Acute kidney injury in cirrhosis


  • Guadalupe Garcia-Tsao,

    Corresponding author
    1. Section of Digestives Diseases, Yale University School of Medicine, New Haven, CT, and VA-Connecticut Healthcare System, West Haven, CT
    • Yale University School of Medicine, Section of Digestive Diseases, One Gilbert Street, TAC, room # S241B, New Haven, CT 06510
    Search for more papers by this author
    • fax: 203-785-7273.

  • Chirag R. Parikh,

    1. Section of Nephrology, Yale University School of Medicine, New Haven, CT and VA-Connecticut Healthcare System, West Haven, CT
    Search for more papers by this author
  • Antonella Viola

    1. Section of Digestives Diseases, Yale University School of Medicine, New Haven, CT, and VA-Connecticut Healthcare System, West Haven, CT
    2. Department of Clinical Medicine, University of Bologna, Italy
    Search for more papers by this author

  • Potential conflict of interest: Nothing to report.


Acute renal failure (ARF), recently renamed acute kidney injury (AKI), is a relatively frequent problem, occurring in approximately 20% of hospitalized patients with cirrhosis. Although serum creatinine may underestimate the degree of renal dysfunction in cirrhosis, measures to diagnose and treat AKI should be made in patients in whom serum creatinine rises abruptly by 0.3 mg/dL or more (≥26.4 μmol/L) or increases by 150% or more (1.5-fold) from baseline. The most common causes of ARF (the term is used interchangeably with AKI) in cirrhosis are prerenal azotemia (volume-responsive prerenal AKI), acute tubular necrosis, and hepatorenal syndrome (HRS), a functional type of prerenal AKI exclusive of cirrhosis that does not respond to volume repletion. Because of the progressive vasodilatory state of cirrhosis that leads to relative hypovolemia and decreased renal blood flow, patients with decompensated cirrhosis are very susceptible to developing AKI with events associated with a decrease in effective arterial blood volume. HRS can occur spontaneously but is more frequently precipitated by events that worsen vasodilatation, such as spontaneous bacterial peritonitis. Conclusion: Specific therapies of AKI depend on the most likely cause and mechanism. Vasoconstrictors are useful bridging therapies in HRS. Ultimately, liver transplantation is indicated in otherwise reasonable candidates in whom AKI does not resolve with specific therapy. (HEPATOLOGY 2008;48:2064-2077.)

The development of acute kidney injury (AKI) in patients with cirrhosis is an ominous event. In a systematic review of 118 studies evaluating predictors of survival in cirrhosis, parameters of liver dysfunction (Child-Pugh score and its components) and parameters of renal dysfunction (creatinine and blood urea nitrogen/azotemia) were both powerful predictors of death in decompensated cirrhosis.1 Higher serum creatinine (SCr) consistently portends worse survival. In fact, SCr is one of three variables that form part of the model of end-stage liver disease (MELD) score that is a good predictor of 3-month mortality and is currently used in determining priority for orthotopic liver transplantation (OLT).2 Pretransplantation creatinine was recently found to be the most powerful predictor of survival post-OLT.3 Therefore, identifying and treating the cause of renal dysfunction in cirrhosis is essential.

This review of AKI in cirrhosis expands several recent reviews on hepatorenal syndrome (HRS)4–6 by covering causes of acute renal failure (ARF) other than HRS and expands recent reviews of ARF in cirrhosis.7, 8


In response to the need for a common definition and classification of ARF, the Acute Kidney Injury Network recently developed and published a consensus definition of “acute kidney injury” (AKI), a new term for ARF.9 AKI is defined as an abrupt (arbitrarily set at 48 hours) reduction in kidney function manifested by an absolute increase in SCr of 0.3 mg/dL or more (≥26.4 μmol/L), equivalent to a percentage increase in SCr 50% or more (1.5-fold from baseline) or a urine output of less than 0.5 mL/kg per hour for more than 6 hours. The AKI definition has three stages that indicate the severity of renal dysfunction and are based on SCr or urine output (Table 1). The urine output criterion was included based on the predictive importance of this measure but with the awareness that urine output may not be measured routinely and accurately in non–intensive care unit settings. These criteria include both an absolute and a percentage change in creatinine to accommodate variations related to age, sex, and body mass index and to reduce the need for a baseline creatinine but do require at least two creatinine values within 48 hours.

Table 1. Classification/Staging System for Acute Kidney Injury (AKI)9
StageSerum Creatinine CriteriaUrine Output Criteria
  • *

    Any patient requiring renal replacement therapy is, by definition, at a stage 3.

1Increase in serum creatinine of more than or equal to 0.3 mg/dL (≥ 26.4 μmol/L) or increase to more than or equal to 150% to 200% (1.5-fold to 2-fold) from baselineLess than 0.5 mL/kg per hour for more than 6 hours
2Increase in serum creatinine to more than 200% to 300% (>2-fold to 3-fold) from baselineLess than 0.5 mL/kg per hour for more than 12 hours
3*Increase in serum creatinine to more than 300% (>3-fold) from baseline (or serum creatinine of more than or equal to 4.0 mg/dL [≥354 μmol/L] with an acute increase of at least 0.5 mg/dL [44 μmol/L])Less than 0.3 mL/kg per hour for 24 hours or anuria for 12 hours

AKI, acute kidney injury; ARF, acute renal failure; ATN, acute tubular necrosis; FENa, fractional excretion of sodium; GFR, glomerular filtration rate; HRS, hepatorenal syndrome; MAP, mean arterial pressure; MELD, model of end-stage liver disease; OLT, orthotopic liver transplantation; RCT, randomized controlled trials; SBP, spontaneous bacterial peritonitis; SCr, serum creatinine; SLKT, simultaneous liver and kidney transplant; TIPS, transjugular intrahepatic portosystemic shunt; UNa, urinary sodium.

