Potential conflict of interest: Nothing to report.
Acute liver failure (ALF) shares striking similarities with septic shock where a decrease in HLA-DR expression on monocytes is associated with disease severity and predicts outcome. We investigated monocyte HLA-DR expression in ALF in relation to inflammatory mediator levels and clinical outcome. Monocyte HLA-DR expression was determined in 50 patients with acetaminophen-induced ALF (AALF) and 20 non–acetaminophen-induced ALF (NAALF). AALF patients were divided into dead/transplanted (AALF-NS, n = 26) and spontaneous survivors (AALF-S, n = 24). Fifty patients with chronic liver disease (CLD) and 50 healthy volunteers served as controls. Monocyte HLA-DR expression was determined by double-color flow-cytometry with monoclonal antibodies detecting HLA-DR and monocyte specific CD14. Serum levels of interleukin (IL) -4, -6, -10, tumor necrosis factor (TNF)-α and interferon (IFN)-γ were concomitantly measured by ELISA. Compared to healthy volunteers (75%) and CLD (67%) monocyte HLA-DR percentage expression was lower in AALF (15%, P < .001) and NAALF (22 %, P < .001). Compared to AALF-S, AALF-NS had lower monocyte HLA-DR % (11% vs. 36%, P < .001) and higher levels of IL-4, IL-6, IL-10 and TNF-α (P < .001). HLA-DR percentage negatively correlated with INR, blood lactate, pH and levels of encephalopathy (r = −0.8 to −0.5, P < .01), IL-10 (r = −0.8, P < .0001), TNF-α (r = −0.4, P = .02). HLA-DR percentage level ≤15% has a 96% sensitivity and 100% specificity and 98% accuracy in predicting poor prognosis. In conclusion, the strong relationship of monocyte HLA-DR expression with indices of disease severity, mediators of inflammation and outcome indicates a key role for this molecule as a biomarker of disease severity and prognosis. (HEPATOLOGY 2006;44:34–43.)
Systemic inflammatory response syndrome (SIRS) is an overwhelming inflammatory process manifested by two or more of the following parameters: temperature >38°C or <36°C; heart rate >90 beats per minute; tachypnoea >20 breaths per minute or PaCO2 <4.3 kPa; white cell count >12 × 109/L or <4 × 109/L; or the presence of >10% immature neutrophils. SIRS occurs following major inflammatory insults such as infection, surgery, and trauma and results from the release of potent inflammatory mediators into the circulation including the proinflammatory cytokines interleukin (IL)-1, IL-6, and tumor necrosis factor α (TNF-α).1–3 Recent data suggest that circulating levels of the anti-inflammatory cytokine IL-10 also correlate with the severity of the systemic inflammatory insult.4, 5 The evidence to date indicates that there is a complex interaction between pro- and anti-inflammatory responses during the evolution of this inflammatory process.
The occurrence of SIRS in acute liver failure (ALF) is well recognized.6–8 Moreover, the existence of SIRS, whether or not it is triggered by infection, is implicated in the progression of encephalopathy and poor outcome in this patient cohort.6, 8 Increased serum levels of proinflammatory cytokines (IL-1, IL-6, IL-8, and TNF-α) are thought to reflect the severity of the inflammatory response evoked in ALF,7, 9–15 with some studies reporting higher IL-6 and TNF-α levels in nonsurviving ALF patients.7, 10, 14, 15
Monocytes are central to SIRS secreting large quantities of proinflammatory cytokines and being responsible for antigen presentation through the surface expression of the HLA class II molecule. Reduced monocyte expression of the HLA class II DR molecule characterizes patients with septic shock with a poor prognosis, while recovery to normal is seen in those who survive.16–25 The profound reduction in HLA-DR expression and a reduced ability in vitro to produce TNF-α following lipopolysaacharide stimulation is reflected in the fact that patients with septic shock usually succumb to repeated bacterial and opportunistic infections.21 Recently, reduced monocyte HLA-DR expression has been reported in patients with septic decompensation of acute-on-chronic liver failure.26
Acetaminophen-induced acute liver failure (AALF) is characterized by progressive vasodilatory shock and multiple organ failure and shares striking similarities with septic shock.27–29 Both conditions result in uncontrolled activation of the inflammatory cascade, despite being triggered by different events.
