Left ventricular-to-right atrial communications are rare types of ventricular septal defects known collectively as the Gerbode defect.¹ Acquired forms of this defect have been described as a complication of infective endocarditis due to septal perforation by the infective process^2–4 usually involving the aortic valve or aneurysm of the interventricular septum.⁵ This rare form of acquired defect has also been described following aortic or mitral valve replacement surgery and has been attributed to excessive debridement of the calcified mitral or aortic annulus during valve replacement, which causes a breakdown of the central fibrous body, leading to a left ventricular-to-right atrium shunt.^6–8 Surgical correction of this acquired shunt, which carries a high operative mortality rate, remained the only therapeutic option until recently. We describe a patient with a chronic left ventricular-to-right atrial (LV-to-RA) fistula following biosprosthetic valve replacement of the mitral valve and a novel transcatheter approach to close this defect, thereby avoiding repeat open-heart surgery.

       Case Report. A 78-year-old female was diagnosed with mitral valve prolapse secondary to

Figure 1

Color Doppler in the four-chamber subcostal view showing the left ventricle-to-right atrial shunt (white arrow).

myxomatous degeneration with extensive mitral annular calcification and coexisting ischemic heart disease. For her symptomatic ischemic heart disease, she had undergone stent placement in the left anterior descending artery (LAD). In 2003 for her severe symptomatic mitral regurgitation, she underwent mitral valve replacement with a bioprosthetic valve (Carpentier Edwards no. 31). Her coronary angiogram prior to surgery showed a patent stent to the LAD, but an additional lesion in the posterior descending artery (PDA) to which the right internal mammary artery (RIMA) was grafted. Postoperatively, the patient did well and showed significant improvement in her symptom class. She was hospitalized 6 months after her surgery with complaints of increasing exertional dyspnea. When she was referred to our department, she was in Class IV dyspnea with frank right ventricular failure Clinical examination on admission revealed evidence of congestive hepatomegaly, engorged neck veins, peripheral edema and a loud 3/6 holosystolic murmur in the right lower parasternal region, as well as an additional clinical and radiologic finding of massive pleural effusion. The echocardiography performed in the course of that hospitalization revealed a normal left ventricular systolic function with a normally-functioning mitral prosthesis, and moderate right ventricular and right atrial enlargement. There was evidence of severe tricuspid regurgitation (TR) with a dilated and noncollapsing inferior vena cava (IVC), as well as prominent hepatic vein flow reversal. Also noted was a large (8 mm) communication between the left ventricular outflow tract (LVOT) and the right atrium

Figure 2

Transthoracic echocardiography (parasternal short axis) showing the anatomical defect (white arrow).

(Figures 1–3). This fistulous tract was seen arising from the posteromedial aspect of the LVOT just below the aortic valve. She was treated for her underlying failure and was taken for cardiac catheterization after her general condition had stabilized. Cardiac catheterization revealed evidence of severe pulmonary hypertension (87/22 mmHg with a mean of 50 mmHg). The patient’s mean right atrial pressure was 16 mmHg with a normal pulmonary wedge pressure. Her oximetry run revealed a step up in the oxygen saturation in the right atrium of 10% with Qp/Qs of 1.6:1, and a PVR/SVR of 0.2. Coronary angiography revealed a patent stent in the proximal LAD and a normally-functioning RIMA to the right PDA. Her LV angiogram, performed in two oblique views, revealed a significant left-to-right shunt with opacification of the right atrium. Following

Figure 3

Transthoracic echocardiography (parasternal short axis) with color flow demonstrating the shunt across the left ventricle-to-right atrial defect (white arrow).

catheterization, in view of its proximity to the aortic valve, it was decided to refer the patient for surgical correction of the shunt. However, due to her underlying general condition and her in situ RIMA to the right PDA, she was refused by three surgical teams. Hence, it was decided to perform percutaneous transcatheter closure of the defect. The risks of the procedure, including the need for a permanent pacemaker, were explained to the patient.

       Procedure. The transcatheter procedure was performed under general anesthesia and under transesophageal guidance. The patient received heparin (100 IU/kg) with additional boluses to maintain activated clotting time (ACT) more than 250 seconds. A temporary pacemaker was inserted from the right femoral vein. A left ventricular angiogram was performed in the left anterior oblique view, which demonstrated the left ventricle to right atrial shunt (Figure 4). Using a 6 Fr Multipurpose catheter (MPA-2, Cordis Corp., Miami, Florida), from the right femoral vein, an attempt was made to enter the defect from the right side of the heart. After many unsuccessful attempts it was then decided to enter the defect from the left side of the defect. Subsequently a straight tipped 0.035 exchange wire (Cook, Inc., Bloomington, Indiana) was passed through a 5 Fr right coronary catheter (Cordis) and the defect was crossed from the left ventricle. The wire was the passed into the right atrium and into the IVC. This wire was then

