Prognostic Value of C-Reactive Protein Levels in Patients With Cirrhosis

Transcription

1 LIVER TRANSPLANTATION 21: , 2015 ORIGINAL ARTICLE Prognostic Value of C-Reactive Protein Levels in Patients With Cirrhosis Vincent Di Martino, 1,2 Caroline Coutris, 1 Jean-Paul Cervoni, 1 Stavros Dritsas, 1,2 Delphine Weil, 1 Carine Richou, 1 Claire Vanlemmens, 1 and Thierry Thevenot 1,2 1 Service d H epatologie et de Soins Intensifs Digestifs, Centre Hospitalier R egional Universitaire H ^opital Jean Minjoz, Besanc on, France; and 2 Universit e de Franche Comt e, Unit e de Formation et de Recherche des Sciences M edicales et Pharmaceutiques, Besanc on, France Identifying cirrhosis with a poor short-term prognosis remains crucial for improving the allocation of liver grafts. The purpose of this study was to assess the prognostic value of a model combining the variation of C-reactive protein (CRP) levels within 15 days, the Model for End-Stage Liver Disease (MELD) score, and the presence of comorbidities in patients with decompensated cirrhosis with a Child-Pugh score > B7 and to test the relevance of this model in patients with compensated cirrhosis. We collected data for cirrhotic patients without hepatocellular carcinoma, extrahepatic malignancy, human immunodeficiency virus infection, organ transplantation, seen between January 2010 and December Multivariate analyses of predictors of 3-month mortality used Cox models adjusted with the age-adjusted Charlson comorbidity index. The prognostic performance [area under receiver operating characteristic curves (AUROCs)] of the 3-variable model was compared to that of the MELD score. The 241 patients who met the inclusion criteria included 109 patients with a Child-Pugh score > B7 who were hospitalized for decompensation. In these patients with severe cases, the 3-month mortality was independently predicted by the MELD score [hazard ratio (HR), 1.10; 95% confidence interval (CI), ; P < 0.001] and a CRP level > 32 mg/l at the baseline and on day 15 (HR, 2.21; 95% CI, ; P ). This model was better than MELD alone (AUROC, versus 0.734; P ). In the whole population with cirrhosis, the 3-month mortality was also predicted by high MELD scores (HR, 1.11; 95% CI, ; P < 0.001) and a CRP level > 10 mg/l at the baseline and on day 15 (HR, 2.89; 95% CI, ; P < 0.001), but the AUROCs of the 3-variable model and the MELD score alone were no longer significantly different (0.89 versus 0.88, not significant). In conclusion, prognostic models incorporating variations in CRP predict 3-month mortality in patients with cirrhosis. Such models are particularly relevant for patients with decompensated cirrhosis but provide a limited increase in prediction in comparison with the MELD score in the whole population with cirrhosis. Liver Transpl 21: , VC 2015 AASLD. Received November 4, 2014; accepted January 29, See Editorial on Page 713 The prognosis of patients with cirrhosis remains difficult to assess despite its substantial improvement during the last decades 1,2 ; this is particularly due to Additional Supporting Information may be found in the online version of this article. Abbreviations: ACCI, age-adjusted Charlson comorbidity index; ACCP/SCCM, American College of Chest Physicians/Society of Critical Care Medicine; AUROC, area under receiver operating characteristic curve; CPP, comite de protection des personnes; CRP, C-reactive protein; HCC, hepatocellular carcinoma; HIV, human immunodeficiency virus; HR, hazard ratio; IL, interleukin; MELD, Model for End-Stage Liver Disease; OR, odds ratio; SIRS, systemic inflammatory response syndrome; TNF-a, tumor necrosis factor a. Potential conflict of interest: Nothing to report. Funding: This study was not supported by any pharmaceutical company or governmental agency. Author contributions: study conception, V.D.M. and J.P.C.; data collection, C.C.; statistical analyses, V.D.M.; manuscript drafting, V.D.M, and C.C.; and critical revision of the manuscript for important intellectual content, J.P.C., T.T., S.D., D.W., C.R., and C.V. Address reprint requests to Vincent Di Martino, M.D., Ph.D., Service d Hepatologie, Centre Hospitalier Regional Universitaire H^opital Jean Minjoz, Besançon Cedex, France. Telephone: ; FAX: ; vdimartino@chu-besancon.fr DOI /lt View this article online at wileyonlinelibrary.com. LIVER TRANSPLANTATION.DOI /lt. Published on behalf of the American Association for the Study of Liver Diseases VC 2015 American Association for the Study of Liver Diseases.

