Hepatic Venous Pressure Gradient Measurements in Children: Correlation with Hepatic Histology and Clinical Indicators of Portal Hypertension

Abstract

Objectives:

In adults, elevated hepatic venous pressure gradients (HVPG) are correlated with the degree of liver fibrosis on histopathology and predict worse outcomes including variceal bleeding and death. We aimed to examine the association between HVPG measurements, histopathologic findings, and clinical indicators of portal hypertension in children.

Methods:

Utilizing retrospective data from two pediatric centers between 2006–2015, we identified children who underwent simultaneous HVPG measurement and transjugular liver biopsy. Medical charts were reviewed for histopathology, imaging, endoscopic and clinical data.

Results:

Forty-one children (median age 11 years) were included in the analysis with diagnoses of acute hepatitis (n=15), chronic liver disease (n=12), hepatic non-cirrhotic portal hypertension (n=4), acute liver failure (n=3) and non-hepatic causes of portal hypertension (n=7). Elevated mean HVPG measurements were found in children with acute liver failure (10mmHg, range 4–12) and chronic liver disease (7mmHg, range 1–12). HVPG measurements did not correlate with the histological severity of fibrosis (ρ = 0.23, p = 0.14) or portal inflammation (ρ = 0.24, p = 0.29), and no difference was found in HVPG when comparing children with and without a history of variceal bleeding (p = 0.43).

Conclusions:

HVPG measurements do not correlate significantly with the degree of hepatic fibrosis on biopsy. Furthermore, HVPG measurements are not associated with the presence of varices or history of variceal bleeding, suggesting the possibility of intra-hepatic shunting in children with advanced liver disease. Therefore, unlike in adults, HVPG measurements may not accurately predict children who are at risk of complications from portal hypertension.

INTRODUCTION

In adults with cirrhosis, the association between elevated measurements of hepatic venous pressure gradient (HVPG) and both the degree of liver fibrosis on histology,^1–3 and the clinical outcomes of esophageal variceal development, variceal bleeding, and increased mortality are well described.^1,4 Comparable studies are not available for children.

In intrahepatic conditions that cause portal hypertension, a gradient exists between the low-pressure hepatic outflow (hepatic veins) and high-pressure portal inflow (portal vein). The measurement of this gradient is determined by advancing a catheter via jugular vein access to the hepatic vein where free hepatic vein pressure (FHVP) is measured and subtracted from wedge hepatic vein pressure (WHVP; transduced portal vein pressure) to calculate the HVPG.¹ In adults, portal hypertension is defined as a HVPG measurement >5 mmHg and clinically significant portal hypertension is defined as the presence of varices, variceal bleeding and/or ascites and occurs at a HVPG measurement ≥10 mmHg. Additionally, at a pressure gradient of ≥12 mmHg, there is an increased risk of esophageal variceal bleeding.^4,5,6

Utilization of HVPG measurements is the standard of care in monitoring adults with advanced liver disease.⁴ In adults, reduction of HVPG by 10% or a decrease to ≤12mmHg using non-selective beta-blockers, decreases the risk of variceal bleeding. Similar to adults, children with variceal bleeding have significant morbidity including the need for intensive care, blood transfusions, risk of sepsis, and mortality. However, normative values for HVPG measurements have not been established in children and it is unknown whether HVPG is a valid marker to identify portal hypertension or predict significant complications. Additionally, the correlation of HVPG measurements with hepatic histology in children has not been established.⁷

We aim to examine whether HVPG measurements in children are correlated with: 1) histopathologic findings and 2) clinical indicators of portal hypertension, thereby revealing whether HVPG measurements could be used to predict the risk of complications from portal hypertension in children.