The purpose of this study was to examine monocyte HLA-DR expression in ALF (including AALF) and investigate whether there is an association between this expression of the molecule with the magnitude of the inflammatory response evoked and eventual outcome.
SIRS, systemic inflammatory response syndrome; IL, interleukin; TNF-α, tumor necrosis factor α; ALF, acute liver failure; AALF, acetaminophen-induced acute liver failure; NAALF, non–acetaminophen-induced acute liver failure; CLD, chronic liver disease; KCH, King's College Hospital; INR, international normalized ratio; AALF-NS, AALF patients who died or received a liver graft; AALF-S, AALF patients who survived with conservative medical management; TDR, total HLA-DR–positive monocytes; MFI, mean fluorescence intensity; IFN-γ, interferon γ.
Patients and Methods
We prospectively studied monocyte HLA-DR expression and serum levels of pro- and anti-inflammatory cytokines in patients with AALF and compared them with non–acetaminophen-induced acute liver failure (NAALF), chronic liver disease (CLD), and healthy controls.
Seventy consecutive patients admitted to the Liver Intensive Care Unit at King's College Hospital (KCH) were entered into the study. Fifty AALF and 20 NAALF patients were investigated. AALF patients were divided further into those who died or received a liver graft (AALF-NS [n = 26]) and those who survived with conservative medical management (AALF-S [n = 24]). Fifty consecutive inpatients with stable CLD undergoing transplant assessment and 50 healthy volunteers served as pathological and healthy controls, respectively (Table 1). Exclusion criteria were: age <18 or >65 years, neoplasia, previous or concurrent immunosuppressive therapy, and clinical or microbiological evidence of sepsis on admission. AALF and NAALF patients were identified for emergency transplantation according to KCH criteria.29The clinical staff making decisions with regard to liver transplantation were not aware of monocyte HLA-DR and cytokine results. The study was approved by the KCH Ethics Committee. If a patient was unable to give informed consent, assent was obtained by the patient's nominated next of kin.
Table 1. Demographic Data of Study Populations
Abbreviations: ALD, alcoholic liver disease; HBV, hepatitis B virus; HCV, hepatitis C virus; PSC, primary sclerosing cholangitis; NA, not applicable.
Timing of Monocyte HLA-DR Measurement in AALF and NAALF Patients.
Monocyte HLA-DR expression was determined in 50 AALF and 20 NAALF patients, respectively, within 48 hours of admission to the KCH Liver Intensive Care Unit. Monocyte count, international normalized ratio (INR), liver and renal function tests, arterial lactate, and clinical and physiological variables such as degree of encephalopathy, APACHE II, mean arterial pressure, and norepinephrine requirements were recorded at the time of sampling for HLA-DR and serum cytokine determination.
Monocyte HLA-DR expression was assessed sequentially in 20 AALF patients. After determination of admission monocyte HLA-DR expression (see above), further measurements were performed between days 3 and 6 in 10 AALF-NS patients and 10 AALF-S patients and between days 7 and 10 in 8 of the 10 AALF-S patients.
Flow Cytometric Analysis of Monocyte HLA-DR Surface Expression.
Flow cytometric analysis was performed immediately after collecting blood samples in EDTA. Fifty microliters of whole blood was incubated with FITC-conjugated anti-CD14 and phycoerythrin-conjugated HLA-DR (BD PharMingen, San Diego, CA) for 20 minutes at room temperature and, after erythrocyte lysis, cells were analyzed using a FACSCalibur cytometer (Becton Dickinson, San Diego, CA), and 3000-5000 monocyte gated events (CD14+) were acquired. Negative controls were isotype-matched mouse monoclonal antibodies (BD PharMingen). Intra- and interassay coefficients of variation were 5% and 8.4%, respectively. The results are expressed as the HLA-DR percentage, the total number of HLA-DR positive monocytes (TDR), and the mean fluorescence intensity (MFI). The HLA-DR percentage is the percentage of monocytes expressing the HLA-DR molecule within the gated CD14 positive population. The TDR is the count of HLA-DR–expressing monocytes as derived from the hematological analyzer (Coulter counter) multiplied by the HLA-DR percentage obtained by FACS analysis. The MFI is the mean density of expressed HLA-DR molecules per individual monocyte.
