgG (2 μg/ml, negative control). The grids were then rinsed in blocking buffer (5 min), incubated with 10 nm gold-conjugated goat anti-rabbit antiserum (1:1,000; Ted Pella, Inc.), rinsed in phosphate-buffered saline, and air dried. Sections were counterstained with uranyl acetate and examined with a Phillips CM10 electron microscope (Phillips/FEI Co., Hillsboro, OR).

Although the use of paraffin-embedded material did not allow ideal tissue preservation, TFAMPs were always associated with cytoplasmic organelles and absent from lipid droplets and from the extracellular spaces (fig. 4). The labeling was most intense in the mitochondria and the endoplasmic reticulum and to a lesser extent in the nuclear membrane and the nucleus. The normal liver tissue exhibited only minimal background labeling, similar to the section in which the primary antibody was replaced by normal rabbit IgG (results not shown).

The use of volatile anesthetics, including halothane, enflurane, isoflurane, and desflurane, has been associated with a severe form of idiosyncratic liver injury that is believed to be caused, in many cases, by immune-mediated processes. 11–13This theory is supported by clinical evidence and immunochemical evidence. For example, in the case of halothane hepatitis, patients often have received multiple halothane anesthetics and have had fever, rash, arthralgia, and eosinophilia, which are signs and symptoms suggestive of an immune-mediated process. In addition, most halothane hepatitis patients have serum antibodies that recognize one or more liver microsomal proteins either in their native state (antigens) or after they have been covalently trifluoroacetyl modified by the trifluoroacetyl chloride metabolite of halothane (neoantigens). 12,14These antibodies, and possibly antigen-specific T cells, have been suggested to have an immunopathological role not only in halothane hepatitis but also in hepatitis caused by other volatile anesthetics. 9In fact, anti-trifluoroacetyl antibodies have been detected in the sera of three patients diagnosed with isoflurane hepatitis. 15,16Serum antibodies from another isoflurane hepatitis patient were found to react with a trifluoroacetyl-labeled 60-kd protein antigen in rat liver microsomes. 17Unfortunately, by the time isoflurane was implicated in the current case, a serum sample could not be obtained to determine whether anti-trifluoroacetyl antibodies were present. Nevertheless, we were able to demonstrate TFAMPs in the patient's liver, an important finding not previously reported in the liver of a patient diagnosed with isoflurane-induced liver injury.

The finding of TFAMPs in the liver of the patient was surprising because isoflurane is oxidatively metabolized to form TFAMPs approximately two orders of magnitude less than halothane. 9,18This result suggests that the patient may have had an increased concentration of cytochrome P450 2E1 in the liver because this isoform of cytochrome P450 is predominantly responsible for the oxidative metabolism of halothane, 19enflurane, 20and isoflurane. 21In this regard, if the patient had a preexisting condition of nonalcoholic steatohepatitis before anesthesia, this could have accounted for this enzyme being increased. 22Alternatively, the patient may have had increased concentrations of cytochrome P450 2A6, which can also oxidatively metabolize halothane to form TFAMPs. 19,23 

The lack of massive infiltration of T or B lymphocytes or macrophages in the zones of necrosis suggests that the pathophysiologic mechanism of isoflurane hepatitis may not involve a significant contribution by cellular immune reactions but instead could be mediated by other mechanisms. The finding of TFAMPs for the first time in the mitochondria, nucleus, nuclear membranes, and rough endoplasmic reticulum after isoflurane anesthesia is consistent with this idea (fig. 4) because it suggests that liver injury may have been initiated at one or more of these sites by TFAMPs. For example, microvesicular steatosis may have been initiated by the TFAMPs inhibiting β oxidation and respiration in the mitochondria. 24In this regard, a recent case report of fatal hepatotoxicity after isoflurane exposure was also associated with microvesicular fatty changes in the liver. 7Moreover, the finding of an active form of cytochrome P450 2E1 in the mitochondria of hepatocytes 25suggests that TFAMPs may have been formed at this site. Alternatively, the TFAMPs in the endoplasmic reticulum may have led to a humoral-based liver injury after they were released from damaged hepatocytes, as has been proposed as a possible mechanism of halothane hepatitis. 13 

In summary, we present a man who developed fulminant hepatic failure after a second exposure to isoflurane for general anesthesia. There was no evidence that the liver injury was caused by hypotension, hypoxia, viral hepatitis, or sepsis. It is unlikely that his liver injury is attributable to the glyburide, which has rarely been associated with hepatitis 26–28or any of his other medications because he had been taking them for many months without adverse effect. The demonstration of TFAMPs in intact and necrotic hepatocytes associated with subcellular organelles suggests that the hepatitis may have been caused by isoflurane, possibly by both immune- and nonimmune-mediated mechanisms of toxicity.