Types of Acute Kidney Injury

Traditionally, three types of ARF/AKI are identified: (1) prerenal azotemia, which results from renal hypoperfusion without a glomerular or tubular lesion; (2) intrinsic renal failure, which results from tubular cell necrosis (ischemic or toxic), glomerulonephritis, or interstitial nephritis; and (3) post-renal failure, which results from urinary tract obstruction causing hydronephrosis. Patients with cirrhosis can develop all types of AKI,8 but they can additionally develop HRS, a type of prerenal AKI that is not responsive to volume expansion and is seen exclusively in patients with severe liver dysfunction

HRS is a unique potentially reversible form of ARF secondary to renal vasoconstriction10 that results from extreme vasodilatation.11, 12 It is therefore a functional disorder, not associated with structural kidney damage. The 1-year and 5-year probabilities of developing HRS in patients with ascites are approximately 20% and 40%, respectively13 and are highest in patients with more marked sodium and water retention and marked activation of vasoconstrictive systems.13, 14

HRS is divided into two types (1 and 2) based on prognosis and clinical characteristics. Survival of patients with HRS-1 is shorter than that of patients with HRS-2 (median survival 1.0 versus 6.7 months).15

HRS-1 is characterized by an abrupt deterioration in renal function that occurs mostly in an inpatient setting and often develops after a precipitating event, particularly spontaneous bacterial peritonitis (SBP).4, 16 HRS-2 is characterized by a steady or slowly progressive course that occurs mostly in an outpatient setting in patients with refractory ascites.17 Because this more “chronic” form of HRS does not meet criteria for AKI, it is not considered in this review. The term HRS will refer to HRS-1, unless otherwise specified.


AKI occurs in approximately 19% of hospitalized patients with cirrhosis,18–22 and the most common cause is prerenal AKI, accounting for approximately 68% of the cases19, 23, 24 (Fig. 1). AKI is mostly secondary to infection,19, 24 hypovolemia (gastrointestinal hemorrhage, aggressive diuresis, or diarrhea), use of vasodilators, and other factors that cause renal vasoconstriction such as nonsteroidal anti-inflammatory drugs or intravenous contrast agents. HRS, which is not volume-responsive, constitutes approximately 25% of the cases of prerenal ARF; that is, it accounts for only approximately 17% of cases of ARF in hospitalized patients with cirrhosis19, 23, 24 (Fig. 1).

Figure 1.

Prevalence and types of acute renal failure (ARF)/acute kidney injury (AKI) in hospitalized patients with cirrhosis. Percentages and numbers were obtained by adding up patients in the references cited. ATN, acute tubular necrosis; GMN, glomerulonephritis; HRS, hepatorenal syndrome. *Infections precipitated this type of AKI in 51% of the cases.19, 24

Acute tubular necrosis (ATN) is more common than HRS as a cause of AKI, accounting for about a third of the cases (Fig. 1). It is mainly caused by an ischemic insult to the renal tubules as a result of a hypotensive event after bleeding or severe sepsis. However, the use of aminoglycosides, which are directly toxic to renal tubules, was found to be the most important predictor of ARF in cirrhosis in a study performed in U.S. veterans.18

Postrenal causes of AKI in cirrhosis are rare and represent less than 1% of the cases.23 Similarly, chronic kidney injury also appears to be a rare diagnosis in hospitalized patients,19 although a small study has recently shown that immune-complex glomerulonephritis is quite common in patients with end-stage hepatitis C–induced cirrhosis.25 A recent review has suggested that an additional subtype of HRS should include patients with chronic kidney injury who develop superimposed AKI.6


ARF/AKI is common in cirrhosis for several reasons. Patients with cirrhosis are prone to intravascular volume depletion secondary to gastrointestinal bleeding, diuretic use, and lactulose-induced diarrhea. Moreover, these patients are often exposed to nephrotoxic agents such as nonsteroidal anti-inflammatory drugs, contrast agents, and aminoglycosides, to which they are particularly susceptible.18, 26 Perhaps most importantly, because of the hyperdynamic circulatory state of cirrhosis, renal blood flow in patients with cirrhosis (particularly in those with more severe liver disease) is very susceptible to events associated with a further decrease in effective arterial blood volume.

As recently reviewed,12 the hyperdynamic circulation in cirrhosis is a progressive vasodilatory syndrome that was first recognized clinically in patients with cirrhosis by observing that they frequently had “warm extremities, cutaneous vascular spiders, wide pulse pressure, and capillary pulsations in the nail beds.”27 Progressive vasodilatation (both splanchnic and systemic) is a key factor in the pathogenesis of many of the complications of cirrhosis, prominently in the kidney.12 Splanchnic and systemic vasodilatation in cirrhosis is a consequence of portal hypertension and is attributable mostly to nitric oxide overproduction, although other molecules also participate in this complex process.12 As shown in Fig. 2, vasodilatation leads to a decreased effective arterial blood volume (relative hypovolemia) and activation of neurohumoral systems such as the renin-angiotensin-aldosterone system; sympathetic nervous system; and non-osmotic release of antidiuretic hormone). Relative hypovolemia initially leads to sodium and water retention (with ascites formation), increased intravascular volume, and increased cardiac output. With progression of cirrhosis, vasodilatation worsens and activated vasoconstrictive systems lead to renal vasoconstriction and decreased renal blood flow11, 28 (Fig. 2). Additionally, as in other forms of high cardiac output syndrome, the heart response becomes insufficient to maintain perfusion pressure (high-output heart failure) and further contributes to a decrease in renal blood flow and renal failure12, 29 (Fig. 2). It has been proposed that the sympathetic nervous system causes a rightward shift in the renal autoregulatory curve that makes renal blood flow critically dependent on renal perfusion pressure and that this further contributes to the development of renal failure.30

Figure 2.

