fax: (886) 3-3288662
Liver Failure and Liver Disease
Adrenal insufficiency in patients with cirrhosis, severe sepsis and septic shock†
Article first published online: 23 MAR 2006
Copyright © 2006 American Association for the Study of Liver Diseases
Volume 43, Issue 4, pages 673–681, April 2006
How to Cite
Tsai, M.-H., Peng, Y.-S., Chen, Y.-C., Liu, N.-J., Ho, Y.-P., Fang, J.-T., Lien, J.-M., Yang, C., Chen, P.-C. and Wu, C.-S. (2006), Adrenal insufficiency in patients with cirrhosis, severe sepsis and septic shock. Hepatology, 43: 673–681. doi: 10.1002/hep.21101
Potential conflict of interest: Nothing to report.
- Issue published online: 23 MAR 2006
- Article first published online: 23 MAR 2006
- Manuscript Accepted: 6 JAN 2006
- Manuscript Received: 29 AUG 2005
- Chang Gung Medical Research Fund. Grant Number: CMRPG63003
- Chang Gung Memorial Hospital, Chia-Yi, Taiwan
Patients with cirrhosis are susceptible to bacterial infection, which can result in circulatory dysfunction, renal failure, hepatic encephalopathy, and a decreased survival rate. Severe sepsis is frequently associated with adrenal insufficiency, which may lead to hemodynamic instabity and a poor prognosis. We evaluated adrenal function using short corticotropin stimulation test (SST) in 101 critically ill patients with cirrhosis and severe sepsis. Adrenal insufficiency occurred in 51.48% of patients. The patients with adrenal insufficiency had a higher hospital mortality rate when compared with those with normal adrenal function (80.76% vs. 36.7%, P < .001). The cumulative rates of survival at 90 days were 15.3% and 63.2% for the adrenal insufficiency and normal adrenal function groups, respectively (P < .0001). The hospital survivors had a higher cortisol response to corticotropin (16.2 ± 8.0 vs. 8.5 ± 5.9 μg/dL, P < .001). The cortisol response to corticotropin was inversely correlated with various disease severity, Model for End-Stage Liver Disease, and Child–Pugh scores. Acute physiology, age, chronic health evaluation III score, and cortisol increment were independent factors to predict hospital mortality. Mean arterial pressure on the day of SST was lower in patients with adrenal insufficiency (60 ± 14 vs. 74.5 ± 13 mm Hg, P < .001), and a higher proportion of these patients required vasopressors (73% vs. 24.48%, P < .001). Mean arterial pressure, serum bilirubin, vasopressor dependency, and bacteremia were independent factors that predicted adrenal insufficiency. In conclusion, adrenal insufficiency is common in critically ill patients with cirrhosis and severe sepsis. It is related to functional liver reserve and disease severity and is associated with hemodynamic instability, renal dysfunction, and increased mortality. (HEPATOLOGY 2006;43:673–681.)
Critical illness is accompanied by the activation of the hypothalamic-pituitary-adrenal (HPA) axis, which is highlighted by increased serum corticotropin and cortisol levels.1–3 The activation of the HPA axis is a crucial component of the host's adaptation to severe stress. Cortisol is essential for the normal function of the immune system, maintenance of vascular tone, and various cellular functions. In patients with severe sepsis, the integrity of the HPA axis can be impaired by a variety of mechanisms.1, 4 Recently, the concept of relative adrenal insufficiency has been used to describe a subnormal adrenal response to adrenocorticotropin in severe illness, in which the cortisol levels, even though high in terms of absolute value, are inadequate to control the inflammatory situation.1 The short corticotropin stimulation test (SST) is most commonly used to evaluate the appropriateness of the adrenal response in this setting. In patients with septic shock, a decreased response to the SST, namely, an absolute increment of the serum cortisol level less than 9 μg/dL, is associated with an impaired vascular reactivity to vasopressors5 and a high mortality.5–9 Patients with cirrhosis share many similar hemodynamic features with patients with septic shock and adrenal insufficiency, namely, increased cardiac output, decreased peripheral vascular resistance, decreased mean arterial pressure, and hypo-responsiveness to vasopressors.1, 10–13 Consistent with observations in septic patients, hemodynamic impairment is closely related to mortality and morbidity in patients with cirrhosis.10, 11, 14 Studies have shown that cirrhosis in animals and humans is characterized by increased levels of endotoxin and various inflammatory cytokines, which can contribute to hemodynamic impairment15–17 and potentially to adrenal insufficiency as well.1 Furthermore, the liver is the primary site of metabolism of adrenal steroid hormone and synthesis of cholesterol, which is the major precursor of steroid.18 Therefore, preexisting liver dysfunction may further disturb the activation of the HPA axis during severe sepsis and septic shock. Moreover, adrenal insufficiency in severe sepsis and septic shock may aggravate hemodynamic impairment in critically ill patients with cirrhosis, leading to a poor prognosis. In fact, Harry et al.19 recently showed adrenal insufficiency is common and may contribute to hemodynamic instability and mortality in patients with acute liver failure. Nevertheless, the impact adrenal insufficiency has on the outcomes of patients with cirrhosis and severe sepsis and septic shock is unknown. Considering the high prevalence of adrenal insufficiency during severe sepsis and septic shock and the major role of the liver in the metabolism of glucocorticoid hormone, we felt compelled to conduct this investigation. Our efforts involved the examination of the adrenal response to corticotropin and the relationship between adrenal function and outcome in patients with cirrhosis and severe sepsis and septic shock.
