Potential conflict of interest: Nothing to report.
Spontaneous bacterial peritonitis (SBP), a severe complication in patients with advanced liver cirrhosis, has been attributed to bacterial translocation from the intestine. Variants of the NOD2 (nucleotide-binding oligomerization domain containing 2) gene have been associated with impaired mucosal barrier function in Crohn disease. We hypothesized that the risk of acquiring SBP is increased in patients with cirrhosis carrying NOD2 variants. We recruited 150 nonselected patients with liver cirrhosis and ascites admitted to our unit, monitored survival, and recorded the development of SBP prospectively and retrospectively. SBP was defined as the presence of polymorphonuclear neutrophil (PMN) cells >250 per μL of ascitic fluid. Patients were genotyped for the NOD2 variants p.R702W, p.G908R, and c.3020insC. During a median follow-up of 155 days, 54 patients (36%) died and SBP was diagnosed in 30 patients (20%). The occurrence of SBP was increased significantly (P = 0.008) in carriers of NOD2 variants (odds ratio [OR] = 3.06). Retrospectively, SBP was observed in 22 additional patients, and the combined prospective and retrospective analysis substantiated the association between NOD2 and SBP (P = 0.004; OR = 2.98). Of note, carriers of NOD2 risk alleles showed a significantly (P = 0.007) reduced mean survival time (274 days) in comparison to patients with wildtype genotypes (395 days). Conclusion: Common NOD2 variants linked previously to impaired mucosal barrier function may be genetic risk factors for death and SBP. These findings might serve to identify patients with cirrhotic ascites eligible for preemptive antibiotic treatment. (HEPATOLOGY 2010.)
Spontaneous bacterial peritonitis (SBP) is a frequent and severe complication of cirrhosis. As a marker of severe hepatic dysfunction, SBP occurs in up to 30% of patients with cirrhosis and ascites.1 The survival of patients with liver cirrhosis who recover from a first episode of SBP is significantly reduced and despite antibiotic treatment, SBP is still associated with in-hospital mortality rates between 15% and 30%.2–4 The term SBP was coined more than 40 years ago by Conn,5, 6 who speculated that the translocation of intestinal bacteria represents a critical event in the development of SBP. However, genetic factors predisposing to bacterial translocation and SBP have not been identified to date. It has long been anticipated that in addition to intestinal bacterial overgrowth and immune dysfunction, patients at risk for SBP demonstrate increased intestinal permeability, a prerequisite for bacterial translocation from the gut,7–9 which is defined as the migration of bacteria from the intestinal lumen to mesenteric lymph nodes or other extraintestinal sites.4
In 2001, variants of the NOD2 (nucleotide-binding oligomerization domain containing 2) gene (Supporting Fig.) were associated with impaired mucosal barrier function in Crohn disease.10–12 Because NOD2 is involved in the intestinal recognition of bacteria and bacterial products, insufficient activation of NF-κB in carriers of NOD2 risk variants may result in deficient elimination of bacteria and enhancement of their translocation from the intestine.13 It has also been shown that NOD2 variants influence survival in sepsis14 and graft-versus-host disease (GvHD),15 but there is no evidence for a correlation with any liver disease. We hypothesized that the development of SBP in patients with liver cirrhosis is also associated with NOD2 risk variants. The aim of the present study was to assess the potential role of NOD2 as a gene conferring susceptibility to SBP or even death in a large series of patients with cirrhosis and advanced liver cirrhosis and ascites. For the study, we selected those three NOD2 variants (p.R702W, pG908R, and c.3020insC; Supporting Fig.) that are known to confer a deficit in NF-κB activation in response to lipopolysaccharide and peptidoglycan,16 providing evidence for a unifying pathomechanism whereby NOD2 variants confer an increased risk for complications of liver cirrhosis.
Between May 2006 and October 2007 we prospectively recruited 150 patients with liver cirrhosis and ascites who were admitted for hospitalization to the Department of Internal Medicine I, University Hospital Bonn, Germany. Criteria for inclusion were liver cirrhosis and ascites detected by ultrasound. Cirrhosis was diagnosed by clinical, laboratory, and ultrasound findings, or histology if available (fibrosis stage 4). All patients were of Caucasian ethnicity. The severity of the underlying liver disease was assessed according to the Model of End-stage Liver Disease (MELD) and the Child-Pugh score.
