Severe Ischemic Early Liver Injury After Cardiac Surgery

Severe Ischemic Early Liver Injury After Cardiac Surgery Jai S. Raman, FRACS, Kazuhiro Kochi, MD, Hiroshi Morimatsu, MD, Brian Buxton, FRACS, and Rinaldo Bellomo, MD Departments of Cardiac Surgery, and Intensive Care and Medicine, University of Melbourne, Austin and Repatriation Medical Centre, Melbourne, Victoria, Australia Background. The epidemiology, pathogenesis and prognosis of severe ischemic early liver injury (SIELI) after cardiac surgery are poorly understood. Accordingly, we studied patients whose alanine transaminase (ALT) concentration acutely increased above 500 IU/l in the immediate postoperative period and compared these patients to two control groups matched for preoperative and immediate postoperative characteristics. Methods. We used a prospective database of 1,800 consecutive cardiac surgical cases to identify the study groups. Group I was made up of 20 patients with ALT levels above 500 IU/L in the acute postoperative stage (SIELI). Preoperative liver tests were normal in all these patients. Group II was obtained by identifying 20 control cases whose age, type of surgery, NYHA classification, and Parsonnet score matched Group I (preoperative controls). Group III was obtained by identifying 20 patients who developed postoperative acute renal failure and shock (ARF/shock; postoperative controls) but no enzyme evidence of hepatic injury. Results. Acute renal failure, a low cardiac index (CI) state, and mortality were more common in SIELI and ARF/Shock patients compared with preoperative controls (all p values less than 0.01). Peak postoperative pulmonary artery occlusion (PAOP) and central venous (CVP) pressures were also higher in SIELI and ARF/Shock Severe early (first 48 hours) ischemic liver injury (SIELI) after cardiac surgery is considered a rare postoperative complication with limited mention in the literature [1]. Its incidence, prognosis, pathogenesis, and clinical associations are poorly understood. This is unfortunate because such injury is likely to significantly complicate the postoperative course of cardiac surgery patients and to have important prognostic implications. Furthermore, an understanding of its associations and clinical course may assist clinicians in its prevention. Accordingly, we sought to investigate the incidence, clinical associations, course, and prognosis of SIELI after cardiac surgery by comparing a cohort of patients with this condition with two groups of cardiac surgery patients matched for preoperative or postoperative characteristics. We now report our findings. Accepted for publication June 7, 2002. Address reprint requests to Dr Raman, Department of Cardiac Surgery, Austin and Repatriation Medical Centre, Studley Road, Heidelberg, Melbourne, Victoria 3084, Australia; e-mail: jai.raman@armc.org.au. patients than controls (all p values less than 0.02). A higher dose of norepinephrine and milrinone were required to maintain blood pressure and cardiac output in SIELI and ARF/shock patients than preoperative controls (all p values less than 0.005). SIELI patients, however, differed from ARF/Shock patients in that they had a higher preoperative NYHA class and a greater incidence of hypertension and diabetes. Stepwise linear regression analysis identified a postoperative low CI and the presence of diabetes as the only predictors of peak ALT value (p less than 0.05). Multivariate logistic regression analysis identified female gender as being associated with a higher likelihood of SIELI (odds ratio: 6.13; 95% CI 1.08 to 34.82) Conclusions. SIELI after cardiac surgery carries a high mortality and is strongly associated with a low cardiac output and increased filling pressures, suggesting that liver ischemia induced by a combination of decreased perfusion and congestion is fundamental to its pathogenesis. A history of cardiac failure, diabetes, hypertension, and female gender may represent risk factors for its development and severity in the setting of a postoperative low cardiac output state. (Ann Thorac Surg 2002;74:1601 6) 2002 by The Society of Thoracic Surgeons Material and Methods The Institutional Ethics Committee waived the need for informed consent for this anonymous retrospective chart analysis. For the purpose of this study, severe ischemic early liver injury (SIELI) was defined as an acute increase in serum alanine amino transferase (ALT) levels to more than 500 IU/L within 48 hours of surgery. Among 1,800 cardiac surgical patients identified in a prospectively collected surgical database between 1997 and 2000, 20 patients were identified. All these patients had normal liver function tests before surgery. During the same period, 20 control cases were chosen from the same database (Group II). These patients were chosen to create a cohort that was matched to SIELI patients for preoperative features (preoperatively matched controls). These features included age, type of surgery, NYHA classification, and Parsonnet score [2]. During the same period, another 20 control cases were chosen from the same database (Group III). These pa- 2002 by The Society of Thoracic Surgeons 0003-4975/02/$22.00 Published by Elsevier Science Inc PII S0003-4975(02)03877-8 18