2 754 DI MARTINO ET AL. LIVER TRANSPLANTATION, June 2015 the development of liver transplantation. The context of the organ shortage makes it crucial to optimize organ allocation by giving priority to the sickest patients; this emphasizes the need for accurate shortterm prognosis assessment. Currently, patients eligible for liver transplantation are sorted according to their Model for End-Stage Liver Disease (MELD) score, 3 which is an indicator of liver function. However, it has been shown that the short-term prognosis of patients with end-stage liver failure largely depends on events that worsen temporarily or are superimposed to liver failure. These events are better taken into account by general prognostic scores numbering organ failures (such as the Acute Physiology and Chronic Health Evaluation and Sequential Organ Failure Assessment scores) than scores specifically dedicated to liver failure such as the Child-Pugh and MELD scores. A key event for prognosis seems to be the existence of systemic inflammatory response syndrome (SIRS), 4,5 which has been shown to increase mortality, 6,7 complications of portal hypertension, and encephalopathy 8,9 in patients with cirrhosis with or without renal failure. It, therefore, seems appropriate to test prognostic models incorporating parameters of liver function and parameters reflecting SIRS. In a previous prospective study, we suggested that C-reactive protein (CRP) was an accurate marker of SIRS and that it was able to predict 6-month mortality in patients with cirrhosis. 1 This study included hospitalized patients with cirrhosis with a Child-Pugh score > B7 and suggested a good prognostic value for a 3-variable model incorporating the MELD score, an assessment of extrahepatic comorbidities, and the variation of CRP levels within 15 days. The aims of this study were, therefore, (1) to provide a validation of this model in a new cohort of patients, (2) to investigate whether CRP is still useful for predicting mortality in the general population of patients with cirrhosis (ie, without clinical decompensation and with a Child-Pugh score of A), and (3) to investigate whether this model is able to predict 3-month mortality to make it more appropriate for the context of liver transplantation. PATIENTS AND METHODS Study Design We conducted a prospective observational cohort study. The protocol for this study was approved by the Besançon University Hospital institutional review board (Comite de protection des personnes (CPP) Est-2). Patients All consecutive patients with cirrhosis admitted between January 2010 and December 2011 to our university hospital were prospectively included if they fulfilled the following criteria: (1) cirrhosis either histologically proven or considered obvious by clinical, biochemical, and morphological criteria; (2) no hepatocellular carcinoma (HCC); (3) no progressive extrahepatic malignancy, organ transplant, or human immunodeficiency virus (HIV) infection; and (4) consent. There were no exclusion criteria regarding age or Child-Pugh score. Patients enrolled in our previous study were not eligible for the present study. Data Collection Ninety-four variables were recorded at admission and during follow-up at day 15 and month 6. They concerned demographics (sex and age); causes and history of liver disease; relevant comorbidities as assessed with the ageadjusted Charlson comorbidity index (ACCI) 10 ;reason for admission [with a specific distinction between planned hospitalizations (for routine paracentesis, endoscopy, or other diagnostic procedures) and unplanned hospitalizations (in the event of decompensation)]; clinical status on admission, including the SIRS criteria; 5 liver biochemistry; assessment of renal function; cultures of blood, urine, and ascites (if any); and measurements of routine serum CRP and procalcitonin. Statistical Analyses Quantitative variables were expressed as means and standard errors or as medians and 95% confidence intervals (CIs) in the event of abnormal distributions. Comparisons of subgroups of patients regarding variations of CRP levels were performed with the chisquare test, Student t test, and analysis of variance. Correlations between CRP levels and MELD scores were evaluated with the Spearman rank test. Statistical analyses were performed separately in the whole study population and in the subset of patients with a Child-Pugh