The recruitment of the patients was approved by the human research ethics committee of the Medical Faculty at the University of Bonn, and all patients provided informed consent for inclusion in the study. After informed consent, we obtained EDTA-anticoagulated blood and serum samples for standard hematological, biochemical and coagulation tests, and performed the index paracentesis. Ascites was analyzed for the following parameters: total and differential cell counts, total protein and albumin concentrations, pH, glucose and cholesterol levels as well as lactic acid dehydrogenase activity according to standard operational procedures of the clinical chemical laboratory. Ascitic cytology for differential white blood and polymorphonuclear neutrophil (PMN) cell counts was performed on a stained smear made from the sediment of 10 mL centrifuged ascitic fluid.17
Following the criteria of the International Ascites Club,1 SBP was diagnosed when the ascites PMN cell count was >250/μL. During the current and subsequent hospital stays, we monitored the occurrence of SBP and causes of death. Following the consensus criteria for the systemic inflammatory response syndrome (SIRS),18 SIRS was diagnosed retrospectively when at least two of the following criteria were present: temperature <36°C or >38°C, heart rate >90 bpm, and white blood cell (WBC) count <4,000/mm3 or >12,000/mm3; because patients were admitted on our general wards without monitoring respiratory rates or PaCO2, these SIRS criteria could not be considered. In addition, we recorded treatment with antibiotics at the time of admission to our unit. Furthermore, we analyzed the patients' medical history in our hospital and identified previous SBP episodes as defined above.
Bacterial (bact) DNA in ascitic fluid was identified by a multiplex real-time polymerase chain reaction (PCR)-based assay for rapid detection and simultaneous automated differentiation of bactDNA (LightCycler SeptiFast Mgrade, Roche Diagnostics, Mannheim, Germany) employing a protocol validated at our institution and published recently.30, 31
All patients were followed until death, liver transplantation, or the end of our observation period 3 months after the inclusion of the last patient. The median follow-up time was 114 days (range 1-575). Patients receiving liver transplantation were censored on the day of transplantation.
Genomic DNA was extracted from EDTA-anticoagulated blood using a membrane-based extraction kit (Qiagen, Hilden, Germany). DNA concentration was calibrated to 5-20 ng/μL, using a NanoDrop ND-1000 spectrophotometer (Peqlab, Erlangen, Germany).
The NOD2 gene variants (rs2066844 [p.R702W], rs2066845 [p.G908R], rs2066847 [c.3020insC]; Supporting Fig.) were genotyped using solution-phase hybridization reactions with 5′-nuclease and fluorescence detection (TaqMan assays) on the 7300 Real-Time PCR System (Applera, Norwalk, CT). PCR reactions contained 20 ng genomic DNA, 1× Platinum qPCR SuperMix-UDG master mix (Invitrogen, Karlsruhe, Germany) 900 nM of each primer, and 200 nM of VIC-labeled and FAM-labeled probes, respectively, in 25-μL reactions. Amplification conditions were 95°C for 10 minutes, followed by 45 cycles at 95°C for 15 seconds and 60°C for 60 seconds. Primer and probe sequences were: p.R702W, MGB_F CTGAGTGCCAGACATCTGAGAAG, MGB_R GCTGCGGGCCAGACA, VIC CCTGCTCTGGCGCC, FAM CTGCTCCGGCGCC; p.G908R, MGB_F TGATCACCCAAGGCTTCAGC; MGB_R GAACACATATCAGGTACTCACTGACAC; VIC ACTCTGTTGCG- CCAGA; FAM CTGTTGCCCCAGAAT; c.3020insC, MGB_F CCAGGTTGTCCAATAACTGCATC; MGB_R CCTTACCAGACTTCCAGGATGGT; VIC TGCAGGCCCCTTG; FAM CTGCAGGCCCTTG. Selected results of TaqMan assays were ascertained by direct BigDye termination cycle sequencing on the ABI PRISM 310 Genetic Analyzer (Applera).