1602 RAMAN ET AL Ann Thorac Surg SEVERE LIVER INJURY 2002;74:1601 6 tients had similar early postoperative features to the SIELI patients (shock requiring vasopressor and inotropic support, prolonged mechanical ventilation, and severe acute renal failure requiring acute renal replacement therapy; ie, multiorgan failure) but did not develop enzyme evidence of ischemic liver injury (postoperatively matched controls). In all patients, induction and maintenance of anesthesia were performed using fentanyl, propofol, or isofluorane. Cardiopulmonary bypass (CPB) was maintained according to an -stat protocol. In blood gas management, the arterial ph was maintained at 7.4, and the arterial PCO 2 was regulated from 35 to 40 mm Hg. Total CPB flow was 2.6 L min 1 m 2 and mean arterial pressure during bypass was kept at greater than 70 mm Hg with the use of norepinephrine if necessary. Minimal core body temperature was 33 C. Preoperative left ventricular function, bypass time, cross-clamp time, postoperative hemodynamics, the changing of biochemical data (including ALT, creatine kinase [CK], and serum creatinine [CR]), ventilation time, severity of renal dysfunction, and mortality were compared between the groups. Postoperative low output state (LOS) was defined by a cardiac index less than 2.1 L min 1 m 2 or by the need for extracorporeal life support (ECMO) or intraaortic balloon pumping during the first 4 postoperative days. Prolonged ventilation was defined as the need for ventilation for more than 48 hours. Severe acute renal failure was defined as either a urine output less than 200 ml/12 hours or anuria in association with an acute rise in urea and creatinine concentration and a clinical decision that renal replacement (continuous hemofiltration in all cases) should be initiated. Statistical Analysis Statistical analysis was performed using the Kruskal- Wallis test and Fisher s exact test for comparison of continuous and nominal variables between the three groups. If the Kruskal-Wallis test was positive, the Mann- Whitney test was applied to binary comparisons with downward adjustment for the number of comparisons performed. Spearman s test was used to study correlations between continuous variables. We used stepwise linear regression analysis to identify the independent variables, which predicted postoperative peak ALT. We used multivariate logistic regression analysis to determine which factors might determine the development of SIELI. A commercially available computerized statistical package (Sigma Stat 2.0; SPSS Inc, Chicago, IL) was used for statistical analysis. The results are expressed as means standard deviation. Statistical significance was defined by a p less than 0.05. Results The mean age for the SIELI patients was 65.8 11.3 years old (range: 42 to 75 years old). There were 8 males and 12 females. Thirteen patients had coronary artery bypass surgery (CABG), 3 had valve surgery with CABG, and 4 patients had valve surgery only. Their mean New York Heart Association functional class was 3.4 7 0.7. In the preoperative control group, CABG was performed in 13 patients, valve surgery with CABG in 4 patients, and valve surgery in 3 patients. In the second control group of patients with acute renal failure and shock, CABG was performed in 9 patients, valve surgery in 6 patients, CABG with valve surgery in 1 patient, repair of aortic dissection in 2 patients, repair of ASD and mitral valve in 1 patient, and CABG with Bentall s procedure in 1 patient. Severe ischemic early liver injury after cardiac surgery patients had a higher mean NYHA class than controls and ARF/shock patients, a higher incidence of diabetes and hypertension and a trend towards more female patients. The preoperative features of these three groups and their differences are summarized in Table 1. Surgery induced marked changes in creatinine, ALT, and CK levels in the acute postoperative stage (Figs 1 3). The mean peak ALT level was significantly higher in SIELI patients than controls and ARF/shock patients (p 0.0001). Cardiopulmonary bypass time and aortic crossclamp time, however, were longest in patients with ARF/shock (p 0.013). Comparative data are summarized in Table 1. A postoperative low output state (LOS) was significantly more common in SIELI and ARF/shock patients than in group II patients. Hemodynamic data are summarized in Table 2. The lowest cardiac index during the first 4 postoperative days was significantly worse in SIELI patients when compared with group II but not group III patients. Both highest pulmonary artery occlusion pressure and central venous pressure were also significantly higher in SIELI and ARF/shock patients than in group II (Table 2). Higher doses of norepinephrine and milrinone were required in SIELI and ARF/shock patients to maintain an acceptable blood pressure and cardiac index (Table 2). Prolonged mechanical ventilation was also more common in Group I and III (p 0.008) and mortality rate higher in these two groups (p 0.008). There were no significant differences, however, when SIELI patients were compared to ARF/shock patients (Table 2). Peak postoperative alanine aminotransferase (ALT) levels correlated with PAOP (p 0.0138), CVP (p 0.019) and negatively with CI (p 0.019). Stepwise linear regression analysis, however, identified a low postoperative CI and the presence of diabetes as the only independent predictors of peak ALT value (p 0.04) (continuous measure of severity). Multivariate logistic regression analysis (MVLRA), however, identified female gender as the only variable associated with a higher risk of developing SIELI (dichotomous outcome; odds ratio: 6.13; 95% CI 1.08 to 34.8). There were no significant differences between survivors and nonsurvivors among patients in Group I (Table 3). MVLRA could not identify independent predictors of survival.

