Potential conflict of interest: Nothing to report.
Presence of bacterial DNA in noninfected patients with cirrhosis and ascites is associated with a marked inflammatory response including activation of the inducible form of nitric oxide synthase and release of nitric oxide, similar to that observed in patients with spontaneous bacterial peritonitis. Although presence of bacterial DNA is associated with an impaired prognosis, no information is available regarding its hemodynamic consequences. Systemic and hepatic hemodynamics before and after a liquid test meal were assessed in a series of 75 noninfected patients with cirrhosis (55 with ascites). Bacterial DNA was measured by polymerase chain reaction. Bacterial DNA was detected only in patients with ascites. Clinical data and liver function were similar in ascitic patients with presence (n = 21) or absence of bacterial DNA (n = 34). Bacterial-DNA(+) patients had significantly lower mean arterial pressure (P = 0.002) and systemic vascular resistance (P = 0.03) than bacterial-DNA(−) patients. Cardiac output, cardiopulmonary pressures, hepatic venous pressure gradient (HVPG), and hepatic blood flow were similar in both groups. Thirty minutes after the test meal, in response to increased blood flow caused by postprandial hyperemia, there was a significantly greater increase in HVPG and impaired hepatic vasorelaxation in bacterial-DNA(+) as compared with bacterial-DNA(−) patients, which indicates hepatic endothelial dysfunction. Indeed, the increase in HVPG after the test meal significantly correlated with serum bacterial DNA concentration. Conclusion: Presence of bacterial DNA, a marker of bacterial translocation, is associated with aggravation of peripheral vasodilation and with worsening of intrahepatic endothelial dysfunction. (HEPATOLOGY 2010;.)
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Portal hypertension is a serious consequence of cirrhosis that can result in life-threatening complications with increased mortality and morbidity.1 The primary factor in the pathophysiology of portal hypertension is increased intrahepatic resistance to portal-collateral blood flow. Portal hypertension is further aggravated by increased portal venous inflow, caused by splanchnic vasodilation. Moreover, insufficient nitric oxide (NO) availability in the hepatic microcirculation is considered an important factor that contributes to increase the hepatic vascular resistance. Because of this, the cirrhotic liver, unlike the normal liver, cannot vasodilate in response to a volume flow load such as that caused by meals, which results in abrupt postprandial increases in portal pressure, a concept known as intrahepatic endothelial dysfunction.2-5
Bacterial translocation (BT), defined as the passage of viable bacteria from the gut to mesenteric lymph nodes (MLNs) and/or other extraintestinal sites, has been associated with a worsening of arterial and splanchnic vasodilation in animal models of cirrhosis and ascites.6, 7 Splanchnic vasodilation is mediated by increased NO production in the splanchnic vasculature that could be stimulated directly by bacterial products, acting either on the endothelial and inducible forms of NO synthase8, 9 or involving the activation of proinflammatory cytokines. Our group has previously demonstrated that the presence of bacterial DNA (bactDNA) in culture-negative MLNs in an animal model of cirrhosis may be considered as a surrogate marker of BT, which is associated with a local inflammatory response similar to that found in culture-positive MLNs10 or in patients with spontaneous bacterial peritonitis.11 Detection of bacterial DNA in the serum closely reflects bacterial DNA in MLNs and is therefore considered as a marker of BT.10 Furthermore, experimental studies have demonstrated an increased intrahepatic vascular tone in cirrhotic livers exposed to endotoxin.12, 13 Therefore, there is a rational basis to hypothesize that BT could aggravate portal hypertension by increasing portal venous inflow and the intrahepatic vascular resistance.
This prospective investigation in a cohort of consecutively admitted patients was undertaken to assess the basal and meal stimulated hemodynamics in noninfected patients with portal hypertension according to the presence of bactDNA.
