Patients with cirrhosis frequently develop sepsis because

Transcription

1 GASTROENTEROLOGY 2005;129: Renal Failure in Patients With Cirrhosis and Sepsis Unrelated to Spontaneous Bacterial Peritonitis: Value of MELD Score CARLOS TERRA,*,, MÓNICA GUEVARA,*,, ALDO TORRE,*,, ROSA GILABERT, JAVIER FERNÁNDEZ,*, MARTA MARTÍN-LLAHÍ,*,, MARIA E. BACCARO,*,, MIQUEL NAVASA,*, CONXITA BRU, VICENTE ARROYO,*,, JUAN RODÉS,*,, and PERE GINÈS*,, *Liver Unit, Hospital Clínic, Universitat de Barcelona, Barcelona; Institut d Investigacions Biomèdiques August Pi-Sunyer (IDIBAPS), Barcelona; Instituto Reina Sofia de Investigación Nefrológica, Barcelona; and Radiology Department, Hospital Clínic, Universitat de Barcelona, Barcelona, Spain Background & Aims: Although renal failure is a common complication of sepsis and patients with cirrhosis frequently develop sepsis, there have been no studies specifically assessing renal function in patients with cirrhosis and sepsis unrelated to spontaneous bacterial peritonitis. The aim of this study was to investigate prospectively the frequency, characteristics, and outcome of renal failure in patients with cirrhosis and sepsis unrelated to spontaneous bacterial peritonitis. Methods: One hundred six consecutive patients with cirrhosis and sepsis were studied prospectively. Patients with spontaneous bacterial peritonitis were excluded. Results: Twenty-nine out of 106 patients (27%) with cirrhosis and sepsis developed acute renal failure as compared with only 8 of 100 patients (8%) from a control group of cirrhotic patients without infection (P <.0001). Renal failure in the sepsis group was reversible in 22 (76%; 21% of all patients) patients and nonreversible in 7 (24%; 6% of all patients) patients. Renal failure was associated with impairment of effective arterial blood volume, without evidence of tubular damage. The occurrence and type of renal failure correlated strongly with mortality (mortality at 3 months: nonreversible renal failure, 100%; reversible renal failure, 55%; no renal failure, 13%). Among variables obtained at diagnosis of sepsis, the Model for End-Stage Liver Disease (MELD) score was the only independent predictive factor of mortality. Conclusions: Renal failure is common in patients with cirrhosis and sepsis unrelated to spontaneous bacterial peritonitis and is associated with arterial underfilling and renal vasoconstriction. Outcome is poor, even in the setting of reversible renal failure. The MELD score is the best prognostic marker of patients with cirrhosis and sepsis. Patients with cirrhosis frequently develop sepsis because of severe disturbances in the defensive mechanisms against bacterial infections. 1 3 On the other hand, cirrhosis is associated with marked alterations in circulatory function with arterial underfilling because of splanchnic arterial vasodilation. 4,5 This circulatory dysfunction may theoretically predispose to the development of renal failure in the setting of sepsis because a major mechanism of renal failure in sepsis is a reduction in systemic vascular resistance with subsequent neurohumoral-mediated renal vasoconstriction. 6,7 Most of the information available on the association between bacterial infections and renal failure in cirrhosis refers to spontaneous bacterial peritonitis (SBP). Renal failure is common in patients with SBP, and its occurrence is associated with a poor outcome The high incidence of renal failure in the setting of SBP is related to a marked circulatory dysfunction and a high local production of cytokines. 10,11,13,16 18 In sharp contrast with the large amount of information on renal failure in SBP, there is little information on renal failure associated with sepsis unrelated to SBP in cirrhosis. Therefore, the current study was undertaken to investigate prospectively the frequency, characteristics, and outcome of renal failure associated with sepsis unrelated to SBP in cirrhosis. Patients and Methods Patient Population All patients with cirrhosis and sepsis unrelated to SBP seen at the liver unit of the Hospital Clínic of Barcelona between January 2002 and June 2003 were included in a prospective study aimed at investigating the frequency, characteristics, pathogenic mechanisms, and outcome of acute renal failure associated with sepsis in cirrhosis. A control group of patients with cirrhosis without infection admitted to the hospital for management of complications of the disease during the same period of time was also studied. Abbreviation used in this paper: SBP, spontaneous bacterial peritonitis by the American Gastroenterological Association /05/$30.00 doi: /j.gastro

2 December 2005 RENAL FAILURE IN CIRRHOSIS AND SEPSIS 1945 Study Group: Patients With Cirrhosis and Sepsis The study group includes 106 patients with cirrhosis and sepsis unrelated to SBP. Patients with SBP were excluded because renal failure in the setting of SBP has been extensively studied in previous investigations, 8 15 and these patients are currently treated in our unit with intravenous albumin to reduce the incidence of renal failure. 