Adrenal insufficiency in patients with cirrhosis and septic shock: Effect of treatment with hydrocortisone on survival

Authors

  • Javier Fernández,

    Corresponding author
    1. Liver Unit, Department of Surgery and Centro de Investigaciones Biomédicas Esther Koplowitz, IMDiM, IDIBAPS, Hospital Clínic, University of Barcelona, Spain
    • Liver Unit, Hospital Clínic, Villarroel 170, 08036, Barcelona. Spain
    Search for more papers by this author
    • fax: (34) 93-4515522

  • Angels Escorsell,

    1. Liver Unit, Department of Surgery and Centro de Investigaciones Biomédicas Esther Koplowitz, IMDiM, IDIBAPS, Hospital Clínic, University of Barcelona, Spain
    Search for more papers by this author
  • Michel Zabalza,

    1. Liver Unit, Department of Surgery and Centro de Investigaciones Biomédicas Esther Koplowitz, IMDiM, IDIBAPS, Hospital Clínic, University of Barcelona, Spain
    Search for more papers by this author
  • Vanessa Felipe,

    1. Liver Unit, Department of Surgery and Centro de Investigaciones Biomédicas Esther Koplowitz, IMDiM, IDIBAPS, Hospital Clínic, University of Barcelona, Spain
    Search for more papers by this author
  • Miguel Navasa,

    1. Liver Unit, Department of Surgery and Centro de Investigaciones Biomédicas Esther Koplowitz, IMDiM, IDIBAPS, Hospital Clínic, University of Barcelona, Spain
    Search for more papers by this author
  • Antoni Mas,

    1. Liver Unit, Department of Surgery and Centro de Investigaciones Biomédicas Esther Koplowitz, IMDiM, IDIBAPS, Hospital Clínic, University of Barcelona, Spain
    Search for more papers by this author
  • Antonio M. Lacy,

    1. Liver Unit, Department of Surgery and Centro de Investigaciones Biomédicas Esther Koplowitz, IMDiM, IDIBAPS, Hospital Clínic, University of Barcelona, Spain
    Search for more papers by this author
  • Pere Ginès,

    1. Liver Unit, Department of Surgery and Centro de Investigaciones Biomédicas Esther Koplowitz, IMDiM, IDIBAPS, Hospital Clínic, University of Barcelona, Spain
    Search for more papers by this author
  • Vicente Arroyo

    1. Liver Unit, Department of Surgery and Centro de Investigaciones Biomédicas Esther Koplowitz, IMDiM, IDIBAPS, Hospital Clínic, University of Barcelona, Spain
    Search for more papers by this author

  • Potential conflict of interest: Nothing to report.

Abstract

Relative adrenal insufficiency is frequent in patients with severe sepsis and is associated with hemodynamic instability, renal failure, and increased mortality. This study prospectively evaluated the effects of steroids on shock resolution and hospital survival in a series of 25 consecutive patients with cirrhosis and septic shock (group 1). Adrenal function was evaluated by the short corticotropin test within the first 24 hours of admission. Patients with adrenal insufficiency were treated with stress doses of intravenous hydrocortisone (50 mg/6 h). Data were compared to those obtained from the last 50 consecutive patients with cirrhosis and septic shock admitted to the same intensive care unit in whom adrenal function was not investigated and who did not receive treatment with steroids (group 2). Incidence of adrenal insufficiency in group 1 was 68% (17 patients). Adrenal dysfunction was frequent in patients with advanced cirrhosis (Child C: 76% vs. Child B: 25%, P = .08). Resolution of septic shock (96% vs. 58%, P = .001), survival in the intensive care unit (68% vs. 38%, P = .03), and hospital survival (64% vs. 32%, P = .003) were significantly higher in group 1. The main causes of death in group 1 were hepatorenal syndrome or liver failure (7 of 9 patients). In contrast, refractory shock caused most of the deaths in group 2 (20 of 34 patients). In conclusion, relative adrenal insufficiency is very frequent in patients with advanced cirrhosis and septic shock. Hydrocortisone administration in these patients is associated with a high frequency of shock resolution and high survival rate. (HEPATOLOGY 2006;44:1288–1295.)