A consecutive series of 79 inpatients with cirrhosis at the Liver Unit and referred to the Hepatic Hemodynamic Laboratory for clinical and hemodynamic evaluation of portal hypertension from August 2004 to November 2007 were considered for this investigation. All patients had cirrhosis diagnosed by clinical, biological, ultrasonographical, or histological criteria. The protocol was approved by the Hospital Clinic of Barcelona Ethics Committee (Comité de Ética y Ensayos Clínicos) and followed the principles of the Declaration of Helsinki. Written informed consent to participate in the study was obtained from each patient.
Inclusion criteria were age between 25 and 80 years and confirmed diagnosis of cirrhosis. Exclusion criteria were evidence of gastrointestinal bleeding; portal vein thrombosis; diffuse or multinodular hepatocellular carcinoma not fulfilling Milano criteria; cardiac, renal or respiratory failure; previous surgical or intrahepatic portosystemic shunt; prescription of vasoactive drugs including beta-blockers and/or investigational drugs; bacterial infection or treatment with antibiotics in the preceding 2 weeks; or positive blood or ascitic fluid culture previous to hepatic hemodynamic study or presence of neutrocytic ascites (polymorphonuclear cell count >250 cells/mm3). Bacterial infection was ruled out by clinical history, physical examination, laboratory analysis, and both blood and ascitic fluid cultures performed in blood culture bottles.14 Other associated morbidities were excluded by clinical history, physical examination, electrocardiogram, and routine biochemical analysis. Per protocol, inclusion in the nonascites group was closed after 20 patients had been included.
Patients were maintained on a sodium-restricted diet and diuretics were withdrawn for 2 days before the hemodynamic study. No large-volume paracentesis was allowed in the preceding 5 days. After fasting overnight, patients were transferred to the Hepatic Hemodynamic Laboratory. Under local anesthesia, an 8-French venous catheter introducer (Axcess; Maxxim Medical, Athens, TX) was placed in the right jugular vein under ultrasonographic guidance (SonoSite Inc, Bothell, WA) using the Seldinger technique. Under fluoroscopic control, a Swan-Ganz catheter (Edwards Laboratory, Los Angeles, CA) was advanced into the pulmonary artery for measurement of cardiopulmonary pressures and cardiac output (CO) by thermal dilution. A 7F balloon-tipped catheter (Medi-Tech; Boston Scientific Cork, Ltd., Cork, Ireland) was then advanced in to the main right hepatic vein to measure wedged and free hepatic venous pressures as previously described.2, 5, 15 All measurements were performed in triplicate in each study period, and permanent tracings were obtained on a multichannel recorder (Marquette Electronics, Milwaukee, WI). Portal pressure was estimated from the hepatic venous pressure gradient (HVPG), the difference between wedged and free hepatic venous pressure. The hepatic vascular resistance (dyne/second/cm−5) was estimated as: HVPG (mm Hg) × 80/hepatic blood flow (HBF) (L/minute).2, 5
Preceded by a priming dose of 5 mg, a solution of indocyanine green (Pulsion Medical Systems, Munich, Germany) was infused intravenously at a constant rate of 0.2 mL/minute. After an equilibration period of at least 40 minutes, four separate sets of simultaneous samples of peripheral and hepatic venous blood were obtained for the measurement of HBF as previously described.16 To avoid interferences from differences in plasma turbidity, the Nielsen correction was used.15 Mean arterial pressure (MAP) was measured every 5 minutes by a noninvasive automatic sphygmomanometer (Marquette Electronics, Milwaukee, WI). Heart rate was derived from continuous electrocardiogram monitoring. Patients with an indocyanine green fractional clearance lower than 0.1 were excluded for measurement of HBF.
After completing baseline hemodynamic measurements, 57 of the patients included received a mixed liquid meal (400 mL) containing 26 g proteins, 74 g carbohydrates, and 19 g lipids for a total of 600 kcal (Ensure Plus; Abbot Laboratories BV, Zwolle, the Netherlands), which was ingested within approximately 5 minutes. The systemic and splanchnic response to the test meal was evaluated at 30 minutes, when maximal postprandial hyperemia and increase in HVPG has been demonstrated to occur.17-20 HVPG was also measured at 15 minutes.
