We included eighty-nine patients with diagnosed liver cirrhosis and severe portal hypertension, undergoing TIPS insertion, in our study. Primary outcome was the increase in portal hepatic venous pressure gradient (PHPG) after TIPS insertion as a read-out of pro-thrombotic milieu generation. All patients were treated between August 1996 and August 2003 in the University of Bonn, Department of Internal Medicine I, Germany. Patients, older than 18 years with clinical signs of liver cirrhosis and a multidisciplinary defined indication for TIPS insertion, were included in our trial. Exclusion criteria were the presence of systemic infection, hepatic encephalopathy (higher than grade I), bilirubinemia (higher than 5mg/dl) or arterial pulmonary hypertension. We used bare stents for TIPS insertion (8–10 mm Wallstent, Boston Scientific, MA, USA), as previously described[11,12]. Control angiography was performed after a mean of 14 days. Here, position and function of the stent was evaluated in patients with significant decrease in portal venous flow or TIPS flow (N = 64). The process flow was previously specified as routine in the center’s protocol to optimize TIPS function[13]. During the procedures, portal and hepatic venous pressures were invasively measured by the use of a pressure transducer system (Combitrans, Braun, Melsungen, Germany) and a multichannel monitor (Sirecust, Siemens, Germany). Per definition, the difference between the portal and hepatic venous pressure was named portal hepatic pressure gradient (PHPG). Once, the right branch of the portal vein was cannulated, we harvested blood from portal and hepatic vein in EDTA tubes (N = 89) to obtain material for the analyses of TXB₂, PGD₂ and PGE₂. All TIPS insertions were performed without general anesthesia. After collection of the blood, we centrifuged the samples at 3000 revolutions per minute for 15 minutes at 4°C. Afterwards, plasma samples were stored at minus 80°C. Clinical patient data were recorded at their admission to the study. For the analyses of common blood parameters (e.g. creatinine or bilirubin levels), we used standard biochemical tests. The study protocol passed and has been approved by the local ethics committee of the University of Bonn (029/13). All patients signed and agreed to all procedures as declared in the study protocol. The final manuscript was reviewed by all authors. Access to the study data was ensured and all authors approved the final manuscript.

Two hundred μL plasma was porcupined with isotopically labeled internal standards (TXB₂-d4, PGD₂-d4, PGE₂-d4), 100 μL EDTA solution (0.15M) and 600 μL ethyl acetate, to quantify levels of thromboxane B₂, prostaglandin D₂ and E₂. Specimens were vortexed, and subsequently centrifuged at 20.000 g for 15 minutes. We removed the organic phase and rerun the extraction by the addition of 600 μL ethyl acetate. After combination of the organic fractions, a sparing stream of nitrogen was used for evaporation at a temperature of 45°C. Reconstitution of the residues was performed by the addition of 50 μL acetonitrile/water/formic acid (20:80:0.0025, v/v/v) and transferred to glass vials. LC-MS/MS analysis was carried out using an Agilent 1290 Infinity LC system (Agilent, Waldbronn, Germany) coupled to the hybrid triple quadrupole linear ion trap mass spectrometer QTRAP 6500+ (Sciex, Darmstadt, Germany) equipped with Turbo V source operating in negative ESI mode. The chromatographic separation was carried out using a Synergi Hydro-RP column (150 × 2 mm, 4 μm particle size and 80 Å pore size; Phenomenex, Aschaffenburg, Germany). A gradient program was employed at a flow rate of 300 μL/min. Mobile phase A was water/formic acid (100:0.0025, v/v) and mobile phase B was acetonitrile/formic acid (100:0.0025, v/v). Seperation of the analytes was performed under gradient conditions within sixteen minutes. The injection volume was 10 μL and the gradient program started with 90% A for 1 min. The mobile phase A was decreased to 60% within 1 min. After 1 min biding, the mobile phase was further decreased to 50% within 1 min and another held for 2 min. Within 2 min, mobile phase A was further decreased to 10% and held for 1 min. In the space of one minute, the initial conditions were restored. The column was re-equilibrated for 6 min. For mass spectrometric, parameters were set as follows: Source temperature 500 °C, ion spray voltage -4500 V, curtain gas 40 psi, nebulizer gas 40 psi and turbo heater gas 60 psi. Both quadrupoles were running at unit resolution. Analyst Software 1.6.3 and Multiquant Software 3.0.2 (both Sciex, Darmstadt, Germany) were used for analysis and quantification. The following precursor-to-product ion transitions were used for quantification: m/z 369.2 → m/z 195.0 for TXB₂, m/z 351.2 → m/z 233.3 for PGD₂ and m/z 351.2 → m/z 315.0 for PGE₂. Peak area of the corresponding internal standard was used for correction of the peak area of each analyte. Calibration curves were constructed using linear regression with 1/x2 weighting. The coefficient of correlation was at least 0.99. Variations in accuracy were less than fifteen percent over the whole range of calibration, except for the lowest limit of quantification, where a variation in accuracy of twenty percent was accepted.

