Outcomes of Esophageal Varices in Adults With Fontan Palliation and Liver Cirrhosis

Background The purpose of this study was to define the risk and outcomes of esophageal varices in adults with Fontan palliation and liver cirrhosis undergoing esophagogastroduodenoscopy (EGD). Method The results of EGD, abdominal ultrasound, and liver biopsy, as well as clinic notes from the hepatologist, were reviewed to determine the diagnosis of cirrhosis and esophageal varices. The incidence of acute gastrointestinal bleeding complication was assessed among patients with esophageal varices using the time of EGD as the baseline. Results Of 149 patients with Fontan palliation and liver cirrhosis, the prevalence of esophageal varices at baseline EGD was 34% (51 of 149). Of 98 patients without esophageal varices at baseline EGD, 27 (27%) underwent subsequent EGD, of whom 11 showed a new diagnosis of esophageal varices. The incidence of a new diagnosis of esophageal varices was 9% per year. Of 62 patients with esophageal varices, 9 (15%) had acute gastrointestinal bleeding complications during 45 (37-62) months of follow-up, yielding an incidence of 5% per year. Of the 9 patients, 8 underwent EGD and variceal banding during the hospitalization for bleeding and 1 patient died of septicaemia. Of the 8 patients who survived to hospital discharge, 2 patients were readmitted for esophageal bleeding within 12 months from the index hospitalization. Higher hepatic vein wedge pressure and hepatic vein pressure gradient were associated with esophageal varices and bleeding complications. Conclusions In this selected sample of adults with Fontan palliation and liver cirrhosis, esophageal varices were relatively common, and patients with esophageal varices had risk of bleeding complications.

Liver cirrhosis is common in adults with Fontan palliation because of chronic hepatic congestion inherent in the Fontan physiology.8e11 In health, the portal vein provides more than 70% of hepatic blood supply, and the portal circulation is unique because it relies on a low transhepatic gradient for hepatic blood flow (typically less than 5 mm Hg) to maintain liver perfusion and to decompress the splanchnic circulation.12e14 In contrast, the structural remodelling in liver cirrhosis results in an increased transhepatic gradient, and this in turn requires a higher portal venous pressure (portal hypertension) to generate the required pressure gradient to perfuse the liver.12e14 A chronic exposure to high portal venous pressure (portal hypertension) results in shunting of blood through the portosystemic circulation leading to formation of esophageal varices, which can sometimes present with life-threatening esophageal variceal bleeding.12e14 In spite of the high prevalence of cirrhosis among adults with Fontan palliation, there are limited data about the prevalence, risk factors, and outcomes of esophageal varices in this poplation. 5,15The purpose of this study was to define the clinical characteristics of patients with esophageal varices, identify correlates of esophageal varices, and delineate longterm clinical outcomes after the diagnosis of esophageal varices in adults with Fontan palliation undergoing esophagogastroduodenoscopy (EGD).