Patients and Methods
Patient Information, Data Collection, and Definitions.
This study was conducted with the approval of the institutional review board. Formal consent was obtained from the next of kin. The study was performed in the intensive care unit (ICU) of two university-affiliated hospitals between June 2004 and May 2005. The study enrolled 101 consecutive patients with cirrhosis and severe sepsis requiring intensive monitoring or treatment. Severe sepsis was defined by the criteria of the American College of Chest Physician/Society of Critical Care Medicine,20 namely, sepsis associated with organ dysfunction, hypoperfusion abnormality, or sepsis-induced hypotension. Hypoperfusion and perfusion abnormalities may include, but are not limited to, lactic acidosis, oliguria, or acute alteration of mental status. Liver cirrhosis was defined histologically or based on clinical, image, and laboratory findings. The diagnosis was made histologically in 16 patients in whom liver biopsy was performed to diagnose hepatocellular carcinoma or confirm cirrhosis. It was clinically diagnosed in 85 patients. The clinical diagnosis of hepatic cirrhosis was made based on the presence of esophageal varices or ascites with no other cause, along with an unequivocal image on sonography or computed tomography. Patients with a history of corticosteroid treatment and those who had received the steroidogenesis-inhibiting agent etomidate were excluded from this study. All patients were treated with a standard treatment protocol for management of severe sepsis and septic shock.21 Albumin (25 g/day) was given to patients with serum albumin below 2.5 g/dL. Albumin level was checked every 3 days. Albumin was re-infused when necessary. After discharge from hospital, most patients were followed at outpatient departments of our hospitals. Survival was analyzed at 3 months after admittance to the ICU. Telephone interviews were made to obtain the information as to survival or date of death of those patients who were not followed at our hospitals (n = 4).
The severity of liver disease on the day of the SST was graded by the Child–Pugh system.22 Meanwhile, illness severity also was assessed by the following scoring systems: organ system failure (OSF), sequential organ failure assessment (SOFA), Model for End-stage Liver Disease (MELD), and Acute Physiology, Age, Chronic Health Evaluation III (APACHE III). These scoring systems have been validated to predict the outcomes of critically ill patients with cirrhosis.22–27 For these scoring systems and physiological evaluations, the most abnormal value for each organ system on the day of the SST was recorded.
Renal dysfunction was defined as a serum creatinine level greater than 1.5 mg/dL on the day of the SST. The cut-off value of 1.5 mg/dL was based on previous investigations showing glomerular filtration rate is distinctly decreased in patients with cirrhosis and a serum creatinine higher than this level.28 Furthermore, this is also the cut-off value used to define hepatorenal syndrome.29
Vasopressor dependency was defined by a need for vasoactive substance(s) to maintain a systolic blood pressure greater than 90 mmHg despite volume expansion. Hemodynamic instability was defined by a mean arterial pressure less than 60 mmHg19, 30 or vasopressor dependency on the day of the SST.
