TLR4/CD14 Variants-Related Serologic and Immunologic Dys-Regulations Predict Severe Sepsis in Febrile De-Compensated Cirrhotic Patients

Genetic variants and dysfunctional monocyte had been reported to be associated with infection susceptibility in advanced cirrhotic patients. This study aims to explore genetic predictive markers and relevant immune dysfunction that contributed to severe sepsis in febrile acute de-compensated cirrhotic patents. Polymorphism analysis of candidate genes was undergone in 108 febrile acute de-compensated cirrhotic patients and 121 healthy volunteers. Various plasma inflammatory/regulatory cytokines, proportion of classical (CD 16-, phagocytic) and non-classical (CD16+, inflammatory) monocytes, lipopolysaccharide (LPS)-stimulated toll-like receptor 4 (TLR4) and intracellular/extracellular cytokines on cultured non-classical monocytes, mCD14/HLA-DR expression and phagocytosis of classical monocytes were measured. For TLR4+896A/G variant allele carriers with severe sepsis, high plasma endotoxin/IL-10 inhibits HLA-DR expression and impaired phagocytosis were noted in their classical monocyte. In the same group, increased non-classical monocyte subset, enhanced LPS-stimulated TLR4 expression and TNFα/nitrite production, and systemic inflammation [high plasma soluble CD14 (sCD14) and total nitric oxide (NOx) levels] were noted. For CD14-159C/T variant allele carriers with severe sepsis, persist endotoxemia inhibited mCD14/HLA-DR expression and impaired phagocytosis of their classical monocyte. In the same group, increased non-classical monocyte subset up-regulated TLR4-NFκB-iNOS and p38MAPK pathway, stimulated TNFα/nitrite production and elicited systemic inflammation. In febrile acute de-compensated cirrhotic patients, TLR4+896A/G and CD14-159C/T polymorphisms-related non-classical and classical monocytes dysfunction resulted in increased severe sepsis risk. Malnutrition, high plasma endotoxin and sCD14 levels, single TLR4+896A/G or CD14-159C/T variant allele carriers and double variant allele carriers are significant predictive factors for the development of severe sepsis among them.