Hyperdynamic circulation in the pathophysiology of hepatorenal syndrome (HRS). Ascites, hyponatremia, and HRS in cirrhosis have a common pathophysiology. The main abnormality is splanchnic and systemic vasodilatation that leads to a decreased effective arterial blood volume (relative hypovolemia) and activation of neurohumoral systems (RAAS, renin-angiotensin-aldosterone system; SNS, sympathetic nervous system; ADH, nonosmotic release of antidiuretic hormone). Relative hypovolemia initially leads to sodium and water retention (with ascites formation), increase in intravascular volume, and increased cardiac output. With progression of cirrhosis, vasodilatation worsens and activated vasoconstrictive systems lead to renal vasoconstriction. Additionally, the increased cardiac output is now insufficient to maintain perfusion pressure (high-output heart failure) and further contributes to a decrease in renal blood flow and renal failure.

Although the “spontaneous” development of renal failure in a patient with cirrhosis and hyperdynamic circulation indicates the presence of HRS, this syndrome often presents after a precipitating event.4 In the setting of hyperdynamic circulation, rapid fluid loss (from gastrointestinal bleeding or diarrhea) or sepsis/systemic inflammatory response syndrome-related vasodilation leads to a further decrease in effective arterial blood volume, renal vasoconstriction, and prerenal AKI (Fig. 3).

Figure 3.

Mechanisms for the development of acute renal failure (ARF)/acute kidney injury (AKI) in cirrhosis. GI, gastrointestinal; SIRS,systemic inflammatory response syndrome; VCS, vasoconstrictive systems (include renin-angiotensin-aldosterone and sympathetic nervous system). *Worsening of the vasodilatation already present in the patient with cirrhosis

This is further complicated by the fact that these same events, through the development of circulatory shock, can lead to intrinsic renal failure from renal tubular necrosis. Intense renal vasoconstriction, as seen in HRS, can in turn lead to tubular ischemia and necrosis as demonstrated by electron microscopy31 or by urine markers of acute tubular necrosis32, 33 (Fig. 3).


SCr is the most established, simple, and inexpensive parameter of glomerular filtration rate (GFR) and is the primary method of detection of all forms of renal failure. However, it has several limitations. First, SCr is not helpful in distinguishing among various causes of renal injury. Second, SCr lags behind renal injury and is therefore a delayed marker of decreased renal function.34 Third, significant renal disease can exist with minimal or no changes in SCr because of renal reserve, enhanced tubular creatinine secretion, or other factors.35, 36 Lastly, SCr is greatly influenced by numerous nonrenal factors such as body weight, race, age, sex, total body volume, drugs, muscle metabolism, and protein intake.35 Although the simplified “modification of diet in renal disease” formula provides a robust estimate of GFR relative to SCr corrected by age, race, and sex in chronic kidney disease when SCr is at a steady state, it is not useful in AKI when SCr is not in equilibrium. In cirrhosis, SCr may be an even poorer reflection of kidney function because of a reduced muscle mass, particularly in patients with severe liver disease. In this setting, the release of creatinine is considerably reduced, and therefore, patients may have a normal SCr in the setting of a very low GFR. Additionally, severe hyperbilirubinema gives a falsely low value of SCr if the chemical rather than enzymatic method is used for measurement.37

As recently reviewed,38 other methods of renal function assessment also have limitations and do not correlate well with GFR. Newer serum markers such as cystatin C are promising; however, they require further validation using “gold standard” measures of GFR such as iodothalamate or inulin clearance. Therefore, and despite its limitations; SCr remains the key biomarker in the diagnosis of ARF in cirrhosis.

Diagnosis of HRS

HRS type 1 has been defined by consensus as a doubling of SCr to a level greater than 2.5 mg/dL (>226 μmol/L) in less than 2 weeks.16, 17 Per the AKI consensus criteria, this level of renal failure would correspond to a stage 2 (that is, a greater than twofold to threefold increase in SCr from baseline)9 (Table 1).

The diagnosis of HRS has been defined in two International Ascites Club consensus conferences in 199616 and in 2005 (Table 2).17 The new definition of HRS: (1) excludes creatinine clearance because it is more complicated to perform and does not increase the accuracy of renal function estimation; (2) includes renal failure in the setting of ongoing bacterial infection (but in the absence of septic shock), indicating that the diagnosis can be established before completing antibiotic therapy; (3) determines that plasma volume expansion should be performed with intravenous albumin rather than saline solution; and (4) excludes minor diagnostic criteria (urinary indices) because these criteria have poor sensitivity and specificity.

Table 2. HRS New Diagnostic Criteria17
  • Cirrhosis with ascites

  • Serum creatinine >1.5 mg/dL (133 μmol/L)

    • HRS-1 doubling of the initial serum creatinine concentrations to a level greater than 2.5 mg/dL (>226 μmol/L) in less than 2 weeks

  • No improvement in serum creatinine (decrease to 1.5 mg/dL or less) after at least 2 days of diuretic withdrawal and expansion of plasma volume with albumin (1 g/kg body weight/day up to a maximum of 100 g/day)

  • Absence of shock

  • No current or recent treatment with nephrotoxic drugs or vasodilators

  • Absence of parenchymal kidney disease as indicated by proteinuria >500 mg/day, microhematuria (>50 red blood cells per high-power field), or abnormal renal ultrasonography

Although HRS is a diagnosis of exclusion, certain patient characteristics are more or less typical of HRS. The presence of ascites is a prerequisite for the diagnosis of HRS because the same mechanisms that lead to ascites formation lead to HRS11 (Fig. 2). Other features that are characteristic of patients with HRS can be gathered from Table 3, which summarizes baseline characteristics of 509 patients with HRS that met International Ascites Club criteria for HRS (only 5% of whom had HRS-2) and that specified, at a minimum, age, Child-Pugh class or score, and mean arterial pressure (MAP).14, 15, 23, 33, 39–48 Patients with HRS have advanced liver disease, as evidenced by the median Child-Pugh score in these patients being 11.2, and their low MAP (median, 74 mmHg), and low serum sodium (median, 127 mEq/L), findings that are consistent with the presence of vasodilatation (low MAP), sodium retention (ascites), water retention (dilutional hyponatremia), and renal vasoconstriction (HRS) (Fig. 2). If these findings are absent, the diagnosis of HRS is unlikely. Of note, SCr levels in HRS are approximately 3.6 mg/dL and rarely exceed 6 mg/dL, and urine output is usually approximately 600 mL/day (Table 3).