Study population
This is a retrospective cohort study of adults (aged !18 years) with Fontan palliation and liver cirrhosis who underwent EGD at the Mayo Clinic between January 1, 2003, and December 31, 2022.The diagnosis of liver cirrhosis was based on 2 criteria: abdominal ultrasound scan showing coarse echotexture consistent with chronic parenchymal disease and a clinical evaluation by the hepatologist confirming the diagnosis of cirrhosis.
All adults with Fontan palliation followed at the Mayo Clinic undergo an initial hepatology evaluation, typically at the time of initial presentation.This evaluation comprises a clinic visit with a Fontan hepatologist, abdominal imaging (ultrasound, computed tomography scan, or magnetic resonance imaging), and laboratory blood tests (comprehensive metabolic panel, complete blood count, hepatitis serology, and alpha-fetoprotein).Based on the outcome of this evaluation, some patients may be referred for EGD, invasive haemodynamic assessment, or liver biopsy.The patients with cirrhosis undergo abdominal ultrasound, comprehensive metabolic panel, and alpha-fetoprotein every 6 months and a clinical evaluation in a hepatology clinic every 1-2 years.The patients without cirrhosis undergo abdominal ultrasound, comprehensive metabolic panel, and alpha-fetoprotein every 12 months and a clinical evaluation in a hepatology clinic every 3-5 years.
The electronic health records were reviewed including clinic notes, procedure notes, echocardiogram, cardiac catheterization report, and laboratory data.The date of EGD was regarded as the baseline encounter, and the clinical data obtained within 6 months from the baseline encounter were used to define the baseline clinical characteristics.
Method: The results of EGD, abdominal ultrasound, and liver biopsy, as well as clinic notes from the hepatologist, were reviewed to determine the diagnosis of cirrhosis and esophageal varices.The incidence of acute gastrointestinal bleeding complication was assessed among patients with esophageal varices using the time of EGD as the baseline.Results: Of 149 patients with Fontan palliation and liver cirrhosis, the prevalence of esophageal varices at baseline EGD was 34% (51 of 149).Of 98 patients without esophageal varices at baseline EGD, 27 (27%) underwent subsequent EGD, of whom 11 showed a new diagnosis of esophageal varices.The incidence of a new diagnosis of esophageal varices was 9% per year.Of 62 patients with esophageal varices, 9 (15%) had acute gastrointestinal bleeding complications during 45 (37-62) months of follow-up, yielding an incidence of 5% per year.Of the 9 patients, 8 underwent EGD and variceal banding during the hospitalization for bleeding and 1 patient died of septicaemia.Of the 8 patients who survived to hospital discharge, 2 patients were readmitted for esophageal bleeding within 12 months from the index hospitalization.Higher hepatic vein wedge pressure and hepatic vein pressure gradient were associated with esophageal varices and bleeding complications.Conclusions: In this selected sample of adults with Fontan palliation and liver cirrhosis, esophageal varices were relatively common, and patients with esophageal varices had risk of bleeding complications.qui une oesophagogastroduod enoscopie (OGD) a et e r ealis ee.M ethodologie : Les r esultats de l'OGD, de l' echographie abdominale et de la biopsie du foie, ainsi que les notes cliniques de l'h epatologue ont et e consult es pour etablir les diagnostics de cirrhose et de varices oesophagiennes.L'incidence des complications h emorragiques gastrointestinales aiguës a et e evalu ee chez les patients pr esentant des varices oesophagiennes en utilisant l'OGD initiale comme r ef erence de d epart.R esultats : Chez les 149 patients ayant fait l'objet d'une intervention de Fontan et pr esentant une cirrhose h epatique, la pr evalence des varices oesophagiennes lors de l'OGD initiale etait de 34 % (51/149).Parmi les 98 patients sans varices oesophagiennes lors de l'OGD initiale, 27 (

Study objectives
The aims of the study were to determine (1) the prevalence of esophageal varices in Fontan patients with liver cirrhosis undergoing EGD, and the correlates of esophageal varices; (2) the incidence of a new diagnosis of esophageal varices in subsequent EGDs in patients without esophageal varices identified in prior EGD; and (3) the incidence and risk factors for acute gastrointestinal (GI) bleeding complication during follow-up.Acute upper GI bleeding complication was defined as hematemesis associated with at least 2 of the following features: decrease in haemoglobin by !2 g/dL, need for blood transfusion, or urgent medical evaluation in the emergency department.16e18

Data abstraction
The EGD reports and the clinic note from the hepatologist were reviewed to determine the presence, size, and location of esophageal varices.Based on these reports, esophageal varices were defined as large (>5 mm) or small ( 5 mm).Abdominal ultrasound reports were reviewed to determine the splenic size, presence of ascites, and presence of liver cirrhosis.Results of viral hepatitis serology and alcohol intake were retrieved from hepatology evaluation.Alcohol intake was classified as none (<1 drink/wk), light/moderate (1-14 drinks/wk), and heavy intake (>14 drinks/wk). 19iver histologic data were abstracted from the pathology reports in the patients who underwent liver biopsy.The liver specimens were stained with trichrome and reticulin stains.Portal fibrosis was assessed using the Batts-Ludwig (stages 0-4) staging system, and sinusoidal fibrosis was staged (0-4) as previously described. 6,9,20Similar to previous studies, 6 we dichotomized the patients into those with no sinusoidal fibrosis (stage 0) vs sinusoidal fibrosis (stages 1-4) and into patients without portal fibrosis (F0) vs with portal fibrosis (F1-4).
Cardiac catheterization was performed on chronic medications in the fasting state and under mild sedation. 21,22As previously described, venous access was obtained via internal jugular vein or femoral vein, and venous catheterization was performed using a 7 F balloon wedge catheter. 21,22Arterial catheterization was performed via a radial arterial or femoral arterial access using a 4 F or 5 F pigtail catheter for a direct measurement of arterial pressure and arterial oxygen saturation.Venous oxygen saturation was directly measured from samples drawn from the branch pulmonary arteries (mixed venous saturation) and pulmonary artery wedge position (pulmonary venous saturation).A saturation of 95% was used as the assumed pulmonary venous saturation in cases where the pulmonary venous saturation was not measured directly.Arterial saturation was measured from radial artery, femoral artery, or ventricular blood samples.Cardiac output was determined by the indirect Fick technique using assumed oxygen consumption and directly measured oxygen contents in the pulmonary artery and systemic circulations. 21,22For the assessment of hepatic haemodynamics, the catheter position in the hepatic vein was confirmed by appearance on fluoroscopy and contrast angiography before the measurement of free hepatic vein pressure and hepatic vein wedge pressure.Hepatic vein pressure gradient was calculated as hepatic vein wedge pressure minus free hepatic vein pressure. 3