Introduction
Acute de-compensated cirrhotic patients are suffered from compromise hepatic immune surveillance, impaired phagocytosis and high severe sepsis-related mortality [1]. Poor response to bacterial challenge, increased susceptibility to bacterial infection, high severe sepsis-related mortality had been reported in advanced cirrhotic patients [2,3]. Fever is the nonspecific clinical manifestations for infection that frequently elicits suspicion of sepsis. So, it is important to search accurate predictors to early identify those febrile cirrhotic patients who are at high severe sepsis risk [2,3].
The circulating non-classical (CD16 + , inflammatory) monocytes produce pro-inflammatory cytokines TNFα and IL-6 and, in vitro, after LPS stimulation. Significantly increased CD16 + monocytes proportion can be found in sepsis and cirrhotic patients [15][16][17]. To initiate effective phagocytosis, HLA-DR expression on monocytes are essential for the presentation of peptides derived from ingested microbes to prevent severe sepsis [1,2]. So, low HLA-DR expression has been reported as poor prognostic markers for ICU mortaility in critically ill cirrhotic patients [18].
Variant allele of TNFα -308G/A affects TNFα transcription and increase septic shock susceptibility [19]. Elevated IL1β level enhances the severity of sepsis [20]. IL-6 stimulates TNFα and IL-1β release to amplify the inflammatory reaction. Both IL-6-597G/A and IL-6-174G/C polymorphism regulated IL-6 transcription rate and plasma IL-6 level, which is elevated in septic patients [21]. High plasma IL-6 helps in detecting severe sepsis in admitted patients with the suspicion of infection [22].
In febrile acute de-compensated cirrhotic patients, current study wants to explore the potential roles of candidate genetic variants and corresponding dys-regulation in the pathogenesis of increased severe sepsis risk. The interactions between genetic variants-related downstream in vivo and in vitro serologic and immunologic pathogenic changes among cases with and without sever sepsis were assessed. was based on liver biopsy results or on clinical (presence of stigma of cirrhosis including spider angioma, plamar erythema, captus medusa, ascites, varcies, splenomegaly, etc), laboratory, and ultrasonographic (including elastrography) data. Fever was defined as body temperature above 39˚C or above 38.5˚C measured consecutively at two occasions at least 1 h apart. With distribution (30:70) of variant and wild-type alleles frequency of all tested gene polymorphisms, we estimated the number of febrile de-compensated cirrhotic patients needed to observe differences in susceptibility of severe sepsis at least 10% (D) and with a common variance of 20 (σ) by literatures [8,10,12] With the type 1 error risk of 5% (α), power of 80% (1-β) and, Type 2 error (β), 20%; it was estimated that 104 patients would be required in total. Notably, the power (80%) calculation used was sufficient to detect 40% (70-30%) difference in the allele frequency of all tested gene polymorphisms.
Exclusion criteria were: human immunodeficiency virus infection, previous transplantation or any other type of immunodeficiency, steroid treatment, pituitary or adrenal disease, hepatocellular carcinoma, severe chronic heart (New York Heart Association function class III or IV) or pulmonary disease (global initiative for chronic obstructive lung disease III or IV), chronic dialysis, acute respiratory distress syndrome, and refusal of patient to participate. Patients gave written informed consent to participate in the study which was approved by the Institutional Review Board Taipei Veterans General Hospital, Taiwan, R.O.C. (IRB number: 201303013AC, approved on 19/April/2013).
Demographic and the baseline clinical evaluation [the Child-Pugh, and model for end stage liver disease (MELD), APACH III scores] were completed within 48 hours of hospitalization. Subjective global nutritional assessment (SGNA) score of >1 (2 to 4) was defined as malnourish [24]. All clinical parameters especially newly developed systemic inflammatory response syndrome (SIRS) [24], sepsis and severe sepsis during admission and during 3-month followup; in-hospital and 3-month mortality and causes of death were carefully collected after enrolled. Sepsis was diagnosed as the presence of SIRS in combination with suspected or proven infection but without any evidence of organ dysfunction or the need for intravenous vasopressor drug support to maintain blood pressure. Severe sepsis was defined as sepsis that was temporally accompanied by the need for intravenous vasopressor drug support (excluding dopamine at ≦5μg/kg/min) to maintain blood pressure (despite adequate fluid resuscitation) along with the presence of perfusion abnormalities, or metabolic acidosis (pH≦7.3) or the development of respiratory, renal, hepatic, or hematological failure. After recruitment, admitted febrile de-compensated cirrhotic patients [25] were divided into severe sepsis group and non-severe sepsis group, which including uncomplicated, SIRS and sepsis cases. Among 108 enrolled febrile de-compensated cirrhotic patients, 9 uncomplicated cases, 18 SIRS cases and 34 sepsis cases were identified as non-severe sepsis group. By contrast, 47 severe sepsis cases were identified as severe sepsis group Retrospectively, we determined the time of the first de-compensation of cirrhosis (ascites, variceal bleeding, encephalopathy or infection) and the period from this time until the first day of the present hospitalization/time of entering current study (pre-study period). Previous incidence and total episode of infections during the pre-study period were recorded.
Additionally, age and sex-matched unrelated 121 healthy controls and 51 afebrile compensated cirrhotic patients with available blood sample for genetic analysis were included as comparison group. Healthy controls were those whose visit our hospital for health check-up without clinical or biochemical evidence of liver, renal and cardiovascular disease. The afebrile compensated cirrhotic patients were identified from ongoing studies on cirrhosis in outpatient department of our hospital.