METHODS

Data source and study population

We retrospectively identified 44 children (0–17 years old) with chronic liver disease defined as ongoing inflammation of the liver for at least 6 months with the potential to progress to cirrhosis and end-stage liver disease, acute liver failure defined as INR > 1.5 with encephalopathy or INR > 2 without encephalopathy, acute hepatitis defined as elevated transaminases >40, hepatic non-cirrhotic portal hypertension defined as causes of portal hypertension other than cirrhosis that are pre-sinusoidal, or non-hepatic causes of portal hypertension defined as pre- and post-hepatic etiologies of portal hypertension at two large tertiary-care pediatric hospitals in the United States, who underwent transjugular liver biopsy with HVPG measurements under the same anesthesia between October 2006 and October 2015. There were no subjects who had a transjugular intrahepatic portosystemic shunt and to our knowledge no patient had a spontaneous portosystemic shunt though only 70% of subjects (n=29) had abdominal CT/MRI with the remainder having abdominal imaging with ultrasound alone and one subject that had no abdominal imaging. Children with pre-hepatic etiologies of portal hypertension (n=3) were excluded as histopathology and HVPG are expected to be normal, though clinical signs of portal hypertension are present.⁷ One child underwent repeat HVPG measurements with liver biopsy and only the first of these data points was recorded, with subsequent measurements included only for safety outcomes. Our final sample included 41 subjects.

The definition of portal hypertension was extrapolated from the adult literature and defined as a HVPG measurement of >5 mmHg.^4,5 As in adults, HVPG may be decreased by treatment with non-selective beta-blockers, we therefore recorded if any children were on beta-blockers.

Medical charts of all subjects were reviewed to provide demographic data, procedural notes, endoscopy findings, relevant imaging, and pathology reports. This project was approved by the Seattle Children’s Hospital and Emory University Institutional Review Boards.

HVPG measurement and transjugular liver biopsy procedure

All procedures were performed under general anesthesia by a board-certified interventional radiologist. There were 8 interventionalists at Seattle Children’s Hospital and 4 interventionalists at Emory University over the course of the study with no difference in technique between operators or changes in equipment. Internal jugular vein access was obtained by placing a 9 French sheath (Cook Medical: Bloomington, IN, USA). Pressure measurements were obtained in the right atrium, free hepatic vein and in the wedged hepatic vein, either with an endhole catheter or a balloon occlusion catheter. Following pressure measurements, transvenous liver biopsy was performed (Liver Access and Biopsy Set, Cook Medical: Bloomington, IN, USA or TLAB Transjugular Liver Biopsy System, Argon Medical Devices: Frisco, TX, USA).

Histopathologic findings

Medical charts of all patients were reviewed for hepatic histology reports. Histopathologic findings were subdivided by the presence of the following: necrosis (mild, moderate, severe), portal inflammation (mild, moderate, severe) and fibrosis (expansion, extension, bridging, cirrhosis).

Clinical signs of portal hypertension

Any clinical evidence of portal hypertension was broadly defined as: the presence of varices on upper endoscopy or CT/MRI imaging⁴, ascites and/or splenomegaly based on imaging, a history of thrombocytopenia not attributable to other etiology, a history of variceal bleeding and/or spontaneous bacterial peritonitis. Analyses were further performed both with and without children with hepatic non-cirrhotic portal hypertension (n=4) in whom HVPG is expected to be normal to slightly increased.^1,9

Statistical methods and analytical software

Descriptive statistics were calculated for all demographic and clinical variables. Spearman’s rank correlation was used to examine the correlations between HVPG measurements, clinical evidence of portal hypertension, and histopathologic findings. In the latter analyses, fibrosis categories of expansion and extension were combined. Significance testing was done at the α = 0.05 level. SAS 9.4 (SAS Institute Inc., Cary, North Carolina, USA) was used for all analyses.

RESULTS

Baseline characteristics of study population

Forty-two liver biopsies were obtained from 41 subjects with a median age of 11 years (range 5 months to 17 years) of which 13 (32%) were female. The most common diagnoses in children undergoing transjugular liver biopsy with concurrent HVPG measurement was: hepatitis (n=15) followed by chronic liver disease (n=12), hepatic non-cirrhotic portal hypertension (n=4), acute liver failure (n=3), and non-hepatic causes of portal hypertension (n=7). Ninety-three percent of subjects (n=38) had a venogram at the time of HVPG measurement, none of which showed evidence of intra-hepatic veno-venous shunting. No patients were on prophylaxis with beta-blockers at the time of HVPG measurement [Supplemental table 1].

