The impact of hepatic steatosis on portal hypertension

Background and aims Studies in animal models have suggested that hepatic steatosis impacts on portal pressure, potentially by inducing liver sinusoidal endothelial dysfunction and thereby increasing intrahepatic resistance. Thus, we aimed to evaluate the impact of hepatic steatosis on hepatic venous pressure gradient (HVPG) in patients with chronic liver disease. Method 261 patients undergoing simultaneous HVPG measurements and controlled attenuation parameter (CAP)-based steatosis assessment were included in this retrospective study. Results The majority of patients had cirrhosis (n = 205; 78.5%) and n = 191 (73.2%) had clinically significant portal hypertension (CSPH; HVPG≥10mmHg). Hepatic steatosis (S1/2/3; CAP ≥248dB/m) was present in n = 102 (39.1%). Overall, HVPG was comparable between patients with vs. without hepatic steatosis (15.5±7.5 vs. 14.8±7.7mmHg; p = 0.465). Neither in patients with HVPG (<6mmHg; p = 0.371) nor in patients with mild portal hypertension (HVPG 6–9mmHg; p = 0.716) or CSPH (HVPG≥10mmHg; p = 0.311) any correlation between CAP and HVPG was found. Interestingly, in patients with liver fibrosis F2/3, there was a negative correlation between CAP and HVPG (Pearson’s ρ:-0.522; p≤0.001). In multivariate analysis, higher CAP was an independent ‘protective’ factor for the presence of CSPH (odds ratio [OR] per 10dB/m: 0.92, 95% confidence interval [CI]:0.85–1.00; p = 0.045), while liver stiffness was associated with the presence of CSPH (OR per kPa: 1.26, 95%CI: 1.17–1.36; p≤0.001). In 78 patients, in whom liver biopsy was performed, HVPG was neither correlated with percentage of histological steatosis (p = 0.714) nor with histological steatosis grade (p = 0.957). Conclusion Hepatic steatosis, as assessed by CAP and liver histology, did not impact on HVPG in our cohort comprising a high proportion of patients with advanced chronic liver disease. However, high CAP values (i.e. pronounced hepatic steatosis) might lead to overestimation of liver fibrosis by ‘artificially’ increasing transient elastography-based liver stiffness measurements.


Introduction
Non-alcoholic fatty liver disease (NAFLD) is becoming the most prevalent liver disease with 17-46% of adults affected in Western countries. [1] NAFLD is the hepatic manifestation of the metabolic syndrome and may progress to liver fibrosis, cirrhosis, portal hypertension (PHT), and hepatocellular carcinoma often resulting in the need for liver transplantation. [2][3][4][5] With increasing severity of liver fibrosis and cirrhosis, portal hypertension (hepato-venous pressure gradient (HVPG) �6 mmHg) might develop and progress to clinically significant portal hypertension (CSPH, HVPG �10 mmHg). [6] CSPH is the main driver for the development of complications such as variceal bleeding [7], ascites[8] and portosystemic encephalopathy. [6,[9][10][11] While the role of liver fibrosis as a mechanical resistance factor for development of PHT is well established, the influence of hepatic steatosis needs further investigation. Francque et al. (2010) demonstrated that a rat model of steatohepatitis induced by methionine-choline-deficient diet (MCDD) develops PHT in the absence of fibrosis.
[12] These results were confirmed in another animal model of diet-induced obesity and metabolic syndrome: Hepatic steatosis induced liver sinusoidal endothelial dysfunction which in turn increased intrahepatic vascular resistance, and thus, portal perfusion pressure even in the absence of liver fibrosis or hepatic inflammation. [13] Another in-situ perfusion model in rats fed with MCDD showed a distortion of the regular hepatic sinusoidal anatomy by a disorganized pattern with multiple interconnections and vascular extensions. Hepatic overexpression of thromboxane synthase and endothelin-1 were found and the authors suggested sinusoidal endothelial dysfunction as main mechanism responsible for increased intrahepatic vascular resistance. [14] Moreover, a recent animal model confirmed elevated portal pressure and hyperdynamic systemic circulation in a rat-model with severe steatosis. [15] Human evidence for an impact of 'simple' steatosis on portal pressure is limited to a small study including non-cirrhotic NAFLD patients reporting an association of steatosis and elevated portal pressure. The degree of steatosis was the only factor being significantly different between patients with HVPG �5 mmHg and �6mmHg. [16] Metabolic liver disease and hepatic steatosis might also influence HVPG in patients after eradication of hepatitis C virus infection. [17] In summary, these findings suggest that hepatic steatosis may impact on sinusoidal structure and function and thus, on portal pressure. Hepatic steatosis may be best evaluated by liver biopsy, however, this procedure is associated with a series of complications and sampling variabilty. [18][19][20] Although hepatic steatosis can be accurately quantified by magnetic resonance imaging, the applicability of this method is limited by its resource-intensiveness. [21,22] In 2010, transient elastography (TE)based controlled attenuation parameter (CAP) was introduced allowing the assessment of hepatic steatosis at the time of liver stiffness (LSM). [23] Consecutively, the accuracy of CAP for diagnosing hepatic steatosis has been confirmed in several liver biopsy controlled studies.
[24-27] Thus, we aimed to clarify the impact of hepatic steatosis assessed by TE-based CAP on HVPG in patients with chronic liver disease.