Blood samples from peripheral vein and hepatic vein were taken at baseline and collected into endotoxin-free tubes (Endo Tube ET; Chromogenix AB, Sweden) centrifuged, and plasma samples stored at −80°C until analysis. All samples were processed in airflow chambers and tubes were never exposed to free air.
Quantification and Identification of Amplified Bacterial DNA Fragments.
Detection of bactDNA was performed as previously described.21 Briefly, a sample of 200 μL of plasma was incubated in a lysozyme-proteinase K buffer for 2 hours and placed into QIAamp Spin Columns (Qiagen, Hilden, Germany). A broad-range polymerase chain for the conserved region of the 16S ribosomal RNA prokaryote gene was carried out using the following universal primers: 5′-AGAGTTTGATCATGGCTCAG-3′ and 5′-ACCGCG ACTGCTGCTGGCAC-3′. Total polymerase chain reaction volume was filtered with QIAquick Spin Columns (Qiagen, Hilden, Germany) before nucleotide sequencing with ABI-Prism Dye Terminator Cycle Sequencing version 2.0 Ready Reaction Kit and ABI-Prism 310 automated sequencer (Applied Biosystems, Foster City, CA), according to the manufacturer's instructions. The identification of sequences was carried out by BLAST at the National Center for Biotechnology Information Web site (www.ncbi.nlm.nih.gov). Technical details of the method, including accuracy, precision, linearity, and reproducibility, are describe elsewhere.21
Quantification of Serum Cytokine Levels.
Enzyme-linked immunosorbent assays for the quantitative measurement of tumor necrosis factor alpha (TNF-α), and interleukin-12 (IL-12) levels as representative cytokines of the proinflammatory immune response were performed in basal plasma of patients using Human Quantikine kits (R&D Systems, Minneapolis, MN) according to the manufacturer's instructions. All samples were tested in triplicate and read 490 nm in a Thermomax microplate reader (Molecular Devices, Sunnyvale, CA). The lower limits of detection of all cytokine assays were 5-10 pg/mL. Standard curves were generated for each plate, and the average zero standard optical densities were subtracted from the rest of standards controls and samples to obtain a corrected concentration for all cytokines.
Measurement of Nitric Oxide Products and Plasma Renin Activity.
The sum of the NO metabolites nitrite (NO2−) and nitrate (NO3−) is widely used as an index of NO metabolite (NOx) generation and is expressed as NOx levels.22 NOx levels in plasmatic samples were calculated by measuring conversion of NO3− to NO2− by the enzyme nitrate reductase via enzyme-linked immunosorbent assay (R&D Systems) based on the Griess reaction that absorbs visible light at 540 nm. All samples were tested in triplicate; standard curves were generated for each plate, and the average zero standard optical densities were subtracted from the rest of standards, controls and samples to obtain a corrected NOx concentration. Plasma renin activity (PRA) was determined by means of radioimmunoassay (Clinical Assay; Baxter, Cambridge, MA) as described.23
In all patients blood cultures were obtained from a venous catheter introducer placed in the right jugular vein. Blood culture bottles (BACTEC 9050 Aerobic Plus F and Anaerobic Plus F bottles, Becton-Dickinson) were incubated in a BACTEC 9240 system (Becton-Dickinson). All bottles were incubated for a minimum of 5 days according to the manufacturer's instructions. When a positive signal was obtained, bottles were removed and an aliquot of the broth was Gram-stained and processed for organism identification.
The Kolmogorov-Smirnov test was used to assess the normality of the distribution of continuous variables. Comparison between groups was performed by ANOVA and Student t test for paired data with normal distribution, whereas the Mann-Whitney U test was used in the nonnormally distributed variables. Qualitative data were compared by chi-squared test with Yates' correction. Results are shown as mean ± standard deviation. Correlation was performed by means of Pearson's coefficient. Statistical significance was established at P < 0.05. Statistical analysis was performed using SPSS 17.0 statistical package (SPSS Inc., Chicago, IL).
