Hepatorenal syndrome: diagnostic accuracy, clinical features, and outcome in a tertiary care center

THE AMERICAN JOURNAL OF GASTROENTEROLOGY © 2002 by Am. Coll. of Gastroenterology Published by Elsevier Science Inc.

Vol. 97, No. 8, 2002 ISSN 0002-9270/02/$22.00 PII S0002-9270(02)04277-6

Hepatorenal Syndrome: Diagnostic Accuracy, Clinical Features, and Outcome in a Tertiary Care Center Kym Watt, M.D., F.R.C.P.C., Julia Uhanova, M.D., M.Sc., and Gerald Y. Minuk, M.D., F.R.C.P.C. Liver Diseases Unit, University of Manitoba, Winnipeg, Manitoba, Canada

OBJECTIVES: Hepatorenal syndrome (HRS) is a common and life-threatening entity that is associated with advanced liver disease. There are limited data on predisposing or precipitating factors related to the development of this syndrome. The aims of this study were: 1) to determine the accuracy of the diagnosis of HRS in an urban tertiary care center; 2) to compare features of HRS with those of nonHRS renal failure; and 3) to document the precipitating factors for HRS in this setting. METHODS: We conducted a retrospective chart review of 46 patients at the Health Sciences Center in Winnipeg, Manitoba, who were diagnosed with HRS within the last 10 yr. In addition to assessing the accuracy of the diagnosis of HRS (as defined by the International Ascites Club), we also developed a database that included clinical and biochemical features of these patients’ liver disease, possible predisposing factors, precipitating events, etiology of cirrhosis, and mortality rates. A cohort of 19 patients with advanced liver disease and renal failure not ascribed to HRS (non-HRS) served as controls. RESULTS: A total of 46 patients were labeled as having HRS during their hospital stay, of whom 27 (59%) fulfilled International Ascites Club criteria. Their mean age was 51.8 ⫾ 12.1 yr, and 67% were male. Demographic and biochemical findings in the HRS patients were similar to those of the non-HRS patients. Hepatic encephalopathy (p ⬍ 0.03) and jaundice (p ⬍ 0.01) more commonly preceeded renal failure in the HRS group than in the controls. The most common predisposing factor for HRS was alcohol abuse. Precipitating factors for HRS included infection in 48% patients, GI bleeding in 33%, and over paracentesis in 27%. Drug-induced renal disease was significantly more common in patients with non-HRS renal failure. Contrary to previous reports, nonsteroidal anti-inflammatory drugs were not common precipitating factors of HRS. Alcohol-related liver disease was the most common underlying etiology of liver disease in all patients who developed renal failure but was proportionally higher in the HRS group. Mortality was almost universal in both groups, regardless of the cause of renal failure. CONCLUSIONS: The diagnostic accuracy (or lack thereof) of HRS highlights the importance of ensuring that diagnostic

criteria are met in studies describing the natural history and/or the results of therapeutic intervention in HRS. Alcohol abuse is a common predisposing factor for the development of HRS. Infection, GI bleed, and aggressive paracentesis were the most common precipitating factors for HRS. (Am J Gastroenterol 2002;97:2046 –2050. © 2002 by Am. Coll. of Gastroenterology)

INTRODUCTION Hepatorenal syndrome (HRS) is a life-threatening, functional renal failure associated with advanced liver disease. Studies by Gines et al. (1) reported an incidence of 18% in 1 yr and 39% at 5 yr in cirrhotic patients with ascites but no renal failure. There are two types of HRS (1, 2). Type I consists of an acute deterioration in renal function over days or weeks, and occurs in an advanced stage of liver disease. Of type I cases, 50% are precipitated by GI bleeding, infection, dehydration from overt paracentesis or diuresis, surgery or drug exposures (1, 3, 4, 5) with the remaining 50% occurring spontaneously (2). Type II HRS occurs in patients with relatively preserved liver function. These patients show a slow but progressive deterioration in their glomerular filtration rate. They also tend to have diureticresistant ascites. In the absence of liver transplantation, the prognosis for both types I and II HRS is poor. Contributing to the poor outcome is the failure to recognize and avoid possible precipitating factors. Thus, the objectives of the present study were: 1) to determine the accuracy of the diagnosis of HRS in an urban tertiary care center; 2) to compare features of HRS with those of non-HRS renal failure (including demographics, biochemical profiles, and nature of the underlying liver diseases); and 3) to document the precipitating factors for HRS in this setting. We also compared mortality rates in liver failure patients with and without HRS.