Patients and definitions
All patients undergoing both HVPG and TE-based CAP measurement between 01.01.2014 and 31.12.2016 at the Medical University of Vienna were included in this retrospective analysis. Exclusion criteria were: (i) unreliable HVPG measurement (presence of intrahepatic venovenous communications, technical problems, measurement during non-selective β-blocker intake, presence of portal venous thrombosis or Budd-Chiari-Syndrome) or CAP measurement (food intake within 6 hours prior to measurement or technical problems), (ii) previous liver transplantation or transjugular intrahepatic portosystemic shunt (TIPS), (iii) hepatocellular carcinoma (HCC), (iv) acute liver failure or (v) antiviral treatment of viral hepatitis C. If patients underwent more than one HVPG and CAP measurement, only the first adequate measurement was included. Patient characteristics and laboratory parameters were assessed by chart review. Since no clear distinction between alcohol and non-alcoholic fatty liver disease could be made due to the retrospective nature of our study, we combined these two etiologies in one liver disease category.

HVPG measurements
HVPG measurements were performed at the Vienna Hepatic Hemodynamic Laboratory according to a standardized procedure using a 7F balloon catheter (Pejcl Medizintechnik, Austria) as previously described.
[28] In brief, the catheter introducer sheet was placed in the right internal jugular vein under local anesthesia and ultrasound guidance using the Seldinger technique. A balloon catheter was then placed under fluoroscopic control in a large hepatic vein. HVPG was calculated as the difference between wedged hepatic vein pressure and free hepatic vein pressure as a mean of three measurements. Treatment with non-selective β-blockers was paused at least three days prior to the measurement.[29, 30] PHT was defined as HVPG �6mmHg, while HVPG values �10mmHg denoted CSPH, respectively.

Liver stiffness and CAP measurements
CAP measurements were done simultaneously to LSM using FibroScan 1 (Echosense, Paris, France) as previously described.
[31] M and XL-probe were used according to the recommendation of the device. Reliability of LSM was determined according to established criteria.

Liver biopsy
Histological specimens were included in this study, if biopsy was performed within one week after CAP/HVPG measurement using either the transjugular or percutaneous route.
[37] Data on fibrosis stage and hepatic steatosis (%) were extracted from medical records. Of note, liver fibrosis was graded according to the Batts-Ludwig or METAVIR score, as applicable. [38,39] Hepatic steatosis was semi-quantitatively assessed and graded as no hepatic steatosis (S0, fat accumulation in <5% of hepatocytes), grade 1 (S1, fat accumulation in 5-33% of hepatocytes), grade 2 (S2, fat accumulation in 34-66% of hepatocytes), and grade 3 (S3, fat accumulation in �67% of hepatocytes). [40] Statistics Statistical analyses were performed using IBM SPSS Statistics 25 (SPSS Inc., Armonk, New York, USA) and GraphPad Prism 6 (GraphPad Software, La Jolla, California, USA). Continuous variables were reported as mean ±standard deviation (SD) or median (interquartile range; IQR), and categorical variables were shown as numbers (n) and proportions (%) of patients. Comparisons of continuous variables were performed using Student t test or Mann-Whitney-U test, as applicable. Group comparisons of categorical variables were performed using either Pearson's chi-squared or Fisher's exact test. Pearson correlation coefficient was calculated to assess a potential correlation between HVPG and CAP. Otherwise, Spearman's rank correlation coefficient was used. Univariate and multivariate binary logistic regression analysis was used to determine factors independently associated with the presence of CSPH. A p-value �0.05 was considered statistically significant.