Bacterial infections are a frequent complication and an important cause of death in patients with advanced cirrhosis.1–3 The mortality rate for patients admitted to the hospital with severe bacterial infections but without septic shock is approximately 30%,4–6 although it may decrease to 10% if circulatory support with intravenous albumin is given at infection diagnosis.7 The development of a complex syndrome characterized by acute deterioration of circulatory, hepatic, and renal function and hepatic encephalopathy is the main cause of death in these patients.6, 7 Prognosis of patients with cirrhosis admitted with septic shock is even worse. Hospital mortality in these patients ranges between 60% and 100%,8, 9 which is higher than that reported in the general population of patients with septic shock (25%–70%).10–12 In our experience, survival of patients with cirrhosis in septic shock has not improved greatly during the past decades despite important advances in antibiotic treatment and general supportive measures.

During the last few years, several studies have shown that septic shock is frequently associated with relative adrenal insufficiency, a condition characterized by an inadequate production of cortisol with respect to the peripheral demands.11–18 It also has been reported that patients with septic shock and relative adrenal insufficiency show resistance to vasoconstrictor drugs,19 a higher incidence of refractory shock,13 and very high hospital mortality rate.11, 15, 17, 18 Finally, various trials12, 20, 21 and two meta-analyses10, 22 show that the administration of low doses of hydrocortisone (stress doses) during several days improves shock reversal and survival in septic shock. Assessment of adrenal function and treatment with stress doses of hydrocortisone in cases with relative adrenal insufficiency is, therefore, recommended for the management of severe sepsis.13, 23

Standard treatment of septic shock in patients with advanced cirrhosis consists of fluid resuscitation with colloids, administration of vasoconstrictors to increase arterial pressure, and dobutamine in cases of cardiac dysfunction, specific treatments for organ failure (renal replacement, mechanical ventilation) and other supportive measures. In June 2004, we decided to introduce the assessment of adrenal function in the management of all patients with cirrhosis admitted to our intensive care unit with septic shock and to treat those patients with relative adrenal insufficiency with stress doses of hydrocortisone. The current article compares the clinical course and hospital mortality of the first 25 patients treated with this new approach with that observed in the last 50 patients receiving the standard treatment.

Abbreviations

SOFA, Sequential Organ Failure Assessment; APACHE-II, Acute Physiology, Age, Chronic Health Evaluation II.

Patients and Methods

Patients.

Two groups of patients with cirrhosis admitted to a liver intensive care unit with septic shock were studied. Group 1 (prospective series, admitted from June 2004 to January 2006) included 25 consecutive patients in whom adrenal function was assessed and who received intravenous hydrocortisone at low doses if relative adrenal insufficiency was diagnosed. Group 2 (retrospective series, admitted from March 2001 to May 2004) included 50 consecutive patients with no assessment of adrenal function and no treatment with hydrocortisone. Diagnosis of cirrhosis was based on histology or on clinical, laboratory, and ultrasonographic data. Septic shock was diagnosed clinically by the presence of data compatible with systemic inflammatory response syndrome, a mean arterial pressure below 60 mmHg during more than 1 hour despite adequate fluid resuscitation (increase in central venous pressure to approximately 14 mmHg), and need for circulatory support with vasopressor drugs.24 Two or more of the following criteria were required to diagnose the presence of systemic inflammatory response syndrome25: temperature greater than 38°C or less than 36°C, heart rate greater than 90 beats/min, tachypnea greater than 20 breaths/min, white cell count greater than 12 × 109/L or less than 4 × 109/L or more than 10% of immature neutrophils. Diagnosis of septic shock was performed in all patients before admission into the intensive care unit. In nine patients in group 1 and in 22 in group 2, it was performed in the standard liver hospitalization unit. In the remaining 16 patients in group 1 and 28 patients in group 2, septic shock was diagnosed in the emergency area. Patients with cirrhosis after liver transplantation, with human immunodeficiency virus infection, or with advanced hepatocellular carcinoma were not included. No patient had history (within 6 months) of steroid treatment, and none received previous treatment with drugs inhibiting steroidogenesis. The change in our protocol of treatment of septic shock in cirrhosis was decided by the Committee of the Intensive Care Units of the hospital. The study was approved by the Ethics Committee.