Pneumonia, urinary tract infection, spontaneous bacterial peritonitis, central catheter-related infection, biliary tract infection, cellulitis, septic arthritis, and abscess were defined as previously described.31–35 Bacteremia was defined as the presence of viable bacteria in the blood,20 as evidenced by a positive blood culture. Spontaneous bacteremia was defined as bacteremia without identified infection focus.34 Culture-negative sepsis was defined as presence of systemic inflammatory response syndrome20 and negative cultures after exclusion of the possibility of non-infection inflammatory conditions as the causes of systemic inflammatory response syndrome.
Blood cultures and appropriate culture specimens from the infection focus were obtained. Hematological and biochemical data also were collected systemically on the day of the SST.
An SST was performed within the first 24 hours of admission to the ICU. Synthetic adrenocorticotropic hormone 250 μg (Synacthen, Novartis Pharma AG, Basel, Switzerland) was given intravenously. Blood samples were obtained immediately before and 30 and 60 minutes after injection. Cortisol levels were measured by a competitive immunoassay using direct chemiluminescent technology (Bayer Corporation, East Walpole, MA). Coefficients of variation for this test were obtained at 3 cortisol concentrations: 6.58% at a mean cortisol concentration of 3.88 μg/dL, 4.22% at 14.17 μg/dL, and 4.98 % at 37.15 μg/dL. The minimal detectable concentration was 0.2 μg/dL. The peak cortisol level was defined as the highest cortisol level obtained after synacthen administration, whether at 30 or 60 minutes. The cortisol response was defined as the difference between the baseline and peak cortisol levels. As previously described,1 the criteria for adrenal insufficiency were defined as follows: baseline value less than 15 μg/dL, or cortisol response less than 9 μg/dL with a baseline value between 15 and 34 μg/dL. Adrenal insufficiency was considered unlikely when the baseline cortisol level is greater than 34 μg/dL.
Descriptive statistics are expressed as mean ± SD. All variables were tested for normal distribution using the Kolmogorov-Smirnov test. The Student t test was used to compare the means of continuous variables and the normal distribution data. Otherwise, the Mann-Whitney U test was used. Categorical data were tested using the chi-squared (χ2 test. Survival was analyzed by the Kaplan-Meier method and compared between groups with the log-rank test. Meanwhile, risk factors were assessed using univariate analysis, and variables that were statistically significant in the univariate analysis were included in the multivariate analysis by applying a multiple logistic stepwise regression procedure to obtain independent risk factors.36 The correlation between the results of the SST and the disease severity scores was analyzed with linear regression using the Pearson method. Moreover, the χ2 test of trend was used to analyze the adrenal insufficiency rates associated with the number of organ system failures. Discrimination was tested using the area under a receiver operating characteristic (ROC) curve37 to assess the ability of the model to distinguish patients who had adrenal dysfunction from those who did not. ROC analysis was also performed to calculate the cutoff values, sensitivity, specificity, overall correctness, and positive and negative predictive values. The best Youden index (sensitivity + specificity −1)38 was also used to determine the best cutoff point. All statistical tests were two-tailed, and the significance level was set at P = .05 or less. Data were analyzed using SPSS 10.0 for Windows (SPSS Inc., Chicago, IL).
One hundred one critically ill patients with cirrhosis were enrolled in this investigation. The mean patient age was 57 years. There were 77 men (76.23%) and 24 women (23.76%). The cause of liver cirrhosis was hepatitis B virus (HBV) in 30 patients, alcohol in 30, hepatitis C virus (HCV) in 11, HBV plus alcohol in 7, HBV plus HCV in 12, primary biliary cirrhosis in 2, HCV plus alcohol in 1, and other causes in 8. Overall in-hospital mortality for the entire group was 59.4%. The number of patients grouped according to the site of infection was as follows: pneumonia, 40 (39.6%); urinary tract infection, 25 (24.7%); spontaneous bacterial peritonitis, 17 (16.8%); spontaneous bacteremia, 16 (15.8%); central catheter-related infection, 4 (3.9%); cellulitis, 4 (3.9%); biliary tract infection, 4 (3.9%); liver abscess, 1 (0.98%); septic arthritis, 1 (0.98%); intra-abdominal abscess, 1 (0.98%); culture-negative sepsis without identified infection focus, 5 (4.9%). Seventeen patients were considered to have more than one site of infection focus.