Table 3. Baseline Characteristics of 509 Patients with HRS in Different Series in the Literature that Specified, at a Minimum, Age, Child-Pugh Score/Class, and MAP
First Author YearN (n HRS-2)Age (years)Sex (% Male)% Alcoholic Cirrhosis (%alc hep)Child-Pugh ScoreMAP (mmHg)Serum Creatinine mg/dL (μmol/L)Serum Sodium (mEq/L)Urine Output (mL/day)Urinary Sodium (mEq/day)
  • All values are expressed as mean values except for Mulkay (median values).

  • *

    Median (IQ range) of studies; MAP, mean arterial pressure; alc hep, alcoholic hepatitis; IQ, interquartile.

Angeli '99 (33) study group5 (0)62-401/4 (B/C)765 (442)1306805
Angeli '99 (33) controls8 (0)61-37.51/7 (B/C)793.6 (319)1284796.6
Gulberg '99 (39)7 (0)5057570/7 (B/C)74>2 (> 177)1244257
Mulkay '01 (40)12 (0)546875 (42)2/10 (B/C)763.4 (301)1275007
Moreau '02 (23)99 (0)567190 (26)Most C782.9 (257)127--
Colle '02 (41)18 (0)477289 (39)13753.2 (283)127655-
Halimi '02 (42)18 (2)607378 (22)11823.2 (283)13148916
Duvoux '02 (43)12 (0)54586711653.9 (345)1256388
Solanki '03 (44) Study group12 (0)5175->9762.9 (257)-627-
Solanki '03 (44) Control group12 (0)5367->9742.2 (195)-569-
Kiser '05 (45) Group 119 (1)5268-11.5614.5 (398)128--
Kiser '05 (45) Group 216 (8)4763-11.2713.6 (319)133--
Kiser '05 (45) Group 38 (2)5150-11613 (265)132--
Alessandria '05 (15)41 (0)56884611713.9 (345)122--
Ruiz del Arbol '05 (14)12 (0)---12704.3 (381)124--
Neri '08 (46) Study group26 (0)5938.511.511.5722.8 (248)126571-
Neri '08 (46) Control group26 (0)604215.411.2682.9 (257)126620-
Sanyal '08 (47) Study group56 (0)517352 (36)12764 (354)131--
Sanyal '08 (47) Control group56 (0)537052 (36)11773.9 (345)132--
Martin-Llahi '08 (48) Study group23 (6)59706110733.6 (319)124607-
Martin-Llahi '08 (48) Control group23 (5)5556.58311684.1 (363)129596-
Median (IQ range)*Total N 509 (24)54 (5159)68 (5772)57 (4078)11.2 (1112)74 (6876)3.6 (2.94.0)127 (125131)596 (489638)7 (612)
319 (257354)

Differential Diagnosis

Differentiation among the three main causes of AKI may be difficult in patients with cirrhosis because the clinical presentations do not match classical paradigms and, as mentioned previously, factors that lead to prerenal azotemia can precipitate HRS and can also precipitate ATN (Fig. 3).

Tubular ability to reabsorb sodium and to concentrate urine is preserved in prerenal azotemia and HRS but is impaired in ATN. Prerenal azotemia and HRS are therefore classically described as sodium avid states with low (<20 mEq/L) urinary sodium (UNa), low (<1%) fractional excretion of sodium (FENa), and elevated (>500 mOsm/kg) urine osmolality. Conversely, patients with ATN have high UNa (>40 mEq/L), high FENa (2%), and urine osmolality under 350 mOsm/kg. However, in patients with HRS, particularly in those on a high dose of diuretics, UNa is consistently greater than 10 mEq/L.49 Conversely, ATN that occurs in patients with sodium avid states, such as cirrhosis, has been described as having a FENa ≤ 1%.7, 50 UNa and FENa are thus less useful in patients with cirrhosis and have been eliminated as diagnostic criteria in HRS.17

Classic descriptions of prerenal azotemia and HRS describe a “bland” urine sediment, and bile-stained granular and epithelial casts are described in ATN. However, these casts may be seen as nonspecific findings in patients with advanced liver disease and jaundice.51, 52 Conversely, ATN has been described morphologically31 and by urine biomarkers such as β-2 microglobulin32 in HRS.33 Thus, urine sediment may not be helpful either in the differential of ARF in cirrhosis.

Differentiation between prerenal azotemia and HRS also may be difficult. By definition, prerenal azotemia improves with volume expansion. However, assessment of exact intravascular volume deficit in a patient who is already total body sodium overloaded is difficult; the rate of fluid administration is unspecified, and thus the rate of response in SCr is highly variable and often incomplete.

The difficulty in establishing an early diagnosis of HRS leads to a significant delay in the initiation of specific therapy and is one of the barriers in the development of new agents.

In noncirrhotic patients, urinary biomarkers such as interleukin-18 have an accuracy of 95% in differentiating ATN from other causes of renal dysfunction such as prerenal azotemia, urinary tract infection, and chronic kidney disease53 and are being investigated in patients with cirrhosis.


Treatment of AKI in cirrhosis depends on its cause. Prerenal azotemia should be managed by treating/discontinuing the precipitant and through volume repletion. ATN should be treated using renal replacement therapy, particularly in the presence of volume overload, hyperkalemia, or metabolic acidosis not responding to medical therapy. Because of a lack of clinical studies, there are no special recommendations for dose, intensity, and duration of dialysis in patients with cirrhosis who develop ATN. Interestingly, a recent study has demonstrated benefit of terlipressin in a consecutive series of patients with cirrhosis with ATN.54 It is likely that mesenteric/systemic vasoconstriction induced by terlipressin will lead to renal vasodilatation with improvement of renal blood flow to the damaged renal tubules. Therapy of the complications of AKI and dialysis therapy are beyond the scope of this review. The remainder of this section deals with treatment of HRS.