Ethics
This study was approved by the ethics committee of the Medical University of Vienna (EK: 1124/2017) and performed in accordance with the ethical guidelines denoted in the Declaration of Helsinki (version 2008). As this study is a retrospective analysis and according to the Ethic's vote no written informed consent was required. Patients' data were pseudo-anonymized before inclusion in this study. (Table 1) Overall, 475 patients underwent simultaneous CAP/LSM and HVPG measurement within the study period. Eighty-nine patients were excluded due to unreliable results of HVPG or CAP measurements, 53 patients due to antiviral therapy of hepatitis C, 45 patients due to previously diagnosed HCC and 20 due to previous liver transplantation or TIPS. Seven patients were excluded due to other reasons (S1 Fig

Severity of portal hypertension according to hepatic steatosis grade (Fig 2)
HVPG values were comparable between patients with no, mild, moderate and severe steatosis in the overall cohort (HVPG in S0: 15  The impact of steatosis on portal hypertension Factors independently associated with the presence of CSPH (

Comparison of HVPG with histological steatosis grade
In 78 patients (29.9%), histological data was available. In these patients, CAP significantly correlated with histological steatosis (Pearson's r = 0.355, p = 0.001). However, there was neither a correlation of the percentage of histological steatosis with HVPG (Spearman's ρ = -0.042, p = 0.714) nor of the histological steatosis grade (Spearman's ρ = -0.007, p = 0.957). Additionally, subgroup analyses among different etiologies, histological fibrosis stages and fibrosis stages determined by TE did not reveal a significant correlation except for a trend towards a negative correlation of HVPG with hepatic steatosis in patients with cirrhosis on liver histology (Spearman's ρ = -0.342, p = 0.054; S1 Table).