score > B7 who were hospitalized for clinical decompensation. This latter population constituted a validation cohort of results that were previously published. For these 2 analyses, the primary outcome measure for prognosis was the 3-month mortality rate. Patients were separated into 3 groups defined by the variation of CRP levels between the baseline and day 15 as defined previously, with relevant thresholds of CRP assessed with the Youden index. Group A (validation cohort) and group A 0 (whole population) were defined by persistently high CRP levels (at the baseline and day 15); group B (validation cohort) and group B 0 (whole population) were defined by high CRP levels at the baseline with a subsequent decrease below the threshold at day 15; and group C (validation cohort) and group C 0 (whole population) were defined by CRP levels below the threshold at the baseline. Patients grouped according to CRP variation and other factors affecting prognosis were evaluated with the Kaplan- Meier model and log-rank test for univariate analyses and with the proportional hazard regression (Cox) Model for multivariate analysis. To include in the model the variation of CRP levels within the first 15 days as a predictive factor of mortality, day 15 was considered as the starting point, and patients who died within the first 15 days were not included in the survival analysis. Patients transplanted during follow-

3 LIVER TRANSPLANTATION, Vol. 21, No. 6, 2015 DI MARTINO ET AL. 755 Figure 1. Flowchart of patients with cirrhosis who were included in the study. up were included and considered alive, with the liver transplant date marking the end of follow-up. The performances of MELD, baseline CRP, and the multivariate model for predicting short-term mortality were compared with areas under the receiver operating characteristic curve (AUROCs). All statistical analyses were performed with the NCSS package for Windows. RESULTS Study Population During the study period, 505 patients with cirrhosis were admitted, but 74 patients were excluded because they participated in the previous study: 138 patients had HCC, 7 patients were HIV-infected, 7 patients were organ recipients, and 16 patients had solid cancer or hematologic malignancies. Because of missing data, 22 additional patients were excluded. Thus, a total of 241 patients with cirrhosis were finally included (Fig. 1). Their characteristics are summarized in Table 1. Briefly, there were 160 men (66.4%); the median age was 60 years (95% CI, years); and the cirrhosis was related to alcohol for 140 patients (58.1%), to metabolic syndrome for 40 patients (16.6%), to hepatitis C virus or hepatitis B virus for 28 patients (11.6%), and to other diseases (Wilson s disease, hemochromatosis, autoimmune disease, or cryptogenic disease) for 33 patients (13.7%). Eighty-eight patients (36.5%) were Child-Pugh A, 64 patients (26.6%) were Child-Pugh B, and 89 patients (36.9%) were Child-Pugh C. Ninety-two patients (38.2%) were hospitalized for an endoscopic procedure or paracentesis (planned hospitalization), and 149 patients (61.8%) were hospitalized for decompensation (unplanned hospitalization). The median MELD score was 15.4 (95% CI, ). The CRP value ranged from 1 to 372 mg/l (median, 21.2 mg/l; 95% CI, mg/l). We identified 69 overt bacterial infections (28.6%), including 20 spontaneous peritonitis cases (8.3%) and 18 bacteremia cases (7.5%). Among the 33 alcoholic patients with a Maddrey discriminant function > 32 who underwent transjugular liver biopsy, 23 patients had histologically proven alcoholic hepatitis. They represented 9.5% of the whole cohort and 21.1% of patients with decompensated cirrhosis. Seventy-three patients (30.3%) had SIRS as defined by the American College of Chest Physicians/Society of Critical Care Medicine (ACCP/SCCM) consensus conference 5 ; 113 patients (46.9%) had extrahepatic comorbidities, including 30 cardiac failures (12.4%), 18 respiratory failures (7.5%), and 15 renal failures unrelated to hepatorenal syndrome (6.2%). The median ACCI 10 was 5.4, and it ranged from The subgroup of patients with a Child-Pugh score > B7 who were hospitalized for decompensation included 109 patients (45.2%). During a 14-month median follow-up (range, 15 days to 38 months), 76 patients (31.5%) died; they