Ann Thorac Surg RAMAN ET AL 2002;74:1601 6 SEVERE LIVER INJURY 1603 Table 1. Comparison Between the Patients With Severe Ischemic Early Liver Injury and Controls SIELI (Group I) Controls (Group II) ARF/Shock (Group III) Age 65.7 10.7 68.5 12.1 70.1 12.5 Male/Female 8/12 13/7 15/5 Parsonnet scores 12.5 6.1* 14.3 6.6 16.9 6.43* NYHA classification 3.4 0.7* 3.2 1.2 2.3 0.81* Diabetes mellitus 13*, 6 4* Hypertension 16*, 10 8* Cardiopulmonary bypass period (min) 146.6 52.8*, 110.7 38.2, 230.4 105.0*, Aortic cross-clamp period (min) 95.9 37.6* 80.5 28.8 142.1 69.7*, Postoperative low output status a 13 4 13 Extracorporeal life support/intraaortic 4/4 1/0 3/7 balloon pumping Preoperative ALT (alanine amino 34.1 21.4 36.0 30.4 49.6 75.0 transferase) (IU/L) Postoperative maximum ALT (IU/L) 2523 1843*, 44.0 2.31 138 135* Preoperative creatinine ( mol/l) 137 47 83 40, 157 127 Postoperative creatinine ( mol/l) 227 104 122 42, 235 125 Postoperative creatine phosphokinase 3640 3956 1363 1341 2298 2010 (IU/L) Mechanical ventilation for 48 hours 15 3 12 Death 13 1, 9 a Postoperative low output state was defined by a cardiac index was less than 2.1 l/min/m 2 or if the patients required extracorporeal life support or intraaortic balloon pumping during the post-operative period. * Indicates a significant difference between SIELI and ARF group. Indicates significant difference between SIELI and preoperative control group. Indicates a difference between ARF and preoperative control group. ALT alanine amino transferase; ARF acute renal failure; NYHA New York Heart Association; SIELI severe ischemic early liver injury after cardiac surgery. Comment There are limited data in the literature about the epidemiology, clinical association and prognosis of patients with SIELI after cardiac surgery [1, 3]. The incidence, associations, probable etiology, and prognosis of this syndrome after cardiac surgery are therefore poorly understood. The first finding of our study was that the incidence of SIELI reached 1.1% (20 patients among 1800) and that its mortality was 65%. This mortality is higher than that reported in other series of ischemic hepatitis [1, 3 6]. However, very few patients with postcardiac surgery ischemic hepatitis are reported in the literature with most Fig 1. Changes in alanine amino transferase levels in the immediate postoperative period. No significant changes occurred in preoperative controls (square) and acute renal failure/shock patients (triangle), but there was a marked and rapid rise in affected patients (circle) to a mean level more than 3000 IU/L that peaked at day 3 and then returned to baseline over several days. (Pre preoperative.) Fig 2. Creatine kinase levels after surgery. There were significantly greater levels in patients with severe ischemic early liver injury after cardiac surgery (circle) compared with preoperative controls (square) but not acute renal failure/shock patients (triangle).