Statistical analysis was performed with SPSS 13.0 (SPSS, Munich, Germany). Data are given as medians and ranges.
Differences of survival between carriers of different genotypes were analyzed by Kaplan-Meier statistics (log-rank test). To test for independence of risk factors on survival, we performed multivariate regression analysis. Candidate variables that entered the univariate analysis were age, gender, serum albumin, serum bilirubin, platelet count, serum creatinine, total protein in serum, MELD score, the presence of any NOD2 risk allele, and SBP. Significant univariate risk factors entered the multivariate regression analysis, which was performed with an incrementally forward stepwise approach. Probabilities were set at 0.05.
An exact test was used to check whether genotype frequencies are consistent with Hardy-Weinberg equilibrium, indicating that alleles are distributed by random mating and remain constant in the given population. Allele and genotype frequencies were compared between cases and controls by Pearson's goodness-of-fit χ2 test and Armitage's trend test, respectively (http://ihg.gsf.de/ihg/snps.html).
Power calculations were performed using PS (http://biostat.mc.vanderbilt.edu/wiki/Main/PowerSampeSize). We calculated that the study was designed to detect a significantly increased odds ratio (OR) of 2 with a power of 80%, based on risk allele frequencies of 0.10 and type I error rates of 0.05. For the survival analysis, the study had a power of 99%, given an expected mortality rate of 30% that we expected on the basis of our previous study with comparable patients.19
Patient Characteristics and SBP Frequency.
Table 1 illustrates that the majority of our patients presented with ascites due to alcoholic liver cirrhosis (63%) or chronic viral hepatitis (17%). All other etiologies were less common. Most patients were males (72%), and the median age of all patients was 57 years (range 23–86). As expected, the majority of patients showed advanced liver disease, with 94% of the patients classifying for Child-Pugh scores B or C. The median MELD score at the time of admission was 16 (range 5–40). C-reactive protein levels at the time of paracentesis did not differ between patients with NOD2 risk alleles (29 ± 31 mg/dL) variants and those without (28 ± 30 mg/dL).
Table 1. Clinical Characteristics of Patients with Cirrhosis and Ascites
INR, international normalized ratio; MELD, Model for End-Stage Liver Disease.
Median age years (range)
57 (23 – 86)
Etiology of cirrhosis n (%)
Chronic hepatitis B/C virus infection
Autoimmune hepatitis, chronic cholestatic
liver diseases and vascular diseases
Cryptogenic liver cirrhosis
Median INR (range)
1.3 (0.9 – 4.7)
Median total serum bilirubin mg/dL (range)
2.03 (0.24 – 38.79)
Median serum creatinine mg/dL (range)
1.31 (0.38 – 14.93)
Median MELD score (range)
16 (5 – 43)
Child-Pugh score n (%)
Antibiotic treatment at time of admission
n (% of total)
Employing PMN cell count >250/μL as a diagnostic criterion, 14 patients (9% of the cohort) were diagnosed with SBP on the first day of admission (index paracentesis), 16 patients (11%) in the follow-up period (median 155 days, range 1–575), and 22 patients (15%) during previous hospital stays.
NOD2 Risk Variants Negatively Affect Survival.
For the survival analysis, patients who were lost to follow-up (n = 2) were excluded and patients who received a liver transplant (n = 7) were right-censored. In total, 54 patients (36%) died. Table 2 summarizes the causes of death for these patients, with acute-on-chronic liver failure and infections representing the most common conditions.
Table 2. Causes of Death
SIRS, systemic inflammatory response syndrome.
Acute-on-chronic liver failure
Total number of deaths
Table 3 summarizes the minor allele frequencies for the three NOD2 variants in the different groups of patients, all of which were in the range or slightly above previously reported frequencies in Caucasian healthy cohorts.20 No patients carried homozygous NOD2 mutations. The NOD2 genotype distributions in the total cohort were consistent with Hardy-Weinberg equilibrium (all P > 0.05). In our cohort, 19 patients carried the variant p.R702W, 10 carried p.G908R, and eight tested positive for c.3020insC (Supporting Fig.).