4 756 DI MARTINO ET AL. LIVER TRANSPLANTATION, June 2015 TABLE 1. Patients Characteristics Whole Population (n 5 241) Child-Pugh Score > B7, Unplanned Hospitalization (n 5 109) Age (years) 60 (59-62) 59 (57-61) Sex (male) 160 (66.4%) 83 (76.1%) Etiology of cirrhosis Alcohol 140 (58.1%) 79 (72.5%) Virus 28 (11.6%) 9 (8.3%) Metabolic 40 (16.6%) 9 (8.3%) Other 33 (13.7%) 12 (11.0%) Histologically proven alcoholic hepatitis 23 (9.5%) 23 (21.1%) Extrahepatic comorbidities 113 (46.9%) 49 (45.0%) ACCI 5.4 ( ) 5.3 ( ) MELD 15.4 ( ) 21.5 ( ) Bacterial Infection 69 (28.6%) 57 (52.3%) Spontaneous peritonitis 20 (8.3%) 20 (18.3%) Pneumonia 16 (6.6%) 16 (14.7%) Bacteremia 18 (7.5%) 15 (13.8%) SIRS 73 (30.3%) 55 (50.5%) SIRS according to heart rate 69 (28.6%) 41 (37.6%) SIRS according to breath rate or 206 (85.5%) 100 (91.7%) partial pressure of carbon dioxide SIRS according to leukocyte count 76 (31.5%) 41 (37.6%) SIRS according to body temperature 18 (7.5%) 14 (12.8%) Baseline CRP (mg/l) 21.2 ( ) 35.2 ( ) Procalcitonin (ng/ml) 0.69 ( ) 0.72 ( ) Planned hospitalization 92 (38.2%) Beta-blockers 135 (56.0%) 70 (64.2%) Child-Pugh A 88 (36.5%) Child-Pugh B 64 (26.6%) 26 (23.9%) Child-Pugh C 89 (36.9%) 83 (76.1%) included 38 deaths at month 3 and 13 additional deaths at month 6. Fifteen patients (6.2%) underwent liver transplantation. Determination of Optimal Cutoffs of CRP Levels for Predicting 3-Month Mortality High baseline CRP levels were significantly associated with 3-month mortality ( mg/l for patients who died versus mg/l for others, P < 0.001). The AUROC was (95% CI, ). According to the Youden index, the best discriminant value of CRP was 10 mg/l (the upper normal range). With this cutoff, the prediction of 3-month mortality was true for 62% of patients, the sensitivity was 89%, and the negative predictive value was 97%. When we considered the changes in CRP levels between the baseline and day 15 with this cutoff, 72 patients belonged to group A 0, 21 belonged to group B 0, and 120 belonged to group C 0. In a similar receiver operating characteristic curve analysis restricted to patients with decompensated cirrhosis and a Child-Pugh score > B7 (validation cohort), the AUROC was (95% CI, ), and the best discriminant value of CRP given by the Youden index was 32 mg/l, which gave the following performance indicators for predicting 3-month mortality: sensitivity, 51%; specificity, 70%; negative predictive value, 75%; and well-classified, 64%. When we considered the changes in CRP levels between the baseline and day 15 with this cutoff, 19 patients belonged to group A, 20 belonged to group B, and 69 belonged to group C. Predictors of 3-Month Mortality in Patients With Decompensated Cirrhosis In univariate analyses restricted to Child-Pugh > B7 patients with cirrhosis hospitalized for decompensation, 3-month mortality was significantly associated with the Child-Pugh C stage, a high MELD score, the presence of SIRS, bacteremia, overt bacterial infection, and particularly spontaneous peritonitis. The survival rate was significantly lower in group A patients versus the others (P ). At 3 months, it was 57.0% (95% CI, 32.8%-81.2%) in group A versus 60.0% (95% CI, 38.5%-81.5%) in group B and 75.2% (95% CI, 65.0%-85.4%) in group C; at 6 months, it was 44.3% (95% CI, 20.0%-68.7%), 60.0% (95%, CI, 38.5%-81.5%), and 64.4% (95% CI, 53.9%-75.9%) in groups A, B, and C, respectively (Fig. 2A). The 3- variable prognostic model (including ACCI, MELD, and group A as defined by CRP variations) was able to predict 3-month mortality (AUROC, 0.789; 95%

5 LIVER TRANSPLANTATION, Vol. 21, No. 6, 2015 DI MARTINO ET AL. 757 TABLE 2. Multivariate Analysis of Factors Associated With 3-Month Mortality in the Whole Population 95% % Variance Variables HR CI (HR) P (Pseudo-r 2 ) ACCI MELD < CRP < group A 0 NOTE: A Cox model was used for 33 events and 208 patients (r ). Figure 2. Survival according to the variation of CRP levels between the baseline and day 15: (A) the validation cohort (Child-Pugh score > B7 and unplanned hospitalization, n 5 109) and (B) the general population (n 5 217). Survival was significantly lower in groups A and A 0 as defined by a persistently elevated CRP level. CI, ). Group A [hazard ratio (HR), 2.21; 95% CI, ; P ] and the MELD score (HR, 1.10; 95% CI, ; P < 0.001) had independent influence on 3-month mortality. There was no correlation between the CRP levels and the MELD score (r 2 < 0.001, P 5 not significant). Predictors of 3-Month Mortality