TLR4/HLA-DR expression on monocytes
The isolated CD16 − and CD16 + monocytes were double-stained with anti-CD16-PE (BD Biosciences, USA) and anti-TLR4-APC (BD Biosciences, USA) antibodies. In a dosefinding preliminary experiment, the most potent stimulation of TLR4 expression on cells analyzed by flow cytometry (FACScan, BD Biosciences) was presented at 100ng/mL of Escherichia Coli LPS (Sigma-Aldrich, St, Louis, MO). Therefore, all the following acute in vitro experiments were divided into un-stimulated and stimulated [LPS, 100ng/mL] groups. The HLA-DR expression on CD16 -(classical) and CD16 + (non-classical) monocytes were measured by staining with anti-HLA-DR-APC, anti-CD14-FITC (BD Biosciences, USA) and anti-CD16-PE antibodies.
Our preliminary experiments revealed that the TLR4 expression was not different between CD16 -(classical) monocytes between groups. It is well-established that classical monocyte are phagocytic with no inflammatory attributes [15][16][17], so the LPS-stimulated extra-and intracellular inflammatory cytokines production and corresponding signals were only evaluated in cultured CD16 + (non-classical) monocytes.

Extracellular and intracellular cytokine assays of CD16 + monocytes
For both stimulated LPS (100ng/mL) and un-stimulated groups, the isolated CD16 + monocytes were incubated for 20 h, cell free supernatants were harvested and analyzed for IL-6, IL-1β, and TNFα production using a commercial ELISA kit (R&D, Minneapolis, MN). Additionally, supernatants nitrite concentration was measured by the Griess reaction.

Phagocytic ability of CD16 -(phagocytic) monocytes
Both for un-stimulated and stimulated (100ng/mL of LPS) groups, CD16monocytes (2×10 5 / well) were incubated in five replicates with Alexa Fluor 488 (AF488)-conjugated Escherichia coli BioParticles (Invitrogen). CD16monocytes incubated with bioiparticles were used as positive controls, while monocytes without bioparticles were used as negative controls. The phagocytic indexes of the CD16monocytes from different individual were calculated by the following formula, Phagocytic index = [(MFI of experiment-MFI of negative controls)/MFI of positive controls-MFI of negative controls]×100%, where MFI stands for mean fluorescence intensity [29].

Statistical analysis
The significance of differences in allelic frequencies between each group was determined by Fisher's exact test. Differences in the distribution of alleles between the groups and deviation from Hardy-Weinberg equilibrium were assessed by Pearson χ2 test and likelihood-ratio χ2 tests of independence; 2×2 tables were used to compare allele distribution between any 2 groups. Continuous variables were expressed as mean and standard deviation (SD). Student 0 s t-test was used to compare continuous variables from two groups. For univariate and multivariate regression analysis, the third quartile of plasma sCD14 (3.7 μg/mL) and endotoxin (>2.3 EU/mL) levels at inclusion of all febrile acute de-compensated cirrhotic patients were used as cut-off values for high-risk group of severe sepsis.

Basal Characteristics
Basically, the enrolled afebrile compensated cirrhotic patients were characterized by lower Child-Pugh class, lower MELD score, relative normal serum sodium concentration, higher serum albumin, lower serum bilirubin and less prothrombin time prolongation than febrile de-compensated cirrhotic patients (S2 Table). At inclusion, cases complicated with severe sepsis had significantly higher plasma sCD14, IL-10, NOx levels, higher proportion of moderate/ massive ascites, lower serum albumin and more episodes of overall previous infection per patients than those non-severe sepsis cases (S2 Table and