Figure 5

Post-deployment left ventricular angiogram in the LAO cranial view showing the device in situ (white arrow) and a trivial shunt.

snared with a Multisnare set no. 7 (PFM, Germany) from the right femoral vein, and an arteriovenous loop was created. A 7 Fr Amplatzer delivery sheath was advanced on this arteriovenous loop from the venous end and positioned in the arch of the aorta. We decided to use a 10–8 mm Amplatzer duct occluder (ADO) device (AGA Medical Corp., Golden Valley, Minnesota). The duct occluder was attached to the delivery cable and advanced into the LV and then into the aortic arch. The distal disk of the ADO was opened in the aortic arch, and the entire assembly was withdrawn into the LV and into the mouth of the defect under fluoroscopic and

Figure 4

Left ventricular angiogram in the left anterior oblique view with cranial angulation demonstrating the opacification of the right atrium across the defect (white arrow) prior to device closure.

echocardiographic guidance. Before delivery of the proximal portion of the device, echocardiography confirmed that there was no impingement of the mitral, tricuspid and aortic leaflets by the device. A left ventricular angiogram was then performed and showed a very trivial opacification of the right atrium, which was predominantly through the device. After reconfirming the absence of a residual shunt by transesophageal echocardiography, the device was delivered. Postdeployment, a left ventricular angiogram was performed once again, showing a very trivial shunt (Figure 5). Fluoroscopic confirmation of device stability was also obtained. During the entire procedure, the patient was hemodynamically stable and had no underlying rhythm disturbances. No additional shunt was seen on echocardiography (Figure 6). Clinical examination revealed disappearance of the holosystolic murmur and postprocedure electrocardiography revealed the presence of an incomplete right bundle-branch block. Transthoracic echocardiography,

Figure 6

Transesophageal echocardiography post-deployment demonstrating the device in situ and absence of a shunt.

performed 24 hours postprocedure, revealed the device in situ and no residual shunt. There was moderate tricuspid regurgitation (severe prior to the procedure), and the pulmonary artery systolic pressures estimated from the TR jet was 65 mmHg. We expect a gradual reduction in the right atrial and right ventricular volumes and in the severity of the patient’s pulmonary hypertension.

       Discussion. Failure of expected clinical improvement after prosthetic valve replacement is usually attributed to prosthetic valve dysfunction, perivalvular leak and residual valvular lesions. Silverman et al,⁹ in their review of this iatrogenic complication, noted the development of such left ventricular-right atrial fistula in 5 patients after mitral valve replacement alone, 2 of whom died following both aortic valve replacement and mitral valve replacement (the other 3 patients died later), and in 7 patients following aortic valve replacement alone (of whom 3 died later). This condition leads to very early (< 6 months) and progressive congestive heart failure if left uncorrected. Anatomically, this fistula after mitral valve replacement is typically located superior to the septal leaflet of the tricuspid valve, near its junction with the anterior leaflet. A probe, if passed through the fistula from the right side, will appear in the aortic outflow tract beneath the annulus in the area of the noncoronary cusp. The defect is produced by injury to the membranous septum. Injury to this structure, either due to technical misadventure or extensive debridement of calcium (our patient had extensive mitral annular calcification) during both aortic valve replacement and mitral valve replacement, can potentially cause this complication. Intraoperative diagnosis of this complication should, therefore, be considered in these circumstances. The operating surgeon can find possible evidence of this complication by palpation of a thrill over the left ventricular outflow tract or the right atrium near the base of the heart. It can be confirmed by digital exploration through the right atrial appendage after the patient has been removed from bypass and visualization of a jet of blood felt coming from the left ventricle during systole. Prevention of this complication entails great caution during debridement of calcium or excision of a valve in the area of a posteromedial commissure. Marsten et al¹⁰ suggested a slightly asymmetric seat for the prosthetic ring to avoid injury to the atrioventricular septum. Increased availability and application of transcatheter interventional techniques have offered unique opportunities to definitively treat congenital or acquired intracardiac shunts in the interventional catheterization laboratory.
       From the closure of the common congenital intracardiac shunts, device therapy has been extended now to the use of periprosthetic valve leaks. The novel use of these devices, as was the case in our patient, has only been mentioned once before and also involved use of the Amplatzer duct occluder.¹¹ The design of the Amplatzer duct occluder allows for assessment of the device’s impact on surrounding cardiac structures prior to release. Sizing of the device is of utmost importance, as larger devices could impinge on the medial prosthetic leaflet, leading to significant restriction of mobility and valve dysfunction. If this occurs, a device of the next smaller size can be substituted. We feel that following the successful and encouraging results achieved in our patient, this form of therapy will find more clinical applications in such surgically high-risk patients.