Surgical Improvement of Hepatic Venous Mixing to Resolve Systemic Arterial Hypoxemia Associated with Post-Fontan Pulmonary Arteriovenous Fistulae

Abstract

Pulmonary arteriovenous fistulae are known to develop in patients who have functional single-ventricle heart disease and interruption of the inferior vena cava with direct hepatic drainage to the heart, in which a bidirectional Glenn shunt is the only source of pulmonary blood flow. The progressive systemic arterial hypoxemia that is associated with pulmonary arteriovenous fistulae can have important clinical consequences. Baffling the hepatic venous return to the pulmonary circulation can alleviate pulmonary arteriovenous fistulae.

Herein, we present the case of a 13-year-old patient with modified Fontan anatomy and pulmonary arteriovenous fistulae, in whom redirection of a previously placed hepatic venous-to-right pulmonary artery conduit was required in order to increase systemic arterial oxygen saturation. Revision of the conduit improved mixing of hepatic venous effluent with blood flow from the bidirectional Glenn shunt. Three years after this revision, the patient's oxygen saturation remained stable at 90%, and his physical activity was markedly improved. We present our rationale for selected redirection of the conduit and discuss other surgical options that can improve hypoxemia that is associated with pulmonary arteriovenous fistulae.

Pulmonary arteriovenous fistula (PAVF) can develop in patients who have undergone placement of a bidirectional Glenn shunt for single-ventricle heart disease that is associated with interruption of the inferior vena cava (IVC) and direct hepatic venous drainage to the heart.^1,2 The progressive systemic arterial hypoxemia that is associated with PAVF can have important clinical manifestations. It has been reported that baffling hepatic venous return to the pulmonary circulation can alleviate PAVF. Here, we present and discuss the case of a 13-year-old modified-Fontan patient with PAVF, in whom redirection of a previously placed hepatic venous-to-right pulmonary artery (PA) conduit was required in order to overcome unfavorable streaming and to increase systemic arterial oxygen saturation levels.

Case Report

We evaluated a 13-year-old boy who had been born with an unbalanced (left-dominant) atrioventricular canal, pulmonary outflow obstruction, and interruption of the IVC with hemiazygous continuation to a left superior vena cava. The hepatic venous confluence drained directly into the right side of a common atrium. At 15 months of age, he underwent placement of a left-sided bidirectional Glenn shunt (Kawashima procedure) and ligation of the main PA. At 3 years of age, his systemic arterial oxygen saturation had fallen to 80% on room air. At age 4 years, his hepatic veins were baffled to the right PA with use of a nonfenestrated extracardiac 16-mm GORE-TEX® tube graft (W.L. Gore & Associates; Flagstaff, Ariz).

Initially after the Fontan procedure was completed, the patient's oxygen saturation increased to percentage levels in the high 80s. However, over the ensuing 2 years, the readings progressively declined to the low 60s; he required supplementary oxygen and developed worsening fatigue and exercise intolerance. Cardiac catheterization showed no pressure drops of consequence across the cavopulmonary and hepatopulmonary pathways (average pressure, 14 mmHg). Angiography showed neither PAVF nor any abnormal systemic venous collateral vessels connecting with the heart. However, the right lung was receiving almost the entire hepatic venous flow through the extracardiac conduit (Fig. 1). Contrast echocardiography was used as a means of detecting diffuse PAVF. Ten mL of hand-agitated saline solution was injected into selected PA branches. Transesophageal echocardiography was used to evaluate bubble contrast returning to the pulmonary veins and left atrium. Selective injection of the saline solution into the right PA showed only faint opacification of the right pulmonary veins. Conversely, injection of the solution into the left PA revealed rapid and dense opacification of the left pulmonary veins and the left atrium. We concluded that revision of the extracardiac conduit would be required in order to distribute hepatic venous return more equally to both sides of the pulmonary circulation.

Fig. 1 Selective angiogram of the hepatic veins shows that the right lung receives almost the entire hepatic venous flow through the extracardiac conduit.