Distribution of variant allele of candidate SNPs between groups
In comparison with healthy controls, higher frequencies of TLR4+896A/G and CD14-159C/T variant alleles were noted among severe sepsis and non-severe sepsis febrile decompensated cirrhotic patients as well as afebrile compensated cirrhotic patients (Table 1 and S3 Table). Higher frequencies of TLR4+896A/G and CD14-159C/T variant alleles were noted in febrile de-compensated cases than afebrile compensated cases (Table 1). In comparison with non-severe sepsis cases, higher frequencies of TLR4+896A/G and CD14-159C/T variant alleles were noted in severe sepsis cases. However, the percentage of carrying variant alleles of the following SNPs including TLR4 3 0 UTR, G/C; CD14 0 UTR, C/A; TNFα -238G/A; IL-1β -31T/C; IL-1β +3954C/T; IL-6 -174G/C; IL-6 -597G/A were not different among healthy controls, severe sepsis febrile de-compensated cirrhotic patients, non-severe sepsis febrile de-compensated cirrhotic patients and afebrile compensated cirrhotic patients (Table 1 and S3 Table).

Characteristics of cases with different genotypes
Significantly, more overall previous episodes of infection (S1F Fig), higher proportion of previous SBP episodes, and severe sepsis cases were observed in TLR4+896A/G and CD14-159C/T variant allele carriers than wild-type carriers ( Table 2). Among TLR4+896A/G and CD14-159C/T variant allele carriers, significantly higher plasma sCD14 and NOx levels were noted than their comparative groups. Plasma IL-10 level was significantly higher in TLR4+896A/G variant allele carrier than their controls but similar between variant and wild-type CD14-159C/T allele carriers (Table 2).
Increased CD16 + (non-classical) monocyte subset was associated with high TLR4/TNFα/iNOS expression and extracellular TNFα/nitrite production The baseline and LPS-stimulated TLR4 expression on the CD16 -(classical) monocytes were not different between variant and wild-type allele carriers of TLR4+896A/G or CD14-159C/T as well as between non-severe sepsis and severe sepsis cases (data not shown).
In variant TLR4+896 allele carriers with severe sepsis, a significantly positive correlation was noted between LPS-stimulated TLR4 expression and corresponding TNFα/nitrite production ( Fig 2D). However, above correlation was loss among variant CD14-159C/T allele carriers with severe sepsis (S3B Fig). In line with increased extracellular LPS-stimulated TNFα/nitrite levels, significantly higher intracellular TNFα/iNOS level confirm the cellular sources of TNFα/nitrite from circulating CD16 + monocytes of these cases (Fig 2E & 2F).
Compared to un-stimulated states, higher LPS-stimulated IL-6/IL-1β levels in CD16 + monocytes supernatant was observed among all cases. Nonetheless, the magnitudes of these LPS-stimulated changes were not different between variant and wild-type TLR4+896A/G or CD14-159C/T allele carriers as well as between non-severe sepsis and severe sepsis cases ( Fig  1C, S2D Fig).

Down-regulated HLA-DR expression on monocytes increase severe sepsis risk
Significantly, higher plasma endotoxin and IL-10 levels were observed in cases carrying variant TLR4+896A/G allele and severe sepsis (Fig 3C). Among variant CD14-159C/T allele carriers, higher plasma endotoxin levels rather than plasma IL-10 levels were noted than wild-type carriers.
Among TLR4+896A/G variant allele carriers with severe sepsis, a significant negative correlation was noted between plasma IL-10 levels and LPS-stimulated HLA-DR expression on CD16monocyte (Fig 3D). Among both TLR4+896A/G and CD14-159C/T variant allele carriers with severe sepsis, a significant negative correlation was found between plasma endotoxin level and LPS-stimulated HLA-DR expression on CD16monocyte (Fig 3E).