1604 RAMAN ET AL Ann Thorac Surg SEVERE LIVER INJURY 2002;74:1601 6 Fig 3. Serum creatinine concentration in patients with severe ischemic early liver injury after cardiac surgery (circle) and acute renal failure/shock (triangle) was significantly greater than in preoperative controls (square) in the immediate postoperative period. (Pre preoperative.) cases representing medical patients with significant cardiac disease. This study reports the outcome of a series of cardiac surgery patients and it highlights the very high mortality associated with the development of liver ischemia after cardiac surgery. Given the high mortality of these patients and the considerable effort invested in their physiologic support (mechanical cardiovascular support, continuous hemofiltration, and prolonged mechanical ventilation), it would be important to be able to identify prognostic markers that suggest a hopeless prognosis. Unfortunately, no such markers were found. Thus, we were unable to obtain data that would allow the early identification of patients whose further treatment might be futile. Being able to predict which patient features are associated with a higher risk of ischemic liver injury would also be useful becuase it might allow the subsequent study of protective strategies. In order to clarify which preoperative, intraoperative or immediate postoperative features powerfully predicted the development of SIELI, we compared affected patients with two groups: a cohort of cases matched for major preoperative features, predicted risk of surgery, and type of surgery; and a second group of patients with a similar degree of postoperative shock, who did not develop SIELI. These comparisons revealed that SIELI patients were more likely to have prolonged CPB, a low cardiac output state, require vasopressor and inotropic support, and reach high filling pressures when compared with preoperatively matched patients but not when compared with patients matched for the presence of postoperative shock. These findings are consistent with the hypothesis that SIELI is essentially secondary to low-cardiac output induced liver ischemia. Indeed, univariate analysis supported this interpretation further by revealing that higher PAOP, CVP, and peak postoperative creatine kinase levels and a lower cardiac index strongly correlated with postoperative ALT levels. Finally, stepwise linear regression analysis identified a low CI as the only independent predictor of peak ALT. These findings, however, do not explain why equally shocked patients (group III) developed acute Table 2. Hemodynamic Features of Affected Subjects and Controls a SIELI (I) Controls (II) ARF/Shock (III) p Value Lowest cardiac index (L/min/m 2 ) 1.86 0.28 2.35 0.46 2.20 0.62 *0.0035; Highest PAOP (mm Hg) 19.1 3.8 15.1 4.7 22.5 5.87 *0.019 0.01 Highest RAP (mm Hg) 16.6 5.7 12.8 0.3 17.1 3.9 *0.009 0.007 Mean radial arterial pressure (mm Hg) 80.1 15.7 80.2 13.4 72.8 9.3 * Peak norepinephrine dose ( g/min) 25.4 22.4 4.95 10.2 34.8 36.4 *0.0004 0.0003 Peak milrinone ( g/kg/min) 0.356 0.290 0.163 0.294 0.352 0.156 *0.005 0.005 a Patients with extracorporeal life support were excluded. PAOP pulmonary artery occlusion pressure; RAP right atrial pressure. * Indicates the p value for a comparison between group I and II; comparison between group I and III; and comparison between group II and III. ARF acute renal failure; PAOP pulmonary artery occlusion pressures; RAP radial artery pressure; SIELI severe ischemic early liver injury after cardiac surgery.

Ann Thorac Surg RAMAN ET AL 2002;74:1601 6 SEVERE LIVER INJURY 1605 Table 3. Comparison Between Survivors and Nonsurvivors Survivors (n 7) Nonsurvivors (n 13) p Value Age 64.5 11.5 65.7 11.1 Male/Female 2/5 6/7 Parsonnet scores 12.2 7.0 12.6 6.2 NYHA classification 3.2 0.8 3.5 0.7 Diabetes mellitus 5 8 Hypertension 5 11 Cardiopulmonary bypass period (min) 144.7 55.7 148.6 56.3 Aortic cross-clamp period (min) 92.8 41.1 96.9 39.1 Postoperative low output status* 5 8 Extracirculatory life support/intraaortic 2/1 2/3 balloon pumping Preoperative ALT (IU/L) 22.3 7.2 41.2 23.3 Postoperative maximum ALT (IU/L) 2210.0 1556.2 2745.1 2045.4 Preoperative creatinine ( mol/l) 85 19 160 96 Postoperative CR ( mol/l) 87 98 245 110 Postoperative creatine phosphokinase (IU/L) 3000 3507 4144 4307 * Postoperative low output state was determined if the cardiac output index was less than 2.1, or if the patients required extracorporeal life support or intraaortic balloon pumping during the first 4 postoperative days. ALT alanine transferase; CR serum creatine; NYHA New York Heart Association. renal failure but not SIELI. Direct comparisons with this group and multivariate logistic regression analysis suggest that greater levels of preoperative cardiac failure, the presence of diabetes, a history of hypertension and female gender might increase the risk that a postoperative shock state will induce SIELI, while a longer cardiopulmonary bypass time and aortic