11 In 83 of these 106 patients (78%), the infection was the main cause of admission to the hospital, whereas the remaining 23 patients (22%) developed the infection during hospitalization. Criteria for exclusion were as follows: (1) Patients with septic shock (n 9) within the first 24 hours of diagnosis of the infection, as defined by a decrease in systolic blood pressure below 90 mm Hg or a reduction of more than 40 mm Hg from baseline, despite adequate fluid resuscitation, accompanied by tachycardia and oliguria (urine output less than 20 ml/h) or anuria in the absence of other causes of shock 19,20 ; these patients were excluded because all studies so far reported assessing renal failure in SBP have excluded patients with septic shock at diagnosis of the infection. (2) Patients treated with liver transplantation (n 14) or with associated infection with the human immunodeficiency virus (n 3); these 2 conditions were excluded because the natural history of cirrhosis is markedly influenced by the associated severe immunosuppression (3) Death within the first 48 hours after diagnosis of sepsis (n 3). Control Group: Patients With Cirrhosis Without Infection The control group includes 100 consecutive patients with cirrhosis who neither had infection at admission to the hospital nor developed infection during hospitalisation. Main causes of admission in this group of patients were gastrointestinal bleeding in 42 patients (42%), ascites in 41 patients (41%), and hepatic encephalopathy in 17 patients (17%). Clinical Management The work-up of a suspected bacterial infection included the following: (1) systematic medical history, with particular emphasis on symptoms of infection; (2) complete physical examination and measurement of body temperature, heart rate, and arterial pressure; (3) laboratory tests including white blood cell count, liver and renal tests, and urine sediment; (4) diagnostic paracentesis in patients with ascites to obtain a sample of ascitic fluid for culture and polymorphonuclear count; (5) chest x-ray; (6) blood and urine cultures; and (7) culture of other organic fluids or secretions (ie, cerebrospinal fluid, sputum) when clinically indicated. Patients were then started on empiric antibiotic therapy intravenously and managed according to widely accepted guidelines. 25 The empiric antibiotic therapy was as follows: (1) community-acquired pneumonia: ceftriaxone plus azitromycin; (2) nosocomial pneumonia: imipenem plus ciprofloxacin; (3) urinary tract infection: ceftriaxone; (4) skin infection: amoxicillinclavulanic acid; (5) biliary tract infection: piperacillin-tazobactam; and (6) no evident source of infection: ceftriaxone for community-acquired infections and ceftriaxone plus teicoplanin for nosocomial infections. The antibiotic dosage was adjusted according to renal function throughout the treatment period. The empiric antibiotic treatment was modified on the basis of the results of cultures and antibiotic susceptibility in vitro. Antibiotic treatment was maintained until the disappearance of signs and symptoms of infection, normalization of white blood cell count, and negative cultures. Patients were given intravenous fluids (dextrose 5%) to ensure an adequate hydration status. Vasopressor drugs were given only to patients who developed signs of multiorgan failure during the course of the infection. Complications of cirrhosis were managed with standard therapy. 26 In patients with infection who also had ascites, diuretics were withdrawn at diagnosis of the infection and were restarted after infection resolution; infected patients with large ascites were treated with paracentesis after the resolution of the infection, except in cases with tense ascites and marked abdominal discomfort, who were treated with paracentesis before resolution of the infection. Intravenous albumin was only given to patients requiring paracentesis. 27 Assessment of Renal Function Renal function was assessed by measuring serum creatinine concentration at admission and throughout the hospitalization. In patients who developed bacterial infection during hospitalization, serum creatinine was also measured at diagnosis of the infection. Serum electrolytes were also determined. In patients developing renal failure, causes of renal failure other than sepsis were specifically sought. 28,29 The existence of volume depletion was ruled out by medical history, exploratory findings, and assessment of central venous pressure when clinically indicated. The administration of nephrotoxic drugs in the preceding days before the development of renal failure was also excluded. The presence of a parenchymal renal disease was ruled out by renal ultrasonograhpy, measurement of urine protein, and examination of urine sediment. In patients developing renal failure, urine output was monitored. Furthermore, blood samples were obtained within the first 48 hours of diagnosis of renal failure to measure plasma renin activity (PRA) and the plasma concentrations of aldosterone, norepinephrine, arginine vasopressin, endothelin, and atrial natriuretic peptide. In 10 patients with sepsis and renal failure, renal vascular resistance was estimated by measuring renal resistive index in interlobar arteries using Doppler ultrasonography. The sonographic studies were performed with ultrasound equipments with color Doppler capabilities: Sequoia 512 (Acuson, Mountain View, CA) or Antares (Siemens, Germany). The renal parenchyma was scanned with a multifrequency (2 4 MHz) convex or sector transducer with the patient in supine or lateral decubitus position. Pulsed Doppler intrarenal Doppler wave forms were obtained from the interlobar arteries at the upper, middle, and lower third of both

3 1946 TERRA ET AL GASTROENTEROLOGY Vol. 129, No. 6 kidneys, using a 3-mm sample size and the low-velocity range and wall filter that provided an arterial waveform without artifacts. The arterial waveforms were analyzed semiquantitatively by the resistive index (RI) (RI peak systolic velocity minimum diastolic velocity/peak systolic velocity). The analysis of results was based on the average of the 3 measurements in each kidney. Previous studies in cirrhosis as well as in other conditions associated with renal failure have shown that the resistive index of renal arteries is an accurate method to estimate renal vascular resistance The analytical procedures for determination of neurohormonal parameters have been described elsewhere. 34 Definitions Cirrhosis. The diagnosis of cirrhosis was based on liver biopsy, when available, or clinical, biochemical, ultrasonography, and endoscopy findings. Type of bacterial infection. Urinary tract infection, cellulitis, biliary tract infection, gastroenteritis, and meningitis were defined according to standard criteria. If a positive blood culture was obtained in patients with these infections, it was considered as secondary to the primary source of infection. Pneumonia was defined according to criteria described elsewhere. 