Management.

Patient assessment and treatment of septic shock was identical in the two groups except for the evaluation of adrenal function and treatment with hydrocortisone. A central line and a urinary catheter were inserted in all cases before admission into the intensive care unit for the diagnosis and early management of septic shock. Ascitic fluid leukocyte concentration, fresh urine sediment and blood, ascitic fluid, and urine cultures and chest X-ray also were obtained before admission into the intensive care unit. Bronchial samples were obtained and cultured in the case of pneumonia. The mean period between the initiation of septic shock and the admission into the intensive care unit was 10 ± 16 hours in group 1 and 11 ± 9 hours in group 2. Physical examination, chest radiographs, and standard laboratory measurements were repeated at admission into the intensive care unit in all cases. Patients were treated according to a standard protocol that includes fluid resuscitation with colloids to maintain a right atrial pressure of approximately 14 mmHg and dopamine (1–20 μg/kg/min) or norepinephrine (0.05–4 μg/kg/min) to achieve a mean arterial pressure over 70 mmHg. Patients with spontaneous bacterial peritonitis additionally received intravenous albumin (1.5 g/kg body weight at diagnosis and 1 g/kg body weight on day 3). Albumin was also prescribed in cases with serum albumin less than 20 g/L (doses were adjusted to increase albumin levels to >25 g/L). Antibiotics were administered in both groups of patients according to a previously defined local protocol on treatment of bacterial infections in cirrhosis. Patients with severe renal failure were treated with continuous venovenous hemofiltration, and those with severe respiratory failure were mechanically ventilated. Sedation was performed using midazolam.

Severity of hepatic failure was estimated by the Child-Pugh and model for end-stage liver disease scores. Severity of illness was assessed by the SOFA (Sequential Organ Failure Assessment) and the APACHE-II (Acute Physiology, Age, Chronic Health Evaluation II) scoring systems.26–28 Data used to calculate these scores were those obtained at admission into the intensive care unit. Renal failure was diagnosed when serum creatinine concentration was greater than 1.5 mg/dL. Type 1 hepatorenal syndrome was defined according to the International Ascites Club.29 Diagnostic criteria of spontaneous bacteremia, spontaneous bacterial peritonitis, pneumonia, urinary tract infections, and other bacterial infections have been previously reported.3 Refractory shock was defined as a shock not responding to therapy with volume expansion and high doses of vasopressors in the intensive care unit. These patients died in the setting of multiple organ failure. Acute respiratory distress syndrome was considered when patients developed acutely bilateral chest radiographic infiltrates and hypoxemia in the absence of signs of cardiac dysfunction.30

In patients from group 1, a synacthen test was performed within the first 24 hours of admission. Synthetic adrenocorticotropic hormone (250 μg, Synacthen, Novartis Pharma AG, Basle, Switzerland) was given intravenously. Blood samples to measure plasma cortisol levels (competitive immunoassay using direct chemiluminescent technology; Advia-Centaur, Bayer, Pittsburgh, PA) were obtained before and 60 minutes after synacthen administration. The coefficient of variation for this test is 7%. Relative adrenal insufficiency was diagnosed when one of the following two criteria was met: (1) Baseline cortisol concentration less than 15 μg/dL; (2) Increase in plasma cortisol after synacthen administration lower than 9 μg/dL in patients with baseline cortisol concentration below 35 μg/dL.31 Patients with adrenal insufficiency received 50 mg intravenous hydrocortisone every 6 hours.12 Treatment was gradually discontinued (reduction of 50 mg/d) when patients did not require vasopressor drugs to maintain arterial pressure.

Statistical Analysis.

Continuous variables were compared by the Student t test or by the Mann-Whitney U-test when indicated. Discontinuous variables were compared by the chi-squared test. Yates' correction was applied when the number of cases in a cell was lower than 5. Probability curves were obtained by the Kaplan-Meier method and compared by the log-rank test. Calculations were performed with the SPSS Statistical Package (SPSS Inc. Version 11.0, 2000, Chicago, IL). Differences were considered significant at .05. Results are given as mean ± standard deviation.

Results

Clinical Characteristics of the Patients at Admission Into the Intensive Care Unit.