Short Corticotropin Stimulation Test.
According to the criteria stated previously, 52 (51.4 %) patients had adrenal insufficiency. Forty-six patients had a cortisol response less than 9 μg/dL. Two of these 46 patients were ruled out to have adrenal insufficiency because of baseline cortisol levels greater than 34 μg/dL. Thirteen patients had a baseline level less than 15 μg/dL. Thirty-nine patients had a cortisol response less than 9 μg/dL and a baseline cortisol level between 15 and 34 μg/dL. Five patients had both a baseline cortisol level less than 15 μg/dL and a cortisol response less than 9 μg/dL. The clinical characteristics and outcomes in patient subgroups stratified by adrenal functions are listed in Table 1. In the multivariate analysis, mean arterial pressure, serum bilirubin level, vasopressor dependency, and bacteremia were independent factors predicting adrenal dysfunction (Table 2). The incidence of adrenal insufficiency increased progressively and significantly with the number of organ system failures (χ2 for trend, P < .001, Fig. 1). Meanwhile, the incremental response to corticotropin was negatively correlated with the SOFA (R = −0.566; P < .001), OSF (R = −0.497; P < .001), MELD (R = −0.436, P < .001), APACHE III (R = −0.457; P < .001), and Child–Pugh (R = −0.33; P = .001) scores. However, there was no correlation among baseline, peak cortisol levels, and disease severity scores. Furthermore, there was no correlation between the basal cortisol level and the response to corticotropin stimulation (cortisol increment).
|All Patients (n = 101)||Adrenal Insufficiency (n = 52)||Normal Adrenal Function (n = 49)||P-value|
|Age (years)||57.0 ± 13.1||55.9 ± 13.3||58.3 ± 12.8||NS (.351)|
|Gender (M/F)||77/24||43/9||34/15||NS (.184)|
|ICU mortality (%)||50 (49.5)||37 (71.4)||13 (26.5)||<.001|
|Hospital mortality (%)||60 (59.4)||42 (80.76)||18 (36.7)||<.001|
|Hepatic encephalopathy (%)||68 (67.3)||36 (69.2)||32 (65.3)||NS (.485)|
|BUN||51.3 ± 44.7||58.1 ± 47.5||43.6 ± 40.3||NS (.103)|
|Serum creatinine (mg/dL)||2.6 ± 2.3||3.2 ± 2.7||1.9 ± 1.6||.004|
|Renal dysfunction (%)||58 (57.4)||41 (78.84)||17 (34.69)||<.001|
|Ascites (%)||83 (82.2)||46 (88.46)||37 (75.5)||NS (.059)|
|Na||139 ± 8||141 ± 9||138 ± 8||.036|
|K||4.1 ± 1.1||4.0 ± 1.2||4.3 ± 1.0||NS (.175)|
|Bilirubin (mg/dL)||10.9 ± 11.6||15.3 ± 13.3||5.8 ± 6.4||<.001|
|Albumin (g/dL)||2.7 ± 2.1||2.9 ± 2.9||2.6 ± 0.6||NS (.539)|
|Prothrombin time prolongation (sec)||13.1 ± 15.1||18.0 ± 18.4||8.2 ± 8.0||.002|
|Platelets (×109/L)||83 ± 59||73 ± 56||93 ± 61||NS (.087)|
|Leukocytes (×109/L)||12.6 ± 9.6||14.3 ± 12.0||11.0 ± 6.0||NS (.070)|
|Hemoglobin (g/dL)||9.7 ± 2.0||9.8 ± 1.9||9.6 ± 2.1||NS (.702)|
|Heart rate (beats/min)||112.52±25.55||113.82 ± 25.54||110.97 ± 25.83||NS (.621)|
|Right atrial pressure (mm Hg)||8.83±3.35||9.28 ± 3.30||7.93 ± 3.36||NS (.167)|
|MAP (mm Hg)||67 ± 16||60 ± 14||75 ± 13||<.001|
|SOFA||10.1 ± 4.7||12.6 ± 4.0||7.2 ± 3.8||<.001|
|APACHE III score||90.9 ± 41.1||109.0 ± 37.5||70.1 ± 34.9||<.001|
|OSF number||3.3 ± 1.8||4.4 ± 1.5||2.3 ± 1.5||<.001|
|MELD score||13.0 ± 6.0||15.2 ± 5.2||10.4 ± 6.0||<.001|
|Child–Pugh score||11.9 ± 2.6||12.7 ± 2.2||11.0 ± 2.7||.001|
|Child–Pugh class C (%)||76 (75.2)||45 (86.53)||31 (63.2)||.013|
|Bacteremia(%)||33 (32.7)||25 (48)||8 (16.32)||.002|
|Vasopressor dependency(%)||50 (49.5)||38 (73)||12 (24.48)||<.001|
|Mechanical ventilation(%)||65 (64.4)||32 (61.5)||33 (67.3)||NS (.754)|
|Parameter||Beta Coefficient||Standard Error||Exp (B) (95%CI)||P|
|Univariate Logistic Regression|
