Patient #1

A female infant was delivered at 38 weeks gestation after a normal pregnancy. A murmur was detected on the first day of life, and a diagnosis of tetralogy of Fallot with pulmonary atresia was confirmed by an echocardiogram. A right-sided classic Blalock-Taussig shunt was constructed at the age of one month, using absorbable sutures as is our practise.^{4, 5} Subsequently, at the age of 2 years, we inserted without complications a left-sided modified Blalock-Taussig shunt made of a Gore-Tex tube of 5 mm diameter (W. L. Gore & Assoc., Flagstaff, Arizona).

After a further period of 1 year and 9 months, she was admitted to our hospital in a cyanotic and febrile state. On questioning, it emerged that she had been admitted twice within three months to another hospital with a fever of unknown origin, her temperature exceeding 39°C. Antibiotics had been administered, with alleviation of the fever, in both instances. Clinical examination on this occasion revealed that the continuous murmur expected from the modified Blalock-Taussig shunt was no longer audible, and the patient showed marked pallor. Her white blood cell count was 4.1 × 10⁹/l, and the C-reactive protein level was normal, at 1 mg/l. Cardiac catheterization showed that the left-sided modified Blalock-Taussig shunt was occluded, while the classical right-sided shunt was narrowed. A filling defect was noted in the left pulmonary artery at the pulmonary end of the left-sided shunt (Fig. 1). The index of the left ventricular end-diastolic volume was 36 ml/m². Deeming this value too small, we decided that a definitive repair was not indicated at this point.

Figure 1. A pulmonary angiogram in our first patients showing the occluded modified Blalock-Taussig shunt. Note the filling defect in the left pulmonary artery at the pulmonary end of the shunt (arrow).

At operation to explant the shunt, we found no sign of infection outside the shunt, which was occluded by vegetations that culture positive for Acinetobacter lwoffii. An old thrombus was found around the pulmonary insertion of the shunt, and was removed. Postoperative pathological examination of the thrombus revealed severe infiltration by neutrophils. There was no indication of infection around the left subclavian artery or the left pulmonary artery, so the shunt was replaced with a Dardic shunt (Bio-Vascular, Inc. New Jersey) of 5 mm diameter placed in the same position. She made an uneventful recovery after 4 weeks of appropriate antibiotics given intravenously, followed by 26 weeks of oral antibiotics as indicated by the cultures and sensitivity studies. We performed definitive repair at the age of 5 years and 11 months of age, when we ligated both shunts.

Patient #2

A male infant was delivered at 35 weeks gestation after a normal pregnancy. Cyanosis was noted on the first day of life, and prostaglandin E₁ was started immediately after an echocardiographic diagnosis of tricuspid atresia with severe pulmonary stenosis. Balloon atrioseptotomy was performed when he was aged 3 months, along with construction of a right-sided classic Blalock-Taussig shunt. He was then managed as an outpatient until a left-sided modified Blalock-Taussig shunt was constructed without complication using a Gore-Tex tube of 4 mm diameter when he was aged 1 year and 2 months.

He was subsequently admitted 11 weeks postoperatively with a 2-day history of pyrexia. On clinical examination, his clinical condition was unchanged, and a murmur was clearly heard from the shunt, but there were no other significant physical signs. Microscopy and culture of urine were normal, and the echocardiographic findings were unchanged. A chest radiograph, however, demonstrated a well-demarcated mass in the left-sided and superior part of the mediastinum (Fig. 2a). His white blood cell count was 30.2 × 10⁹/l, and the C-reactive protein level was elevated, at 370 mg/l. A blood culture was taken, and broad-spectrum antibiotics were initiated.

After 5-days of treatment, the white blood cell count was 17.9 × 10⁹/l, and the C-reactive protein level had fallen to 127 mg/l, but the pyrexia had not resolved. The chest radiograph showed an increase in size of the mediastinal mass (Fig. 2b). The blood culture, nonetheless, proved negative, but we decided to remove the left-sided modified Blalock-Taussig shunt.