With the exception of serum albumin, which was lower in group 1, there were no significant differences in demographic, clinical, and laboratory data and the cause of septic shock between the two groups (Table 1). In 16 patients (64%) from group 1 and in 38 (76%) from group 2, the bacteria responsible for septic shock were identified. Positive cultures in blood, ascitic fluid, bronchial, and urine samples were obtained in 12, 1, 5, and 2 patients, respectively, in group 1 and in 23, 12, 12, and 5 patients, respectively, in group 2. The most frequent bacteria isolated in the study were gram-negative bacilli (56% of culture-positive infections in group 1 vs. 57% in group 2). Gram-positive cocci plus gram-negative bacilli were isolated in one patient from group 1 (6%) and in five patients from group 2 (13%). Escherichia coli was by far the most frequently isolated organism (seven patients in group 1 and 13 in group 2). No significant differences were found between the two groups in the degree of hepatic failure, estimated by the Child-Pugh and Model for End-Stage Liver Disease scores, in the severity of the illness, estimated by the SOFA and APACHE-II scores, number of patients needing mechanical ventilation, and vasopressor requirements. Three patients from group 1 and six from group 2 had acute respiratory distress syndrome at admission to the intensive care unit. Seven patients from group 1 and nine from group 2 had diabetes mellitus, and two and four patients, respectively, had hepatocellular carcinoma. Cirrhosis with superimposed acute alcoholic hepatitis was histologically diagnosed in three patients from group 1 and in four from group 2.

Table 1. Clinical Characteristics at Admission to the Intensive Care Unit in Group 1 (Prospective Series) and in Group 2 (Retrospective Series)
 Group 1 (n = 25)Group 2 (n = 50)
  • *

    : Arbitrarily defined as a daily alcohol intake over 20 g in patients with alcoholic cirrhosis.

  • Within 12 hours after admission.

  • Mean doses of dopamine: 6.9 ± 3.3 μg/kg/min in group 1 and 10.4 ± 12.3 μg/kg/min in group 2.

  • §

    Mean doses of norepinephrine: 0.5 ± 0.2 μg/kg/min in group 1 and 0.4 ± 0.1 μg/kg/min in group 2.

  • #

    Mean doses of dopamine: 14.7 ± 9.8 μg/kg/min in group 1 and 17.0 ± 33.5 μg/kg/min in group 2; mean doses of norepinephrine: 0.2 ± 0.1 μg/kg/min in group 1 and 0.9 ± 1.2 μg/kg/min in group 2.

  • **

    P < .05. Doses of vasopressor drugs were not significantly different between groups.

Age (years)57 ± 1156 ± 10
Alcoholic cirrhosis (%)7660
Active alcoholism (%)*4028
TemperatureC)36.4 ± 1.336.5 ± 1.3
Mean arterial pressure (mmHg)55 ± 955 ± 9
Heart rate (beats/min)98 ± 27108 ± 25
Right atrial pressure (mmHg)14 ± 612 ± 6
Hematocrit (%)27 ± 729 ± 7
Leukocyte count109/L)10068 ± 788311568 ± 7347
Platelet count109/L)78760 ± 3939483100 ± 49312
Aspartate aminotransferase levels (IU/L)133 ± 171191 ± 328
Alanine aminotransferase levels (IU/L)62 ± 6192 ± 118
Serum bilirubin (mg/dL)9.9 ± 8.410.0 ± 10.4
Serum albumin (g/L)23 ± 527 ± 5**
Prothrombin index (%)38 ± 1339 ± 21
Serum creatinine (mg/dL)2.1 ± 1.62.8 ± 2.0
Renal failure (%)5664
Serum sodium (mEq/L)130 ± 7134 ± 7
Serum potassium (mEq/L)4.1 ± 1.14.0 ± 0.9
Ascites (%)8484
Hepatic encephalopathy (%)6068
Child-Pugh (score)11.6 ± 1.911.2 ± 2.2
Child-Pugh (% A/B/C)0/16/846/10/84
MELD (score)26.7 ± 7.928.4 ± 9.9
APACHE II (score)25.1 ± 5.624.4 ± 8.0
SOFA (score)12.4 ± 3.012.7 ± 3.7
Etiology of septic shock (%)  
 Pneumonia4444
 Spontaneous bacterial peritonitis2434
 Bacteremia164
 Other infections1618
Nosocomial infection (%)3644
Mechanical ventilation (%)6862
Treatment with dopamine (%)4866
Treatment with norepinephrine (%)§2816
Treatment with dopamine and norepinephrine (%)#2418

Incidence of Relative Adrenal Insufficiency in Group 1.