Blood collected in EDTA from 30 AALF and 20 NAALF patients, respectively, at the same time as initial blood sampling for determination of monocyte HLA-DR expression, as well as from 22 CLD patients, was spun at 2000g for 10 minutes at 4°C; the serum obtained was immediately stored at −80°C. Levels of cytokines (IL-4, IL-6, IL-10, interferon γ [IFN-γ], and TNF-α) were measured via sandwich ELISA.30, 31 The sensitivity of cytokine assays is 5 pg/mL for TNF-α; 10 pg/mL for IL-4, IL-6, and IL-10; and 25 pg/mL for IFN-γ. When optical density values were below the lowest value of the reference curve, a value of 0 pg/mL was assigned. Because nonparametric statistical comparisons were used throughout the study, this did not interfere with the statistical analysis.
Differences between groups were identified via Kruskal-Wallis and Mann-Whitney U tests. Correlations were analyzed using the Spearman rank test. The results are expressed as the median and interquartile range. Cutoff values for the identification of nonsurviving patients requiring transplantation were determined using receiver operating characteristic analyses. A P value of less than .05 was considered statistically significant.
Table 1 illustrates the demographic data of the study population. CLD patients had significantly higher age than AALF patients, NAALF patients, and normal controls. Figure 1A-C shows the percentage, total number, and MFI of HLA-DR–expressing monocytes.
Levels of HLA-DR expression were significantly lower in AALF and NAALF patients compared with CLD patients and healthy controls. Monocyte HLA-DR expression in patients with CLD was stratified according to Child-Pugh class (A, n = 6; B, n = 20; C, n = 24). There was no difference detected in levels of HLA-DR expression when comparing patients with CLD according to Child-Pugh class (class A: HLA-DR 77%, MFI 315, TDR 0.26; class B: HLA-DR 69.5%, MFI 332, TDR 0.38; class C: HLA-DR 53.5%, MFI 250, TDR 0.36). HLA-DR expression was significantly lower in both AALF and NAALF than in patients with the most advanced CLD (Child-Pugh class C).
Whereas monocyte HLA-DR percentage and MFI were similar in AALF and NAALF patients (P = .35 and P = .5, respectively), monocyte TDR expression was significantly lower in AALF patients (P = .002), the total monocyte count being significantly lower in AALF patients compared with NAALF patients (0.24 [range: 0.1-0.6] vs. 0.6 [range: 0.4-1.4]; P = .005). There was no significant difference in HLA-DR expression with respect to age or sex in either AALF (P = .73, P = .97, respectively) or NAALF patients (P = .13, P = .72). AALF patients had significantly higher aspartate aminotransferase, alanine aminotransferase, and creatinine levels (P < .001 and P = .02, respectively) and lower bilirubin levels (P = .002) than NAALF patients (Table 2).
Table 2. Comparison of Indices of Liver Dysfunction and Monocyte HLA-DR Expression Between NAALF and AALF Patients
NOTE. All data are expressed as median (interquartile range). All clinical, hematological, biochemical, and physiological parameters were measured at the time of sampling for monocyte HLA-DR determination.
Abbreviations: OLT, liver transplantation; APACHE II, acute physiological and chronic health evaluation system; NA, not applicable; INR, international normalized ratio; AST, aspartate aminotransferase; ALT, alanine aminotransferase.
Number of patients
Patients fulfilling OLT criteria
Correlation of Monocyte HLA-DR Expression With Severity of Hepatic Injury in AALF and NAALF Patients.
In the undivided AALF patient group, INR negatively correlated with HLA-DR percentage (r = −0.6; P < .001), MFI (r = −0.5; P < .001), and TDR expression (r = −0.6; P < .001). Arterial blood lactate negatively correlated with monocyte HLA-DR percentage (r = −0.8; P =< .001), MFI (r = − 0.5; P < .001), and TDR expression (r = −0.6; P < .001). Arterial pH positively correlated with monocyte HLA-DR percentage (r = 0.6; P < .001), MFI (r = 0.5; P < .001), and TDR expression (r = 0.7; P < .001). Aspartate aminotransferase levels showed only a weak negative correlation with HLA-DR percentage (r = −0.3; P = .03), MFI (r = −0.3; P = .04), and TDR expression (r = −0.4; P = .002). The degree of encephalopathy negatively correlated with HLA-DR percentage (r = −0.5; P < .001), MFI (r = −0.6; P =< .001), and TDR expression (r = −0.6; P < .001).
