Plasma α-Glutathione S- Transferase: A Sensitive Indicator of Hepatocellular Damage During Polymicrobial Sepsis

Hypothesis  Since studies have found the liver enzyme α-glutathione S-transferase (αGST) to be a marker of hepatic injury after hemorrhagic shock, αGST also may serve as a sensitive indicator of hepatocellular damage during the early stage of polymicrobial sepsis.

Design, Interventions, and Main Outcome Measures  Male adult rats were subjected to the cecal ligation and puncture (CLP) model of polymicrobial sepsis or sham operation, followed by fluid resuscitation with isotonic sodium chloride solution. Systemic blood samples were taken at 2, 5, 10, or 20 hours after CLP or sham operation. Plasma levels of αGST and lactate were determined using an enzyme immunoassay and enzymatic assay, respectively. Additional animals were examined for morphologic alterations in liver ultrastructure of septic animals using electron microscopy.

Results  A similar level of αGST (mean ± SEM, 30.5 ± 3.5 µg/L) was found in the sham group at all measured time points. Although plasma levels of αGST did not change at 2 hours after CLP, they were elevated by 249% at 5 hours after the onset of sepsis and continued to increase throughout the septic course. Plasma lactate levels were significantly increased only at 20 hours after CLP (P<.001). Previous studies have shown that liver transaminase levels did not increase at 5 hours, but at 10 and 20 hours after CLP. In addition, electron microscopy revealed structural changes in hepatocyte morphology at 5 and 20 hours after CLP that were indicative of hepatocellular injury.

Conclusion  Since plasma αGST levels increased earlier than plasma lactate and liver transaminase levels, αGST may be a more sensitive indicator of early liver injury and should be used in monitoring hepatocellular damage during the progression of sepsis.

SEPSIS AND septic shock, with ensuing multiple organ failure, continue to be the most common causes of death in surgical intensive care units, despite advances in the care and treatment of critically ill patients.^1-3 Because of its important role in host defense mechanisms, the liver has been studied extensively during various adverse circulatory conditions and is thought to be a major organ responsible for initiating multiple organ failure during sepsis.^4,5 Moreover, abnormality or damage of hepatic tissue during sepsis may lead eventually to liver failure, which can produce deleterious alterations in host metabolism.

Previous studies performed in our laboratory using the sepsis model of cecal ligation and puncture (CLP) have indicated that conventional serum liver enzymes used as markers of hepatic damage (eg, alanine aminotransferase [ALT] and aspartate aminotransferase [AST]) are capable of detecting hepatic damage at 10 hours after the onset of sepsis.⁶ However, further studies performed in our laboratory using the indocyanine green (ICG) clearance technique have shown that hepatocellular function is depressed much earlier than the occurrence of hepatocellular damage, which is indicated by transaminase levels.^7,8 Since hepatocellular dysfunction has been demonstrated to be an early event in the pathophysiologic processes of sepsis,⁴ hepatic damage may occur earlier than previously thought but remain undetected using conventional assessment of transaminase levels. In this regard, recent studies have indicated that the cytosolic liver enzyme α-glutathione S-transferase (αGST) functions as an early and sensitive indicator of hepatocyte damage caused by various adverse conditions, including hemorrhagic shock,⁹ ischemia and reperfusion,¹⁰ liver transplant rejection,¹¹ and acetaminophen overdose.¹² In contrast to transaminases, αGST is more uniformly distributed throughout the liver,¹³ possesses a higher cytosolic concentration in the hepatocyte,¹⁴ and has a significantly shorter half-life in circulation.¹³ Thus, we hypothesized that assessment of plasma αGST levels may reflect early alterations in hepatocellular integrity following the onset of sepsis with a greater degree of sensitivity. Our objectives, therefore, were to determine whether αGST levels can be used to assess hepatocellular damage during polymicrobial sepsis and to compare changes in αGST levels with alterations in hepatocyte ultrastructure.

Polymicrobial sepsis was induced in male Sprague-Dawley rats (weight, 275-325 g) using cecal ligation and puncture (CLP) according to the method of Chaudry et al.¹⁵ Briefly, food was withheld from the rats overnight (approximately 16 hours) before the induction of sepsis, but water was allowed ad libitum. The rats were anesthetized with methoxyflurane (Mallinckrodt Veterinary, Inc, Mundelein, Ill) inhalation, their abdomens were shaved, and a 4-cm ventral midline incision was made. The cecum was then exposed and isolated using ligation with silk ligature just distal to the ileocecal valve to avoid intestinal obstruction. The cecum was punctured twice at opposite ends with an 18-gauge needle, and patency of the puncture holes was established by forcing out a small amount of the cecal contents. The ligated and punctured cecum was returned into the abdominal cavity. The abdominal incision was then closed in 2 layers, and the animals received 3 mL per 100 g of body weight of isotonic sodium chloride solution subcutaneously (ie, fluid resuscitation). Animals undergoing sham operation (sham group) underwent the same surgical procedure except that the cecum was neither ligated nor punctured. The experimental protocol and the care of the animals were performed in accordance with the Animal Welfare Act and National Institutes of Health guidelines. This project was approved by the Institutional Animal Care and Use Committee of Rhode Island Hospital, Providence.