Seventeen of the 25 patients (68%) had adrenal insufficiency. Five patients had a baseline cortisol level less than 15 μg/dL, five had an increase in plasma cortisol after synacthen less than 9 μg/dL, and seven had both a baseline cortisol level less than 15 μg/dL and a cortisol increment after synacthen less than 9 μg/dL. Mean baseline value of cortisol and mean peak value and increment of cortisol after synacthen administration in patients with and without relative adrenal insufficiency were 12.2 ± 6.7 versus 25.0 ± 9.5 μd/dL, 19.0 ± 5.1 versus 37.7 ± 9.4 μg/dL and 6.8 ± 3.1 versus 12.8 ± 4.7 μg/dL, respectively. No differences were seen in clinical and laboratory data, degree of hepatic failure, severity of illness, and type of infection at admission in the intensive care unit between patients having the two diagnostic criteria of relative adrenal insufficiency and those with only one. No significant differences were observed when patients with and without relative adrenal insufficiency were compared (Table 2). There was a tendency, however, for a lower heart rate at admission into the intensive care unit in patients with adrenal insufficiency and for a higher incidence of adrenal insufficiency in patients with cirrhosis with Child-Pugh grade C (16 of 21 patients, 76%) than in those with grade B (1 of 4 patients, 25%) (P = .08). Excluding parameters related to hepatic function, no significant differences in the baseline clinical characteristics at admission into the intensive care unit were observed between Child-Pugh grades B and C patients included in group 1 (data not shown).

Table 2. Clinical Characteristics at Admission to the Intensive Care Unit of Patients With and Without Adrenal Insufficiency (Group 1)
 Adrenal Insufficiency (n = 17)Normal Adrenal Function (n = 8)
  • *

    Arbitrarily defined as a daily alcohol intake over 20 g in patients with alcoholic cirrhosis.

  • Within 12 hours after admission.

  • Mean doses of dopamine: 6.3 ± 3.6 μg/kg/min in patients with adrenal insufficiency and 8.2 ± 2.7 μg/kg/min in patients without adrenal insufficiency.

  • §

    Mean doses of norepinephrine: 0.6 ± 0.4 μg/Kg/min in patients with adrenal insufficiency and 0.5 ± 0.1 μg/kg/min in patients without adrenal insufficiency.

  • #

    Mean doses of dopamine and norepinephrine: 14.7 ± 9.8 μg/kg/min and 0.2 ± 0.1 μg/kg/min, respectively. Doses of vasopressor drugs were not significantly different between patients with and without adrenal insufficiency.

Age (years)58 ± 955 ± 14
Alcoholic cirrhosis (%)7775
Active alcoholism (%)*2962
Temperature (°C)36.5 ± 1.436.1 ± 1.0
Mean arterial pressure (mmHg)57 ± 952 ± 7
Heart rate (beats/min)90 ± 25113 ± 24
Right atrial pressure (mmHg)15 ± 713 ± 5
Hematocrit (%)27 ± 528 ± 9
Leukocyte count109/L)9494 ± 585411287 ± 11498
Platelet count109/L)83290 ± 4013169130 ± 38517
Aspartate aminotransferase levels (IU/L)140 ± 209119 ± 67
Alanine aminotransferase levels (IU/L)58 ± 6571 ± 54
Serum bilirubin (mg/dL)8.9 ± 6.312.0 ± 12.1
Serum albumin (g/L)23 ± 521 ± 6
Prothrombin index (%)39 ± 1237 ± 16
Serum creatinine (mg/dL)1.9 ± 1.12.5 ± 2.4
Renal failure (%)5362
Serum sodium (mEq/L)131 ± 6129 ± 9
Serum potassium (mEq/L)4.2 ± 1.14.0 ± 1.0
Ascites (%)8288
Hepatic encephalopathy (%)6550
Child-Pugh (score)11.7 ± 1.611.5 ± 2.5
Child-Pugh (% B/C)6/9437/63
MELD (score)26.1 ± 7.128.1 ± 10.0
APACHE II (score)24.0 ± 5.927.4 ± 4.4
SOFA (score)12.1 ± 3.312.9 ± 2.4
Etiology of shock (%)  
 Pneumonia4738
 Spontaneous bacterial peritonitis2425
 Bacteremia637
 Other infections230
Nosocomial infection (%)3537
Mechanical ventilation (%)6575
Treatment with dopamine (%)4750
Treatment with norepinephrine (%)§1850
Treatment with dopamine and norepinephrine (%)#350