Liver Transplantation (OLT)

OLT is the only definitive therapy for HRS, because it is the only therapy associated with improvement in survival. However, it is important to reverse HRS because improving renal function pretransplantation is associated with improved posttransplantation outcomes.55–59 It has been shown that renal insufficiency, including HRS, has a negative impact on OLT outcomes, with a higher mortality and higher posttransplantation end-stage renal disease.3, 55–58 Patients who are transplanted with HRS have more complications and a higher in-hospital mortality rate than those undergoing transplantation without HRS.56 Conversely, the outcome of OLT in patients with HRS treated with vasopressin analogs before transplantation is similar to that of patients undergoing transplantation without HRS.60

With the implementation of the MELD score in the allocation of organs in the United States, the presence of HRS increases the possibility of obtaining an organ for transplantation. However, although survival in HRS-2 (the more benign, chronic type of HRS) correlates with the MELD score, all patients with HRS-1 have a very poor outcome,15 suggesting that the development of HRS-1 per se should indicate a high priority for transplantation, independent of MELD.

Simultaneous liver and kidney transplant (SLKT) is increasingly considered in patients with renal dysfunction undergoing OLT.59 There is specific concern that some patients who undergo SLKT may have reversible renal failure. There is also concern that liver grafts are placed prematurely in those with end-stage renal disease. Thus, to assure allocation of transplants only to those truly in need, the transplant community met in March 2006 to review post-MELD data on the impact of renal function on liver waitlist and transplant outcomes and the results of SLKT.61 This consensus conference resulted in an evaluation algorithm that has the goal of confirming the presence of kidney disease with structural damage (preferably on biopsy) that would merit an SLKT. In the setting of chronic kidney disease, a measured creatinine clearance (or preferentially an iothalamate clearance) of 30 cc/minute or less was considered the appropriate threshold for SLKT.62 HRS alone should not be a reason for SLKT; however, this is now being considered in patients with HRS who become dialysis dependent and in whom there is no recovery after 6 to 8 weeks of dialysis (the usual recovery time for ATN).61 The need for dialysis should theoretically be prevented by early treatment of HRS with the “bridging” therapies outlined below, particularly the use of vasoconstrictive therapy.

Vasoconstrictors Plus Albumin

Because the main pathogenic mechanism in HRS is splanchnic and systemic vasodilatation (Fig. 2), vasoconstrictors should ameliorate vasodilatation and improve effective arterial blood volume, renal vasoconstriction, and renal flow. Vasoconstrictors have been used in conjunction with intravenous albumin with the intention of increasing effective blood volume. Because albumin dialysis is associated with an increase in MAP attributable to the ability of albumin to bind vasodilators, it is conceivable that an improvement of renal function in patients with HRS treated with vasoconstrictors and albumin is attributable to the additive effects of both compounds in producing vasoconstriction. The need for albumin has only been examined in a nonrandomized small study that showed that treatment with terlipressin and albumin was associated with a significant decrease in SCr and an increase in MAP, changes that were not observed in a nonconcurrent group of patients treated with terlipressin alone.63

Vasoconstrictors used in the treatment of HRS in small numbers of patients for periods greater than 3 days are associated with increases in MAP, GFR, and serum sodium and decreases in SCr and plasma renin activity (Table 4). Terlipressin is the preferred vasopressin analog because of a lower incidence of side effects and because its administration does not require continuous intravenous infusion.23, 40–42, 44, 46–48, 63, 64

Table 4. Changes in Mean Arterial Pressure, Parameters of Renal Function and Plasma Renin Activity in Prospective Proof-of-Concept Studies of Vasoconstrictors + Albumin Administered for More Than 3 Days
Author (Year)Vaso-constrictornMAP (mmHg)Serum Creatinine (mg/dL)Serum Sodium (mEq/L)Glomerular Filtration Rate (cc/min)Urine Output (cc/day)Urine Sodium (mEq/L)Plasma Renin Activity (ng/mL/h)
  • All values are expressed as mean values except for Mulkay (median values). Changes in bold lettering were statistically significant.

  • Or, ornipressin,; O+M, octreotide plus midodrine; T, terlipressin; N, noradrenaline.

  • *

    In responders, serum creatinine 4.6 →1.3 (↓ 72%) and serum sodium 124 →134 (↑8%).

  • Calculated by MDRD.

  • ng/L.

Guevara'98 (69)Or4698320%30.777%122 →131 ↑7%941356%4501383207%2 →7 ↑250%190.896%
Gulberg'99 (39)Or7748920%2 →?*124 →?*14.636.9153%425911112%730329%-
Angeli'99 (33)O+M5769728%51.864%1301397%10.446.1343%6801540126.5%546820%17477%
Uriz'01 (64)T9688018%3.91.561.5%1221317%824200%710111557%3 →7 ↑133%233.585%
Mulkay'01 (40)T12768714.5%3.41.653%1271356%1133200%5001160132%738443%-
Ortega'02 (63)T13707913%3.61.558%1231316.5%1030200%571105785%39200%20480%
Duvoux'02 (43)N12657312%3.91.659%1251315%1543187%6381982211%852550%56516471%

Noradrenaline, in continuous infusion, has also been shown to ameliorate the hemodynamic/renal abnormalities in HRS,43 as has the combination of midodrine plus octreotide, which has the advantage of oral/subcutaneous administration.33, 65, 66 However, octreotide alone is no more effective than placebo67 and did not improve SCr.45 It is therefore uncertain whether the effectiveness of the combination octreotide/midodrine is attributable to midodrine alone or to the combination. Experimental studies showing that octreotide potentiates the effect of vasoconstrictors68 would suggest the latter.