Table 3. Distribution of Risk Alleles of NOD2 Variants
Figure 1 demonstrates that patients who tested positive for any of the NOD2 risk alleles showed a significantly (P = 0.007, log-rank test) reduced survival in comparison to patients with wildtype genotypes at all three positions. The mean survival time in patients who carried any NOD2 risk allele was 274 days in comparison to 395 days in patients who carried wildtype genotypes.
By univariate analysis, we demonstrated that carrier status for any NOD2 risk allele, male gender, MELD score, the development of SBP, and serum albumin were risk factors for death (all P < 0.05), whereas age or other risk indicators (platelet count, serum bilirubin, serum creatinine, total protein) were not (Table 4A). Multivariate regression analysis demonstrated that the detection of any NOD2 risk variant was the strongest independent predictor of death with an OR of 4.32 (Table 4B).
Table 4. Regression Analysis of Risk Factors for Death
(A) Univariate Analysis
(including all significant parameters from the univariate analysis)
Overall, eight (5.3%) patients fulfilled the criteria of SIRS at time of admission. Of these, three patients had NOD2 risk variants, and five patients carried no NOD2 variants. During follow-up, nine patients died due to SIRS (Table 2; four patients with and five patients without NOD2 variants).
NOD2 Risk Variants Are Associated with SBP.
Table 5A demonstrates that patients carrying a risk allele of any of the three NOD2 variants developed SBP, defined as PMN cell count >250/μL, more frequently at index paracentesis or during follow-up (OR = 3.06, 95% confidence interval [CI] = 1.31–7.15, P = 0.008). In addition to these 30 patients (Table 5A), SBP was documented in 22 additional patients before inclusion in the study. Table 5B shows that the combined prospective and retrospective analysis of both groups together substantiated the association between NOD2 and SBP and indicated a similarly increased risk of SBP (OR = 2.98; 95% CI = 1.39–6.40, P = 0.004). Although the risk allele frequencies of all NOD2 variants tended to be higher in patients with SBP (Table 3), no statistically significant differences were observed for individual NOD2 SNPs.
Table 5. Numbers of Carriers of Any NOD2 Risk Allele in Patients With Cirrhosis With and Without SBP (PMN Cell Count >250/μL)
In ascites samples from 15 patients (10%), we detected bactDNA, and four of these patients carried NOD2 risk alleles. However, contingency table analysis did not support an association between NOD2 genotypes and the presence of bactDNA.
In this study we report a significant association of SBP with common risk variants of the NOD2 gene in patients with advanced liver cirrhosis and ascites. Importantly, the study demonstrates that the NOD2 variants confer a substantially increased risk for death in patients with cirrhosis.
Increased bacterial translocation in cirrhosis has been attributed to structural changes of the intestinal mucosa including dilatation of intercellular space and vascular congestion as well as mucosal oxidative damage.4, 8, 21 Furthermore, innate and adaptive immune responses normally limit the penetration of bacteria across the epithelium, but activation and transmigration of PMN cells cause the release of proinflammatory cytokines such as tumor necrosis factor alpha (TNF-α) and interferon gamma (IFN-γ) as well as nitric oxide, all of which modulate bacterial translocation.4, 8, 9, 22 In the present study we identified NOD2 variants as novel inherited risk factors for SBP in liver cirrhosis. In our prospective analysis, the OR for SBP was about 3 (Table 5A), and this high ratio is in the range of other disease risks conferred by the NOD2 variants.10, 23 Although the results were corroborated by the combined prospective and retrospective analysis that took both occurrence of SPB during follow-up and previous history of SBP into account and resulted in a similar OR (Table 5B), our study warrants further evaluation and needs to be replicated in an independent population. Because about a third of the patients received antibiotics on admission (Table 1), it is likely that we underestimated the frequency of SBP and biased the result against genetic effects.