Clinical Course.

The septic shock resolved in 24 of the 25 patients (96%) from group 1 (maintenance of systemic hemodynamics without vasopressor drugs). The patient not recovering from septic shock did not have adrenal insufficiency. Duration of treatment with hydrocortisone in patients with adrenal insufficiency in group 1 was 5.6 ± 2.2 days. In contrast, in only 29 of the 50 patients (58%) from group 2 the septic shock resolved (P = .001). The resolution of shock was more rapid in group 1 than in group 2. It occurred within the first 48 hours after admission into the intensive care unit in 12 patients (48%) in group 1 and in only nine (18%) in group 2 (P = .01). In the whole series of patients (group 1 and 2), there was a tendency for a higher incidence of refractory shock in patients with cirrhosis with Child-Pugh grade C (33%) compared with that observed in grade A or B patients with cirrhosis (8%) (P = .1). Development of acute respiratory distress syndrome within the intensive care unit was similar in the two groups (seven patients, 28%, in group 1 and nine, 18%, in group 2) as was the incidence of new infections during hospitalization (eight patients, 32%, in group 1 and 11 patients, 22%, in group 2), infections due to multiresistant bacteria (four patients, 16%, in group 1 and five cases, 10%, in group 2) and gastrointestinal hemorrhage (two patients, 8%, in group 1 and five patients, 10%, in group 2). Two patients in group 1 who were mechanically ventilated developed severe fungal infections (invasive pulmonary aspergillosis in one case and pneumonia by Candida spp in the other). Eight patients (32%) from group 1 and 33 (66%, P = .01) from group 2 had renal failure when discharged from the intensive care unit or at the time of death.

Twenty-three patients received intravenous albumin attributable to spontaneous bacterial peritonitis (six in group 1 and 17 in group 2) and 20 patients due to severe hypoalbuminemia (7 and 13, respectively). No significant differences were found between patients treated and not treated with albumin with respect to the incidence refractory shock (30% vs. 28%), intensive care unit survival (51% vs. 53%), and hospital survival (58% vs. 56%). No significant differences were observed when patients with septic shock related and not related to spontaneous bacterial peritonitis, with culture-positive and culture-negative infections, or with infections caused by gram-positive or gram-negative bacteria were compared (data not shown).

Survival.

Eight patients from group 1 (32%) and 31 from group 2 (62%) (P = .03) died in the intensive care unit. Causes of death are given in Table 3. No patient from group 1 died as a consequence of refractory shock. The main causes of death in this group were type 1 hepatorenal syndrome and liver failure. The two patients treated with hydrocortisone developing severe fungal infection 8 and 11 days after admission died as a consequence of the infection. In contrast, refractory shock was by far the main cause of death in group 2. Other causes of death in these patients were related to the underlying cirrhosis (type 1 hepatorenal syndrome, liver failure, and uncontrolled variceal bleeding).

Table 3. Causes of Death in the Intensive Care Unit in Groups 1 and 2
 Group 1 (n = 25)Group 2 (n = 50)P
Refractory shock (n)020.001
Type-1 hepatorenal syndrome (n)23NS
Liver failure (n)44NS
Variceal bleeding (n)04NS
Fungal infection (n)20NS

One additional patient in group 1 and three in group 2 died during hospitalization after being discharged from the intensive care unit. Causes of death were type 1 hepatorenal syndrome (one case per group), liver failure, and massive variceal bleeding . Figure 1 shows that the probability of hospital survival was significantly higher in patients from group 1 than in those from group 2. Differences in survival between groups were specially marked within the first days in the intensive care unit.