|Serum creatinine (mg/dL)||0.281||0.106||1.324 (1.075-1.631)||.008|
|Renal dysfunction||1.717||0.440||5.566 (2.350-13.180)||<.001|
|Bilirubin (mg/dL)||0.093||0.026||1.098 (1.044-1.154)||<.001|
|Prothrombin time prolongation (sec)||0.080||0.028||1.083 (1.024-1.145)||.005|
|MAP (mm Hg)||-0.083||0.018||0.921 (0.888-0.954)||<.001|
|Child–Pugh points||0.279||0.089||1.321 (1.109-1.321)||.002|
|Child–Pugh class C||1.168||0.452||3.215 (1.327-7.790)||.010|
|APACE III score||0.029||0.007||1.029 (1.016-1.043)||<.001|
|SOFA score||0.374||0.078||1.454 (1.248-1.693)||<.001|
|OSF number||0.818||0.167||2.267 (1.634-3.143)||<.001|
|MELD score||0.150||0.040||1.162 (1.073-1.258)||<.001|
|Vasopressor dependency||2.358||0.473||10.571 (4.186-26.700)||<.001|
|Multivariate Logistic Regression|
|Vasopressor dependency||2.239||0.737||9.383 (2.212-39.803)||.002|
SOFA had the best discriminating power to predict adrenal insufficiency. The area under ROC curve for SOFA, OSF, APACHE III, MELD, and Child–Pugh scores are 0.851 ± 0.039 (mean ± SEM) (95% CI: 0.775-0.927), 0.824 ± 0.042 (mean ± SEM) (95% CI: 0.742-0.906), 0.778 ± 0.046 (mean ± SEM) (95% CI: 0.688-0.868), 0.728 ± 0.051 (mean ± SEM) (95% CI: 0.628-0.828), and 0.681 ± 0.053 (mean ± SEM) (95% CI: 0.578-0.785) respectively.Table 3 shows the predictive values of the chosen cutoff points, which give the best Youden index, for prediction of adrenal insufficiency. The incidences of adrenal insufficiency below and above the chosen cutoff scores are shown in Fig. 2.
The ICU and hospital mortality rates for the patients who had adrenal insufficiency were significantly higher than for those with normal adrenal function (71.4% vs. 26.5%, and 80.7 % vs. 36.7 %, respectively, P < .001) (Table 1). Follow-up to 90 days or the time of death was complete for the entire groups. The cumulative rates of survival at 90 days were 15.3% and 63.2% for the adrenal insufficiency group and normal adrenal function group, respectively (P < .0001) (Fig. 3). The patients with adrenal insufficiency had a significantly lower mean arterial blood pressure (Table 1). Hemodynamically unstable patients (n = 54) had a significantly lower cortisol response to corticotropin administration (7.99 ± 6.28 vs. 15.83 ± 7.24 μg/dL, P < .001). Moreover, the incidence of adrenal insufficiency was higher in those patients who were hemodynamically unstable (79.61% vs. 19.14 %, P < .001). On the day of the SST, 58 (57.43%) patients had renal dysfunction. The incremental increase of cortisol levels was significantly higher in the patients with normal renal function (15.37 ± 7.81 vs. 8.87 ± 6.54 μg/dL, P < .001), whereas no difference was seen between the baseline and peak cortisol levels. The incidences of adrenal insufficiency and hemodynamic instability were higher in the patients with impaired renal function when compared with those who with normal renal function (68.96% vs. 27.9%, P < .001; 70.7% vs. 30.2%, P < .001, respectively). Patients with renal dysfunction had higher Child–Pugh scores (12.6 ± 2.1 vs. 10.9 ± 2.7, P = .002). Sixty-eight (67.3%) patients presented with hepatic encephalopathy on the day of the SST. Baseline and peak cortisol levels were higher in the hepatic encephalopathy group (18.6 ± 9.42 vs. 31.1 ± 21.77 μg/dL, P < .001; 32.28 ± 13.2 vs. 41.7 ± 23.99 μg/dL, P = .038, respectively), whereas no difference was seen in cortisol increment. Sixty-five (64.36%) patients were mechanically ventilated on the day of the SST. There was no difference in SST between those patients on ventilators and those who were not.