MATERIALS AND METHODS The study consisted of a retrospective chart review, covering the years 1988 –1998, of patients coded as having HRS

AJG – August, 2002

Hepatorenal Syndrome in a Tertiary Care Center

2047

Table 1. International Ascites Club Criteria Major Low GFR (creatinine ⬎132 ␮mol/L or CrCI ⬍40 ml/min) Absence of shock, nephrotoxins, fluid loss, ongoing infection No improvement with volume expansion Proteinuria ⬍ 500 mg/day and/or normal ultrasound Minor Uvolume ⬍500 ml/day UNa ⬍10 mEq/L Uosm ⬎Posm Urbc ⬍50/HPF Sna ⬍130 mEq/L CrCl ⫽ creatinine clearance; GFR ⫽ glomerular filtration rate; HPF ⫽ high power field; Na ⫽ sodium; osm ⫽ osmolality; rbc ⫽ red blood cells.

at the Health Sciences Center, a tertiary care health facility in Winnipeg, Manitoba, Canada. The diagnostic criteria for HRS were those derived from the International Ascites Club recommendations (2, 5). Major criteria were as follows: 1) a low GFR with a serum creatinine greater than 132 mmol/L or a creatinine clearance of less than 40 ml/min; 2) absence of shock, nephrotoxins, continued fluid losses, or ongoing bacterial infection; 3) no improvement with plasma volume expansion; and 4) proteinuria less than 500 mg/day and a normal ultrasound examination of the kidneys (Table 1). Minor criteria were as follows: 1) urine volume less than 500 ml/day; 2) urine sodium less than 10 mEq/L; 3) urine osmolarity greater than plasma osmolarity; 4) urine red blood cells less than 50 per high power field; and 5) serum sodium less than 130 mEq/L (Table 1). For the purposes of this study, inclusion criteria required major criteria numbers 1–3 as well as either the proteinuria criteria and the normal ultrasound or, if one of those tests were not performed, three of the five minor criteria. Statistical Analysis Statistical analyses included analysis of variance, Student t tests for parametric data, and the ␹2 test for nonparametric data using the MCSS 2000 –Number Cruncher Statistical Systems (NCSS, Kaysville, UT).

RESULTS A total of 48 cases were coded in hospital records as HRS throughout the 10 yr indicated. Two of these patients were excluded, as their charts were unretrievable. Accuracy of Diagnosis Of the 46 patients who constituted the study population, 27 (59%) satisfied the criteria for HRS described above. Seven (26%) of these patients were thought to have type II HRS. In all, 19 of the 46 patients (41%) had an alternative cause for their renal failure. These included the following: shock resulting in acute tubular necrosis in three patients; prerenal failure in five patients (in three patients, prerenal failure reversed with fluid resuscitation, and in two it progressed to

Figure 1. Prevalence of jaundice, ascites, esophageal varices, and encephalopathy in 27 HRS patients and 19 non-HRS patients with combined liver and renal failure. varices ⫽ esophageal varices; enceph ⫽ encephalopathy.

acute tubular necrosis); and renal parenchymal disease in 11 patients (including three patients with microscopic hematuria and proteinuria, and eight patients with possible drugrelated interstitial nephritis or acute tubular necrosis). These 19 patients served as the cohort of cirrhotic patients who developed renal failure through other mechanisms (nonHRS). Demographics and Clinical Features There were no significant differences in the age (51.8 ⫾ 12.1 yr vs 55.2 ⫾ 14.5 yr) or sex distribution (67% vs 58% male) of HRS and non-HRS patients. Hospital length of stays were also similar in the two groups (16.8 ⫾ 10.2 days and 19.0 ⫾ 14.6 days, respectively). As shown in Figure 1, HRS patients were more commonly encephalopathic (96%) and jaundiced (85%) than were non-HRS patients (68% and 56%, respectively; p ⬍ 0.01 and 0.03, respectively). Biochemical Findings There were no significant differences between the two groups with respect to liver function at the onset of renal failure (Table 2).