Clinical outcomes of 12 uncontrolled studies including 258 patients with HRS are summarized in Table 5.23, 33, 39–43, 45, 63–65, 69 Complete response (mostly defined as a decrease in SCr to <1.5 mg/dL) was observed in 60% (156/258) overall, and in 65% (110/169) of patients who received terlipressin.23, 40–42, 63, 64 Interestingly, HRS recurred in only a minority of “responders” (22% or 16/72) once therapy was discontinued. Median survival is approximately 41 days (compared with 14 days in a group of untreated patients.13).

Table 5. Clinical Outcomes of Therapy with Vasoconstrictors (Plus IV Albumin) in Uncontrolled Studies
Author and YearN (n HRS-2)Definition of Complete Response (CR)VasoconstrictorComplete Response n (%)Days to CR (median value and range)HRS Recurrence: n/responders (%)Median Survival (days)
  • All prospective except for Moreau, Colle, Halimi, and Kiser (retrospective).

  • Or, ornipressin; M+O, midodrine plus octreotide; T, terlipressin; N, noradrenaline; V, vasopressin; V+O, vasopressin plus octreotide; sCr, serum creatinine; CrCl, creatinine clearance; NA, not available; BL, baseline; CR, complete response.

  • *

    Median survival for whole series (N = 21; 13 with albumin, 8 without albumin).

  • No mention of albumin administration.

  • Mean.

Guevara '98 (69)8 (?)↓sCr<1.5 mg/dLOr6 (75)7 (4–15)2/6 (33)60
Gulberg '99 (39)7 (0)2-fold ↑CrCl to a value >40 mL/minOr4 (57)14 (8–27)2/4 (50)90
Angeli '99 (33)5 (0)↓sCr ≤2 mg/dLM+O5 (100)20 (5–20)0/5 (0)NA
Uriz '01 (64)9 (3)↓sCr<1.5 mg/dLT7 (78)6 (6–15)0/7 (0)39
Mulkay '01 (40)12 (0)↓sCr<2 mg/dLT11 (92)(2–12)4/9 (44)42
Ortega '02 (63)13 (4)↓sCr ≤1.5 mg/dLT10 (77)7 (4–14)1/10 (10)40*
Moreau '02 (23)99 (0)↓sCr<1.5 mg/dL or of at least 20% from BLT58 (58)NANA21
Colle '02 (41)18 (0)↓sCr<1.5 mg/dL or of at least 20% from BLT11 (61)NA7/11 (64)24
Halimi '02 (42)18 (0)↓sCr at least of 30% from BLT13 (72)NANANA
Duvoux '02 (43)12 (0)↓sCr <1.5 mg/dL or ↑CrCl to >40 mL/minN10 (83)7 (5–10)0/10 (0)60
Wong '04 (65)14 (0)↓sCr<1.5 mg/dL for 3 consecutive daysM+O10 (71)160/10 (0)NA
Kiser '05 (45)19 (1)↓sCr ≤1.5 mg/dL without dialysisV+O8 (42)7NANA
Kiser '05(45)16 (8) O0 (0)7.5NANA
Kiser '05 (45)8 (2) V3 (38)6NANA

Five randomized controlled trials (RCTs) of terlipressin in HRS compared with albumin alone46, 48 or with a placebo44, 47, 70 have been published. The study by Hadengue et al70 was a proof-of-concept crossover study that showed that terlipressin (but not placebo) was associated with an increase in GFR. All other studies investigated clinical outcomes (Table 6) and, except for the study by Sanyal et al.,47 they were all open-label studies. In all of them, HRS reversal was higher in the terlipressin group (54/117 or 46%) compared with the control group (13/117 or 11%). In the only placebo-controlled, double-blind multicenter trial that included the largest number of patients,47 HRS reversal occurred in 34% of the patients, a rate significantly greater than that of placebo-treated patients (13%) but lower than that reported in uncontrolled studies. The low rate of recurrence of HRS in treated patients is confirmed in these studies (2/29 or 7%). Both controlled and uncontrolled studies agree that survival is significantly better in terlipressin “responders.”23, 45, 47, 48 Additional information from recent RCTs46 confirms that terlipressin plus albumin has a beneficial effect in the treatment of HRS. Because survival was not improved in the two largest RCTs,47, 48 liver transplantation is still the therapy of choice for HRS, but terlipressin appears to be the “bridging” therapy of choice.

Table 6. Clinical Outcomes of Therapy with Terlipressin (Plus Albumin) in Randomized Controlled Trials
Author YearDefinition of Complete Response (CR)GroupnComplete Response n (%)Days to Achieve CR (median and range)HRS Recurrence (%)Median Survival (days)
  • sCr, serum creatinine; NA, not available.

  • *

    Terlipressin-responders only.

  • Three-month survival.

  • ∼30% of patients had type 2 HRS.

Solanki '03 (44)“Improvement in renal function”Terlipressin125 (42)15?“Some”8.5
Control120 (0)--9.0
Neri '08 (46)↓sCr ≤1.5 mg/dLTerlipressin2621 (81)NANA90*
  Control265 (19)NANA15
Sanyal '08 (47)↓sCr ≤1.5 mg/dL without dialysisTerlipressin5619 (34)6 (2–14)1/19 (5)24
Placebo567 (13)3 (2–7)1/7 (14)31
Martin-Llahi '08 (48)↓sCr<1.5 mg/dLTerlipressin239 (39)11 (6–16)1/10 (10)27%
Control231 (4)NANA19%

Specifics such as the optimal time for initiation of vasoconstrictive therapy, dose and duration, failure criteria, and the influence and dose of albumin remain to be determined. Vasoconstrictor therapy has been used at different doses and with different side effects (Table 7); patients that receive higher doses appear to have a higher rate of adverse events.71

Table 7. Vasoconstrictors in HRS: Doses Used and Adverse Events
DrugDose RangeObserved Adverse Events
  1. GI, gastrointestinal; AF, atrial fibrillation; MI, myocardial infarction; AE, adverse events.