Median HVPG measurements by disease process

HVPG measurements varied by disease process. Children with acute liver failure had the highest median HVPG measurement of 10 mmHg (range 4–12 mmHg), while children with chronic liver disease had a median HVPG measurement of 7 mmHg (range 1–12 mmHg). Median HVPG measurements were ≤5 mmHg in children with hepatic non-cirrhotic portal hypertension (4mmHg, range 3–12mmHg), hepatitis (3mmHg, range 0–19mmHg) and non-hepatic causes of portal hypertension (4mmHg, range 0–20mmHg) [Figure 1].

Figure 1.

Median HVPG measurement by underlying disease process Children with acute liver failure and chronic liver disease have elevated HVPG measurements while children with hepatic non-cirrhotic portal hypertension, hepatitis and non-hepatic causes of portal hypertension all had normal median HVPG measurements. ALF = acute liver failure, CLD = chronic liver disease, other = non-hepatic causes of portal hypertension.

HVPG measurement and histopathologic findings

Elevated HVPG measurements (≥5 mmHg) were found in children with bridging fibrosis or cirrhosis, and severe portal inflammation [Table 1]. Two subjects with mild necrosis and one subject with moderate necrosis on histopathology were also noted to have elevated HVPG measurements. However, all children with evidence of severe portal inflammation on histopathology who had HVPG measurements ≥5 mmHg also had bridging fibrosis or cirrhosis. Of the 6 children with portal inflammation alone, only one had an elevated HVPG measurement of 12mmHg. Of the 7 children with cirrhosis, 5 had elevated HVPG measurements (range 9–12mmHg) and 2 had normal HVPG measurements (1 and 4mmHg). Of the children with cirrhosis but normal HVPG measurements, one subject was found to have a high-grade stenosis of the middle hepatic vein on venogram (WHVP 30mmHg, FHVP 29mmHg) while the second subject had pulmonary arteriovenous malformations and normal venogram findings (WHVP 21mmHg, FHVP 17mmHg).

Table 1.

Median HVPG measurement by histopathologic finding

Histopathologic finding	Mediana HVPG measurement (mmHg), (range)	p-value
Necrosis (n=5)	10 (3–16)	-
Mild (n=3)	12 (3–16)
Moderate (n=1)	10
Severe (n=1)	4
Portal inflammation (n=21)	5.5 (0–16)	p=0.29
Mild (n=12)	4.5 (0–16)
Moderate (n=6)	4.3 (2–12)
Severe (n=3)	7 (4–12)
Fibrosis (n=23)	6 (1–16)	p=0.14
Expansion (n=7)	4 (2–16)
Extension (n=3)	3 (2–6)
Bridging (n=6)	6.3 (3–12)
Cirrhosis (n=7)	10 (1–12)

^aMedian not calculated in categories where n=1

Despite the trend towards higher HVPG with advancing fibrosis, HVPG measurements did not correlate with the histologic finding of fibrosis (ρ = 0.23, p = 0.14) even when subjects with acute liver failure, hepatic non-cirrhotic portal hypertension, and non-hepatic causes of portal hypertension were excluded from the analysis (ρ = 0.17, p = 0.39) and when the two subjects with cirrhosis but elevated FHVP and therefore normal HVPG were also excluded from the analysis (ρ =0.31, p = 0.05) [Table 2]. HVPG measurements also did not correlate with the histological finding of portal inflammation (ρ = 0.24, p = 0.29). Correlations between HVPG measurements and necrosis level could not be pursued due to limited patient sample (n=5).

Table 2.

Median HVPG measurement by clinical evidence of portal hypertension

Findings of clinical portal hypertension	Median HVPG measurement (mmHg), (range)	p-value
Varices on endoscopy		0.91
Present (n=7)	3 (1–12)
Absent (n=8)	5 (2–19)
Varices on imaginga		0.81
Present (n=3)	4 (2–11)
Absent (n=11)	7 (1–20)
History of variceal bleeding		0.43
Yes (n=4)	4 (0–20)
No (n=37)	7.5 (3–12)
Ascites		0.84
Present (n=19)	6 (0–16)
Absent (n=22)	4 (1–20)
Splenomegaly		0.48
Present (n=27)	6 (0–20)
Absent (n=14)	4 (0–12)
Thrombocytopenia		0.38
>150,000 (n=13)	3 (0–12)
≤150,000 (n=28)	6 (0–20)

^aImaging defined as CT or MRI only, excludes endoscopic imaging

HVPG measurement and clinical signs of portal hypertension

Thirty-five subjects had clinical signs of portal hypertension. Of these subjects, 10 had evidence of varices either on upper endoscopy (n=7) or imaging (n=3) and 4 had a history of variceal bleeding. Twenty-seven children had splenomegaly on imaging and 19 had ascites; there were no documented episodes of spontaneous bacterial peritonitis [Table 2].