Table 2. Biochemical Features and Etiology of Liver Disease in HRS and Non-HRS Patients

Biochemistry Albumin Bilirubin INR Liver disease Viral VOD HIV Cryptogenic Cancer Alcohol

HRS (n ⫽ 27)

Non-HRS (n ⫽ 19)

p Value

20.3 ⫾ 6.1 291.2 ⫾ 227.1 3.23 ⫾ 1.49

20.0 ⫾ 7.9 238.5 ⫾ 230.1 3.16 ⫾ 1.69

0.902 0.444 0.874

3 (11%) 0 1 (4%) 1 (4%) 2 (7%) 24 (89%)

1 (5%) 3 (16%) 1 (5%) 1 (5%) 1 (5%) 14 (74%)

0.499 0.033 0.804 0.804 0.778 0.188

HRS ⫽ hepatorenal syndrome; INR ⫽ international normalized ratio; VOD ⫽ veno-occlusive disease.

2048

Watt et al.

AJG – Vol. 97, No. 8, 2002

Table 3. Alcohol-Related Liver Disease and Renal Failure HRS Liver failure (n ⫽ 545) Alcohol related (n ⫽ 102) (n ⫽ 212*) Other etiology (n ⫽ 443) (n ⫽ 333*)

Renal Failure (HRS and Non-HRS)

27 24

46 OR ⫽ 45.14, CI ⫽ 13.3–153.5 *OR ⫽ 14.4, CI ⫽ 4.17–47.26 p ⫽ 0.00001

3

38

OR ⫽ 32.39, CI ⫽ 14.42–72.3 *OR ⫽ 8.87, CI ⫽ 4.05–19.44* p ⫽ 0.00001

8

* Corrected for 25% under-reporting of alcohol abuse.

Nature of Underlying Liver Disease Alcohol abuse was the most common cause of liver disease in the two groups (HRS 89%, non-HRS 74%, p ⫽ 0.18). Six patients had acute alcoholic hepatitis. To determine whether HRS is more commonly associated with alcohol-induced liver disease than with other causes of liver disease, we compared the proportion of patients with a history of alcohol abuse and HRS (24 of 102, 24%) to the total number of nonalcoholic patients with liver failure and HRS during the same period (three of 443, 0.6%). As shown in Table 3, patients with alcohol-induced liver failure more often had HRS than did patients with other forms of liver failure (relative risk ⫽ 34.75, OR ⫽ 45.1, CI ⫽ 13.3–153.5, p ⫽ 0.00001). Significant results were also found when renal failure, both HRS and non-HRS, was assessed with respect to alcohol-related liver disease (Table 3). To exclude the possibility of under-reporting alcoholinduced liver disease, we randomly reviewed 50 charts that did not identify alcohol as the cause of the liver disease. Of these, we found that approximately 25% had sufficient evidence to implicate alcohol abuse. Despite this correction factor, the prevalence of HRS in alcohol-related liver disease remained significant (relative risk ⫽ 12.57, OR ⫽ 14.4, CI ⫽ 4.17– 47.26, p ⫽ 0.00001) (Table 3). Veno-occlusive disease was more common among nonHRS patients than among those with HRS, but the possibility of a type 1 error cannot be excluded.

DISCUSSION The results of this study indicate the following: 1) a large percentage (approximately 40%) of patients with advanced liver disease and renal failure are mistakenly diagnosed as having HRS; 2) standard demographic features do not distinguish HRS patients from liver failure patients with other causes of renal failure, but encephalopathy and jaundice are more common in HRS patients; 3) alcohol-induced liver disease is more often associated with HRS than are other causes of liver failure; 4) infection, GI bleed, and overt paracentesis are the most common precipitating causes of HRS; and 5) mortality is almost universal in patients with combined liver and kidney failure, regardless of the cause. The reason for the large percentage of misdiagnosed HRS cases is unclear. Misdiagnosis was as common before 1996 (when diagnostic criteria were developed) as afterward, suggesting that many physicians are unaware of the criteria that exist for defining HRS, and may continue to assume that renal failure in the setting of liver disease is in fact HRS. The study design that we used precluded our determining how many cases of HRS remain undiagnosed. This would have required an extensive review of all charts derived from patients with liver failure in this institution during the 10-yr study period. However, it is likely that the inclusion of

Precipitating Causes of HRS Precipitating causes of HRS were identified as bacterial infection in 13 patients (48%), GI bleeding in nine (33%), excessive paracentesis in seven (27%), drugs in two (7%), surgery in two (7%), and miscellaneous factors in three (11%) (Fig. 2). Drug-induced renal failure was more common in non-HRS patients than among those with HRS (52.1% vs 7.1%, p ⬍ 0.00005) (Fig. 2). Mortality The mortality rate of HRS patients was 93% and in nonHRS patients was 89%. Although not precisely indicated, the majority of these deaths were ascribed to multiorgan failure. All but one of the four surviving discharged patients died within 1 month of discharge.