in the General Population In the univariate analysis, 3-month mortality was significantly associated with group A 0 [odds ratio (OR), 7.33; 95% CI, ; P < 0.001]. The 3-month survival was 69.2% (95% CI, 58.3%-80.2%) in group A 0, 75.4% (95% CI, 56.6%-94.2%) in group B 0, and 96.6% (95% CI, 93.4%-99.9%) in group C 0 (P < 0.001; Fig. 2B). Other significant predictors were a Child- Pugh score > B7 (OR, 23.26; 95% CI, ; P < 0.001), a MELD score > 18 (OR, 10.38; 95% CI, ; P < 0.001), histologically proven alcoholic hepatitis (OR, 2.67; 95% CI, ; P ), an overt bacterial infection (OR, 11.20; 95% CI, ; P < 0.001), a positive blood culture (OR, 15.31; 95% CI, ; P < 0.001), pneumonia (OR, 6.57; 95% CI, ; P < 0.001), spontaneous peritonitis (OR, 8.87; 95% CI, ; P < 0.001), SIRS (OR, 14.53; 95% CI, ; P < 0.001) and all SIRS criteria considered separately, cardiac failure (OR, 3.30; 95% CI, ; P ), and unplanned hospitalization (OR, 30.12; 95% CI, ; P < 0.001). The use of beta-blockers had no significant impact on 3-month mortality. Group A 0 patients were more often younger than 60 years (42.7% versus 26.7% P ), more often had a Child-Pugh score of B or C (50.7% versus 3.9% P < 0.001), a MELD score > 18 (56.0% versus 21.2%, P < 0.001), SIRS (72.7% versus 16.5%, P < 0.001), a bacterial infection (62.1% versus 21.1%, P < 0.001), positive blood cultures (88.2% versus 29.1%, P < 0.001), spontaneous peritonitis (85.0% versus 28.5%, P < 0.001), and alcoholic hepatitis (68.2% versus 29.8%, P < 0.001), and had more often been hospitalized for decompensation (49.6% versus 6.4% P < 0.001) than others. There was no significant relationship between group A 0 and sex, extrahepatic morbidities, and the etiology of cirrhosis. CRP and MELD were positively correlated (r , P < 0.001), conversely to that observed in the subset of more severe patients. Through Cox regression multivariate analysis adjusted by ACCI, the MELD score (HR, 1.113; 95% CI HR, ; P < 0.001) and group A 0 according to the CRP variations (HR, 2.894; 95% CI HR, ; P < 0.001) were significant predictors of 3- month mortality (Table 2). The AUROC of the 3- variable model was (95% CI, ). Sensitivity Analyses Sensitivity analyses were performed to investigate whether group A 0 was able to predict the 6-month mortality when we excluded patients with histologically proven alcoholic hepatitis or when we separately considered patients with or without bacterial infection. When we restricted the survival analysis to

6 758 DI MARTINO ET AL. LIVER TRANSPLANTATION, June 2015 Performance of the 3-Variable Prognostic Model Versus the MELD Score Alone In the general population, the prognostic performance of the 3-variable Cox model (incorporating MELD, ACCI, and group A 0 as performed in the previous study) was (AUROC; 95% CI, ). In comparison with the performance of the MELD score alone (AUROC, 0.876; 95% CI, ), the difference was not significant (P ; Fig. 3). In contrast, in the restricted population (Child-Pugh score > B7 and unplanned hospitalization), the same model predicted the 3-month mortality significantly better than the MELD score alone (AUROC, versus 0.734; P ; Fig. 3). Predictors of 6-Month Mortality All the previous analyses were performed for discriminating survivors from nonsurvivors at 6 months. The same predictors were identified with similar cutoffs of CRP levels. The impact of extrahepatic comorbidities was slightly stronger (data not shown). Figure 3. Prediction of the 3-month mortality: comparison of MELD score and 3 variables model. In the whole population of inpatients with cirrhosis, the MELD score accurately predicts 3 month mortality (AUROC ) whereas a Cox model (1) including MELD, Age-adjusted Charlson comorbidity index (ACCI) and the variation of CRP levels between baseline and day 15 provided marginal and non-significant improvement for the same prediction (AUROC ). In severe patients (as defined by Pugh >B7 and unplanned hospitalization for clinical decompensation), the MELD score has lower variability and its power for predicting 3 month mortality decreases (AUROC ). In that population, a 3 variables Cox model (MELD, ACCI, CRP variation within 15 days, (2)) provides significant improvement for predicting 3-month mortality (AUROC , p ). (1) Indexp 5 Exp( *(ACCI) *(groupA, i.e., CRP>10mg/L at baseline and day-15) *(MELD) (2) Index 5 Exp( *ACCI *(groupA, i.e., CRP32mg/L at