Some major centers have adopted one-stage total correction for the management of ductus-dependent pulmonary atresia with a ventricular septal defect (PA/VSD). However, this strategy is associated with relatively high mortality and late intervention rates [1]. Moreover, it is difficult to obtain small-sized valved extracardiac conduits in some countries and nearly impossible to use an appropriately sized conduit in small neonates. In this situation, palliative surgery is inevitable.

A modified Blalock-Taussig (BT) shunt is currently used most frequently for the palliation of PA/VSD. Unfortunately, the mortality and morbidity associated with BT shunts are also high [13]. We evaluated our results from BT shunt operations performed in patients with ductus-dependent PA/VSD and observed an unexpectedly high incidence of inter-stage mortality [4]. Consequently, we changed our palliation strategy for these patients from a BT shunt to a right ventricle-to-pulmonary artery (RV-PA) shunt in 2011. Here, we retrospectively evaluated our results from RV-PA shunt operations performed since then.

Materials and Methods


Between August 2011 and August 2015, 13 patients with ductus-dependent PA/VSD underwent the RV-PA shunt procedure for palliation. Patients with an arborization anomaly of the pulmonary artery or major aortopulmonary collateral arteries were excluded. The mean age of the patients at surgery was 17.9 ± 15.3 (range 5–60 days), and their mean body weight was 2.9 ± 0.6 (range 2.2–4.0) kg. During the same period, a modified BT shunt was performed in a premature low-birth-weight baby, RVOT patch widening in one patient, and total correction in a 4-month-old infant.

Surgical Techniques

The RV-PA shunt procedure was performed under mild hypothermic cardiopulmonary bypass through a median sternotomy. Gore-Tex tube grafts (Gore-Tex; WL Gore and Associates, Flagstaff, AZ, USA) were used in all patients, with 4-, 5-, and 6-mm Gore-Tex tube grafts used in 1, 10, and 2 patients, respectively. First, in the beating heart, we made an anastomosis between the main pulmonary artery or the patch used for the left pulmonary artery angioplasty and the distal end of the Gore-Tex tube graft using an 8-0 polypropylene suture. Then, we induced cardioplegia and made a small ventriculotomy at the right ventricular outflow tract (RVOT). Some hypertrophic muscle bundles were excised, and the proximal anastomosis was followed using a continuous fine Gore-Tex suture (CV6, Gore-Tex; WL Gore and Associates, Flagstaff, AZ, USA) under the arrested heart. When the oxygen saturation was too high after bypass weaning (>95%), the RV-PA shunt was banded with a same-sized tube graft strip (3 mm in width). Excluding one patient who had postoperative extracorporeal membrane oxygenation (ECMO), the cardiopulmonary bypass time was 91.3 ± 35.5 min and aortic cross-clamp time was 24.3 ± 5.8 min. Concomitant procedures were required in three patients, all being left pulmonary artery angioplasties. All patients were given intravenous heparin 5 mg/h soon after the bleeding had stopped to avoid shunt thrombosis. The heparin was replaced by oral aspirin when oral intake resumed.

Data Collection and Follow-up

All records were collected retrospectively after approval of the institutional review board of Pusan National University Yangsan Hospital. Twelve patients were followed in the Pediatric Cardiology Department over a mean duration of 26.0 ± 16.8 months. The timing of definitive surgery was usually dependent on the patient’s systemic oxygen saturation.


Hospital Course

There was no case of early hospital mortality. One patient required postoperative ECMO because of failure of cardiopulmonary bypass weaning. This patient had persistent severe hypoxemia after changing the shunt for a larger one, probably because of high pulmonary vascular resistance. This patient recovered without any sequelae after 5 days of ECMO support and underwent definitive surgery 17.1 months after the RV-PA shunt procedure. Another patient who required ECMO at the intensive care unit (ICU) experienced sudden massive bleeding from the RVOT suture site in the ICU 2 days postoperatively. Cardiopulmonary bypass was required to repair the suture line dehiscence, and ECMO was applied subsequently. The ECMO was weaned without any problems after 5 days. The patient also had successful definitive repair at the age of 11.5 months. Overall, the mean postoperative ventilation time, ICU stay, and hospital stay were 6.4 ± 7.9 days, 7.7 ± 7.5 days, and 21.2 ± 14.4 days, respectively. The mean systemic oxygen saturation was 90.2 ± 3.5% at discharge (Table 1).

Table 1 Patient data

Inter-stage Course

There were two inter-stage deaths at 34 and 47 days after operation, respectively. One patient died of aspiration at home, and the other patient died of RVOT pseudoaneurysm rupture. No autopsy was performed in either patient. Five patients (41.7%) required catheter intervention, for juxtaductal left pulmonary artery stenosis in three patients, right pulmonary artery stenosis in one, and shunt inflow stenosis in one. Two patients (15.4%) required re-operation because of shunt inflow stenosis and RVOT pseudoaneurysm, respectively (Table 1).