Figure 1.

Probability of hospital survival of patients from group 1 (prospective series; continuous line) and group 2 (retrospective series; dotted line).

Mortality was unrelated to adrenal function in group 1. Five of 17 patients (29%) with adrenal insufficiency and three of eight (38%) without adrenal insufficiency died in the intensive care unit.

Discussion

Activation of the hypothalamic-pituitary-adrenal axis is an important feature in host adaptation to severe sepsis and to other acute illnesses such as trauma, major surgery, or burns.13, 31–34 In the case of severe sepsis, activation of the system is initiated by the action of cytokines on the hypothalamus promoting the release of corticotropin-releasing hormone.31 This hormone stimulates corticotropin secretion by the pituitary gland, which in turn increases cortisol secretion by the adrenal glands. In addition, levels of cortisol-binding protein decrease rapidly,13, 31 leading to an increase in the circulating plasma levels of free cortisol, which is the active component of the system.35 During acute illness, the negative feedback of cortisol on the release of corticotropin-releasing hormone and corticotrophin is depressed,13, 33, 36 and this is important to keep a sustained activation of the hypothalamic-pituitary-adrenal axis. Finally, cytokines increase the affinity of cortisol to glucocorticoid receptors.31 Therefore, during sepsis an integrated multi-level response exists to optimize the effect of cortisol in the peripheral tissues.

Cortisol has several beneficial effects in severe sepsis. It is a powerful immunosuppressive hormone and modulates the inflammatory response and cytokine production.13, 37–39 By this mechanism it inhibits the production of nitric oxide and other mediators of septic shock.39 A direct inhibitory effect of cortisol on the inducible form of nitric oxide synthase has also been demonstrated.40 Not surprisingly, cortisol is important in the maintenance of vascular tone and permeability in severe sepsis. Finally, cortisol increases the vascular and cardiac response to the renin-angiotensin and the sympathetic nervous systems,13, 19 the two most important homeostatic mechanisms in the maintenance of arterial pressure during severe sepsis.

During the last few years, evidence has been presented that patients with severe sepsis frequently develop relative adrenal insufficiency.11–18 Although plasma cortisol levels in these patients are normal or increased, they are inappropriately low to control the inflammatory response. Diagnosis of relative adrenal insufficiency is not possible on clinical grounds.31 In patients with severe sepsis, it is frequently associated with shock refractory to vasopressor drugs, multiple organ failure, and death.11, 13, 15–19 However, making the diagnosis before development of refractory shock is important. Therefore, as performed in the current study, diagnosis of relative adrenal insufficiency in critically ill patients relies on the measurement of plasma cortisol levels before and after adrenal stimulation with synthetic corticotrophin.11, 13, 31 Evidence supports the use of a supplemental low dose of corticosteroids (stress dose, 50 mg hydrocortisone every 6 hours) in patients with septic shock, especially when biochemical evidence of relative adrenal insufficiency exists. Several randomized controlled trials in patients with septic shock have shown improvements in hemodynamics, reduction in the need of vasopressors, and a significant decrease in mortality.10, 12, 20–22

Data on adrenal function in patients with liver diseases are scarce. Two studies have been reported assessing adrenal function in patients with liver disease. The first was performed in 45 patients with fulminant hepatic failure.41 No data on bacterial infection were given. Sixty-two percent showed relative adrenal insufficiency, which was associated with a higher incidence of cardiovascular instability requiring vasopressor support, more severe illness, and a higher rate of hospital mortality or liver transplantation. The second study was performed in 101 patients with cirrhosis and severe sepsis requiring intensive monitoring or treatment.8 According to the criteria used in the current study, 50 of these 101 patients had septic shock. Adrenal insufficiency was diagnosed in 52% of the patients and was significantly associated with septic shock, disease severity, renal failure, and hospital mortality. None of the patients included in these investigations were treated with hydrocortisone. These studies, therefore, suggest that relative adrenal insufficiency may develop in patients with acute or chronic liver failure and that it is a frequent event in patients with cirrhosis in septic shock. Only one study exists on the use of stress doses of hydrocortisone in patients with liver disease. It consists of a retrospective comparative study in 40 patients with hypotensive acute or chronic liver failure requiring norepinephrine as vasopressor support.42 Twenty patients were treated with a continuous infusion of 300 mg/d of hydrocortisone for 4 to 5 days. Twenty patients not receiving hydrocortisone admitted during the same period and matched for proportion of acute and chronic liver failure, age, sex, and severity of illness served as controls. Adrenal function was assessed in patients receiving hydrocortisone. Baseline cortisol and the response to corticotropin were below normal in nearly 30% and 70% of patients, respectively. Hydrocortisone reduced norepinephrine requirements. However, it was associated with a significantly higher incidence of infections by resistant bacteria (methicillin-resistant Staphylococcus aureus, vancomycin-resistant enterococci, and Pseudomonas aeruginosa or gentamicin-resistant Klebsiella). No difference was seen in hospital survival.