There was no significant difference in HVPG measurement between children with or without splenomegaly (p=0.48), with or without ascites (p=0.84), with or without varices (p=0.91) or with or without thrombocytopenia (p=0.38). However, in children with any sign of clinical portal hypertension (splenomegaly, varices, ascites), overall median HVPG measurements were elevated (5.5 mmHg, range 0–20mmHg). A sensitivity analysis found that median HVPG measurements were not significantly different when children with hepatic non-cirrhotic portal hypertension (n=4) were removed from the analyses (5.8mmHg, range 0–20mmHg). Median HVPG measurements were normal in children who had no clinical evidence of portal hypertension (3.5 mmHg, range 2–12 mmHg).

Of the 11 children with HVPG measurements ≥10mmHg: one child had grade 2 esophageal varices, 2 children had a no abnormalities on upper endoscopy and 8 children did not have an upper endoscopy within 1 year of HVPG measurement. Two children with HVPG measurements ≥10mmHg had a prior history of variceal bleeding; one child had congenital hepatic fibrosis and the other child had idiopathic chronic liver disease.

In children with a history of variceal bleeding (n=4), HVPG measurements were 3, 4, 11 and 12mmHg. The two children with a history of variceal bleeding, but normal HVPG measurements also did not show cirrhosis on histopathology. There was no difference in HVPG measurements between subjects with a history of variceal bleeding and those without (p = 0.43).

Safety outcomes

Vascular thrombosis of the middle hepatic vein on ultrasound imaging 9 days following transjugular liver biopsy and HVPG measurement occurred in one subject (2.4% complication rate) who had a previous history of high-grade stenosis of the middle hepatic vein.

DISCUSSION

This retrospective analysis of children undergoing simultaneous HVPG measurement and transjugular liver biopsy demonstrates elevated median HVPG measurements in children with acute liver failure and chronic liver disease. Correspondingly, elevated median HVPG measurements were found in children with necrosis and bridging fibrosis or cirrhosis on liver histopathology. However, HVPG measurements did not correlate with worsening degrees of fibrosis or portal inflammation on liver biopsy. HVPG measurements were elevated in children with splenomegaly, varices, and/or ascites, but did not differ in children with and without a history of variceal bleeding. To our knowledge, this is the first study to examine the association between HVPG measurements, histopathologic findings, and clinical indicators of portal hypertension in children.

HVPG measurements were elevated in children with acute liver failure and chronic liver disease, while children with hepatic non-cirrhotic portal hypertension, acute hepatitis and non-hepatic etiologies of splenomegaly and ascites had normal HVPG measurements. Data on HVPG measurements in children with liver disease are sparse. Only two previous pediatric studies have examined HVPG measurements in children describing values ranging from 7–20 mmHg (median 13 mmHg) in children with cirrhosis and 2–33 mmHg (mean 11.3 ± 7.2 mmHg) in children with chronic liver disease.^10,11 Woolfson et al. additionally examined HVPG measurements in children with acute liver failure ranging from 0–14mmHg (median 9 mmHg). We report similar ranges for HVPG measurements in children with chronic liver disease and acute liver failure. Consistent with previous reports, HVPG measurements were both feasible and safe, even in the most critically ill children with acute liver failure.^10,11

HVPG measurements rely on accurate FHVP and WHVP measurements. Woolfson et al. noted that all children with cirrhosis (n=7) had elevated HVPG measurements. However, in our study, 2 of 7 children with cirrhosis had normal HVPG measurements, both of whom had significantly elevated FHVP (17 and 29mmHg), one with a history of pulmonary arteriovenous malformations and the other with hepatic venous outflow obstruction, respectively. The finding of an elevated FHVP should warrant further evaluation, and may result in HVPG findings underestimating the severity of portal hypertension. For example, subjects with cardiac hepatopathy have previously been described as having elevated FHVP and WHVP with normal HVPG.¹²