After a midline sternotomy and the institution of cardiopulmonary bypass, the patient underwent alteration of the hepatic venous-to-right PA conduit. The distal end of the hepatic venous conduit was disconnected from the right PA, and the residual opening in that artery was closed. A portion of the conduit's free end was removed. The remaining part of the original conduit was then connected to the left PA (just below the Kawashima anastomosis) with use of a 16-mm interpositional GORE-TEX tube graft, which coursed beneath the ascending aorta (Fig. 2). The anastomosis was performed with use of running 5-0 Prolene sutures. The patient was weaned from cardiopulmonary bypass without incident. His postoperative course was uncomplicated, and he was discharged from the hospital after 10 days with an arterial oxygen saturation level of 72% on room air. After 4 months, the oxygen saturation increased to a mid-80% level, and, 1 year postoperatively, to as high as 94%. Three years after the revision of the conduit, the patient was doing well: the arterial oxygen saturation remained stable at 90% or higher, and his physical activity was markedly improved.

Fig. 2 Drawing of the modified conduit that courses beneath the ascending aorta.

LPA = left pulmonary artery; LSVC = left superior vena cava; RPA = right pulmonary artery

Discussion

Pulmonary arteriovenous fistulae can develop when no hepatic venous blood reaches the lung vasculature. On the basis of experience with other hepatopulmonary diseases, investigators have postulated that an as-yet undefined biochemical agent in hepatic venous blood provides protection against the development of PAVF.^1,3 It is known that the surgical inclusion of hepatic venous blood in the pulmonary circulation can facilitate the regression of PAVF.^3,4 Baskett and colleagues⁵ successfully treated a patient who had PAVF by baffling the hepatic veins to the azygous vein, which connected the interrupted IVC to a right superior vena cava. Pike and co-authors⁶ reported the resolution of PAVF after they rerouted hepatic venous return to a hemiazygous vein that connected to a left superior vena cava.

In our patient, the conventional extracardiac hepatic venous-to-right PA conduit resulted in unfavorable streaming, with most of the hepatic venous blood entering the right PA. Consequently, PAVFs were detected chiefly in the left lung, in comparison with the right lung. Redirecting the hepatic venous blood flow to achieve adequate perfusion of both pulmonary arteries resulted in marked improvement in the patient's systemic arterial oxygen saturation. Instead of connecting the original hepatic venous-to-right PA conduit to the hemiazygous vein, we chose to connect it to a region beneath the anastomosis of the left superior vena cava and the left PA. We believed that the uncurbed passage of hepatic venous blood to the left PA would yield better mixing and provide a less thrombogenic and more energy-sparing blood-flow pathway than would the alternative—the nearly right-angle course necessary for attachment of the conduit to the hemiazygous vein.

We acknowledge the potential for conduit obstruction with either type of hepatic venous-to-left PA connection. Our patient's case may be special, in that sufficient room was available for us to pass the interpositional graft inferior to the ligated main PA and still position it along the underside of the ascending aorta without inviting obstruction. In other patients, this type of Fontan completion may not be feasible, given the often small space behind the aortic root. Regardless, careful clinical follow-up is required in order to check for signs of hepatic congestion after the redirection of hepatic venous return. We intend to determine the permanence of the observed improvement in our patient's oxygen saturation levels (the reversal of PAVF).

In summary, patients with single-ventricle complex, interruption of the IVC, and direct hepatic venous drainage to the heart are at risk of developing PAVF. During the placement of a bidirectional Glenn shunt, the surgeon can retain a degree of antegrade pulmonary blood flow by not fully tying off the main PA, or by instead taking alternative measures to provide hepatic venous effluent to the pulmonary circulation. Moreover, early incorporation of all hepatic venous return to the pulmonary arteries—with a conduit that is positioned to enable adequate mixing with the blood flowing through a bidirectional Glenn shunt—can provide protection against PAVF, and even reverse the condition.