The current study, therefore, represents the first investigation assessing adrenal function and the effect of supplemental treatment with low doses of hydrocortisone in a homogeneous population of patients with cirrhosis in septic shock. An important limitation of the study is the design. We did not perform a prospective randomized therapeutic controlled trial. However, when we decided to assess adrenal function and to treat relative adrenal insufficiency in patients with cirrhosis in septic shock, evidence was seen that relative adrenal insufficiency is an important problem in patients with septic shock that may benefit from low doses of hydrocortisone. Consequently, we preferred to perform a prospective study in the first 25 patients admitted to our intensive care unit with septic shock and to compare the clinical course of these patients with that of a series of 50 consecutive patients admitted with septic shock just before the initiation of the prospective study. The only difference between these two series of patients was the assessment of adrenal function and the administration of hydrocortisone in cases with adrenal insufficiency in the patients included in the prospective investigation.

Our study confirms that relative adrenal insufficiency is a frequent feature in patients with cirrhosis in septic shock. Sixty-eight percent of our patients in the prospective series showed low baseline plasma cortisol concentration, low response to corticotropin, or both. This prevalence of relative adrenal insufficiency is among the highest previously reported in septic shock.11–18 Our data also confirm that in the setting of a septic shock, diagnosing adrenal insufficiency based on clinical data is very difficult. Only the presence of severe hepatic failure (Child-Pugh grade C) was an indicator of adrenal dysfunction in our patients. The mechanism of adrenal insufficiency in septic shock is not well established. It may be related to a reduction in adrenal blood flow. Very high levels of inflammatory cytokines directly inhibit adrenal cortisol synthesis. Finally, pre-existing conditions of the hypothalamic-pituitary-adrenal axis could also be important.13, 31 In patients with decompensated cirrhosis and ascites, the cytokine response to endotoxin is very much increased,5 and the blood perfusion to extra-splanchnic organs is generally reduced.43 Furthermore, as indicated before, they may present relative adrenal insufficiency before the infection. Therefore, finding a very high incidence of relative adrenal insufficiency in patients with cirrhosis with septic shock is not surprising.

The most important finding of the study is that treatment of septic shock in cirrhosis with a low dose of hydrocortisone is associated to a marked increase in the rate of shock reversal and hospital survival. It changed the natural course of septic shock in these patients. Whereas the main cause of death in the retrospective series was refractory shock, in the prospective series it was type 1 hepatorenal syndrome and liver failure. Infection by resistant bacteria was not a major problem in our patients receiving hydrocortisone. However, two patients died as a consequence of a severe fungal infection. Treatment with hydrocortisone, therefore, is not free of complication in cirrhosis and should be used when indicated. Our results suggest that all patients with relative adrenal insufficiency should be treated with hydrocortisone. In patients with severe hepatic failure (Child-Pugh grade C), in whom the incidence of adrenal insufficiency is extremely high, treatment could be initiated immediately after diagnostic testing and can be stopped if results do not indicate the presence of adrenal insufficiency.

In summary, our results show that adrenal insufficiency is extremely frequent in patients with advanced cirrhosis and septic shock and that treatment with low doses of hydrocortisone is associated with a marked increase in shock reversal and hospital survival. The design of the study was not optimal, and the number of patients investigated prospectively was relatively small. Further investigations on adrenal function in patients with cirrhosis and septic shock are therefore needed to confirm our findings.

Ancillary