baseline and day-15) *MELD). patients without alcoholic hepatitis and without infection, the 3-month survival was 85.2% in group A 0 versus 96.4% in other patients (P ). Similar results were obtained when we removed from analysis only the patients with alcoholic hepatitis or only the patients with bacterial infection. In a sensitivity analysis restricted to infected patients, the difference in 3-month survival was still significant between group A 0 patients and others (50.1% versus 79.8%, P ). DISCUSSION This study provided 2 relevant results. First, it confirmed that the variation of CRP level is a useful tool for predicting short-term mortality in hospitalized patients with cirrhosis with a Child-Pugh score > B7. In this population, the relevant cutoff of CRP was 32 mg/l, which was close to the value that was found in the first set of patients 1. We thus provided a convincing internal validation of the previously published results. Second, we demonstrated that our prognostic model still works in patients with less severe cirrhosis, which is reflected by lower Child-Pugh scores and planned hospitalizations. In that population, the relevant cutoff of CRP was 10 mg/l, that is, the upper normal value. However, the increase in predicting 3- month mortality provided by our model appears to be marginal in comparison with that already given by the MELD score alone. Our study had the advantage of prospectively collecting numerous data from an unselected population and thus avoiding selection bias. The sample size was adequate to allow multivariate analyses and to compare the performance of different prognostic models. In addition, the assessment of comorbidities was performed with the ACCI (quantitative index) rather than a qualitative indicator (presence or absence of extrahepatic comorbidities), and this enabled us to more accurately analyze the prognostic consequences of comorbidities in patients with cirrhosis. Such information is often missing in the literature. Another advantage of this study is the expansion of the analysis of the prognostic value of CRP to patients with less severe cirrhosis; this allowed us to generalize our findings to the general population of patients with cirrhosis. We confirmed the deleterious impact of alcoholic hepatitis, bacterial infection, and SIRS, as already

7 LIVER TRANSPLANTATION, Vol. 21, No. 6, 2015 DI MARTINO ET AL. 759 described in the literature Interestingly, the prognostic relevance of CRP variations, although still significant, was much lower in the whole population versus the population of patients with severe cirrhosis. This was clearly illustrated when we compared the performance of our 3-variable model and that of the MELD score (Fig. 3). This finding was probably the consequence of a higher variability of MELD scores in the whole population; this increased its power. We also found that CRP and MELD were correlated in the whole population and were not correlated in the subset of patients with severe cirrhosis. This makes our prognostic model particularly useful in the setting of end-stage liver disease. In this context, systemic inflammation, a condition able to compromise survival, may be frequent but poorly predicted by variations of the MELD score. The measure of a simple and reproducible marker of SIRS such as CRP thus seems interesting. Conversely, in patients with compensated cirrhosis, the prognostic value of CRP variations is statistically significant, but the clinical significance probably remains limited. SIRS is common in patients with severe cirrhosis. It is the consequence of overt or occult bacterial infections 14 favored by an increased permeability of the gastrointestinal tract, a decrease in serum concentrations of fibronectin, complement, and opsonins, and an impairment of phagocytic function and chemotaxis of the neutrophils. 15 The activation of the inflammatory response is promoted by the bacterial translocation of gram-negative bacilli and grampositive cocci from the gastrointestinal tract to the mesenteric lymph nodes and/or other extra-intestinal sites. 16 Bacterial translocation is the consequence of bacterial growth, 17,18 structural 19 and functional changes (exudative enteropathy) 20 of the mucosa, and a deficiency of mucosal T lymphocytes 21 and immunoglobulin A. 22 Bacterial translocation may increase the production of NO in the splanchnic circulation and aggravate portal hypertension and abnormal liver flow. 23,24 It also activates peritoneal macrophages, which produce proinflammatory cytokines [interleukin-2 (IL2), IL12, and tumor necrosis factor a (TNF-a)]. Such a proinflammatory response is exacerbated by IL6. 