Microbiological information was available for all patients. Seventy-one patients had at least one positive microbiological culture. Positive cultures were obtained from the blood in 33 (32.7%) patients, from urine in 22 (21.78%), from sputum in 35 (34.65%), from ascites in 9 (8.91%) patients, from a CVP catheter tip in 4 (3.96%). A positive blood culture was independently predictive for adrenal dysfunction. Patients with bacteremia had a higher incidence of adrenal insufficiency when compared to non-bacteremic patients (72.7% vs. 41.1 %, P = .002).
Table 4 lists the patient demographic data, clinical characteristics, and results of the SST for both survivors and non-survivors. As shown in Table 4, the response to corticotropin was significantly higher in those who survived, whereas the baseline cortisol level was higher in those who died. The peak cortisol level was not different between survivors and non-survivors. In multivariate analysis, APACHE III score and cortisol response to corticotropin challenge were independent factors to predict hospital mortality (Table 5).
|All Patients (n = 101)||Hospital Survivors (n = 41)||Hospital Non-survivors (n = 60)||P-value|
|Age (years)||57.0 ± 13.1||56.7 ± 13.6||57.2 ± 12.8||NS (.862)|
|Sex (male/female)||77/24||31/10||46/14||NS (.902)|
|Hepatic encephalopathy (%)||68 (67.3)||18 (43.9)||50 (83.3)||<.001|
|BUN||51.3 ± 44.7||330 ± 27.1||63.8 ± 49.9||<.001|
|Serum creatinine,(mg/dL)||2.6 ± 2.3||1.7 ± 1.6||3.3 ± 2.5||<.001|
|Renal dysfunction (%)||58 (57.4%)||13 (31.7%)||45 (75%)||<.001|
|Ascites (%)||83 (82.2%)||26 (63.4%)||57 (95%)||<.001|
|Na||139 ± 8||140 ± 5||139 ± 10||NS (.828)|
|K||4.1 ± 1.1||3.9 ± 0.9||4.2 ± 1.2||NS (.151)|
|Bilirubin (mg/dL)||10.9 ± 11.6||5.0 ± 6.2||14.9 ± 12.8||<.001|
|Albumin (g/dL)||2.7 ± 2.1||2.7 ± 0.6||2.8 ± 2.8||NS (.791)|
|PT prolongation (sec)||13.1 ± 15.1||6.9 ± 5.3||17.3 ± 18.0||<.001|
|Platelets (×109/L||83 ± 59||87 ± 65||80 ± 54||NS (.578)|
|Leukocytes (×109/L)||12.6 ± 9.6||10.3 ± 6.0||14.3 ± 11.2||.039|
|Hemoglobin (g/dL)||9.7 ± 2.0||9.78 ± 2.0||9.66 ± 2.0||NS (.764)|
|Heart Rate (beats/minute)||112.52 ± 25.55||106.44 ± 28.11||116.49 ± 23.19||NS (.084)|
|Right atrial pressure (mm Hg)||8.83 ± 3.35||8.12 ± 4.03||9.10 ± 3.06||NS (.336)|
|Mean arterial pressure (mm Hg)||67 ± 16||78 ± 11||59 ± 14||<.001|
|SOFA||10.1 ± 4.7||6.6 ± 3.1||12.5 ± 4.2||<.001|
|APACHE III score||90.9 ± 41.1||54.6 ± 23.4||115.7 ± 30.7||<.001|
|OSF number||3.3 ± 1.8||2.0 ± 1.0||4.4 ± 1.6||<.001|
|MELD score||13.0 ± 6.0||8.8 ± 5.1||15.8 ± 4.8||<.001|
|Child–Pugh score||11.9 ± 2.6||10.2 ± 2.6||13.1 ± 1.7||<.001|
|Child–Pugh class C (%)||76 (75.2)||21 (51.2)||55 (91.7)||<.001|
|Baseline cortisol (μg/dL)||27.0 ± 19.5||22.0 ± 18.5||30.4 ± 19.6||.032|
|Peak cortisol (μg/dL)||38.6 ± 21.5||38.1 ± 22.6||39.0 ± 20.9||NS (.855)|