Discussion
Early innate immune responses to LPS are critical for determining resistance to bacterial infection; the same responses are important driving forces behind the pathophysiology of sepsis in infected individuals. Unregulated responses to bacterial agonists can cause immune-suppression, which may lead to secondary infections, or to over-stimulated inflammatory responses that can cause severe sepsis, shock, and death in cirrhosis and non-cirrhosis [5][6][7][30][31][32]. Guarner-Argente et al. reported a trend towards a higher incidence of bacterial infections in de-compensated cirrhotic patients whose carrying TLR4 D299G variant genotype compared to wild-type carriers [10]. Appenrodt et al. has demonstrated an association between NOD2 (nucleotide-binding oligomerization domain containing 2) variants and the SBP-related mortality in de-compensated cirrhotic patients [33]. However, the roles and immunologic mechanisms of genetic polymorphism on the increased severe sepsis risk had never been explored in febrile acute de-compensated cirrhotic patients.
Thoughtfully, the correlation among the plasma sCD14/endotoxin/NOx levels and various immune regulatory effectors (LPS-stimulated TLR4, mCD14, NFκBp65 and iNOS expression and corresponding intracellular and extracellular IL-6, TNFα, nitrite levels on cultured classical and non-classical monocytes) activities were surveyed in our current study. Notably, current study revealed that variant alleles of TLR4+896A/G and CD14-159C/T modulate immunologic protein abundance (increase plasma sCD14 levels/up-regulated TLR4 expression on non-classical inflammatory monocyte and down-regulated HLA-DR and mCD14 expression on classical monocyte) and function (impaired phagocytosis of classical monocytes) in our febrile acute de-compensated cirrhotic patients with severe sepsis.
It was suggested that chronic endotoxemia is associated with increased plasma IL-10 levels in cirrhotic patients [30,34]. In de-compensated cirrhotic patients, high IL-10 levels had been reported to be associated with decreased monocyte phagocytic ability [1,9,30]. In line with previous studies, carriers of variant allele of TLR4 +896A/G in our febrile acute de-compensated cirrhotic patients did show increased LPS-stimulated IL-10 production and complicated with higher frequency (59%) of severe sepsis compared to wild-type allele carriers (32%).
Membrane bound CD14 (mCD14) are pattern recognition receptors generate early innate immune response against bacterial pathogens. Decreased mCD14 expression on activated classical (phagocytic) monocyte may decrease the pathogen elimination ability of individuals. Previous study had revealed elevated levels of sCD14 in patients with sepsis [39]. Additionally, low monocyte mCD14 and high plasma sCD14 level can predict 28-day mortality in patients with community acquired infections [27,28]. Notably, in our febrile acute de-compensated cirrhotic patients with severe sepsis, low mD14 expression is characterized by the decreased LPSstimulated phagocytic ability of classical monocyte, whereas high plasma sCD14 level indicated uncontrolled sepsis among CD14-159C/T variant allele carriers.
Reduced monocyte HLA-DR expression has been reported in patients with septic de-compensation of acute-on-chronic liver failure [30]. In advanced cirrhotic patients, high plasma IL-10 levels can negatively regulated their HLA-DR expression on monocytes [30]. In our TLR4+896A/G variant allele carriers with severe sepsis, a negative correlation was noted between plasma IL-10 and LPS-stimulated CD16 -(classical) monocyte HLA-DR expression. Due to persistent endotoxemia, acute de-compensated cirrhotic patients were suffered from the down-regulation of monocyte HLA-DR expression and immune paralysis [1,30].
Accordingly, it was reasonable to observe a significant negative correlation between plasma endotoxin levels and LPS-stimulated CD16 -(classical) monocyte HLA-DR expression in our TLR4+896A/G or CD14-159C/T variant allele carriers with severe sepsis. It had been reported that iNOS-derived NO can down-regulate HLA-DR expression and suppress systemic monocyte activation [40]. In our TLR4+896A/G or CD14-159C/T variant allele carriers with severe sepsis, the significantly higher LPS-stimulated iNOS expression and nitrite production on their cultured non-classical (inflammatory) monocytes and significantly higher plasma NOx levels compared to their control groups were observed.