Statistical analysis
Data were presented as median (interquartile range [IQR]), and count (%).Between-group comparisons were performed using the unpaired t test, Wilcoxon rank-sum test for continuous variables, and Fisher exact test.Logistic regression analysis was used to assess the correlates of esophageal varices.The cumulative incidence of acute GI bleeding complication was assessed using the Kaplan-Meier method and Cox regression.Missing data were addressed using conditional amputation. 23All statistical analyses were performed with the BlueSky Statistics software (version 7.10; BlueSky Statistics LLC, Chicago, IL), and a P value of <0.05 was considered to be statistically significant for all analyses.
At the time of the diagnosis of esophageal varices, 16

New diagnosis of esophageal varices on subsequent EGD
Of 98 patients without esophageal varices on baseline EGD, 27 (27%) underwent subsequent EGD, and the median interval between the first and the second EGD was 51 (IQR: 43-86) months.The EGDs were performed as part of the routine Fontan monitoring protocol, and none of the patients had a prior history of GI bleeding.Of the 27 patients, 11 (41%) had a new diagnosis of esophageal varices on the second EGD.All 11 patients had small esophageal varices involving the lower third of the esophagus, and 6 patients had associated portal gastropathy.The incidence of a new diagnosis of esophageal varices was 9% per year.At the time of the diagnosis of esophageal varices, 4 of the 11 (36%) patients were on selective b-blocker therapy, and nonselective bblocker therapy was initiated in another 7 patients.
All 9 patients with GI bleeding complications were hospitalized.The median decline in haemoglobin concentration was 2.4 (IQR: 1.7-3.1)g/dL, and 8 of the 9 patients required blood transfusion.Of the 9 patients, 8 underwent EGD and variceal banding during the index hospitalization (hospitalization for acute GI bleeding).One of the 9 patients with variceal bleeding died during the index hospitalization for esophageal bleeding because of septicaemia and hepatic encephalopathy.Of the 8 patients who survived to hospital discharge, 2 patients were readmitted for GI bleeding within 12 months from the index hospitalization.

Discussion
In this study, we assessed the prevalence, correlates, and outcomes of esophageal varices among adults with Fontan palliation and liver cirrhosis who underwent EGD.The main findings were as follows: (1) The prevalence of esophageal varices at baseline EGD was 34%.(2) Among patients without esophageal varices at baseline EGD, the annual incidence of new diagnosis esophageal varices on subsequent EGDs was 9% per year.(3) The correlates of esophageal varices diagnosis (at baseline and subsequent EGD) were higher hepatic vein wedge pressure and higher hepatic vein pressure gradient.(4) Among patients with the diagnosis of esophageal varices, the annual incidence of acute GI bleeding complication was 5% per year, and the 5-year cumulative incidence of major upper GI bleeding complication was 26%.(5) The correlates of acute GI bleeding complications were large esophageal varices, higher hepatic vein wedge pressure, and higher hepatic vein pressure gradient.
Chronic liver disease is relatively common in adults with Fontan palliation, and the prevalence of liver cirrhosis has been estimated at 20%-30% in this population.1e7 Despite the high prevalence of cirrhosis among adults with Fontan palliation, there are limited data about the prevalence, risk factors, and outcomes of esophageal varices in this population.1e5,9,24 The available data suggest that the presence of ACEI/ARB, angiotensin-converting enzyme inhibitor/aldosterone receptor antagonist; AVV, atrioventricular valve; CHD, congenital heart disease; EF, ejection fraction; HV, hepatic vein; IAC, intra-atrial conduit; IQR, interquartile range; MRA, mineralocorticoid receptor antagonist; NYHA, New York Heart Association; PAWP, pulmonary artery wedge pressure; PVR, pulmonary vascular resistance; Qs, systemic blood flow; WU, Woods unit.esophageal varices, in combination with other features such as ascites, splenomegaly, and thrombocytopenia, was associated with adverse outcomes. 5To the best of our knowledge, this is the first systematic analysis of the disease burden of esophageal varices among adults with Fontan palliation and liver cirrhosis.Although the analysis was based on a selected population, it provides clinically relevant estimates of disease prevalence and bleeding risk of esophageal varices in this population.
These findings have important clinical implications with regards to monitoring of adults with Fontan physiology.Currently, the only definitive treatment for Fontan-associated liver disease is heart-liver transplant. 25,26However, this therapy is often reserved for patients with advanced or end-stage disease. 25,26For the majority of patients with Fontan physiology who do not meet criteria for heart-liver transplant, medical management centres around the treatment of reversible haemodynamic lesions (surgical or transcatheter therapy), treatment of complications of portal hypertension using medical therapy (diuretics and b-blockers), and paracentesis, and close surveillance for liver-related complications such as hepatocellular carcinoma and esophageal varices. 4,27The results of the current study demonstrating esophageal varices in one-third of adult Fontan patients with liver cirrhosis suggest that a baseline EGD should be performed in all adults Fontan patients with cirrhosis.Fontan patients without liver cirrhosis were excluded from the current study, and hence we are unable to comment on the diagnostic yield or clinical benefit of screening for esophageal varices in that subset of patients.
Another important clinical implication of the results of this study is the relatively high incidence of esophageal variceal bleeding in this population.In the non-Fontan population, the detection of esophageal varices provides opportunity for prophylactic therapy such as use of nonselective b-blocker therapy, somatostatin analog, and esophageal variceal banding. 12,28,29These therapies have been shown to decrease portal venous pressure and reduce the risk of acute GI bleeding based on data from the non-Fontan population. 12,28,29In the current study, we did not observe any association between b-blocker therapy and esophageal varices,  but these results should be interpreted with caution because the study was not designed to test the efficacy (or lack thereof) of these prophylactic therapies.