Using a commercial enzyme‐linked immunosorbent assay kit (Cusabio), lipopolysaccharide (LPS) serum levels were measured. Therefore, a specific antibody for LPS was precoated onto a microplate, and 100 μL of sample or standards was plated for 2 hours at room temperature. After incubation, samples were read at 450 nm. Values were expressed as picograms per milliliter; intra‐ and inter-assay coefficients of variation were 8% and 10%, respectively.

GraphPad Prism 5 for Windows (GraphPad Software, Inc.) or BIAS® for Windows was used for the performance of statistical analyses. Wilcoxon matched-pairs test was used for paired intra-individual comparisons, namely portal versus hepatic vein prostanoid concentrations. Group differences of unrelated groups were assessed by means of χ² contingency tables or Wilcoxon-Mann-Whitney U tests, as appropriate. Associations of outcomes with continuous variables were assessed in linear regression models. After univariate analyses, multivariate analyses were performed for significant associations using a P value >0.1. Variables with P >0.1 were discharged from the model. Survival analyses were performed using Cox proportional hazards regression analysis. Correlations were assessed by the use of Spearmen/Kendall rang correlation. P values < 0.05 were considered to be statistically significant.

We included eighty-nine patients with a mean age of fifty-nine, undergoing BMS TIPS insertion, in the study. Refractory ascites (N = 41 patients; 46%) was the major indication for TIPS insertion, followed by secondary prophylaxis of variceal bleeding (N = 36 patients; 40%) and treatment of hepatorenal syndrome (N = 6 patients; 7%). In six patients (7%) indication was presence of both, esophageal variceal bleeding and ascites. Development of cirrhosis was induced in most patients by alcohol abuse (51%), followed by viral hepatitis (7%) and primary biliary cirrhosis (3%). Here, patients presented a mean MELD score of twelve points (range 6–33), and most patients were classified as Child-Pugh B at the time of study inclusion. The mean survival time after TIPS insertion was 770 days. Please find the entire patients characteristics in Table 1.

Quantification of baseline TXB₂ levels during TIPS insertion revealed significantly lower concentrations in the portal than in the hepatic vein (231.2 pg/mL vs. 726.9 pg/mL; P<0.0001). In contrast, concentrations of PGD₂ (295.7 pg/mL vs. 169.4 pg/mL; P<0.0001) and of PGE₂ (872.6 pg/mL vs. 693.4 pg/mL; P<0.0001) were significantly higher in the portal than in the hepatic vein (Fig 1, Table 1). Since albumin binding of prostanoids has been shown to reduce their bioavailability and activity in patients with liver cirrhosis[14], we assessed correlations of TXB₂, PGD₂ and PGE₂ with serum albumin concentrations. Here, albumin concentrations correlated inversely with PGD₂ and PGE₂ concentrations in the portal vein (P = 0.04 each), but neither with portal vein concentrations of TXB₂, nor with hepatic vein concentrations of TXB₂, PGD₂ or PGE₂. Patient age did not affect portal or hepatic vein prostanoid levels. Correlation of white blood cell count and portal vein TXB₂ levels revealed a significant correlation (P = 0.012). Furthermore, portal and hepatic vein levels of TXB₂ correlated with platelet count (PV: P = 0.014; HV: P = 0.002).

The mean portal hepatic venous pressure gradient (PHPG) before TIPS insertion was 20.6 mmHg, and it decreased to a mean of 9 mmHg after TIPS insertion. At control angiography, performed in 64 patients due to significant decrease in portal venous flow or TIPS flow, mean PHPG measurement was 14.6 mmHg (Table 3). Out of 64 patients, 38 had shunt dysfunction due to partial thromboses and received angioplasty. Next, uni- and multivariable regression analysis was performed to identify predictors of PHPG increase after TIPS insertion (Table 4). Of note, baseline TXB₂ (P = 0.004) concentrations in the portal vein showed an independent correlation with PHPG increase at short time follow-up. An additional predictor of PHPG increase after TIPS insertion was baseline PHPG (P = 0.001). In contrast, no statistically significant association between portal vein concentrations of PGD₂ and PGE₂, or hepatic vein concentrations of TXB₂, PGD₂ and PGE₂ was observed.

In a subset of 17 patients, we analyzed portal vein levels of lipopolysaccharide (mean 57.4 pg/ml +/- 9.6pg/ml). Spearmen’s rank correlation did not show significant correlations between portal vein levels of lipopolysaccharide and portal vein levels of PGD₂ (R = -0.26; P = 0.31), PGE₂ (R = -0.23; P = 0.37), or TXB₂ (R = -0.02; P = 0.95).

COX regression analysis of survival after TIPS insertion revealed baseline creatinine (P<0.0001) concentrations as a negative predictor of survival (S1 Table). In detail, median survival of patients with baseline creatinine concentration ≥1.2 mg/dL was 236 days compared to 820 days in patients with creatinine concentration <1.2 mg/dL (P = 0.0005)(S1 Fig). However, no significant association between prostanoid concentrations in the portal or hepatic vein and patient survival after TIPS insertion was observed.