35 Spontaneous bacteremia was defined in the presence of positive blood cultures without evident source of infection. Culture-negative sepsis was defined as presence of fever ( 38 C) and/or leukocytosis with immature (band) forms and negative cultures after exclusion of conditions other than infection that could be responsible for the systemic inflammatory response. The infections were considered community acquired when they developed outside the hospital or within the first 48 hours of hospitalization. All other infections were considered nosocomial. Renal failure. Renal failure was defined as previously described. 11 The cut-off level of serum creatinine used to define renal failure was 1.5 mg/dl. This value was chosen because patients with cirrhosis and a serum creatinine level greater than 1.5 mg/dl have an average glomerular filtration rate of approximately 30 ml/min. 36 In patients without preexisting renal failure, renal failure was diagnosed whenever there was an increase of serum creatinine of 50% or greater, with a final value above 1.5 mg/dl in 2 consecutive measurements within a 24-hour period. In patients with preexisting renal failure before infection, an increase in the serum creatinine level by more than 50% from baseline was required for the diagnosis of renal failure. Renal failure was classified into 2 categories: (1) reversible, when serum creatinine decreased during hospitalization to a value of less than 30% over baseline; or (2) nonreversible, when serum creatinine did not return to baseline values and remained elevated (at least 30% over baseline values) over 1.5 mg/dl until the end of hospitalization or death. 11,37,38 Hyponatremia. Hyponatremia was defined as a serum sodium concentration below 130 meq/l. 39 Follow-up After discharge from the hospital, most patients were seen at the outpatient clinic of the Hospital Clínic of Barcelona for continued care. Survival was analyzed at 3 months after entering the study. Patients not followed up at our institution (n 17) were contacted by phone to know whether they were alive or not at the end of the 3-month follow-up period. Statistical Analysis The analysis of survival was performed using the Kaplan Meier method. Clinical and analytical variables were analyzed as possible predictors of survival in a univariate analysis, and survival curves were compared with the log-rank test. A multivariate analysis of survival was performed using a Cox regression method. Patients treated with liver transplantation during follow-up (3 in the sepsis group and none in the control group) were considered censored at time of surgery. Patients with advanced tumors who died during the 3-month follow-up period (n 5) were excluded from the analysis of survival. Comparisons of variables between patients with and without renal failure were made using the Student t or Mann Whitney tests for continuous data and the 2 test for categorical data. The accuracy of the MELD score in predicting 3-month survival was calculated using the receiving operating curve (ROC). Statistical analysis was performed using the SPSS 10 for Windows (SPSS Inc, Chicago, IL). Results are expressed as mean SD, with P.05 considered as statistically significant. Results Characteristics of Patients Table 1 shows the demographic, clinical, and biochemical data of all patients included in the study. As expected, patients with sepsis had higher body temperature and leukocyte count compared with values in patients without infection. Moreover, patients with sepsis had greater frequency of hepatic encephalopathy, lower albumin and serum sodium levels, higher serum creatinine, and higher Child Pugh and MELD scores compared with patients without infection. Table 2 shows the type of infection and responsible bacteria in patients with sepsis. The resolution of the infection was obtained in 98 patients (93%). In the remaining 8 patients, infection did not resolve, and patients died with signs of active infection. Twenty-three patients (22%) required a modification of the initial antibiotic therapy (in 10 cases because of the results of cultures and in 13 cases because of persistent signs of infection in the setting of negative cultures). Renal Failure Incidence. Renal failure developed in 29 of the 106 patients (27%) with sepsis compared with only 8 of

4 December 2005 RENAL FAILURE IN CIRRHOSIS AND SEPSIS 1947 Table 1. Demographic, Clinical, and Biochemical Characteristics of the 106 Patients with Sepsis at the Time of Diagnosis of Infection and the 100 Patients With Cirrhosis Without Infection Characteristic Study group (n 106) Control group (n 100) P value Age (y) Sex (male/female) 54/52 63/37.08 Etiology of cirrhosis (%).2 Alcohol a 40 (38) 37 (37) Hepatitis C 52 (49) 52 (52) Other b 14 (13) 11 (11) Hepatocellular carcinoma, n (%) 27 (25) 21 (21).4 Ascites, n (%) 55 (52) 56 (56).5 Hepatic encephalopathy, n (%) 35 (33) 19 (19).02 Temperature ( C) Mean arterial pressure (mm Hg) Heart rate (beats/min) Albumin (g/l) Bilirubin (mg/dl) Prothrombin time Ratio (%) INR Child-Pugh A/B/C, n 16/51/39 30/47/23 Score MELD score AST (IU/L) ALT (IU/L) Alkaline phosphatase (IU/L) Gama glutamyl transpeptidase (IU/L) Serum creatinine (mg/dl) Serum sodium (meq/l) Hyponatremia, c (%) 23 (22) 10 (10).02 Serum potassium (meq/l) Leukocyte count ( 10 3 /mm 3 ) a Associated with hepatitis C virus infection in 13 patients in the sepsis group and in 6 patients in the control group. b Study group: autoimmune (4), primary biliary cirrhosis (3), 1 -antitripsin deficiency (2), secondary biliary cirrhosis (2), primary sclerosing cholangitis (1), nonalcoholic steatohepatitis (1), cryptogenic (1); Control group: autoimmune (2), primary biliary cirrhosis (4), secondary biliary cirrhosis (1), cryptogenic (4). c Serum sodium lower than 130 meq/l. Table 2. Type of Infection and Responsible Bacteria in the 106 Patients With Cirrhosis and Sepsis Included in the Study Number of patients (%) Community/ nosocomial Positive cultures (%) Type of infection Pneumonia 34 (32) 30/4 9 (26) Urinary tract 30 (28) 21/9 28 (93) Celullitis 14 (13) 14/0 8 (57) Spontaneous bacteremia 11 (10) 5/6 11 (100) Biliary tract a 8 (8) 7/1 4 (50) Culture-negative sepsis b 6 (6) 4/2 0 (0) Other c 3 (3) 2/1 3 (100) Total 106 (100) 83/23 63 (59) Isolated bacteria Gram-negative bacteria 35 (33) 26/9 Escherichia coli 24 (22) 18/6 Klebsiella pneumoniae 4 (4) 4/0 Pseudomonas aeruginosa 3 (3) 3/0 Other d 4 (4) 1/3 Gram-positive bacteria 28 (26) 18/10 Streptococcus pneumoniae 8 (8) 6/2 Staphylococcus aureus 7 (7) 3/4 Enterococcus faecalis 6 (6) 2/4 Other e 7 (7) 7/0 a Acute cholangitis in 7 patients and acute cholecystitis in 1. b See definition in Patients and Methods section. c Meningitis (1), liver abscess (1), and secondary peritonitis due to perforated gastric ulcer (1). d Hafnia alvei (2), Klebsiella oxytoca (1), Providencia stuartii (1). e Listeria monocitogenes (2), Staphylococcus epidermidis (2), Strectococcus agalactiae (1), Streptococcus bovis (1), Streptococcus pyogenes (1). pneumonia, 29%; urinary tract infection, 10%; and other infections, 18% (P.03, overall). Characteristics of renal failure in patients with sepsis. Table 3 shows the characteristics of renal function and systemic hemodynamics in the 29 patients with the 100 patients (8%) without infection (P.0001). In 17 of the 83 patients (20%) admitted to the hospital because of sepsis, renal failure was already present at the time of diagnosis of sepsis. In the remaining 12 patients, renal failure developed at variable time intervals after the diagnosis of sepsis (range, 1 14 days). In the control group of patients without infection, renal failure was present at admission in 6 patients, and, in the remaining 2 patients, renal failure developed during hospitalization. The probability of developing renal failure in the 2 groups is shown in Figure 1. The incidence of renal failure in patients with sepsis, classified according to the type of infection, was as follows: culture-negative sepsis, 66%; spontaneous bacteremia, 45%; cellulitis, 35%; Figure 1. Probability of development of renal failure in the 106 patients with cirrhosis and sepsis (solid line) and in the 100 patients with cirrhosis without infection (dashed line) included in the study.

5 1948 TERRA ET AL GASTROENTEROLOGY Vol. 129, No. 6 Table 3. Characteristics of Renal Function and Systemic Hemodynamics in Patients With Cirrhosis and Sepsis With and Without Renal Failure Parameter Renal failure (n 29) No renal failure (n 77) P value Renal function Serum creatinine (mg/dl) BUN (mg/dl) Creatinine clearance (ml/min) Renal resistive index a NA Serum sodium (meq/l) Serum potassium (meq/l) Serum osmolality (mosm/kg) NA Urine volume (ml/day) Urine sodium (meq/l) Urine osmolality (mosm/kg) NA Proteinuria (mg/day) microglobulin ( g/g/creatinine) NA Systemic hemodynamics Mean arterial pressure (mm Hg) (60 110) Plasma renin activity (ng/ml/h) b (0.1 57) Aldosterone (ng/dl) b ( ) Norepinephrine (pg/ml) b ( ) Arginine vasopressin (ng/l) c (0.3 30) NA Endothelin (pmol/l) c 18 5 (7 27) NA Atrial natriuretic peptide (fmol/ml) b (22 140) NOTE. Values are mean SD (range). Normal values of 2 -microglobulin in patients with cirrhosis and ascites without renal failure are g/g creatine 37 ; normal values of renal resistive index in healthy subjects are Normal values of vasoactive factors in healthy subjects in our laboratory are plasma renin activity ng/ml/h; aldosterone ng/dl; norepinephrine pg/ml; arginine vasopressin mg/l; endothelin 5 3 pmol/l; atrial natriuretic peptide fmol/ml. a Available in 10 patients. b Available in 21 patients with renal failure and 18 patients without renal failure. c Available in 21 patients with renal failure. sepsis who developed renal failure and in the 77 patients with sepsis who did not develop renal failure. Renal failure was characterized by marked reduction of glomerular filtration rate and high renal vascular resistance in the setting of preserved tubular function, the latter indicated by low urine sodium, urine osmolality higher than plasma osmolality, and normal urinary excretion of 2 -microglobulin, a sensitive marker of tubular damage. 40 Oliguria was uncommon, and most patients had preserved urine volume. Proteinuria was lower than 500 mg/day in all patients. In patients with renal failure, the renal resistive index, which estimates the degree of vasoconstriction of the renal circulation, was markedly increased above normal values (Figure 2). Finally, patients with renal failure had lower arterial pressure and greater activation of the renin-angiotensin and sympathetic nervous systems and a higher level of atrial natriuretic peptide compared with patients without renal failure. Taken together, these findings indicate that renal failure in patients with cirrhosis and sepsis is characterized by a very severe renal vasoconstriction with preserved tubular function in the setting of a marked impairment of effective arterial blood volume. In the sepsis group, renal failure was reversible in 22 of the 29 patients (76%) and nonreversible in the remaining 7 patients (24%). In 6 of these 7 patients with nonreversible renal failure, sepsis did not resolve despite modification of antibiotic therapy, and patients died with active infection and multiorgan failure. By contrast, in patients with reversible renal failure, lack of resolution of the infection was only seen in 1 of the 22 patients (P.001). In the control group of patients without infec- Figure 2. Individual values of renal resistive index, as assessed by Doppler ultrasonography, in 10 patients with renal failure associated with sepsis. The shaded area represents normal values (mean 2 SD).