cross-clamp time appear to increase the probability that the shock state will induce acute renal failure. These observations, however, must be taken with a great deal of caution, because of the relatively small size of the patient groups under study. One limitation of our investigation is that the inclusion of patients in the SIELI group rests upon a clinical rather than histologic diagnosis and, as such, its correctness might be called into question. The syndrome of so-called ischemic hepatitis has been described under a variety of circumstances associated with poor cardiac output and, in some cases, low mean arterial blood pressure and venous congestion [4 8]. Its defining features are the clinical context in which it occurs and the particular pattern of enzyme change. Typically, there is a greater than tenfold increase in ALT levels and within 48 hours of the injury and a simultaneous increase in lactate dehydrogenase (LDH). Such elevations in ALT can reach values of more than 3000 IU/L and the LDH levels can reach values more than 10,000 IU/L. Both enzymes return to baseline levels within several days with the LDH doing so in 2 to 3 days and the ALT more typically taking 5 to 7 days. This pattern is characteristic and differs from that of infectious hepatitis or anesthetic agent induced hepatitis. All patients reported in our studies followed this pattern. The other typical clinical feature is that ischemic hepatitis is rarely associated with the typical features of severe liver failure (encephalopathy, prolonged INR, hypoglycemia and so on) [5 7]. Histopathology reveals centrilobular hepatic necrosis, which is characteristic and quite distinct from the findings seen in acute viral or druginduced hepatitis [9, 10]. However, biopsy is rarely performed because the clinical syndrome is so characteristic. None of our patients had a liver biopsy. However, all patients who underwent autopsy had evidence of the typical histologic features of ischemic hepatitis. Furthermore, in all our cases, the presence of an acute low cardiac output syndrome, the need for high doses of vasopressors, the need for heavy inotropic support, the correlation of ALT levels with high cardiac filling pressures, and a low cardiac index are all consistent with the diagnosis of so-called ischemic hepatitis. Given our findings, it might seem appropriate to use the term ischemic hepatitis from the outset rather than the acronym SIELI. However, the term is misleading by implying that inflammation (hepatitis) is a major histologic and pathogenetic aspect of this syndrome. This problem has been previously pointed out [6]. Acute hepatic infarction may be a more correct term [6]. Ischemic liver injury may be better. We have used the term severe, because some liver and splanchnic ischemia probably occurs in many patients during and after cardiac surgery [11] but does not lead to major clinical complications. We have also used the word early to make clear that the pathogenesis of the syndrome cannot relate to other types of liver injury reported. In summary we have reported a series of patients with severe ischemic early liver injury (so-called ischemic hepatitis or shock liver ) after cardiac surgery and have described the incidence of this syndrome, its outcome, clinical associations, likely pathogenetic features, and

1606 RAMAN ET AL Ann Thorac Surg SEVERE LIVER INJURY 2002;74:1601 6 possible risk factors. This information should assist clinicians in its correct identification and help them appreciate its clinical course and prognosis. This study was supported by the Anaesthesia and Intensive Care Trust Fund of the Austin and Repatriation Medical Centre. References 1. Michalopoulos A, Alivizatos P, Geroulanos S. Hepatic dysfunction following cardiac surgery: determinants and consequences. Hepatogastroenterology 1997;44:779 83. 2. Parsonnet V, Dean D, Bernstein AD. A method of uniform stratification of risk for evaluating the results of surgery in acquired adult heart disease. Circulation 1989;79(Supp I): 3 11. 3. Bynum TE, Boitnott JK, Maddrey WC. Ischemic hepatitis. Dig Dis Sci 1979;24:129 35. 4. Editorial. Ischemic hepatitis. Lancet 1985;1:1019 20. 5. Gitlin N, Serio KM. Ischemic hepatitis: widening horizons. Am J Gastroenterol 1992;87:831 6. 6. Cohen JA, Kaplan MM. Left-sided heart failure presenting as hepatitis. Gastroenterology 1978;74:583 7. 7. Gibson PR, Dudley FJ. Ischemic hepatitis: clinical features, diagnosis and prognosis. Aust NZ J Med 1984;14:822 5. 8. Rawson JS, Achord JL. Shock liver. Southern Med J 1985;78: 1421 5. 9. Lefkovitch JH, Mendez L. Morphologic features of hepatic injury in cardiac disease and shock. J Hepatol 1986;2:313 27. 10. De la Monte SM, Arcidi JM, Moore GW, et al. Midzonal necrosis as a pattern of hepatocellular injury after shock. Gastroenterology 1984;86:627 31. 11. Jakob SM, Ruokonen E, Takala J Assessment of the adequacy of systemic and regional perfusion after cardiac surgery. Br J Anaesth 2000 May;84(5):571 7.