At 2, 5, 10, or 20 hours after CLP or sham operation (5-8 rats/group), whole blood was drawn into heparinized syringes using cardiac puncture. Immediately, the plasma was separated by centrifugation at 2200 rpm for 15 minutes at 4°C, divided into aliquots, and stored at −70°C until assayed. Plasma levels of αGST were quantified (in micrograms per liter) using a rat αGST enzyme immunoassay (Hepkit-Alpha, Biotrin International, Dublin, Ireland) according to the manufacturer's directions. This particular immunoassay has a high degree of sensitivity (minimum detectable concentration, 0.3 µg/L) and specificity, and requires only a small amount of plasma sample (10 µL) for conducting the assay. Plasma lactate concentration was determined enzymatically using a diagnostic kit (Sigma-Aldrich Corporation, St Louis, Mo).

Animals (3 per group) were killed using an overdose of pentobarbital sodium at 5 or 20 hours after CLP or sham operation. Immediately after the death of the experimental animals, the liver tissue was removed and submerged in 2% paraformaldehyde and 2% glutaraldehyde containing 2.5% dextrose and 0.01% calcium chloride in 0.1 mol/L phosphate buffer (pH, 7.4). The tissues were then cut in approximately 1-mm³ cubes and kept in the above fixative at 4°C for at least 3 hours. After washing with 0.1-mol/L phosphate buffer, the tissues were postfixed for 1.5 hours in 0.1-mol/L phosphate buffer containing 1% osmium tetroxide and dehydrated through a graded series of ethanol alcohol solutions. After infiltration with Epon-Araldite resins (Electron Microscopy Sciences, Fort Washington, Pa), the tissues were embedded in plastic capsules and polymerized in an oven at 70°C. Ultrathin sections were prepared using an ultramicrotome (Reichert Ultracut E; Leica Inc, Deerfield, Ill). The thin sections were then mounted on formvar- and carbon-coated single-slotted grids (Electron Microscopy Sciences) and stained with 2% uranyl acetate and Reynold lead citrate before examination under an electron microscope (JEOL 100CX; JEOL Ltd, Cambridge, Mass).

All data are expressed as mean ± SEM and compared using unpaired t test between the CLP and sham groups at each time point. Differences in values were considered significant if P≤.05.

Levels of αGST were detectable in plasma taken from normal animals, with a mean level of 31.6 ± 3.6 µg/L (n = 6). Plasma levels of the enzyme in the sham groups were highly consistent at all time points (Figure 1). The mean level in the sham groups across all time points was 30.5 ± 3.5 µg/L. At 2 hours after CLP, αGST levels did not differ from levels in the sham group. However, at 5 hours after the onset of sepsis, plasma αGST levels increased significantly by 249% and continued to increase, with peak levels (229.4 ± 26.2 µg/L) occurring at 20 hours after CLP (Figure 1).

Plasma lactate levels were detectable at all time points in the sham groups. Although lactate levels in animals at 2 to 10 hours after CLP were increased, the elevation was not significant. However, plasma lactate level was elevated by 114% at 20 hours after the onset of sepsis (P<.001) (Figure 2).

Electron microscopy of hepatic tissue taken from the sham groups revealed normal hepatocytes with intact ultrastructure and no notable alterations in cellular integrity (Figure 3). However, examination of specimens taken at 5 hours after CLP demonstrated apparent changes in ultrastructure, particularly in the bile canaliculi and surrounding areas (Figure 4). These alterations include marked dilation of bile canaliculi, reduction of the canalicular microvilli, and development of occluding lamellated membranes in the lumen. These changes are indicative of metabolic alterations and probably result from the accumulation and altered metabolism of bile acids. Also apparent in hepatocytes at 5 hours after CLP is the presence of microcrystals in lysosomes (Figure 4). This feature may be attributed to bile acid accumulation. Examination of hepatocytes at 20 hours after CLP revealed ultrastructure changes similar to those observed in early sepsis (ie, dilation of bile canaliculi and reduction of canalicular microvilli) with the addition of swelling of the Golgi apparatus and regional ribosomal detachment from the rough endoplasmic reticulum (Figure 5).