6 December 2005 RENAL FAILURE IN CIRRHOSIS AND SEPSIS 1949 tion, renal failure was reversible in 5 of the 8 patients (62%) and nonreversible in the remaining 3 patients (38%). Figure 3 shows changes in serum creatinine throughout hospitalization in patients with sepsis with reversible and nonreversible renal failure, as well as in patients with sepsis who did not develop renal failure. Table 4 shows a comparison of renal function and systemic hemodynamics parameters obtained at the time of diagnosis of renal failure in patients with reversible vs those with nonreversible renal failure. Patients with nonreversible renal failure had lower creatinine clearance and urine volume, higher values of plasma renin activity, and higher plasma levels of norepinephrine, arginine vasopressin, and endothelin compared with patients with reversible renal failure. By contrast, no significant differences were found between the 2 groups in mean arterial pressure and heart rate. These data suggest the existence of a greater reduction in effective arterial blood volume in patients with nonreversible as compared with that of patients with reversible renal failure. Forty-two of the 106 patients (39%) with sepsis had hyponatremia compared with 18 of the 100 patients (18%) of the control group without infection (P.001). In 23 of the 42 patients with sepsis, hyponatremia was present at the diagnosis of sepsis, and, in the remaining 19 patients, it developed during hospitalization. In patients with sepsis, hyponatremia occurred in close association with renal failure: 21 of the 29 patients (72%) with renal failure had hyponatremia as compared with Figure 3. Mean ( SE) values of serum creatinine at different time points throughout hospitalization in patients who developed nonreversible renal failure (triangles), reversible renal failure (squares), and patients who did not develop renal failure (circles). End is end of hospitalization or death during hospitalization. a, P.01 vs no renal failure; b, P.01 vs reversible renal failure; c, P NS vs basal value in the same group. Table 4. Characteristics of Renal Function and Systemic Hemodynamics in Patients with Sepsis and Reversible and Nonreversible Renal Failure at the Time of Diagnosis of Renal Failure Parameter Reversible n 22 Nonreversible n 7 P value Renal function Serum creatinine (mg/dl) BUN (mg/dl) Creatinine clearance (ml/min) Serum sodium (meq/l) Urine volume (ml/ 24h) Systemic hemodynamics Heart rate (bpm) Mean arterial pressure (mm Hg) Plasma renin activity (ng/ml/h) Aldosterone (ng/dl) Norepinephrine (pg/ ml) Arginine vasopresin (ng/l) Endothelin (pmol/l) Atrial natriuretic peptide (fmol/ml) NOTE. Values are median standard deviation. only 21 of the 77 patients (27%) who did not develop renal failure (P.0001). Survival Twenty-nine out of the 106 patients with sepsis died during the 3-month follow-up period compared with only 6 of the 100 patients from the control group without infection, the 3-month probability of survival in the 2 groups being 73% and 94%, respectively (P.001). Causes of death in the group of patients with sepsis were persistent infection in 8 patients, liver and renal failure in 7, gastrointestinal bleeding in 5, hepatocellular carcinoma in 4, liver failure in 2, rectal cancer in 1, and unknown in the remaining 2 patients. Causes of death in patients from the control group were liver failure in 4 patients and gastrointestinal bleeding and acute pancreatitis in 1 patient each. Among patients with sepsis, there were no differences in 3-month survival with respect to the type of infection. In patients with sepsis, mortality was strongly associated with the development of renal failure. In fact, 19 of the 29 (66%) patients who developed renal failure died during follow-up compared with only 10 of the 77 (13%) patients who did not develop renal failure (P.0001).

7 1950 TERRA ET AL GASTROENTEROLOGY Vol. 129, No. 6 Figure 4. The area under the ROC curve for the MELD score with 3-month mortality as the end point (c-statistic 0.83). The solid line represents the ROC based on chance alone and has a c-statistic of 0.5. Among patients with renal failure, the type of renal failure influenced mortality. All patients with nonreversible renal failure (7 out of 7, 100%) died during follow-up (median survival time, 15 days), compared with 12 of the 22 (55%) patients with reversible renal failure (P.0001). The existence of hyponatremia in patients with sepsis was also associated with impaired survival. In fact, 40% of patients with hyponatremia died during follow-up compared with only 19% of patients without hyponatremia (P.05). To identify factors predicting prognosis in patients with sepsis, variables obtained at the time of diagnosis of the infection were analysed for prognostic value. In univariate analysis, the variables with prognostic value were as follows: leukocyte count, temperature, serum creatinine, blood urea nitrogen (BUN), serum sodium, hyponatremia, serum potassium, albumin, bilirubin, prothrombin time, Child Pugh score, MELD score, ascites, and hepatic encephalopathy. In multivariate analysis, only the MELD score at diagnosis showed independent prognostic value. The value of the MELD score with higher sensitivity and specificity to predict 3-month survival was 20 (Figures 4 and 5). The relationship between the MELD score and 3-month probability of survival in patients with sepsis is shown in Figure 6. Discussion The main findings of the current study are as follows: (1) Renal failure is a frequent complication of patients with cirrhosis and sepsis unrelated to SBP; (2) the development of renal failure in these patients is Figure 5. Probability of survival of patients with cirrhosis and sepsis divided according to the MELD score obtained at diagnosis of sepsis. Patients with tumors (n 5) who died during the 3-month follow-up period were excluded from the analysis (see Patients and Methods section). associated with an impaired survival; and (3) the MELD score at diagnosis of sepsis is the best parameter in the assessment of prognosis in these patients. The incidence of renal failure in the current series of patients with cirrhosis and sepsis was 27% (29 out of the 106 patients included in the study), a value significantly higher compared with an 8% incidence in a control group of patients with cirrhosis without infection admitted to the hospital for the management of complications of cirrhosis. Renal failure occurred in 2 distinct clinical types: reversible and nonreversible. Reversible renal failure was the most common type (three fourths of patients with renal failure; 21% of all patients with sepsis) and was characterized by a marked impairment of renal function occurring early during the course of the infection and recovering after resolution of the infection. By contrast, nonreversible renal failure was less common (one fourth of patients with renal failure; 7% of all patients with sepsis) and was related in most patients to persis- Figure 6. Relationship between MELD score and 3-month probability of survival in patients with cirrhosis and sepsis included in the current study.