|Cortisol increment (μg/dL)||11.6 ± 7.8||16.2 ± 8.0||8.5 ± 5.9||<.001|
|Bacteremia (%)||33 (32.7)||12 (29.3)||21 (35.0)||NS (.546)|
|Vasopressor dependency (%)||50 (49.5)||10 (24.4)||40 (66.7)||<.001|
|Mechanical ventilation (%)||65 (64.4)||23 (56.1)||42 (70.0)||NS (.152)|
|Parameter||Beta Coefficient||Standard Error||Exp (B) (95%CI)||P|
|Univariate Logistic Regression|
|Baseline cortisol||−0.030||0.015||0.971 (0.943-0.999)||.043|
|Cortisol increment||0.166||0.038||1.181 (1.097-1.271)||<.001|
|Serum creatinine (mg/dL)||−0.408||0.132||0.665 (0.513-0.862)||.002|
|Renal dysfunction||−1.866||0.449||0.155 (0.064-0.373)||<.001|
|Bilirubin (mg/dL)||−0.112||0.031||0.894 (0.841-0.950)||<.001|
|Leukocytes (×109/L)||0.000||0.000||1.000 (1.000-1.000)||.039|
|Prothrombin time prolongation (sec)||−0.133||0.040||0.875 (0.809-0.947)||.001|
|MAP (mm Hg)||0.112||0.023||1.118 (1.069-1.170)||<.001|
|Child–Pugh score||−0.588||0.121||0.555 (0.438-0.704)||<.001|
|Child–Pugh class C||−2.238||0.551||0.107 (0.036-0.314)||<.001|
|APACE III score||−0.099||0.021||0.906 (0.869-0.945)||<.001|
|SOFA score||−0.470||0.096||0.625 (0.518-0.755)||<.001|
|OSF number||−1.144||0.223||0.319 (0.206-0.493)||<.001|
|MELD score||−0.276||0.056||0.759 (0.680-0.848)||<.001|
|Hepatic encephalopathy||−1.855||0.468||0.157 (0.063-0.392)||<.001|
|Vasopressor dependency||−1.825||0.455||0.161 (0.066-0.394)||<.001|
|Multivariate Logistic Regression|
|APACHE III score||−0.117||0.034||0.890 (0.833-0.950)||<0.001|
|Cortisol increment||0.127||0.059||1.135 (1.012-1.273)||0.030|
This study showed that impaired adrenal function, as evidenced by the SST, is common in patients with cirrhosis and severe sepsis. Adrenal insufficiency is associated with increased mortality. Mean arterial pressure, serum bilirubin, vasopressor dependency and bacteremia were independent factors predicting adrenal insufficiency in critically ill patients with cirrhosis and severe sepsis.
Consistent with the results from patients with acute liver failure,19 the cortisol response to corticotropin is inversely correlated with various disease severity scores. Of these disease severity scores, the APACHE III score acted as an independent factor to predict hospital mortality. In fact, as is consistent with previous observation in patients with septic shock,6 the cortisol response was also an independent prognostic factor in patients with cirrhosis and severe sepsis. Furthermore, we found a negative although weak correlation between the cortisol increment and Child–Pugh scores, suggesting that adrenal dysfunction is related to liver function reserve. This is further supported by the fact that the bilirubin level is an independent predicting factor for adrenal dysfunction. In our patients, adrenal dysfunction was related to the degree of liver failure and multiple organ system dysfunction and may have contributed to mortality.