Figure 2. The chest radiograph of our second patient (a) is shown at the time of admission. A subsequent film taken five days later (b) shows an increase in the size of the mass in the left superior mediastinum.

Before we removed the left-sided shunt, we fashioned a right-sided bidirectional cavopulmonary shunt using a temporary bypass from the superior caval vein to the right atrium. When we performed a left thoracotomy, we found a large amount of clear fluid in a thick-walled cavity beside the shunt. No obviously infected fluid was seen in the thoracic cavity, nor were signs of infection evident around the left subclavian artery and the left pulmonary artery. No lymphatic leaks or distended lymphatic vessels were encountered. It was clear that the fluid had been leaking through the Gore-Tex graft itself, since an obvious transudation was clearly visible on its surface. When we clamped the shunt, the patient developed severe hypoxia, so we decided that the shunt was still needed, both to avoid desaturation and to permit growth of the pulmonary artery. We therefore replaced the original shunt with a new Gore-Tex tube of 4 mm diameter placed in the same position. No vegetations were found in the explanted shunt, but it proved culture-positive for Streptococcus viridans. The patient made an uneventful recovery after 3 weeks of appropriate intravenous antibiotic therapy, followed by 15 weeks of oral antibiotics as indicated by the cultures and sensitivity studies. When he was 2 years and 11 months of age, we performed a total cavo-pulmonary connection, and ligated the original shunt.

Discussion

Infection of a modified Blalock-Taussig shunt is a very rare, and also very serious, complication.^{6, 7} Most experienced vascular surgeons faced with such an infection elect to remove the entire graft, and construct a new shunt through uninfected tissues on the other side of the thorax to supply pulmonary arterial flow. While this approach has offered the best chance of eradicating infection, it obviously involves major, high-risk surgical procedures. Both of our patients, however, had already had shunts constructed on the opposite side to the infected shunts. We reasoned, as had others,⁸ that if the infection of the modified Blalock-Taussig shunt was localized, and there was no sign of infection at the anastomotic site, simple replacement of the graft might prove adequate for treating these patients, who already had an alternative source of pulmonary arterial supply.

The clinical features, laboratory tests, and investigations such as computed tomography or magnetic resonance angiography, can facilitate diagnosis of infection of such shunts. In our cases, the final proof of infection came from culture of the excised graft. In patients with a modified Blalock-Taussig shunt, nonetheless, infection should always be considered when the patient has a fever of unknown origin. Once antibiotic therapy has failed, there is the risk of rupture at the site of anastomosis of the shunt. In such cases, therefore, early surgical intervention becomes mandatory. Should febrile patients also have a seroma around the shunt, replacement of the graft should also be considered.⁷

Despite evidence supporting the need for short periods of antibiotic prophylaxis, the need for prolonged prophylaxis after cardiovascular surgery remains questionable.⁹ Usage of antibiotics to treat infection of the graft, nonetheless, remains important.¹⁰ In our patients, we administrated antibiotics over a period of 30 weeks for our first patient, and for 18 weeks for the second patient. Further studies are needed to clarify the optimal duration of antibiotic therapy.

Even though we achieved success in both our patients by replacing the graft to control the infection, several aspects of this approach warrant further discussion. Although one of our patients was febrile and leukocytotic on admission, repeated blood cultures were negative. The infection, therefore, was almost certainly in an early phase, as was also the case in our second patient. Since we did not encounter sepsis of the grafts, with positive blood cultures, or signs of infection at the site of anastomosis such as pseudoaneurysm or hemothorax, we can only speculate that patients with either septicemia or pseudoaneurysm would have responded in the same fashion.

To our knowledge, replacement of an infected prosthetic graft with a new tube inserted on the same side has not previously been reported. Since the new shunt was not removed from the patients at the time of definitive repair, long-term follow up will be needed to clarify the efficacy of the operation.

Acknowledgement

We thank Michael Kopcak for his editorial advice.