8 December 2005 RENAL FAILURE IN CIRRHOSIS AND SEPSIS 1951 tence of the infection and multiorgan failure. Patients outcome was much worse in nonreversible renal failure compared with that of patients with reversible renal failure (see below). In patients with renal failure, there was no evidence of significant tubular dysfunction because urine sodium was low, urine osmolality was high, and urinary excretion of markers of tubular damage was not increased. Therefore, it appears that the existence of acute tubular necrosis as cause of renal failure in this population can be ruled out, at least at the time of diagnosis of acute renal failure. This lack of data suggestive of tubular injury together with the existence of findings indicative of severe arterial underfilling (lower arterial pressure and greater activation of the reninangiotensin system and sympathetic nervous system compared with patients without renal failure) supports the concept that a functional renal vasoconstriction is the major determinant of renal failure in patients with cirrhosis and sepsis unrelated to SBP, although other factors may also play a pathogenic role (see below). The development of renal failure in patients with cirrhosis and sepsis was associated with a marked impairment of survival, even in patients in whom renal failure was reversible. Mortality of patients who developed renal failure was 45% during hospitalization and increased up to 60% at 3 months after diagnosis of the infection. This poor outcome is in keeping with that reported for patients with hepatorenal syndrome or patients developing renal failure in the setting of other complications of cirrhosis, such as SBP or gastrointestinal hemorrhage. 9,11,37,38,41 43 Although the poor prognosis of patients with nonreversible renal failure was anticipated, the observation that reversible renal failure had a negative impact on the long-term outcome of patients with cirrhosis and sepsis unrelated to SBP was an unexpected finding. Three-month mortality rate of patients with sepsis and reversible renal failure was 55% compared with 13% of patients who did not develop renal failure. This occurred despite a similar rate of infection resolution in these 2 groups (95% vs 99%, respectively), which suggests that poor prognosis was not directly related to sepsis itself. In fact, patients with cirrhosis and sepsis without renal failure had a 3-month survival probability, which was similar to that of the control group of patients with cirrhosis without infection (87% vs 94%, respectively, P.1). The reason for the high mortality in patients with reversible renal failure remains speculative. It could be that the occurrence of reversible renal failure identifies a subgroup of patients with particularly advanced liver disease who are more prone to rapid development of other complications of cirrhosis, thus increasing the mortality risk. Alternatively, and not mutually exclusive, the circulatory dysfunction associated with the development of renal failure may have deleterious effects on organs other than the kidneys. Whatever the reason for the increased mortality, these findings indicate that the development of reversible renal failure in the setting of sepsis in cirrhosis identifies a subgroup of patients with a particularly high risk of early death, which could benefit from expedited transplantation. Considering the widespread use of the MELD score for assessing prognosis in patients with cirrhosis and organ allocation in liver transplantation, 44,45 the usefulness of the MELD score in predicting prognosis in patients with cirrhosis and sepsis unrelated to SBP was investigated. The MELD score at diagnosis of sepsis was the best predictor of prognosis among a large number of parameters related to liver disease, renal failure, and characteristics of the infection that were analyzed for prognosis assessment. This high prognostic value of the MELD score is likely due to the fact that it combines the prognostic value of liver failure with that of renal failure, which, as stated before, is a very important prognostic factor in this setting. Because hyponatremia has recently been shown to be a marker of prognosis independent of the MELD score in patients with cirrhosis awaiting liver transplantation, the prognostic value of hyponatremia in patients with sepsis was also investigated in our study. Although the occurrence of hyponatremia was very common and correlated with mortality in a univariate analysis, it did not show prognostic value in multivariate analysis. A group of patients with cirrhosis and renal failure associated with SBP was not included in this study because all patients with SBP are currently treated with intravenous albumin in our unit, a method that has reduced dramatically the incidence of renal failure in this setting. 11 Nevertheless, a comment on the characteristics of renal failure in patients with cirrhosis and sepsis unrelated to SBP compared with those of renal failure reported in patients with SBP seems pertinent. Although, in both settings, the characteristics of reversible renal failure appear to be similar, this does not seem to be the case for the nonreversible form of renal failure. Patients with SBP frequently (approximately 20%) develop nonreversible acute renal failure that occurs despite the resolution of the infection. 9 This form of renal failure appears to be less common in patients with sepsis (7%) and is almost always associated with persistent infection. Whether this may be related to differences in the intensity and/or location of the inflammatory reaction in SBP vs sepsis (mainly splanchnic in the former vs nonsplanchnic in the latter) or other unrecognized factors is not