Normal HVPG measurements in children with cirrhosis may also be secondary to intrahepatic veno-venous shunts. Miraglia et al. first described the finding of communicating vessels between hepatic veins in children following Kasai for biliary atresia; in these cases, HVPG measurements underestimate portal pressure.¹¹ Six children were included in our retrospective analysis with biliary atresia, however all children had undergone liver transplantation and intrahepatic veno-venous shunting was not specifically noted. While the majority of children in our study had an abdominal CT or MRI, spontaneous portosystemic shunts could not be ruled out in the 30% of subjects that had abdominal imaging with ultrasound alone. However, the literature remains sparse on this subject with Simon-Telera et al reporting elevated HVPG measurements in adult subjects with spontaneous portosystemic shunts compared to subjects without shunts while Park et al found no significant difference in HVPG between adults with varices and spontaneous portosystemic shunts versus those with varices but no spontanteous portosystemic shunts.^13,14

Unlike in adults, HVPG measurements did not correlate with the degree of liver fibrosis in children.¹⁵ Notably, there was a high degree of variability in HVPG measurements, ranging from normal to elevated in every category of liver fibrosis from expansion to cirrhosis. In subjects with only fibrous expansion on liver histopathology but HVPG measurements >5mmHg, biopsy sample variability with liver fibrosis heterogeneity, or differences in WHVP values depending on which hepatic vein was used, may be explanations though typically the right hepatic vein was used with the exception of a patient with a left lateral liver transplant.^5,16

HVPG measurements were higher in children with any signs of clinical portal hypertension including ascites and/or splenomegaly and/or varices compared to children without any clinical signs of portal hypertension. Surprisingly, children with varices on either imaging or endoscopy had normal median HVPG measurements. Even in subjects with a history of variceal bleeding, a potentially life-threatening complication of portal hypertension, 50% of children showed no evidence of cirrhosis on histopathology and had normal HVPG measurements. Furthermore, of the children with HVPG measurements ≥10mmHg who underwent endoscopy, 66% showed no evidence of varices. In adults, elevated HVPG is a valid predictor of poorer clinical outcomes including variceal bleeding and mortality risk.^1,4 In contrast, our findings suggest that the use of HVPG measurements alone may not capture all children at risk for adverse clinical outcomes secondary to portal hypertension.

Limitations of this study include the small number of subjects by disease category restricting the ability to perform additional sub-analyses. However, our study is similar in size to previous publications examining HVPG measurements in children,^10,11 reflecting both the prevalence of pediatric liver disease and potentially lower utilization of HVPG measurements in an era where normative data and the predictive value of these measurements are not yet clear. While females additionally comprised a smaller proportion of subjects in our study, no previous studies have reported gender differences in HVPG measurements in adults or children. Limitations of this study also included its retrospective nature, which did not allow for a standardized technique for transjugular liver biopsy though all were performed by expert interventionalists, or a centralized reading of pathology specimens though all were read by experienced pathologists.

In summary, our findings suggest that HVPG measurements are both feasible and safe in even the most critically ill children with liver disease. HVPG measurements, however, did not correlate significantly with the degree of fibrosis on liver biopsy and were not significantly different between children with and without a history of variceal bleeding. Further prospective studies are needed to assess whether adult HVPG thresholds for portal hypertension and clinically significant portal hypertension should be applied to children, and which combination of biomarkers including HVPG measurement, histopathologic analysis, imaging, and laboratory data, accurately predict children at risk of complications from portal hypertension.

What Is Known.

• Hepatic venous pressure gradient (HVPG) measurements are safe and feasible in children

• Variceal bleeding is a significant contributing factor to morbidity and mortality in children with portal hypertension

What Is New.

• HVPG measurements do not correlate significantly with the degree of fibrosis on liver biopsy in children

• HVPG measurements do not differ significantly between children with and without a history of variceal bleeding

• HVPG measurements alone do not capture all children at risk for adverse clinical outcomes secondary to portal hypertension and additional biomarkers are likely needed

Abbreviations:

FHVP

Free Hepatic Vein Pressure

HVPG

Hepatic Venous Pressure Gradient

WHVP

Wedge Hepatic Vein Pressure