Definitive Repair

All patients who survived the RV-PA shunt underwent total correction at a mean interval of 13.1 months. The mean age at definitive repair was 13.7 ± 5.0 months, and the mean body weight was 8.2 ± 1.9 kg. The mean systemic oxygen saturation before the definitive operation was 84.4 ± 8.5%. A home-made tricuspid expanded polytetrafluoroethylene valved conduit was used in all patients: 12 mm in two patients, 14 mm in seven and 16 mm in one. There have been no deaths since the definitive operations.


The RV-PA shunt has been widely used as a palliative procedure in patients with hypoplastic left heart syndrome and has been used for resuscitation of a diminutive pulmonary artery in patients with PA/VSD with major aortopulmonary collateral arteries. In 2008, Bradley et al. [1] introduced the RV-PA shunt as a palliative procedure in patients with a biventricular heart with a single source of pulmonary blood flow. Traditionally, palliation for such patients was provided using a modified BT shunt. Although the modified BT shunt is a simple procedure that has been used for over 50 years, it poses a risk when performed, especially in low-birth-weight neonates. It is also associated with significant inter-stage mortality [3, 5]. In the study by Bradley et al. [1], there was no case of early mortality among four patients weighing less than 2.5 kg or two patients weighing less than 2 kg. This implies that the RV-PA shunt can be performed safely in low-birth-weight babies. The RV-PA shunt has some clear advantages over the modified BT shunt, although it requires cardiopulmonary bypass. There is no diastolic runoff in the systemic arterial system, which is observed after a systemic-pulmonary arterial shunt operation. This suggests that the coronary blood flow is not compromised by the lower diastolic pressure [6]. Pulmonary blood flow occurs only during cardiac systole, and the absence of diastolic runoff in the arterial system allows the placement of a larger diameter shunt (usually 5 or 6 mm in diameter) compared with the modified BT shunt, which can prevent early or late shunt occlusion. In addition, more even growth of both pulmonary arteries can be expected than that with a modified BT shunt. Zheng et al. [7] compared pulmonary artery growth between the RV-PA shunt and a systemic-to-pulmonary artery shunt and concluded that the RV-PA shunt demonstrates better pulmonary artery growth. However, the RV-PA shunt also has some disadvantages compared with the modified BT shunt; besides a need for cardiopulmonary bypass, the RV-PA shunt requires a ventriculotomy and RV volume load through a valveless conduit. Unlike in hypoplasitc left heart syndrome, the ventriculotomy is not a disadvantage in this biventricular lesion, which inevitably requires a ventriculotomy during the definitive repair. In fact, in our experience, the RV-PA shunt procedure can limit the ventriculotomy size. Only a small ventriculotomy is usually required during the RV-PA shunt procedure, and we found no need to extend the previous ventriculotomy during the definitive repair in almost all cases. We used the dilated fibrotic RVOT tissue around the proximal end of the conduit. We usually closed the VSD through the right atrium, rather than through the ventriculotomy, to reduce the ventriculotomy size to the extent possible. We believe that the ventriculotomy in the RV-PA shunt procedure is much smaller than that in early total correction. Regarding the RV volume load through the valveless conduit, the long-term effect of volume loading during the palliation period is not clear. Further studies are needed to obtain more information on this issue.

Is the RV-PA shunt really a safe procedure? We cannot definitively say “yes” because we observed some serious early and late complications in this study. First, we had a case with persistent hypoxemia that required ECMO. The hypoxemia persisted even after changing the shunt with a larger one. We believe that the persistent hypoxemia was probably caused by the high pulmonary vascular resistance seen in neonates. Second, we observed RVOT pseudoaneurysms or dehiscence of the suture line, which are both serious complications. We believe that those complications resulted from an improper fit of the hard Gore-Tex tube graft to the very friable neonatal RVOT muscle and unavoidable high postoperative RV pressure. Zheng et al. used autologous pericardium in many of their patients who were older infants and children. They did not report RVOT pseudoaneurysms [7]. Quintessenza et al. [8] introduced a method of constructing the RV-PA conduit in the Norwood operation, known as the “dunk technique.” We believe that this technique can be used for the RV-PA shunt in PA/VSD to reduce the RVOT incision and decrease the chance of an RVOT pseudoaneurysm. Third, we observed two cases of shunt inflow stenosis, which may have been caused by a technical error or turbulence-induced endothelial hypertrophy. This complication can be avoided by an appropriate ventricular incision and muscle bundle excision. Fourth, longer postoperative ventilator time is thought to be another risk of the RV-PA shunt. We compared the postoperative ventilation times between patients with an RV-PA shunt and those with a BT shunt (n = 22) between July 2003 and April 2010. The postoperative ventilation time was much longer in the former than in the latter patients (6.4 ± 7.9 vs. 2.3 ± 2.0 days, p = 0.026).

In conclusion, much attention should be paid to the development of serious complications, such as persistent hypoxemia, RVOT psuedoaneurysm, suture line dehiscence, and shunt inflow stenosis when performing the RV-PA shut procedure, although the RV-PA shunt procedure is an option for palliation of ductus-dependent PA/VSD.