Limitations
This is a retrospective, single-centre cohort study, and it is therefore prone to selection and ascertainment bias.The use of b-blocker therapy or the criteria for esophageal variceal banding were not standardized.Finally, we did not adjust for potential confounders in the statistical model because of the small sample size.These limitations may limit the generalizability and reproducibility of the results.

Conclusions
Esophageal varices were relatively common in adults with Fontan palliation and liver cirrhosis and were associated with the risk of acute GI bleeding complications.These data support the need for screening for esophageal varices among patients with Fontan palliation and liver cirrhosis.Further studies are required to assess opportunities to prevent the development of varices, the diagnostic yield of less invasive screening strategies such as abdominal computerized tomography scan, and to determine the optimal therapy to prevent life-threatening bleeding in patients with esophageal varices.It is important to emphasize that these data were derived from a selected patient population with liver cirrhosis, and hence the prevalence of esophageal varices (and in turn the yield of EGD) maybe lower in other cohorts.b-Blocker therapy was modeled as a categorical variable using the patients who were not on any therapy as the reference group.AVV, atrioventricular valve; CI, confidence interval; EGD, esophagogastroduodenoscopy; HR, hazard ratio; HV, hepatic vein; PAWP, pulmonary artery wedge pressure; PVR, pulmonary vascular resistance; Qs, systemic blood flow.
31%) patients were on b-blocker therapy (selective b-blocker n ¼ 14 and nonselective n ¼ 2).Nonselective b-blocker therapy was initiated in another 29 (57%) patients.Of the other 6 patients, 4 patients had a prior history of intolerance to an initial trial of b-blocker therapy for atrial arrhythmias, mostly because of chronotropic incompetence and severe fatigue.The remaining 2 patients have no prior history of bblocker use as can be determined from the review of medical records.There was no statistically significant difference in the proportion of patients on b-blocker therapy at the time of EGD among the patients with esophageal varices vs patients without esophageal varices (16 of 51 [31%] vs 39 of 98 [40%], P ¼ 0.3).

Figure 1 .
Figure 1.Forest plot showing correlates of esophageal varices based on univariable logistic regression analysis (top left).The forest plot showing correlates of esophageal variceal bleeding based on univariable Cox regression analysis (bottom left).The Kaplan-Meier curve showing the cumulative incidence of esophageal variceal bleeding (right).The variceal size was modelled as a binary variable, and large varices were defined >5 mm in size.CI, confidence interval; HR, hazard ratio; HV, hepatic vein; OR, odds ratio.

Table 1 .
Baseline characteristicsBetween-group comparisons were based on the unpaired t test and Wilcoxon rank-sum test for continuous variables and the Fisher exact test for categorical variables.

Table 2 .
Hepatic indicesBetween-group comparisons were based on the unpaired t test and Wilcoxon rank-sum test for continuous variables and the Fisher exact test for categorical variables.

Table 3 .
Univariable b-Blocker therapy was modeled as a categorical variable using the patients who were not on any therapy as the reference group.AVV, atrioventricular valve; CI, confidence interval; HV, hepatic vein; OR, odds ratio; PAWP, pulmonary artery wedge pressure; PVR, pulmonary vascular resistance; Qs, systemic blood flow.

Table 4 .
Univariable Cox regression models for correlates of upper GI bleeding