The genetic study indicates that the local immune system in carriers of the NOD2 variants may be incapable of limiting bacterial translocation from the intestine. NOD2 is expressed in intestinal epithelial and mononuclear cells and represents an intracellular “pattern recognition receptor” that senses the muramyl dipeptide component of bacterial cell walls and leads to activation of the proinflammatory NF-κB signaling pathway.13, 24 The NOD2 risk variants associated with Crohn disease reduce the ability of NOD2 to activate NF-κB25, and a decrease in NF-κB-induced inflammatory response and intestinal production of antimicrobial peptides such as α-defensins26 may favor bacterial translocation and systemic inflammation. Consistent with this paradigm, the NOD2 risk variants might also be associated with gastrointestinal GvHD after allogeneic stem cell transplantation, another condition with increased bacterial translocation.15 The presence of bacterial DNA (bactDNA) in blood was reported in 40% of patients with cirrhosis and considered to be evidence of bacterial translocation.27, 28 Recently, the presence of bactDNA in blood and ascitic fluid was shown to define subgroups of patients with poor prognosis, with acute-on-chronic liver failure representing the most common cause of death.22, 29 Using a multiplex real-time PCR-based assay for rapid detection and differentiation of bactDNA validated recently by us and other groups,30, 31 we did not detect an association between NOD2 variants and the presence of bactDNA in ascites.32 However, these data have to be compared to ascites analyses based on other DNA tests,21, 26, 27 and it has yet to be resolved whether bactDNA represents a reliable surrogate marker for local and/or systemic infection in patients with cirrhosis.
In the present study the three NOD2 variants were plainly associated with risk of death in our patients with advanced cirrhosis (OR 4.3), with acute-on-chronic failure again being the most frequent cause of death (Table 2). Accordingly, we note that the NOD2 variants might impair survival not only by impaired mucosal barrier function but by extraintestinal mechanisms, because NOD2 is also expressed in hepatocytes and immune cells in liver, stimulating cellular responses to muramyl dipeptide and hepatic IFN-γ and NF-κB signaling.33 As highlighted above, the NOD2 variants have been linked to mortality in GvHD, partially explained by pulmonary failure,15 and in sepsis,14 a frequent cause of death in patients with advanced cirrhosis.34 SIRS is associated with poor in-hospital outcome in patients with advanced cirrhosis,35 and in the present study the frequency of NOD2 risk alleles tended to be higher in patients with SIRS as compared to the total cohort, but the small number of SIRS patients precluded statistical analysis.
Because survival rates of SBP have improved in recent years, prevention has come into focus, in particular in patients who do not yet meet the conventional criteria for increased SBP risk, such as a previous episode of SBP, gastrointestinal hemorrhage, or low ascites protein levels.1, 9, 36 Recently a randomized study that included patients with low ascites protein concentrations37 and a meta-analysis38 have indicated that antibiotic prophylaxis reduces SBP occurrence and improves short-term survival in high-risk patients with cirrhosis and ascites. Reducing the bacterial load in the gut by selective intestinal decontamination decreases the frequency of SBP, but increases the risks for infections with antibiotic-resistant bacteria39, 40 or posttransplant fungal infections41; thus, prophylaxis is currently restricted to patients at highest risk of SBP, as defined above.9 Now that we might have identified a new subpopulation of patients with cirrhosis at high risk for both SBP and death, this clearly sets the stage for a prospective study of primary prophylaxis of SBP to see whether long-term survival can be further improved in genetically defined at-risk patients.
The authors thank all patients for participating in this study and providing blood samples, and Stephanie Schwartz and Hildegard Keppeler for excellent technical assistance. Contributions: B.A. and F.G. contributed equally to data acquisition and analysis. M.G. and L.T. helped in study design, collected data and recruited patients as part of their doctoral theses. T.S. and F.L. were responsible for study concept, design, and supervision. B.A. and F.G. drafted the article, which was edited by T.S. and F.L. This study was presented in part as Presidential Poster of Distinction at the Annual Meeting of the American Association for the Study of Liver Diseases (AASLD), San Francisco, November 4, 2008, and published in abstract form (HEPATOLOGY 2008;48:A1676).