The characteristics of the liver enzyme αGST make it a more sensitive marker for the assessment of hepatocellular damage compared with transaminases (ie, AST and ALT) or plasma lactate. Predominantly located in hepatocytes, αGST has been found to constitute as much as 3% of the hepatocellular cytosolic proteins.^14,16 It is distributed uniformly throughout the liver compared with the transaminases, which are found predominantly in the periportal hepatocytes, and therefore may be useful in detecting centrilobular damage.¹³ Moreover, αGST has a relatively small molecular weight (approximately 50 kd, compared with 95 kd for AST)¹⁴ and is released readily and rapidly into circulation following hepatocellular damage. In addition, αGST has a half-life of less than 1 hour, in contrast to 48 hours for ALT,¹² and therefore allows early recognition of the time at which active liver damage has ceased, thus allowing better monitoring of alterations and investigations into the mechanism of hepatocellular damage.

The studies using animal models known to produce alterations in hepatocellular function have indicated that the cytosolic enzyme αGST is an advantageous and sensitive marker in the detection of early hepatocellular damage.^9,10,14,16 In this regard, Redl et al⁹ reported that the measurement of αGST levels is more sensitive than that of ALT levels for the detection of posttraumatic hepatocellular injury after hemorrhagic shock in baboons. In that study, αGST levels increased significantly at the end of the shock period, whereas ALT levels did not increase during the entire shock and resuscitation procedure.⁹ Clarke et al¹⁶ investigated the toxic effects of carbon tetrachloride (an acute, well-characterized hepatotoxin) and found that αGST is a more sensitive marker of hepatotoxic effects compared with AST, as αGST is detected earlier and in a dosage-dependent manner after exposure to carbon tetrachloride. In addition, αGST has been demonstrated to be a sensitive and early marker compared with transaminases in the clinical realm in acute and chronic conditions such as hepatitis,¹⁷ acetaminophen overdose,^12,18,19 liver allograft rejection,¹¹ and halothane anesthesia.²⁰ Since hepatocellular damage in the centrilobular region may not be detectable using transaminase measurement, measurement of αGST appears more appropriate in situations where monitoring alterations in hepatocellular damage are essential, such as titrating immunosuppressive therapy¹⁷ or determining whether treatment of liver allograft rejection has been successful.¹¹ Although it has been demonstrated that measurement of αGST may function as a more sensitive indicator of hepatocellular damage after adverse conditions (such as hemorrhage and ischemia and reperfusion) compared with transaminases, its evaluation during sepsis compared with standard methods remained unknown. Therefore, our objectives were to determine whether αGST measurement can be used to assess early hepatocellular damage during polymicrobial sepsis and to compare this technique with other techniques used to assess deteriorations in hepatocyte integrity during adverse conditions.

As we indicated, the baseline plasma levels of αGST in the sham groups were detectable and remained consistent at all time points examined. However, in the CLP groups a significant elevation in αGST levels was observed at 5 hours after the onset of sepsis, and these levels continued to increase throughout the entire septic episode, with peak levels occurring at the last measured time point of 20 hours. Furthermore, it can be speculated that plasma levels of αGST would have continued to increase with the progression of liver damage during the late stage of sepsis. In contrast to previous findings about transaminases in our laboratory,⁶ plasma αGST is a more sensitive indicator of early hepatocellular damage following CLP. Our study indicates that αGST measurement can detect liver injury as early as 5 hours after the onset of sepsis, whereas previous work in our laboratory has demonstrated that ALT and AST are not significantly elevated until 10 hours after CLP.^6,21 Moreover, plasma lactate levels, which have been used as a global measure of tissue hypoxia and particularly as an indicator of hepatic injury, did not increase significantly until 20 hours after the onset of sepsis. Therefore, it appears that plasma αGST, compared with transaminases or lactate, is released and detected more readily after hepatocellular damage during the early stage of sepsis.

Indeed, the results of this experiment suggest that αGST measurement is a more sensitive and advantageous method of detecting liver injury compared with measurement of transaminases and lactate. Measurement of this enzyme can detect hepatocellular injury when measurement of liver transaminases may not detect damage that is occurring or has already occurred. This finding will prove to be critical in the clinical realm to determine whether hepatocellular damage is occurring and whether a particular treatment is effective. Previous studies have indicated that administration of the pharmacologic agent pentoxifylline early after the onset of sepsis maintains hepatocellular function during sepsis.²² Furthermore, studies have shown that the beneficial effects of pentoxifylline are lost once septic shock is established, indicating that the delay in administration may be ineffective or even deleterious.²³ Thus, αGST will be useful given the therapeutic window of pharmacologic interventions. Therefore, plasma αGST levels may function as a sensitive and useful marker of hepatic damage during sepsis to elucidate the efficacy of various treatments and the mechanism underlying hepatocellular damage induced by sepsis.