Patients with cirrhosis are susceptible to bacterial infection,39, 40 which can lead to circulatory dysfunction, renal failure, hepatic encephalopathy, and decreased survival. Actually, cirrhosis is associated with an increased risk of sepsis and sepsis-related death.41 Patients with cirrhosis are characterized by hyperdynmic circulation, which is closely related to the complications of liver cirrhosis. The hemodynamic impairment can be made worse during sepsis, leading to multiple organ failure and mortality. In this regard, our observations show that adrenal dysfunction may contribute to cardiovascular derangement and mortality in patients with cirrhosis and severe sepsis. This phenomenon may be of clinical relevance because it represents a risk factor that can be readily identified and potentially modified by steroid supplement. Like the vascular hyporeactivity to vasopressor in cirrhosis and portal hypertension, patients with occult adrenal dysfunction have an impaired responsiveness to norepinephrine.5 Indeed, hydrocortisone replacement may reverse the blunt response to vasopressor and improve the outcomes of septic patients with adrenal dysfunction.42, 43 These hemodynamic effects of hydrocortisone in this setting may result from inhibition of cytokines and nitric oxide,44 which also mediates the hemodynamic impairment in liver cirrhosis and portal hypertension.10 Considering the recent interest in the anti-inflammatory treatment in various liver diseases, a subset of patients with cirrhosis in some clinical settings may benefit from glucocorticoids because of their anti-inflammatory effects and benefits in patients with cirrhosis and adrenal dysfunction. In fact, steroid administration can reduce vasopressor dependency in acute liver failure with shock.45 However, the risks and benefits of glucocorticoid treatment in critically ill patients with cirrhosis and adrenal dysfunction have never been evaluated. A prospective randomized study is required to elucidate this issue.
An association also was found between adrenal insufficiency and renal dysfunction in our patients. Whether this phenomenon is related to more pronounced hemodynamic impairment or more decompensated liver disease in this patient group is not clear. Renal dysfunction has been reported to be closely related to elevated cytokine levels and circulatory disturbance in patients with cirrhosis and bacterial infection.32, 46, 47 Whether adrenal dysfunction directly contributes to uncontrolled inflammatory processes leading to arterial vasodilatation and renal dysfunction is unknown.
Several mechanisms may be responsible for the impaired adrenal function in our patients. An increased level of cytokines has been shown to be a major determinant of the magnitude of cortisol response to corticotropin.48 Additionally, coagulopathy, which is common in patients with cirrhosis, may lead to adrenal hemorrhage and adrenal insufficiency.1 Finally, cholesterol is the most important precursor of steroid hormones, and impaired synthesis of cholesterol in patients with cirrhosis and sepsis may have an impact on the cortisol response.
In our study, bacteremia was an independent factor predicting impaired adrenal function. The presence of viable bacteria in the blood may reflect a higher bacterial load in a more immunocompromised host. The baseline and peak cortisol levels as well as the cortisol increment were all significantly lower in the bacteremic group, implying altered adrenal synthesis and responsiveness in this specific subset of patients. Considering the increased bacteremic events associated with the various invasive procedures and the serious complications they may cause, further investigations into the pathophysiology of impaired adrenal function may improve the treatment strategy in this clinical setting.
In conclusion, adrenal dysfunction is common in patients with cirrhosis and severe sepsis. Adrenal dysfunction is associated with hemodynamic instability, renal dysfunction, and increased mortality. It occurs more frequently in patients with more severe liver disease and correlates with disease severity scores. Further clarification is needed in terms of whether glucocorticoid supplements in this subset of patients can improve hemodynamic impairment, multiple organ dysfunction and outcomes.
- 10The biology of portal hypertension. In: AriasIM, BoyerJL, ChisariFV, FaustoN, SchachterD, SchafritzD, eds. The Liver: Biology and Pathobiology. 4th ed. New York: Lippincott Williams & Wilkins, 2001: 679–697., , .
- 18Metabolism of adrenal cortical steroid. In: ChristyNP. The Human Adrenal Cortex. New York: Harper & Row, 1971: 81–189..
- 20Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. The ACCP/SCCM Consensus Conference Committee. American College of Chest Physicians/Society of Critical Care Medicine. Chest 1992; 101: 1644–1655., , , , , , et al.
- 31Manual of Clinical Problems in Infectious Diseases. 4th ed. Philadelphia: Lippincott Williams & Wilkins, 1999: 21–237., , , , .