Terlipressin (23;40–42;44;46–48;63;64;70)0.5–2.0 mg intravenously every 4–6 hours• GI: abdominal cramps associated with increased bowel movements, diarrhea, nausea, vomiting, intestinal ischemia
  • Cardiac: arrhythmia (tachycardia, bradycardia, extrasystolia, AF), chest pain, MI
  • Peripheral: livedo reticularis, finger ischemia, cutaneous necrosis at the infusion site, scrotal necrosis, skin lymphangitis
  • Others: arterial hypertension, dyspnea, bronchospasm, respiratory acidosis
Ornipressin (39;69)2–6 U/h (continuous intravenous infusion)Abdominal cramps, intestinal ischemia with or without bleeding, tongue ischemia, cardiac arrhythmia
Vasopressin (45)0.01–0.8 U/min (continuous intravenous infusion)None reported
Noradrenaline (43;73)0.5–3.0 mg/h (continuous intravenous infusion)Chest pain with or without ventricular hypokinesia
Octreotide + Midodrine (33;65)100–200 μg subcutaneously three times a day
 7.5–12.5 mg orally three times a dayDiarrhea, tingling, goosebumps
 25 μg → 25 μg/h (continuous intravenous infusion) 
 2.5 mg/day orally 

Terlipressin needs to be compared with other vasoconstrictors (Table 7). Two recent small, open-label RCTs comparing noradrenaline with terlipressin showed that neither HRS reversal nor the rate of side effects was different between groups.72, 73 This is interesting from a pathophysiological perspective because noradrenaline does not induce splanchnic vasoconstriction, suggesting that the main effect is peripheral vasoconstriction.

Transjugular Intrahepatic Portosystemic Shunt

Three small prospective but uncontrolled studies have assessed the role of transjugular intrahepatic portosystemic shunt (TIPS) in HRS,65, 74, 75 including one performed in five patients whose renal function had improved on octreotide/midodrine.65 These studies show a decrease in SCr in most patients, even in a few with organic renal failure,76 but slower than that obtained using terlipressin plus albumin. Recurrence of HRS is rare as long as the shunt remains patent, but hepatic encephalopathy is a frequent complication. Post-TIPS resolution of HRS appears to improve survival. Long-term success was demonstrated in the study that explored sequential treatment with vasoconstrictors and albumin followed by TIPS.65 Notably, a great majority of patients included in these three studies had alcoholic cirrhosis, many with active alcoholism, and therefore the improvement observed could have resulted from improvement in an acute-on-chronic process. Also, all three studies excluded patients with a Child-Pugh score equal to or greater than 12.

Given the paucity of data, the efficacy of TIPS should be further explored in RCTs.

Extracorporeal Albumin Dialysis

In a small RCT, the molecular adsorbent recirculating system, a modified dialysis method using an extracorporeal albumin-containing dialysate, was shown to improve 30-day survival in eight patients with HRS compared with five patients treated with intermittent venovenous hemofiltration alone.77 Because extracorporeal albumin-containing dialysate incorporates a standard dialysis machine or a continuous venovenous hemofiltration monitor and GFR was not measured, the decrease in SCr observed in most patients could be related to the dialysis process. However, clear beneficial effects on MAP and on hepatic encephalopathy were observed. Extracorporeal albumin-containing dialysate is still considered an experimental therapy, and its use in patients with type 1 HRS cannot be recommended outside of prospective studies.

Approach to the Patient with Cirrhosis and ARF

We propose that, as per the recent AKI consensus (Table 1), a diagnosis of HRS (type 1) be considered whenever there is an increase in SCr of 0.3 mg/dL or more (≥26.4 μmol/L) or an increase of 150% or more (1.5-fold) from baseline. Delaying therapy until higher SCr levels are reached would not seem warranted because baseline SCr is a predictor of HRS reversal,47, 48, 65 and the probability of HRS reversal decreases by 39% for each 1-mg/dL increase in baseline creatinine.78 Treatment for HRS should therefore be initiated earlier, with only 1.5-fold increases in creatinine (Fig. 4).

Figure 4.

Approach to the patient with cirrhosis and acute renal failure. IV, intravenously; QD, every day; BID, twice per day; ATN, acute tubular necrosis; HRS, hepatorenal syndrome. *If compatible clinical characteristics, in other words, ascites, low MAP, and hyponatremia (see text for approach to patient with suspected HRS).

The setting in which AKI occurs is essential. In patients in whom there is a clear history of septic or hypovolemic shock, or in whom there is a recent history of nephrotoxin administration, the most likely cause is ATN. In all other patients, the first step is to discontinue diuretics, lactulose (because it is a frequent cause of diarrhea), vasodilators, and any potential nephrotoxins. Secondly, intravascular volume should be expanded with intravenous albumin at a dose of 1 g/kg body weight up to a maximum of 100 g.17 This dose can be repeated in 12 hours if the SCr has not normalized. A reduction in SCr indicates that AKI is attributable to prerenal azotemia. Intravenous albumin is preferred over saline solution as a volume expander because sodium load is significantly lower with albumin, and it will not worsen fluid retention, and because it will not be associated with a dilutional decrease in SCr. Third, and concurrently, investigations to rule out precipitants of AKI should be undertaken, specifically diagnostic paracentesis (to rule out SBP), blood and urine bacteriological cultures, and chest x-ray (to rule out bacterial infections other than SBP) and patients treated accordingly.