In addition to the assessment of liver injury using plasma αGST measurement, electron microscopic examination of hepatic tissue in early and late sepsis revealed marked alterations in hepatocyte ultrastructure and integrity, including dilation of the bile canaliculi, reduction of the canalicular microvilli, and the development of lamellated membranes in the lumen. These changes can be attributed to the metabolic injury due to the accumulation of bile acids in the hepatocyte. Since hepatocytes play a crucial role in bile production by promoting the uptake, transformation, and excretion of blood components into the bile canaliculi, the observed alterations indicate a disruption of this process. Thus, results of electron microscopy also indicate that alterations in hepatocellular integrity occur in early sepsis, with the injury continuing in late sepsis.

Because the sepsis model of CLP results in an acute and continuous flow of bacteria from the punctured cecum into the peritoneal cavity, blood cultures yield positive findings within 1 hour after the onset of sepsis and circulating levels of endotoxin are elevated.^24,25 Cecal bacteria and bacterial products within the peritoneal cavity probably seed the portal circulation and expose the liver to toxic products. Our preliminary results indicate that portal levels of endotoxin increased significantly at 10 hours after CLP, when systemic levels of αGST also increased significantly. Moreover, plasma levels of endotoxin in systemic blood increased to the same extent as those in portal blood at 10 hours after the onset of sepsis. This suggests that intestinal endotoxin translocation may not play a major role in increasing circulating levels of endotoxin in the sepsis model of CLP. However, hepatocellular dysfunction and damage have been found to occur early after CLP, despite increased cardiac output, decreased peripheral resistance, and increased hepatic microvascular perfusion.⁴ In addition, proinflammatory cytokines are up-regulated early after the onset of sepsis,^4,26 and previous studies from our laboratory have indicated that tumor necrosis factor α infusion at a dose that does not affect hemodynamic measurements produces hepatocellular dysfunction.²⁷ Although bacteria and bacterial products are present in the portal circulation, the mechanism of hepatocellular dysfunction and damage appears to be multifactorial and may involve more than the presence of pathogens in portal circulation.

Aside from the use of plasma liver enzymes to detect alterations in hepatocyte integrity, the ICG clearance technique has been used in several studies to determine hepatocellular dysfunction during the early stage of sepsis.^7,8,21,26 Indocyanine green is a nontoxic tricarboxycyanine green dye that is bound to albumin and cleared from circulation exclusively by the liver through an energy-dependent membrane transport process involving the organic anion transporting protein, which has recently been cloned and characterized functionally.^4,28,29 This technique is extremely sensitive and can detect hepatocellular abnormalities as early as 1.5 hours after the onset of sepsis by assessing the maximal velocity of ICG clearance and the efficiency of the active transport by using a fiberoptic probe and in vivo hemoreflectometer.⁸ This technique is distinct from assessment of plasma liver enzymes, as it measures hepatocellular function, rather than injury, and therefore can reflect hepatocellular dysfunction as opposed to hepatocellular damage.⁴ However, the ICG clearance technique is relatively invasive, requires expensive equipment, and may not be readily available in most laboratories or at the bedside. In view of this limitation, we propose that serial measurements of plasma αGST could be a sensitive and useful marker of assessing hepatic damage during various stages of sepsis.

The fact that plasma αGST levels increase progressively from 5 to 20 hours after CLP indicates that hepatocellular damage is indeed progressing as the septic episode continues and that levels of this enzyme correlate well with the severity of damage. However, because αGST is a cytosolic enzyme that is rapidly released from damaged hepatocytes, and because of the short half-life of αGST (less than 1 hour), we would expect to see its levels return to normal if liver damage has ceased as a result of various interventions such as resecting the septic focus. In this regard, recent studies from our laboratory have indicated that CLP followed by cecal excision at 20 hours after the onset of sepsis was associated with a mortality rate of 50%.³⁰ However, if the necrotic cecum was excised at 10 hours after CLP, no mortality was observed. One would expect that the elevated levels of circulating αGST will eventually return to the reference range in the survivors following CLP and cecal excision. In addition, although most studies performed using measurement of this enzyme to detect hepatocellular damage have involved models of liver damage, one would also expect that liver failure would occur if levels of αGST increased or remained persistently elevated.

In summary, our results indicate that the use of αGST is an advantageous method of detecting early hepatocellular damage during polymicrobial sepsis. Moreover, αGST measurement appears to be a more sensitive and early indicator of hepatic injury than the standard method of measuring transaminase levels and thus should be used to monitor hepatocellular damage as well as to investigate the mechanism responsible for producing hepatocellular damage during sepsis.

Box Section Ref ID

This investigation was supported by grant R29 GM 53008 from the National Institutes of Health, Bethesda, Md (Dr Wang). Dr Wang is the recipient of Independent Scientist Award KO2 AI 01461, National Institutes of Health.

We thank Zheng F. Ba for performing the plasma lactate assays.

Reprints: Ping Wang, MD, Center for Surgical Research, Rhode Island Hospital, Middle House II, 593 Eddy St, Providence, RI 02903 (e-mail: pwang@lifespan.org).