Two out of the 79 patients initially evaluated were excluded due to positive blood cultures (Streptococcusviridans and Staphyloccusepidermidis, respectively); one patient was excluded due to a previously unknown hypertrophic myocardiopathy, diagnosed after Swan-Ganz catheterization, and one patient was excluded because he was receiving an investigational drug. Therefore, 75 patients were finally included in the study, 55 with ascites and 20 without. bactDNA was only detected in patients with ascites (in 38%; 21/55 patients). bactDNA was from gram-negative bacteria (GNB) in 16 cases and from gram-positive cocci (GPC) in the remaining five cases. Bacterial species identified by automatic nucleotide sequencing were: Escherichia coli (n = 11), Klebsiellapneumoniae (n = 5), Enterococcusfaecalis (n = 2), and Staphylococcusaureus (n = 3). Figure 1 shows the flow chart of the study.
Patients with ascites were divided into two investigational groups according to the presence or absence of bactDNA. The two groups showed distinct immune and hemodynamic profiles (see below), but were otherwise not clinically distinguishable (Table 1). The mean age and sex distribution in both groups were similar. Cirrhosis was secondary to alcohol abuse and to hepatitis C virus infection in the majority of patients in each group. The Child-Turcotte-Pugh score and the MELD score were similar regardless of the bactDNA status. Presence and size of esophageal varices was similar in both groups of patients (Table 1).
Table 1. Clinical Characteristics of the Patients
Patients With Cirrhosis and Ascites(n = 55)
Patients Without Ascites (n = 20) All bact DNA (−)
bactDNA(+) (n = 21)
bactDNA(−) (n = 34)
NOTE: Child-Turcotte-Pugh score is represented as median and range. Other results are expressed as mean ± SD. No significant differences were observed between groups in any of the reported parameters.
Abbreviations: bactDNA, bacterial DNA; INR, international normalized ratio; MELD, model for end-stage liver disease; SD, standard deviation.
P < 0.05 in bactDNA(+) versus patients with cirrhosis without ascites.
P < 0.05 in bactDNA(−) versus patients with cirrhosis without ascites.
Three out of 75 patients had small hepatocellular carcinoma, fulfilling the Milano criteria for liver transplantation. bactDNA was detected in one of them (E. coli).
All included patients gave signed informed consent to the study, but 18 patients did not consent to the test meal and postprandial hemodynamic measurements. Therefore, complete baseline data is available from all 75 patients and the baseline and postprandial data is available for 57 cases.
Patients with bacterial DNA had more profound systemic vasodilation, as shown by significantly lower MAP and systemic vascular resistance (SVR) than bactDNA(−) patients (Table 2). There were no statistical differences in CO, heart rate, and stroke volume between bactDNA(+) and bactDNA(-) patients. Baseline HVPG and HBF were similar in both groups (Table 2).
Table 2. Baseline Systemic and Splanchnic Hemodynamic, Inflammatory Markers, and Endogenous Vasoactive Systems According to bactDNA Status and Presence of Ascites
In the whole series, the test meal induced a significant increase in HBF (12% ± 8%, P < 0.001) and HVPG (17% ± 15%, P < 0.001), but not in MAP, CO, SVR, and heart rate. The increase in HVPG was mainly due to an increase in wedged hepatic venous pressure (Table 3). Interestingly, the test meal induced an almost double increase in HVPG in bactDNA(+) than in bactDNA(−) patients, either with or without ascites (ΔHVPG = 16% ± 9% versus 9% ± 6%; P = 0.008 versus 9% ± 8%; P = 0.02, respectively) (Fig. 2). This was regardless of bactDNA being from GNB or GPC (Table 4). The increase of HBF after food intake in the three groups of patients was similar (19% ± 12% in bactDNA(+) versus 23% ± 17% in ascitic bactDNA(−), P = 0.5; versus 12% ± 13% in nonascitic bactDNA(−), P = 0.3) (Fig. 2).