Figure 2. Precipitating causes of renal failure in 27 HRS and 19 non-HRS patients with combined liver and renal failure. Infect ⫽ infection; GIB ⫽ GI bleed; PC ⫽ paracentesis; Sx ⫽ surgery.

AJG – August, 2002

misdiagnosed HRS cases would be far greater than the exclusion of true cases of HRS. Also unclear is why HRS patients were more often encephalopathic than non-HRS patients. It is conceivable that the higher prevalence of alcohol abuse in HRS subjects contributed to impaired cognitive function. It is also possible, although unlikely, that prerenal azotemia is more encephalopathogenic than renal parenchymal or postrenal disorders. Because the precipitants associated with HRS (e.g., GI bleeding, overt paracentesis, and infection) are also common precipitants of encephalopathy, this finding may reflect an association rather than a cause and effect. However, until the pathogenesis of hepatic encephalopathy is determined, the latter explanation remains purely speculative. The higher prevalence of jaundice in patients with HRS may relate to predicted differences in bilirubin clearance in HRS versus non-HRS renal failure. Under normal conditions, conjugated bilirubin is filtered through glomeruli and then reabsorbed by renal tubules. In HRS, because tubular function is normal, the process should remain intact. However, in non-HRS renal failure, the tubules may not be as efficient in bilirubin reabsorption, resulting in higher losses and possibly contributing to the difference between the two groups. It is also conceivable that hyperbilirubinemia per se may have contributed to the renal dysfunction in each group. Numerous studies have reported an association between renal failure and obstructive jaundice, including studies in which bilirubin levels correlated with decreases in creatinine clearance (6, 7). One mechanism is thought to involve bilirubin induction of vasoconstrictor hormonal activity, which is normally counterbalanced by an increase in prostaglandin synthesis (6, 8). In HRS, there is an imbalance between these same mediators (9, 10). How much each contributes to renal dysfunction remains to be determined. Although ethanol has been documented to precipitate tubular necrosis in rats (11, 12), no studies have reported the effects of ethanol on renal function in human patients with liver disease. Nonetheless, certain properties of ethanol might explain the strong association of alcohol-induced liver disease with HRS and with renal failure in general. For example, ethanol inhibits the synthesis of prostaglandins and other factors that regulate renal parenchymal blood flow (8). However, arguing against this explanation was the finding that the mean hospital length of stay for HRS patients exceeded 2 wk, a period sufficiently long for prostaglandin synthesis to be restored. In addition, ethanol is neurotoxic, and there are data suggesting that the autonomic nervous system (and perhaps a specific hepatorenal reflex) are involved in the pathogenesis of HRS (13). Finally, recent studies have implicated increased serum tumor necrosis factor–␣ levels in the development of renal dysfunction, and levels of this cytokine are significantly elevated in patients with alcohol-induced liver disease (14). The precipitating causes of HRS in our study (infection, GI bleed, overt paracentesis) were in keeping with those described by Gines et al. (1) and somewhat conflicting with

Hepatorenal Syndrome in a Tertiary Care Center

2049

those of other groups who identified surgery and drug/toxin exposure as more common precipitating factors (15, 16). Indeed, only two of our patients had undergone recent surgery and only one had recent exposure to nonsteroidal anti-inflammatory drugs, the agents most often implicated in precipitating HRS. On the other hand, drugs were more commonly precipitating causes of parenchymal renal failure in cirrhotic patients. The drugs most commonly implicated were aminoglycosides, amphotericin, and trimethoprim-sulfamethoxazole. These are well known nephrotoxins and should be easily recognized as precipitants of non-HRS renal failure. It should also be noted that we did not find as strong an association between HRS and esophageal varices as has been reported by others (1). The retrospective nature of the study makes this point difficult to interpret. As predicted, mortality rates were high in HRS patients, exceeding 95% (17). This finding emphasizes the need for a better understanding of the pathophysiological mechanisms involved and, thereby, more effective therapies for this disorder. In conclusion, the results of this study highlight the importance of ensuring that specific diagnostic criteria are met in studies describing the natural history and/or effects of therapeutic interventions in HRS. The results also describe a strong association between HRS and alcohol-induced liver disease, a finding that could provide insights into the pathogenesis of HRS. Finally, the results emphasize the need for more effective therapy in this often fatal disorder.