9 1952 TERRA ET AL GASTROENTEROLOGY Vol. 129, No. 6 known and would require specific investigation in future studies. A final issue that deserves discussion is the pathogenic mechanisms leading to renal failure in approximately one third of patients with cirrhosis and sepsis unrelated to SBP. Cirrhotic patients are predisposed to develop renal failure because of a variety of reasons, including volume depletion, shock, intrinsic renal diseases (ie, glomerulonephritis), and nephrotoxicity. In other patients, renal failure develops in the absence of any of these mechanisms, a condition known as functional renal failure or hepatorenal syndrome. 28 The pathogenesis of hepatorenal syndrome is incompletely understood and involves factors related to portal hypertension and abnormalities in the arterial circulation. 29,49 On the other hand, the pathogenesis of renal failure in the setting of sepsis without liver disease is very complex and involves many potentially important factors, including bacterial products such as endotoxin, hemodynamic factors related to underfilling of the arterial circulation, increased production of proinflammatory products (chemokines, cytokines), vasoactive hormones, vasodilator factors (ie, nitric oxide), and disseminated intravascular coagulation. 6 The relative contribution of these factors to renal failure in patients with cirrhosis and sepsis unrelated to SBP is not known and would require specific investigations. Nevertheless, it is likely that the existence of portal hypertension and circulatory abnormalities characteristic of cirrhosis increases the risk of development of renal failure in the setting of an acute inflammatory response such as that occurring in sepsis. In conclusion, the results of this study show that renal failure develops frequently in patients with cirrhosis and sepsis unrelated to SBP. Renal failure is associated with a very poor outcome, even in patients in whom renal function recovers after resolution of sepsis. The MELD score is the best prognostic marker in patients with cirrhosis and sepsis and should be used to assess prognosis in this patient population. Given the beneficial effect of intravenous albumin on renal function and survival in patients with cirrhosis and SBP, 11 the effect of albumin administration in patients with cirrhosis and sepsis unrelated to SBP deserves investigation. References 1. Rimola A, Soto R, Bory F, et al. Reticuloendothelial system phagocytic activity in cirrhosis and its relation to bacterial infections and prognosis. Hepatology 1984;4: Garcia-Tsao G. Current management of the complications of cirrhosis and portal hypertension: variceal hemorrhage, ascites, and spontaneous bacterial peritonitis. Gastroenterology 2001; 120: Navasa M, Rodés J. Bacterial infections in cirrhosis. Liver Int 2004;24: Schrier RW, Arroyo V, Bernardi M, et al. Peripheral arterial vasodilation hypothesis: a proposal for the initiation of renal sodium and water retention in cirrhosis. Hepatology 1988;8: Arroyo V, Colmenero J. Ascites and hepatorenal syndrome in cirrhosis: pathophysiological basis of therapy and current management. J Hepatol 2003;38(Suppl 1):S69 S Schrier RW, Wang W. Acute renal failure and sepsis. N Engl J Med 2004;351: Schrier RW, Wang W, Poole B, et al. Acute renal failure: definitions, diagnosis, pathogenesis, and therapy. J Clin Invest 2004; 114: Rimola A, Navasa M. Infections in liver disease. In: Bircher J, Benhamou JP, McIntyre N, Rizzeto M, Rodés J, eds. Oxford textbook of clinical hepatology. Volume 2, 2nd edition. Oxford: Oxford University Press, 1999: Follo A, Llovet JM, Navasa M, et al. Renal impairment after spontaneous bacterial peritonitis in cirrhosis: incidence, clinical course, predictive factors and prognosis. Hepatology 1994;20: Navasa M, Follo A, Filella X, et al. Tumor necrosis factor and interleukin-6 in spontaneous bacterial peritonitis in cirrhosis: relationship with the development of renal impairment and mortality. Hepatology 1998;27: Sort P, Navasa M, Arroyo V, et al. Effect of intravenous albumin on renal impairment and mortality in patients with cirrhosis and spontaneous bacterial peritonitis. N Engl J Med 1999;341: Hampel H, Bynum GD, Zamora E, et al. Risk factors for the development of renal dysfunction in hospitalized patients with cirrhosis. Am J Gastroenterol 2001;96: Ruiz-del-Arbol L, Urman J, Fernández J, et al. Systemic, renal, and hepatic hemodynamic derangement in cirrhotic patients with spontaneous bacterial peritonitis. Hepatology 2003;38: Perdomo Coral G, Alves de Mattos A. Renal impairment after spontaneous bacterial peritonitis: incidence and prognosis. Can J Gastroenterol 2003;17: Such J, Hillebrand DJ, Guarner C, et al. Nitric oxide in ascitic fluid is an independent predictor of the development of renal impairment in patients with cirrhosis and spontaneous bacterial peritonitis. Eur J Gastroenterol Hepatol 2004;16: Fernández J, Navasa M, Garcia-Pagan JC, et al. Effect of intravenous albumin on systemic and hepatic hemodynamics and vasoactive neurohormonal systems in patients with cirrhosis and spontaneous bacterial peritonitis. J Hepatol 2004;41: Bories PN, Campillo B, Azaou L, Scherman E. Long-lasting NO overproduction in cirrhotic patients with spontaneous bacterial peritonitis. Hepatology 1997;25: De las Heras D, Fernández J, Ginès P, et al. Increased carbon monoxide production in patients with cirrhosis with and without spontaneous bacterial peritonitis. Hepatology 2003;38: Fernández J, Navasa M, Gomez J, et al. Bacterial infections in cirrhosis: epidemiological changes with invasive procedures and norfloxacin prophylaxis. Hepatology 2002;35: American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference. Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Crit Care Med 1992;20: Gane EJ, Portmann BC, Naoumov NV, et al. Long-term outcome of hepatitis C infection after liver transplantation. N Engl J Med 1996;334: Sanchez-Fueyo A, Restrepo JC, Quinto L, et al. Impact of the recurrence of hepatitis C virus infection after liver transplantation on the long-term viability of the graft. Transplantation 2002;73: Benhamou Y, Bochet M, Di Martino V, et al. Liver fibrosis progression in human immunodeficiency virus and hepatitis C virus