In the NAALF patient group, there was no correlation between HLA-DR expression and the indices of acute hepatic dysfunction (INR, arterial blood lactate, arterial pH, degree of encepahalopathy, creatinine). A positive correlation between HLA-DR percentage and TDR expression with bilirubin levels was detected (r = 0.5, P = .04 and r = 0.6, P = .007, respectively).
ALF Outcome Subgroup Analysis.
We further divided the AALF-NS group into patients who underwent transplantation (n = 14) and patients who died without receiving a liver graft (n = 12). As shown in Fig. 2, the median monocyte HLA-DR expression in AALF patients who died (10.5% [range: 8%-12%]) or underwent transplantation (11.5% [range: 5%-14%]) was significantly lower compared with AALF-S patients (36% [range: 27.5%-43%]; P < .001). Differences in MFI and TDR expression were equally significant between the above subgroups. There was no difference in HLA-DR expression between patients who received transplantation and nonsurviving patients who did not receive transplantion (P = .78).
NAALF patients were divided into spontaneous survivors (n = 4), those who died after transplantation (n = 2), those who died without receiving a liver graft (n = 4), and those who underwent transplantation (n = 10). No significant differences in HLA-DR expression were detected between survivors and nonsurvivors, with or without transplantation.
AALF Subgroup Analysis.
AALF-NS patients had significantly lower total monocyte count, percentage, absolute number, and MFI of HLA-DR–expressing monocytes and significantly higher vasopressor requirements, level of encephalopathy, APACHE II score, blood lactate, and INR than AALF-S patients (Table 3). One patient in the AALF-S group survived without liver transplantation despite fulfilling transplantation criteria. This patient had a HLA-DR expression of 72.5%, a MFI of 213, and a TDR of 0.07, levels that were “normal” compared with other AALF patients who did not survive. The decision to treat with conservative medical management was made 7 days after initial fulfillment of the transplantation criteria, when no organ had become available and it was determined that outcome with medical management would be comparable to outcome with transplantation.
Table 3. Comparison of Indices of Liver Dysfunction and Monocyte HLA-DR Expression in AALF Patients Subdivided According to Whether They Died/Required OLT (AALF-NS) or Survived (AALF-S)
NOTE. All data are expressed as median (interquartile range). All clinical, hematological, biochemical, and physiological parameters were measured at time of sampling for monocyte HLA-DR determination.
Abbreviations: OLT, liver transplantation; APACHE II, acute physiological and chronic health evaluation system; NA, not applicable; INR, international normalized ratio; AST, aspartate aminotransferase.
ROC statistical analysis was used to evaluate the validity of monocyte HLA-DR expression to predict a poor outcome in AALF patients. A monocyte HLA-DR cut-off of 15%, a MFI of 52, and a TDR of 0.035 were obtained for the identification of nonsurviving patients and compared to the KCH criteria applied to the same population (Table 4). KCH criteria, and the decision to list for transplantation, were applied within 24 hours of monocyte HLA-DR determination.
Table 4. Performance of Monocyte HLA-DR Expression and KCH Criteria in the Identification of Nonsurviving AALF Patients
No. of Patients
Abbreviations: AUROC, area under receiver operating characteristic curve; OLT, liver transplantation; NA, not applicable.
Table 4 shows that diagnostic test accuracy of HLA-DR expression in predicting outcome was comparable to the current KCH criteria when undertaken 24-48 hours after admission to a specialist unit. HLA-DR expression of <15% showed a better sensitivity and specificity than the KCH criteria. MFI criteria had a comparable sensitivity but a lower specificity, and TDR criteria had a lower sensitivity but a similar specificity to the KCH criteria.
Longitudinal Data in AALF Patients.