ACKNOWLEDGMENT We thank Ms. S. Zdanuk for prompt and accurate typing of the manuscript. Reprint requests and correspondence: G.Y. Minuk, M.D., F.R.C.P.C. Liver Diseases Unit, John Buhler Research Centre, 803F-715 McDermot Avenue, Winnipeg, MB, Canada R3E 3P4. Received Feb. 8, 2001; accepted Dec. 14, 2001.

REFERENCES 1. Gines A, Escorsell A, Gines P, et al. Incidence, predictive factors, and prognosis of the hepatorenal syndrome in cirrhosis with ascites. Gastroenterology 1993;105:229 –36. 2. Bataller R, Gines P, Guevara M, Arroyo V. Hepatorenal syndrome. Semin Liver Dis 1997;17:233– 47. 3. Badalamenti S, Graziani G, Salerno F, Ponticello C. Hepatorenal syndrome: New perspectives in pathogenesis and treatment. Arch Intern Med 1993;153:1957– 67. 4. Rodes J, Bosch J, Arroyo V. Clinical types and drug therapy of renal impairment in cirrhosis. Postgrad Med J 1975;51: 492–7. 5. Arroyo V, Gines P, Gerbes A, et al. Definition and diagnostic criteria of refractory ascites and hepatorenal syndrome in cirrhosis. Hepatology 1996;23:164 –73. 6. Wait R, Kahng K. Renal failure complicating obstructive jaundice. Am J Surg 1989;157:256 – 63.

2050

Watt et al.

7. Uemura M, Tsujii T, Fukui H, et al. Urinary prostaglandins and renal function in obstructive jaundice. Scand J Gastroenterol 1989;24:705–15. 8. Levy M, Wexler MJ, Fechner C. Renal perfusion in dogs with experimental hepatic cirrhosis: Role of prostaglandins. Am J Physiol 1983;245:F521–9. 9. Zipser R. Role of renal prostaglandins and the effects of NSAIDs in patients with liver disease. Am J Med 1986; 81(suppl 2B):95–103. 10. Arroyo V, Gines P, Rimola A, Gaya J. Renal function abnormalities, prostaglandins and effects of NSAIDs in cirrhosis with ascites. An overview with emphasis on pathogenesis. Am J Med 1986;81(suppl 2B):104 –22. 11. VanThiel DH, Williams WD Jr, Gavaler JS, et al. Ethanol—its nephrotoxic effect in the rat. Am J Pathol 1977;89:67– 83. 12. Bilotta J, Pazik M, Greizerstein H, Acara M. Renal effect of

AJG – Vol. 97, No. 8, 2002

13. 14. 15. 16. 17.

ethanol in isolated perfused rat kidney. Eur J Pharmacol 1984; 98:109 –12. Gentile S, Marmo R, Peduto A, et al. Autonomic neuropathy in liver cirrhosis: Relationship with alcoholic etiology and severity of the disease. Ital J Gastroenterol 1994;26:53– 8. Akriviadis E, Botla R, Briggs W, et al. Pentoxifylline improves short term survival in severe acute alcoholic hepatitis: A doubleblind, placebo controlled trial. Gastro 2000;119:1637– 48. Gines A, Escorsell A, Gines P, et al. Natural history and predictive factors of hepatorenal syndrome in cirrhosis with ascites. J Hepatol 1992;16(suppl 1):S112. Korula J. Hepatorenal syndrome. In: Kaplowitz N, ed. Liver and biliary diseases. Baltimore: Williams & Wilkins, 1992:542–51. Epstein M. Hepatorenal syndrome: Emerging perspectives of pathophysiology and therapy. J Am Soc Nephrol 1994;4: 1735–53.