If SCr does not improve or continues to worsen despite these measures, the differential diagnosis is between intrinsic renal failure (ATN), HRS, and postrenal failure. Treatment of HRS can be initiated before completion of antibiotic therapy in patients with bacterial infection whose SCr does not improve despite a clear amelioration in the signs of infection. Because assessment of volume status may be uncertain, and to ensure that volume has been adequately expanded, intravascular volume should be assessed by measuring central venous pressure. A normal or increased central venous pressure indicates that the cause of renal failure is not volume-related. To rule out postrenal failure, a renal ultrasound should be obtained, although this is a rare cause of AKI in cirrhosis (Fig. 1). To rule out intrinsic renal failure, urinary sediment should be analyzed. Finding granular or epithelial casts suggests ATN but is not definitive. The differentiation between ATN and HRS is the most difficult, and studies using urine biomarkers of ATN are awaited. It has been suggested that the response to vasoconstrictors plus albumin may be used to establish this differential.7

Approach to the Patient with HRS

Once the diagnosis of HRS is suspected, specific treatment with vasoconstrictors plus albumin should be initiated. Best evidence supports the use of terlipressin, which should be started at a dose of 0.5 mg intravenously every 6 hours. If there is no early response (>25% decrease in creatinine levels) after 2 days of therapy, the dose can be doubled every 2 days up to a maximum of 12 mg/day (in other words, 2 mg intravenously every 4 hours). Starting at lower doses will minimize potential severe adverse events from this therapy.71

A more rational method to adjust the dose of vasoconstrictors is by monitoring MAP (an indirect indicator of vasodilatation), a method that has been used for adjusting the dose of midodrine plus octreotide,33 an alternative to terlipressin in sites such as the United States, where terlipressin is not available. In this study, the doses of octreotide and midodrine were titrated to obtain an increase in MAP of at least 15 mmHg; midodrine was administered orally at an initial dose of 7.5 mg three times daily and, if necessary, increased to 12.5 mg three times daily; octreotide was administered subcutaneously at an initial dose of 100 μg three times daily and, if necessary, increased to 200 μg three times daily.33 Other alternatives to terlipressin are vasopressin or noradrenaline. Noradrenaline is used as a continuous intravenous infusion at an initial dose of 0.5 mg/hour, adjusted to achieve an increase in MAP of at least 10 mm Hg or an increase in 4-hour urine output to more than 200 mL. If these goals are not reached, the dose is increased every 4 hours in steps of 0.5 mg/hour, up to the maximum dose of 3 mg/hour.43, 73 Vasopressin is also used as a continuous intravenous infusion, starting at a very low dose of 0.01 U/min and titrating up to a dose of 0.5 U/min, depending on changes in MAP and urine output as well as the presence of ischemic side effects. In the study that assessed vasopressin for HRS, the mean vasopressin dose in the group of responders was 0.23 ± 0.19 U/min.45

Albumin is administered together with the vasoconstrictor. The maintenance dose, once the diagnosis of HRS is established and vasoconstrictors are initiated, is of 25 to 50 g/day. Albumin may be discontinued if serum albumin concentration is greater than 45 g/L and should be withdrawn in case of pulmonary edema. Because this complication is uncommon, catheterization to monitor central venous pressure is not mandatory, but careful monitoring of the cardiopulmonary function is recommended.17

Treatment can be stopped if SCr does not decrease by at least 50% after 7 days at the highest dose, or if there is no reduction in creatinine after the first 3 days. In patients with early response, treatment should be extended until reversal of HRS (decrease in creatinine below 1.5 mg/dL) or for a maximum of 14 days.17 Vasoconstrictor therapy should be restarted if HRS recurs after discontinuation of therapy. Once creatinine normalizes, TIPS should be considered, particularly if transplantation is not foreseeable in the near future and the patient has refractory ascites.65


Prophylaxis of AKI in cirrhosis consists of measures that will prevent/treat volume depletion or vasodilatation. Measures that prevent volume depletion include careful use of diuretics with close weight and laboratory follow-up, preventing weight loss of more than 1 kg/day as well as avoidance of diarrhea with the use of lactulose by adjusting its dose to obtain two to three semiformed bowel movements/day. The use of albumin after large-volume paracentesis (LVP) is a measure that can theoretically prevent the development of renal dysfunction by preventing the development of postparacentesis circulatory dysfunction,79 an entity that has been shown to be secondary to vasodilatation80 and is associated with development of renal dysfunction and a higher mortality.81

Measures to prevent ATN include avoiding the use of aminoglycosides and nonsteroidal anti-inflammatory drugs, particularly in patients with ascites, and aggressively treating hypovolemia/hypotension when it occurs.

It is uncertain whether HRS can really be prevented. It has been suggested that in the setting of SBP, the administration of albumin prevents HRS. This is based on a non–placebo-controlled study in patients with SBP, in which the intravenous administration of albumin decreased the risk of HRS by 66% compared with antibiotic treatment alone.82 However, the benefit of albumin is observed mainly in patients with SCr greater than 1.0, urea greater than 30 mg/dL, or a serum bilirubin greater than 4 mg/dL,82–84 that is, patients who already have some degree of renal dysfunction at the time of the diagnosis of SBP.

On the other hand, a recent placebo-controlled RCT that included hospitalized patients with low (<1.5 g/L) ascites protein who also had advanced liver failure or “renal dysfunction” (defined as SCr ≥ 1.2 mg/dL or blood urea nitrogen ≥ 25 mg/dL, or serum sodium level ≤ 130 mEq/L) demonstrated that oral norfloxacin was associated with a reduction in the 1-year probability of HRS (28% versus 41%) and an improvement in survival at 3 but not 12 months.85 Norfloxacin may act by ameliorating or preventing vasodilatation by preventing bacterial translocation and overt infection.86 Notably, 79% of the patients met “renal dysfunction” criteria at baseline, suggesting that patients with a more altered hemodynamic status were those more likely to benefit from antibiotics and would probably benefit from vasoconstrictors.

In both studies,82, 85 the relation between prevention of HRS and improved short-term survival is further proof that renal failure is an important determinant of death in patients with decompensated cirrhosis1 and that early treatment or prevention are essential.