10 December 2005 RENAL FAILURE IN CIRRHOSIS AND SEPSIS 1953 coinfected patients. The Multivirc Group. Hepatology 1999;30: Di Martino V, Rufat P, Boyer N, et al. The influence of human immunodeficiency virus coinfection on chronic hepatitis C in injection drug users: a long-term retrospective cohort study. Hepatology 2001;34: Dellinger RP, Carlet JM, Masur H, et al. Surviving Sepsis Campaign guidelines for management of severe sepsis and septic shock. Crit Care Med 2004;32: Cardenas A, Ginès P. Management of complications of cirrhosis in patients awaiting liver transplantation. J Hepatol 2005; 42(Suppl 1):S124 S Tito L, Ginès P, Arroyo V, et al. Total paracentesis associated with intravenous albumin management of patients with cirrhosis and ascites. Gastroenterology 1990;98: Arroyo V, Ginès P, Gerbes AL, et al. Definition and diagnostic criteria of refractory ascites and hepatorenal syndrome in cirrhosis. International Ascites Club. Hepatology 1996;23: Ginès P, Guevara M, Arroyo V, Rodés J. Hepatorenal syndrome. Lancet 2003;362: Maroto A, Ginès A, Salo J, et al. Diagnosis of functional kidney failure of cirrhosis with Doppler sonography: prognostic value of resistive index. Hepatology 1994: Platt JF, Ellis JH, Rubin JM, et al. Renal duplex Doppler ultrasonography: a noninvasive predictor of kidney dysfunction and hepatorenal failure in liver disease. Hepatology 1994;20: Platt JF, Rubin JM, Ellis JH. Acute renal failure: possible role of duplex Doppler US in distinction between acute prerenal failure and acute tubular necrosis. Radiology 1991;179: Platt JF, Ellis JH, Rubin JM, et al. Intrarenal arterial Doppler sonography in patients with nonobstructive renal disease: correlation of resistive index with biopsy findings. Am J Roentgenol 1990;154: Uriz J, Ginès P, Ortega R, et al. Increased plasma levels of neuropeptide Y in hepatorenal syndrome. J Hepatol 2002;36: Bartlett JG, Dowell SF, Mandell LA, et al. Practice guidelines for the management of community-acquired pneumonia in adults. Clin Infect Dis 2000;31: Bataller R, Ginès P, Guevara M, et al. Hepatorenal syndrome. Semin Liver Dis 1997;17: Guevara M, Fernandez-Esparrach G, Alessandria C, et al. Effects of contrast media on renal function in patients with cirrhosis: a prospective study. Hepatology 2004;40: Cardenas A, Ginès P, Uriz J, et al. Renal failure after upper gastrointestinal bleeding in cirrhosis: incidence, clinical course, predictive factors, and short-term prognosis. Hepatology 2001; 34: Gines P, Berl T, Bernardi M, et al. Hyponatremia in cirrhosis: from pathogenesis to treatment. Hepatology 1998;28: Silkensen JR, Kasiske BL. Laboratory assessment of kidney diseases: clearance, urinalysis and kidney biopsy. In: Breener BM, Rector FC, eds. The kidney. Volume 1, 7th edition. Philadelphia: Saunders, 2004: Ginès A, Escorsell A, Ginès P, et al. Incidence, predictive factors, and prognosis of the hepatorenal syndrome in cirrhosis with ascites. Gastroenterology 1993;105: Moreau R, Durand F, Poynard T, et al. Terlipressin in patients with cirrhosis and type 1 hepatorenal syndrome: a retrospective multicenter study. Gastroenterology 2002;122: Alessandria C, Ozdogan O, Guevara M, et al. MELD score and clinical type predict prognosis in hepatorenal syndrome: relevance to liver transplantation. Hepatology 2005;41: ) 44. Wiesner R, Edwards E, Freeman R, et al. United Network for Organ Sharing Liver Disease Severity Score Committee. Model for end-stage liver disease (MELD) and allocation of donor livers. Gastroenterology 2003;124: Freeman RB Jr, Wiesner RH, Harper A, et al. UNOS/OPTN Liver Disease Severity Score, UNOS/OPTN Liver and Intestine, and UNOS/OPTN Pediatric Transplantation Committees. The new liver allocation system: moving toward evidence-based transplantation policy. Liver Transpl 2002;8: Ruf AE, Kremers WK, Chavez LL, et al. Addition of serum sodium into the MELD score predicts waiting list mortality better than MELD alone. Liver Transpl 2005;11: Heuman DM, Abou-Assi SG, Habib A, et al. Persistent ascites and low serum sodium identify patients with cirrhosis and low MELD scores who are at high risk for early death. Hepatology 2004;40: Biggins SW, Rodriguez HJ, Bacchetti P, et al. Serum sodium predicts mortality in patients listed for liver transplantation. Hepatology 2005;41: Wong F, Blendis L. New challenge of hepatorenal syndrome: prevention and treatment. Hepatology 2001;34: Received December 3, Accepted September 7, Address requests for reprints to: Pere Ginès, MD, Liver Unit, Hospital Clínic de Barcelona, C/Villarroel 170, Barcelona, Catalunya, Spain. pgines@clinic.ub.es; fax: (34) Supported in part by grants from the SAF and Instituto de Salud Carlos III (CO3/2), by a fellowship grant from Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES; to C.T.), by a fellowship grant from Fondo de Investigación Sanitaria (to A.T.), and by a fellowship grant from Hospital Clínic (to M.M-L.). The authors thank Raquel Cela, RN, and the nursing staff of the Liver Unit and Dr Wladimiro Jimenéz for their technical assistance. C.T. and M.G. contributed equally to this work.