To analyze changes in monocyte HLA-DR expression over time, we studied ten patients in AALF-S at day 1-2 and performed two further measurements between days 3-6 and days 7-10. The median HLA-DR percentage, MFI, and TDR were 33.5% (range: 26%-40%), 63 (range: 30-128), and 0.12 (range: 0.08-0.2), respectively, on day 1-2. As depicted in Fig. 3A-C, there was a significant increase in HLA-DR percentage (43% [range: 28%-65%]; P = .04), HLA-DR MFI (241 [range: 80-306]; P = .03), and TDR (0.6 [range: 0.4-0.9]; P = .01) by days 3-6. HLA-DR percentage (43% [range: 30%-60%]; P = .03), HLA-DR MFI (250 [range: 110-306]; P = .03), and TDR (0.65 [range: 0.4-1.8]; P = .02) persisted on days 7-10. There was no microbiological evidence of sepsis during the study period.
We also determined monocyte HLA-DR expression in 10 AALF-NS patients at day 1-2 and performed a second measurement between days 3-6. A further measurement was not available, because these patients either died or proceeded to transplantation. The median HLA-DR percentage, MFI, and TDR were 12% (range: 5%-14%), 31 (range: 16-45), and 0.02 (range: 0.008-0.08), respectively, on day 1-2 and 13% (range: 4.5%-17%), 33.5 (range: 19-48), and 0.04 (range: 0.004-0.08), respectively, on days 3-6. As shown in Fig. 4A-C, HLA-DR expression remained persistently low in this cohort of patients. There was no microbiological evidence of sepsis during the study period.
Serum Cytokine Determination in AALF, NAALF, and CLD Patients.
We measured serum levels of TNF-α, IL-4, IL-6, IFN-γ, and IL-10 in 30 AALF patients (15 AALF-NS, 15 AALF-S) and 20 NAALF patients (corresponding to the time of monocyte HLA-DR determination within 48 hours of admission) and compared them with the levels in 22 patients with CLD. As shown in Figs. 5A-E, higher serum levels of TNF-α, IL-4, IL-6, IFN-γ, and IL-10 were detected in AALF-NS compared with AALF-S and CLD. Serum IL-6 levels were higher in AALF-NS patients compared with CLD. NAALF patients had higher IL-6 and IL-10 levels and lower IFN-γ levels compared with AALF-S and CLD patients. When comparing AALF-NS and NAALF patients, no significant difference in levels of TNF-α, IL-6, or IL-10 were detected; however, higher levels of IFN-γ and IL-4 were noted in AALF-NS patients.
Correlation of Serum Cytokines in AALF Patients and Markers of Acute Hepatic Injury.
Table 5 gives the correlation coefficients and significance values of markers of acute hepatic injury and vasopressor requirements with monocyte HLA-DR percentage and serum TNF-α, IL-4, IL-6, IFN-γ, and IL-10. Significant positive correlations are seen between levels of TNF-α, IL-4, IL-6, and IL-10 with the degree of encephalopathy. Serum IL-6 and IL-10 show the strongest correlation with INR, arterial blood lactate, arterial pH, and vasopressor dose.
Table 5. Correlation Coefficients (r) of Conventional Markers of Acute Hepatic Injury and Vasopressor Requirement With Monocyte HLA-DR Percentage and Serum TNF-α, IL-4, IL-6, IFN-γ, and IL-10 Levels in AALF Patients
Abbreviations: INR, international normalized ratio; NS, not significant; AST, aspartate aminotransferase.
Correlation Between HLA-DR Expression and Serum Cytokines.
Table 6 shows the correlation coefficients and significance values between the 5 serum cytokines and concomitantly measured HLA-DR expression. Serum IL-10 levels had the strongest negative correlation with HLA-DR expression.
Table 6. Correlation Coefficients (r) Between Serum Cytokines and Concomitant Monocyte HLA-DR Measurements in 30 AALF Patients (15 AALF-NS, 15 AALF-S)
Correlation Between Serum Cytokines in AALF Patients.
In all AALF patients, a positive correlation between serum levels of IL-10 and TNF-α (r = 0.60, P = .001) and IL-6 (r = 0.63; P < .001) was detected.