Our study revealed that LPS-stimulated IL-1β/IL-6 production in cultured supernatant of non-classical monocyte and plasma IL-1β/IL-6 levels were not significantly different between our TLR4+896A/G or CD14-159C/T variant and wild-type allele carriers as well as between severe sepsis and non-severe sepsis cases. Actually, recent meta-analysis reported a lack of association between IL-6 -174G/C polymorphism and sepsis risk [41]. In our study, the lack of association between the TNFα-308G/A, TNFα-238G/A, IL-6-174G/C, IL-6-597G/A, IL-1β-31T/C and +3954C/T polymorphisms and severe sepsis risk suggested that these inflammatory cytokines might influence sepsis progression via mechanisms other than regulations by these polymorphisms. Probably, the phenotypic heterogeneity of the sepsis syndrome including the infection location, and the amount of time passed since the onset of infection, as well as other individual parameters including background and environmental factors, contributed to some dis-concordance with previous reports.
Serum CRP level, MELD score and APACH score, were well known as a predictor of severity of inflammation/disease. However, the parameters revealed non-significance to predict severe sepsis in current study. In fact, the study about the predictor for severe sepsis in febrile de-compensated cirrhotic patients is limited. It had been suggested that the concentration of lipopolysaccharide binding protein (LBP) is associated inversely with disease severity scores and outcomes in critically ill cirrhotic patients with severe sepsis [42]. Another study reported that serum LPB predicts severe bacterial infection in cirrhotic patients with ascites [43]. In non-cirrhotic patients, serum C-reactive protein (CRP) level had been reported as the risk factor for severe sepsis in children with cancer and febrile neutropenia [44]. Nonetheless, CRP is primarily produced in the liver and its specificity as diagnostic and prognostic tool for infection is limited in cirrhotic patients with compromised liver function, which will underestimate the severity of acute infection. MELD score is a representative maker for cirrhosis severity as well as criteria in donor liver allocation systems. In liver transplant recipients, MELD score >23 is reported to predict ICU stay >10 days without modify survival [45]. The APACHE is ICU-specific prognostic scores. In cirrhotic patients admitted to ICU, MELD score predict short mortality better than APACH scores [46]. In our admitted febrile de-compensated cirrhotic patients, the lack of roles of serum CRP level, MELD score and APACH score in prediction svere sepsis might be due to different studied population contrast to other non-cirrhotic critically ill patients.
Using multi-modalities comprehensive approaches, our data showed the contribution on the TLR4+896A/G and CD14-159C/T polymorphism-related immune dysfunction including increased non-classical (inflammatory) monocyte proportion-related LPS hyper-inflammatory response and decreased classical (phagocytic) monocyte proportion-related impaired phagocytosis in febrile acute de-compensated cirrhotic patients complicated with severe sepsis. In addition to regular predictive factors such as malnutrition, high plasma endotoxin and sCD14 levels for development of severe sepsis in high risk groups, the novel predictive factors were either single or double TLR4+896A/G and CD14-159C/T variant alleles carriers observed in current study. Accordingly, treatment strategies should change from uniform management to rapid stratifications and sub-categorization, with subsequent aggressive targeted therapeutic intervention in those most at risk. Supporting Information S1 Fig. (A-E) Schematic representative SNPs that explored within various cytokines genes in current study; the comparison of (F) previous episode of infection during the pre-study period and (G) duration of first de-compensation to entering study (pre-study period) between cases with and without severe sepsis. Ã P < 0.05 or ÃÃ P < 0.001 vs. TLR4/CD14 variant alleles carriers/severe sepsis cases. Pre-study period: first de-compensation of cirrhosis and the period from this time until the first day of the present hospitalization/time of entering current study. (TIF) S2 Fig. (A-C). Plasma TNFα, IL-6 and IL-1β levels of all cases; (D). LPS-stimulated IL-1β production; (E,F). LPS-stimulated p38MAPβ and ERK1 mRNA expression on CD16 + (non-classical) monocytes of all cases; # p<0.05 vs. un-stimulated group.