Table 3. Postprandial Changes in Splanchnic and Systemic Hemodynamics
Table 4. Systemic and Splanchnic Hemodynamics, Inflammatory Markers and Endogenous Vasoactive Systems in bactDNA(+) Patients From Gram-Negative Bacteria (bactDNA+GNB) or bactDNA(+) Patients from Gram-Positive Bacteria (bactDNA+GPB)
Estimated hepatic vascular resistance decreased after meal in bactDNA(−) patients (−27% ± 34%), whereas it remained unchanged in bactDNA(+) patients (−6% ± 28%), this difference approaching statistical significance (P = 0.08).
Among bactDNA(+) patients there was a significant correlation between the postprandial increase of HVPG and bactDNA concentration (Fig. 3). However, no correlation was found between serum bactDNA concentration and baseline splanchnic or systemic hemodynamic parameters. Also no statistically significant differences were observed in systemic and splanchnic hemodynamic parameters between those with presence of GNB-driven or GPC-driven bacterial genomic fragments (Table 4).
Laboratory and Hemodynamic Parameters.
bactDNA(+) patients showed significantly higher mean circulating values of proinflammatory cytokines (TNF-α, IL-12), NOx, and plasma renin activity (PRA) than did ascitic bactDNA(−) patients and patients without ascites (Table 2). TNF-α levels were inversely correlated with MAP (r = −0.34, P < 0.05) and SVR (r = −0.33, P < 0.05)
The present study shows that the presence of bactDNA in patients with ascites and portal hypertension aggravates the systemic circulatory dysfunction, further exacerbating the peripheral vasodilation, which is related to increased TNF-α levels. Moreover, presence of bacterial DNA was associated with a more severe intrahepatic endothelial dysfunction, as suggested by a greater increase in HVPG in response to the postprandial hyperemia induced by a standardized test meal. The behavior of patients without ascites was analogous to that observed in patients with ascites without evidence of bacterial translocation.
Previous investigations from our group have shown that translocation of bacterial DNA is a frequent event in patients with ascites,21 which is supported by the information provided by the current study. In fact, none of the nonascitic patients evaluated showed the presence of bactDNA, whereas this was found in 38% of our patients with ascites, a value similar to that reported.11, 24 The fact that bactDNA was not detected in patients with cirrhosis without ascites is in line with previous studies in experimental models of cirrhosis linking bacterial translocation with the presence of ascites.25
It has been suggested that bacterial translocation to MLNs, in the absence of systemic spread of viable bacteria, may lead to an increase of systemic proinflammatory cytokines and increased NO synthesis in the splanchnic circulation, worsening the hemodynamic abnormalities in cirrhosis.6, 26-28 In agreement with this, in this prospective study we show that the presence of bactDNA in serum, a surrogate marker of BT, identifies a group of patients with cirrhosis with a different hemodynamic and immunological profile. bactDNA(+) patients exhibit a marked inflammatory response, as represented by increased levels of TNF-α and IL-12. These levels are significantly higher than those of patients without traces of BT, confirming previous investigations from our group.11, 29 This proinflammatory state in bactDNA(+) is associated with a marked systemic hemodynamic derangement, characterized by lower MAP and lower SVR, a situation likely related to the increased levels of NOx and representing an aggravation of the peripheral vasodilatation of cirrhosis and ascites (Table 2).