25 In patients with cirrhosis who have ascites, Albillos et al. 26 found that high concentrations of lipopolysaccharide-binding protein stimulate the secretion of TNF-a and IL6. The clinical diagnosis of SIRS may be difficult in the event of cirrhosis: hypersplenism can hide hyperleukocytosis or worsen leukopenia; subclinical encephalopathy may lead to an increase in respiratory rate and induce hypocapnia; and tachycardia is frequently associated with hyperkinetic syndrome but can conversely be hidden by beta-blockers. CRP is synthesized in the acute phase of inflammation in response to IL6, and its concentration is not modified by liver impairment. 27 It seems to be a good marker of SIRS and is easily available, inexpensive, and more accurate than clinical parameters of SIRS for identifying systemic inflammation in patients with cirrhosis. We have shown that a simple measurement of CRP at the baseline and at day 15 provides a powerful tool for predicting short-term mortality in hospitalized patients with cirrhosis. Interestingly, high baseline CRP levels could remain elevated over time despite the resolution of bacterial infection in some patients. This suggests that systemic inflammation may become persistent and act as an autonomous disease. In the event of bacterial infection, the value of serial CRP measures for predicting the mortality of patients with cirrhosis was underlined by other authors: prognosis was significantly better in patients who quickly recovered from bacterial infection and experienced a reduction in CRP levels, 28 as also observed in our series. Our findings are also in keeping with those of a multicenter study that was recently published, 29 which found that high CRP levels, as well as hyperleukocytosis, were strong predictors of death in patients with cirrhosis. However, this study did not investigate the relevance of serial measures of CRP and its strong prognostic value out of the context of sepsis. The applicability of our findings to the context of liver transplantation seems good. It may allow for better sorting of patients with intermediate MELD scores on the waiting list. Although it requires a 15-day follow-up to become available, our model helps avoid the prioritization of infected patients for liver transplantation in contrast to what a single measure of CRP would produce. In conclusion, CRP levels and particularly their variation constitute a true prognostic marker in patients with cirrhosis, particularly powerful in the event of advanced liver failure. CRP allows the identification of patients with cirrhosis who have a poor short-term prognosis. A model combining the MELD score and CRP variations may sort the candidates for liver transplantation better than the MELD score alone. Such a model warrants external validations. REFERENCES 1. Cervoni JP, Thevenot T, Weil D, Muel E, Barbot O, Sheppard F, et al. C-reactive protein predicts short-term mortality in patients with cirrhosis. J Hepatol 2012;56: D Amico G, Garcia-Tsao G, Pagliaro L. Natural history and prognostic indicators of survival in cirrhosis: a systematic review of 118 studies. J Hepatol 2006;44: Wiesner R, Edwards E, Freeman R, Harper A, Kim R, Kamath P, et al.; for United Network for Organ Sharing Liver Disease Severity Score Committee. Model for endstage liver disease (MELD) and allocation of donor livers. Gastroenterology 2003;124: Malik R, Mookerjee RP, Jalan R. Infection and inflammation in liver failure: two sides of the same coin. J Hepatol 2009;51: American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference: definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Crit Care Med 1992;20:

8 760 DI MARTINO ET AL. LIVER TRANSPLANTATION, June Navasa M, Follo A, Filella X, Jimenez W, Francitorra A, Planas R, et al. Tumor necrosis factor and interleukin-6 in spontaneous bacterial peritonitis in cirrhosis: relationship with the development of renal impairment and mortality. Hepatology 1998;27: Thabut D, Massard J, Gangloff A, Carbonell N, Francoz C, Nguyen-Khac E, et al. Model for end-stage liver disease score and systemic inflammatory response are major prognostic factors in patients with cirrhosis and acute functional renal failure. Hepatology 2007;46: Rolando N, Wade J, Davalos M, Wendon J, Philpott- Howard J, Williams R. The systemic inflammatory response syndrome in acute liver failure. Hepatology 2000;32(pt 1): Shawcross DL, Sharifi Y, Canavan JB, Yeoman AD, Abeles RD, Taylor NJ, et al. Infection and systemic inflammation, not ammonia, are associated with Grade 3/4 hepatic encephalopathy, but not mortality in cirrhosis. J Hepatol 2011;54: Charlson M, Szatrowski TP, Peterson J, Gold J. Validation of a combined comorbidity index. J Clin Epidemiol 1994;47: Arvaniti V, D Amico G, Fede G, Manousou P, Tsochatzis E, Pleguezuelo M, Burroughs AK. Infections in patients with cirrhosis increase mortality four-fold and should be used in determining prognosis. Gastroenterology 2010; 139: Cazzaniga M, Dionigi E, Gobbo G, Fioretti A, Monti V, Salerno F. The systemic inflammatory response syndrome in cirrhotic patients: relationship with their inhospital outcome. J Hepatol 2009;51: Mathurin P, Duchatelle V, Ramond MJ, Degott C, Bedossa P, Erlinger S, et al. Survival and prognostic factors in patients with severe alcoholic hepatitis treated with prednisolone. Gastroenterology 1996;110: Navasa M, Rimola A, Rodes J. Bacterial infections in liver disease. Semin Liver Dis 1997;17: Mookerjee RP, Malaki M, Davies NA, Hodges SJ, Dalton RN, Turner C, et al. Increasing dimethylarginine levels are associated with adverse clinical outcome in severe alcoholic hepatitis. Hepatology 2007;45: Cirera I, Bauer TM, Navasa M, Vila J, Grande L, Taura P, et al. Bacterial translocation of enteric organisms in patients with cirrhosis. J Hepatol 2001;34: Shindo K, Machida M, Miyakawa K, Fukumura M. A syndrome of cirrhosis, achlorhydria, small intestinal bacterial overgrowth, and fat malabsorption. Am J Gastroenterol 1993;88: Garcia-Tsao G, Wiest R. Gut microflora in the pathogenesis of the complications of cirrhosis. Best Pract Res Clin Gastroenterol 2004;18: Viggiano TR, Gostout CJ. Portal hypertensive intestinal vasculopathy: a review of the clinical, endoscopic, and histopathologic features. Am J Gastroenterol 1992;87: Stanley AJ, Gilmour HM, Ghosh S, Ferguson A, McGilchrist AJ. Transjugular intrahepatic portosystemic shunt as a treatment for protein-losing enteropathy caused by portal hypertension. Gastroenterology 1996; 111: Albillos A, Hera Ad Ade L, Reyes E, Monserrat J, Mu~noz L, Nieto M, et al. Tumour necrosis factor-alpha expression by activated monocytes and altered T-cell homeostasis in ascitic alcoholic cirrhosis: amelioration with norfloxacin. J Hepatol 2004;40: Saitoh O, Sugi K, Lojima K, Matsumoto H, Nakagawa K, Kayazawa M, et al. Increased prevalence of intestinal inflammation in patients with liver cirrhosis. World J Gastroenterol 1999;5: Bellot P, Garcıa-Pagan JC, Frances R, Abraldes JG, Navasa M, Perez-Mateo M, et al. Bacterial DNA translocation is associated with systemic circulatory abnormalities and intrahepatic endothelial dysfunction in patients with cirrhosis. Hepatology 2010;52: Frances R, Mu~noz C, Zapater P, Uceda F, Gascon I, Pascual S, et al. Bacterial DNA activates cell mediated immune response and nitric oxide overproduction in peritoneal macrophages from patients with cirrhosis and ascites. Gut 2004;53: Rosenbloom AJ, Pinsky MR, Bryant JL, Shin A, Tran T, Whiteside T. Leukocyte activation in the peripheral blood of patients with cirrhosis of the liver and SIRS. Correlation with serum interleukin-6 levels and organ dysfunction. JAMA 1995;274: Albillos A, de la Hera A, Gonzalez M, Moya JL, Calleja JL, Monserrat J, et al. Increased lipopolysaccharide binding protein in cirrhotic patients with marked immune and hemodynamic derangement. Hepatology 2003;37: Bota DP, Van Nuffelen M, Zakariah AN, Vincent JL. Serum levels of C-reactive protein and procalcitonin in critically ill patients with cirrhosis of the liver. J Lab Clin Med 2005;146: Ha YE, Kang CI, Joo EJ, Joung MK, Chung DR, Peck KR, et al. Usefulness of C-reactive protein for evaluating clinical outcomes in cirrhotic patients with bacteremia. Korean J Intern Med 2011;26: Moreau R, Jalan R, Gines P, Pavesi M, Angeli P, Cordoba J, et al.; for CANONIC Study Investigators of the EASL- CLIF Consortium. Acute-on-chronic liver failure is a distinct syndrome that develops in patients with acute decompensation of cirrhosis. Gastroenterology 2013;144: