Results

The youngest patient died (patient 1). He was moribund at the time of surgery, with multiorgan failure. Echocardiography had shown multiple fistulae from the aortic root to the right and left atria and right ventricle (fig 3). These were confirmed at surgery, together with destruction of the central fibrous body and mitral annulus. He died at operation because there was no tissue to support a mitral valve prosthesis.

Two patients have required reoperation. Mitral valve repair was performed in patient 4 in addition to aortic root replacement at the original operation, but four months later he developed severe mitral regurgitation without evidence of recurrent endocarditis, and required mitral valve replacement (21 mm St Jude prosthesis). Patient 4 outgrew her homograft, and 8.3 years later underwent pulmonary autograft replacement of the aortic homograft. She still has a partial right hemiparesis, but has normal cognitive development and good exercise tolerance. All surviving patients are in sinus rhythm and New York Heart Association class I at follow up of 1.5 to 9 years.

Discussion

The diagnosis of infective endocarditis may be difficult, as may be decisions regarding the nature and timing of surgical intervention. Endocarditis should be considered in all children with severe septicaemia or embolic phenomena. An important feature of this series was the misleading presentation of three of the five patients, reflected in their initial referral to our hospital, but not to the cardiothoracic Unit, with non-cardiac diagnoses.

Death from unrecognised endocarditis may be more common than realised. In a recent paediatric series from North America, seven of 13 deaths were in patients with infective endocarditis which was not diagnosed before death.1 Six of these seven patients had not had echocardiography. Aortic root abscess was not diagnosed in any patient without previous cardiac surgery in a study of 42 infants and children with infective endocarditis seen in our Unit in the pre-echocardiography era, 1965–83.2

In the present series, the diagnosis was made using transthoracic echocardiography. However, there was no evidence of aortic root abscess, nor of aortic valve vegetations or valve dysfunction, on the first echocardiogram in four of the five patients. This shows that serial echocardiography may be required to establish the diagnosis. It is possible that transoesophageal echocardiography might have led to an earlier diagnosis. However, all but one of our patients was aged 3 years or less, and excellent transthoracic echocardiographic images were available. Transoesophageal echocardiography was available only in the last patient in the series and was not required for diagnosis or preoperative assessment. However, transoesophageal echocardiography may be the optimal diagnostic technique in older children and adults, and the recent availability of small probes has made the technique increasingly attractive in infants when transthoracic echocardiographic images are suboptimal, especially to demonstrate the subtle early appearances of aortic root abscess.

The evolution of the echocardiographic appearance of aortic root abscess involves a localised increase in echodensity in the tissue surrounding the aortic root (fig 1) before cavity formation with echolucency occurs in the perivalvar tissue (fig 2).3 ,4The evolution from an apparently normal aortic root appearance to cavity formation occurred within two days in two of our patients (3 and 4).

The timing and nature of surgery for aortic root abscess is controversial. From experience in the pre-echocardiographic era, some have considered that immediate surgical intervention is not required, because in most patients the abscess cavity will diminish in size and eventually resolve with antibiotic treatment alone.5However, experience in children has been especially limited. In a large series of 98 infants and children with infective endocarditis seen over a 13 year period, 30 underwent acute surgery, only two of whom had an aortic root abscess.6 One of these underwent debridement and aortic valve repair but required subsequent aortic root replacement, whereas the other had aortic root replacement followed by aortic valve replacement and then repeat aortic root replacement after a false aneurysm developed. The authors suggested that it is vastly preferable to complete antibiotic treatment before surgery if satisfactory control of systemic sepsis and haemodynamic status can be maintained. Our experience suggests that this is rarely possible in the presence of genuine aortic root abscess—all of our patients had aortic regurgitation with haemodynamic compromise and evidence of uncontrolled infection at the time of surgery. The rate of haemodynamic deterioration which we observed was especially alarming. There has been no known case of false positive or false negative diagnosis of aortic root abscess in our unit during this study, and no patient with aortic root abscess has been treated without surgery. Medical treatment of aortic root abscess may be theoretically attractive, but successful non-surgical management was considered sufficiently rare as recently as 1993 to justify a case report of an elderly woman who declined further surgery for prosthetic valve endocarditis and aortic root abscess.7

The development of aortic root abscess was considered a rare complication of aortic valve endocarditis in the pre-echocardiography era, but demonstration of aortic root abscess in 24 of 59 adults (41%) with fatal aortic valve endocarditis8 suggests that it was considerably underdiagnosed. In that necropsy study, risk factors for aortic root abscess were recent onset aortic regurgitation, development of atrioventricular block or pericarditis, and, importantly, a short duration of symptoms before major disability or death. This latter risk factor suggests the need for urgent surgical intervention, which remains our preference. If initial medical management is chosen, rigorous echocardiographic assessment on at least a daily basis is advisable, with recourse to urgent surgery in the presence of expanding tissue involvement, increasing severity of aortic regurgitation, or other signs of uncontrolled infection.

In this series, the median time to surgery after detection of aortic root abscess was one day. Surgery was deferred in the patient with recent cerebrovascular accident because of concern about the risk of cerebral haemorrhage during cardiopulmonary bypass in a child with extensive left temporo-parietal cerebral infarction on computed tomography. This may have been excessively cautious. In adult patients with left heart endocarditis, the only risk factor that correlated with perioperative stroke was haemorrhagic infarction on preoperative scans, whereas the timing of valve replacement, ischaemic cerebral infarction, and the size and number of cerebral infarcts were not risk factors.9

The only death occurred in the youngest patient with multiorgan failure, who was moribund by the time endocarditis and aortic root abscess with multiple fistulae and extensive tissue destruction was detected on his first echocardiogram, and this delay in diagnosis precluded successful surgery. Another patient required mitral valve replacement four months after the first operation because of failure of his mitral valve repair, although there was no evidence of persistent infection. Mitral valve repair in active endocarditis can be successful and should be attempted unless there is destruction of the subvalvar apparatus.10 All four survivors in the study remain in sinus rhythm, and avoidance of heart block may be a benefit of early intervention.

Homograft aortic valve or root replacement has been performed successfully in adult patients with active native aortic valve endocarditis complicated by aortic root abscess.11 The use of alternative prostheses may be determined by non-availability of homograft tissue.12 However, the use of mechanical bioprostheses does not preclude a good surgical result. Of 110 mainly young adult patients undergoing early surgery for aortic valve endocarditis complicated by heart failure, 35% of whom had aortic root abscess, in-hospital mortality was only 5%.13 In another series of 21 adults with native or prosthetic aortic valve endocarditis complicated by aortic root abscess, diagnosed preoperatively by echocardiography in seven, successful acute surgical repair was achieved in 20 (95%) by debridement of infected tissue and reconstruction of the left ventricular outflow tract with autologous pericardium, together with mechanical or porcine aortic valve replacement.14

The other alternative to homograft aortic root replacement for aortic root abscess is pulmonary autograft aortic root replacement (Ross operation). The patients in this series presented in an era before the widespread use of pulmonary autograft aortic root placement in children. In favour of the Ross operation, it could be argued that autologous tissue implanted to the aortic root in the presence of active infection might be less susceptible to progressive sepsis. However, infection of the homograft tissue was not a problem in this series of patients. Our preference for homograft aortic root replacement was also dictated by the fact that the Ross operation is a long and technically demanding procedure, and that expeditious surgery was required in the presence of active infection. Small children will outgrow homograft aortic root replacement. Consequently, our preference to date has been to reserve pulmonary autograft aortic root replacement for subsequent replacement of the homograft aortic root in patients presenting with aortic root abscess. However, increasing surgical experience with the Ross operation in children and accumulating evidence of good long term outcome following this procedure could mean that this will become the procedure of choice in the future, as has been advocated in adults,15 but we believe that it would be premature to make this conclusion at present.