Interestingly, the CO was slightly, albeit not significantly, higher in bactDNA(+) patients. The bactDNA(+) patients also showed an enhanced activation of endogenous vasoactive systems (PRA and NOx) associated with more profound disturbance in the hyperdynamic circulatory state. These observations are at odds with the suggestion that progression of circulatory dysfunction leads in advanced stages to a decrease in CO due to so-called cirrhotic cardiomyopathy.30, 31 In our study, despite the marked differences in TNF-α and other proinflammatory mediators, we could not demonstrate any decrease in CO in bactDNA(+) patients. These findings suggest that a much greater increase in proinflammatory cytokines would be required to cause the impaired cardiac contractility reported in patients developing hepatorenal syndrome. On the other hand, our findings further show that such worsened systemic hemodynamics does not translate to any elevation of baseline HVPG, which is somewhat contrary what could be anticipated from the results in experimental models.12, 13, 32
The worsening of systemic hypotension and peripheral vasodilatation observed in bactDNA(+) patients may be mediated by the increased release of proinflammatory cytokines, as suggested by the highly significant correlations observed between MAP and SVR with the plasma levels of TNF-α, which is a widely known stimulator of inducible and endothelial NO synthase activity.32, 33 Systemic hemodynamic parameters and plasma bactDNA concentration, however, were not statistically related, which may be due in part to the fact that translocation of other bacterial products besides bacterial DNA could play a role in the inflammatory response and in the hemodynamic disturbances of cirrhosis. Moreover, monocytes in cirrhosis seem to be primed by bacterial products for release of cytokines.34-37
A second important finding of our study was that bactDNA(+) patients had a more profound abnormality of the intrahepatic circulation, in the sense of worse hepatic endothelial dysfunction, as suggested by their greater postprandial increase in HVPG as compared with patients without detectable traces of bactDNA. This may be attributable to the effect of translocation of bacterial products in the regulation of hepatic vascular tone, because similar findings have been documented in experimental studies showing an increased intrahepatic vascular tone in cirrhotic livers exposed to endotoxin.12, 13, 38 Moreover, experimental studies have demonstrated that short-term exposure to synthetic oligonucleotides containing CpG motifs or fragments of bacterial DNA results in marked deterioration of the hepatic microcirculation.39 This is likely to occur at sinusoidal and postsinusoidal sites, as suggested by the much lower decrease in hepatic vascular resistance in response to increased liver perfusion in bactDNA(+) patients as compared with patients with undetectable bacterial DNA fragments.
The systemic and splanchnic hemodynamic disturbances were independent of the bacterial species detected. Patients with presence of bactDNA from GNB showed a similar inflammatory response than those with bactDNA from GPC. This is in line with previous studies showing an important role of GPC in the inflammatory response observed in advanced cirrhosis. In fact, Riordan et al.40 demonstrated a significant correlation between Toll-like receptor 2 (TLR-2) expression on blood mononuclear cells and serum TNF-α levels, suggesting a relevant role of gram-positive microbial components in the inflammatory status and systemic circulatory dysfunction in cirrhosis. These findings, however should be taken with caution due to the relatively low number of patients with bactDNA from GPC in our series, which accounted for one-fourth of the bactDNA(+) cases. However, it should be pointed out that these patients with bactDNA(+) from GPC, although a relatively small fraction of the total, would have not been detected if bacterial translocation would have been looked for by other techniques, such as measuring lipopolysaccharide or lipopolysaccharide binding protein.29, 41
In summary, our results support the hypothesis that presence of plasma bactDNA, a surrogate marker of bacterial translocation, contributes to the systemic hemodynamic derangements in patients with cirrhosis with ascites. The results of the current study gives further support to the possibility of exploring selective intestinal decontamination in patients with cirrhosis with bactDNA(+) as a adjunctive therapy for portal hypertension. Moreover, this study also supports the idea that bacterial translocation could worsen intrahepatic endothelial dysfunction in cirrhosis, which determines a greater postprandial increase in HVPG. The relevance of this latter finding is unknown, although it has been suggested that clinical or subclinical bacterial infections may contribute to acute variceal bleeding and early rebleeding.42-44
We thank Ms. M.A. Baringo, L. Rocabert, and R. Saez for their expert technical assistance, and C. Esteva for editorial support.