Our data reveal a striking reduction in the expression of surface HLA-DR on monocytes in acute liver failure. There is a reduction in percentage and total number not only in HLA-DR–expressing monocytes but also in HLA-DR expression on the individual cell. These findings echo those of septic shock where a profound decrease of HLA-DR positive monocytes was first reported. Our data also show that the degree of monocyte HLA-DR expression within the first 48 hours of acetaminophen-induced acute liver failure is a marker of prognosis and reflects disease severity, with the levels of HLA-DR–expressing monocytes being particularly low in those patients with a worse outcome and showing a strong correlation with markers of acute hepatic dysfunction. In sequential analyses, monocyte HLA-DR expression remained low in the AALF patients with a poor prognosis and increased in those who made a spontaneous recovery, again indicating that monocyte HLA-DR expression mirrors the clinical evolution of the disease process. The fact that these alterations occur in the absence of positive microbiology emphasizes the fact that changes in HLA-DR expression are directly related to the disease process and are not a consequence of sepsis. In view of its association with prognosis, monocyte HLA-DR expression may aid decision-making in patients, who remain critically unwell for prolonged periods of time, awaiting a liver graft, in whom the survival benefit of transplantation may be similar to that of medical management.
A reduction in monocyte HLA-DR expression was also seen in ALF secondary to nonacetaminophen causes, indicating that this reduction is common to all forms of ALF. In NAALF, however, the degree of HLA-DR expression did not correlate with markers of hepatic injury nor predicted outcome, which likely reflects the smaller number of patients and the heterogeneous nature of this patient group. The usefulness of HLA-DR expression testing in NAALF requires investigation in larger groups of patients suffering from ALF of viral and nonviral etiology.
Given the link between severity of acute hepatic dysfunction and monocyte HLA-DR expression in AALF, the question arises as to whether the reduced surface expression of this molecule is of pathogenic significance. Functional analyses of monocyte function in ALF11, 14 and conditions characterized by reduced monocyte HLA-DR surface expression clearly demonstrate that these cells with reduced HLA-DR expression are “dysfunctional” as a consequence of their impaired ability at presenting antigen and at producing TNF-α following lipopolysaccharide stimulation,21, 32, 33 the latter defect being directly linked to the development of recurrent bacterial and opportunistic infections.21
Because cytokines are known to directly influence the expression of monocyte HLA-DR, we measured both proinflammatory (TNF-α, IL-6) and anti-inflammatory (IL-10) cytokines in the circulation of our AALF patients and found elevated serum levels of both. TNF-α, IL-6 on the one side, and IL-10 on the other have opposite effects on the inflammatory response, negatively regulating each other's synthesis. Hence, it may be surprising that both pro- and anti-inflammatory cytokines were elevated and correlated with each other and with the severity of hepatic injury and clinical outcome.
These findings are similar to experimental endotoxic shock, where both TNF-α and IL-10 are elevated following injection of endotoxin.34–36 This concomitant rise can be observed within minutes after lipopolysaacharide administration, suggesting that a homeostatic mechanism is operative from the very start of the inflammatory response, with anti-inflammatory cytokines attempting to counteract the damaging effects of inflammatory cytokines. That IL-10 indeed plays a role in reducing proinflammatory cytokine secretion and protects against severe liver injury has been demonstrated in animal models of liver failure37–39 where administration of IL-10 prevents lipopolysaacharide/galactosamine and concavalin-A–induced acute hepatoxicity.
In conclusion, TNF-α increases monocyte HLA-DR surface expression, whereas IL-10 decreases it. However, despite this opposite effect, both TNF-α and IL-10 correlated negatively with the levels of surface monocyte HLA-DR expression in AALF patients, IL-10 having the strongest correlation. A possible explanation is that acute liver injury stimulates TNF-α production by Kupffer cells, followed by further inflammatory cell recruitment, which exacerbates the initial liver injury. If the process is not interrupted, IL-10, secreted to counteract the damaging effect of the inflammatory cytokines, may overshoot its target, leading to a severe inhibition of HLA-DR expression with consequent deactivation of monocytes. This would facilitate the development of sepsis, followed by multiple organ dysfunction and death.
We are grateful for the assistance provided by the Liver Intensive Care Unit staff, Professor G. Mieli-Vergani, and Dr. R. Bowen.