Discussion
Hepatic fibrosis has been widely acknowledged as the main 'mechanical' contributor to PHT, due to its profound impact on the structural component of intrahepatic resistance. However, some animal studies have recently been investigating the influence of hepatic steatosis on the development and progression of PHT.[12-16] Portal pressure was found to be increased in the absence of liver fibrosis in three rat-models with diet-induced NAFLD. Furthermore, these results were confirmed in a small study on humans including 50 patients with NAFLD, of which 14 patients had elevated HVPG (�6mmHg).[12, [14][15][16] However, we could not confirm these findings in our study using CAP as a validated non-invasive marker for hepatic steatosis. In our cohort, we neither observed a significant correlation between CAP and HVPG in the overall study population, nor in the subgroups of patients with or without CSPH. This finding indicates that hepatic steatosis is no major factor contributing to portal venous pressure in patients with chronic liver disease. In multivariate analysis, after correcting for other factors potentially associated with CSPH, such as age, the presence of diabetes [41], liver stiffness, and MELD score, higher CAP values were even protective for the presence of CSPH. Of note, the majority of our patients had advanced chronic liver disease (ACLD) as indicated by a median liver stiffness of 28.0kPa. Importantly, previous studies have demonstrated the disappearance of hepatic steatosis with liver fibrosis progression in NAFLD patients (i.e. 'burnt-out NASH') [42], a phenomenon which might also occur in other etiologies of ACLD and therefore may have potentially attenuated the association between CAP and HVPG in our study cohort of ACLD patients. Of note, hepatic steatosis on histology was neither correlated with HVPG in the overall cohort nor in subgroup analyses, despite a trend towards a negative correlation in patients with cirrhosis. Although this finding would suggest a 'protective' effect of hepatic steatosis in these patients, the low sample size (n = 36) limits this finding. Noteworthy, the abovementioned animal studies need to be interpreted with caution. Since these animals developed very pronounced hepatic steatosis in a short period of time, it is unclear whether these findings can be extrapolated to humans, in whom lifestyle-induced hepatic steatosis persists over a long period of time and is substantially less pronounced. Thus, these experimental studies might considerably overestimate the relevance of hepatic steatosis in the human setting. Moreover, animals, if at all, only showed mild hepatic inflammation/liver fibrosis. [12,14] In our clinical setting, other factors contributing to portal hypertension, such as increases in intrahepatic resistance due to hepatic inflammation/fibrosis, as well as splanchnic vasodilatation and the hyperdynamic circulation in patients with CSPH might have had a much stronger impact on portal pressure than hepatic steatosis itself. [43] So far, only one human study evaluated the effect of simple hepatic steatosis on HVPG in humans and found a significant positive correlation between hepatic steatosis and HVPG, however, this study only included 14 patients with The impact of steatosis on portal hypertension elevated HVPG with a mean pressure of 8.8 ± 2.6mmHg and only one patient had liver cirrhosis. [16] This is in contrast to our study population with most patients having progressed to cirrhosis and three quarters of patients having CSPH. Insulin resistance (IR) is an important driver for the development of portal hypertension and hepatic steatosis. [44] Thus, in patients where IR is the key driver of steatosis, IR-induced endothelial dysfunction might aggravate sinusoidal vasoconstriction and therefore contribute to the development of early portal hypertension. However, this factor does not seem to have a strong impact on HVPG in patients with ACLD and established PHT. Interestingly, we observed a significant negative correlation between HVPG and CAP in various subgroups, especially in patients with liver fibrosis stages F2/3, as assessed by TE. One explanation could be an 'artificial' increase of TE-based liver stiffness in these patients by pronounced hepatic steatosis, resulting in an overestimation of liver fibrosis. Recently, there has been a controversy regarding the impact of CAP on the association between LSM and fibrosis [45,46]. Petta et al. (2015) [45] observed increased LSM in patients with F0-2 in patients with severe steatosis. Furthermore, the same group reported a higher false positive rate for the diagnosis of liver fibrosis stage F3/4 and also decreasing diagnostic accuracy for discriminating F0-2 from F3-4 among patients with advanced steatosis according to CAP values. [46] As a result, the 'true' liver fibrosis stage as assessed by histology would be lower in our study, which could explain the indirect correlation between CAP and HVPG values where HVPG was lower than expected from LSM values. This study has several limitations that need to be considered when interpreting its results: First, nearly half of the patients undergoing HVPG and CAP measurements needed to be excluded due to potential confounding factors such as antiviral therapy or non-selective beta-blocker therapy, which left a limited number of patients for the final analyses. Secondly, the majority of our patients had other etiologies than (N)AFLD and suffered from ACLD representing a selection bias towards high proportions of patients with cirrhosis and CSPH in whom hepatic steatosis may be of limited relevance for PHT. Moreover, the proportions of patients with moderate or severe steatosis was limited. In the early study period only the M-probe was available to obtain CAP, which leaves the possibility of elevated LSM and CAP values in obese patients. It has to be considered that CAP might have limited diagnostic value for assessing hepatic steatosis in patients with ACLD. [47,48] Finally, hepatic steatosis might imply a certain degree of presinusoidal PHT through hepatocyte ballooning and sinusoidal distortion, which may not be fully captured by HVPG measurement. This limitation could only be overcome by direct measurement of portal pressure (gradient), which is rarely performed unless the patient is undergoing TIPS. However, as all ACLD patients undergoing TIPS have CSPH and even more advanced disease than our cohort, this would lead to an even stronger selection towards patients in whom the effect of hepatic steatosis on PHT is expected to be less pronounced. In conclusion, hepatic steatosisas assessed by CAP-was not associated with an increase in portal pressure in our cohort comprising a high proportion of patients with ACLD. However, hepatic steatosis might lead to overestimation of liver fibrosis by 'artificially' increasing TE-based LSM.