Twenty years experience of surgical aortic valvotomy for critical aortic stenosis in early infancy☆

The most appropriate first treatment approach for neonates with critical aortic stenosis remains controversial. In the past surgical aortic valvotomy with cardiopulmonary bypass (CPB) pioneered by Lillehei et al. [1] in 1956 was standard. But various operative techniques have been used including hypothermic circulatory arrest, inflow occlusion, transventricular valvotomy with blunt dilatation of the valve. These operations remained the only therapeutic options until 1983, when Lababidi [2] reported the interventional use of aortic balloon valvotomy. Both surgical valvotomy and transcatheter balloon valvotomy are associated with mortality and morbidity and with residual or recurrent valve dysfunction. Both are palliative procedures: sooner or later, reintervention is likely. Actually the choice between the two approaches varies according to the local expertise and/or preference. We favor open valvotomy in preference to percutaneous balloon valvuloplasty. This study evaluates results of this policy by estimating survival of patients and freedom from reintervention for aortic valve stenosis or regurgitation. Between March 1983 and March 2003, 34 consecutive patients younger than 3 months of age (29 girls, 5 boys) underwent open valvotomy for critical aortic stenosis in Sankt Augustin, Germany. Median age was 21 days (range 2–62 days). Twenty-six patients were neonates. Median weight at operation was 3.3 kg (range 1.9–6.3 kg). Thirty-two patients fulfilled Turley et al.'s [3] definition of ‘isolated’ aortic stenosis. This excludes major associated malformations other than coarctation of the aorta, in particular lesions not amenable for biventricular repair. This study also includes two neonates with associated subvalvular aortic stenosis and a small ventricular septal defect. Criteria for diagnosis of hypoplastic left heart syndrome (HLHS) in our unit are based on echocardiographic findings of aortic valve atresia or hypoplasia with aortic valve annulus diameter less than 4.5 mm, mitral valve atresia or hypoplasia with mitral valve annulus diameter less than 7 mm, and absent or hypoplastic left ventricle with a long axis at least 80% smaller than the long axis of the right ventricle. Until 1996, HLHS neonates were referred to an overseas Center in accordance with parents’ wish. Norwood palliation has been offered in our unit intermittently from 1997 and constantly from 2001. Thus, in the period between January 2001 and March 2003, 19 patients underwent Norwood procedure, whereas two underwent surgical aortic valvotomy. In the study period, primary aortic balloon dilatation was performed on 30 occasions: for premature neonates weighing less than 1.5 kg, in case of conditions contraindicating cardiopulmonary bypass, cerebral hemorrhage in particular, and when the surgeon was not readily available. Fourteen patients were directly admitted in the intensive care unit because of cardiorespiratory failure. Ductal dependency was present in five patients who required prostaglandin E1 treatment. The echocardiography with Doppler study showed a preoperative mean peak aortic valve gradient of 72.4 ± 20.9 mmHg. This gradient was over 50 mmHg in all patients except two with impaired left ventricular function (shortening fraction < 28%). Associated major cardiovascular anomalies were endocardial fibroelastosis (n = 8), dilatative cardiomyopathy (n = 1), aortic coarctation (n = 3) with hypoplastic aortic arch in one patient, subvalvular aortic stenosis (n = 2), and ventricular septal defect (n = 1). One of the three aortic coarctations was postoperatively detected. Three patients (8.8%) had a monocuspid, 16 patients (47.1%) a bicuspid, and 15 patients (44.1%) a tricuspid aortic valve. At operation, the stenosis was found to result from fusion of the zone of apposition between leaflets in all patients and from thickening and myxomatous changes of leaflets in 31 patients (91%). Obstruction was mainly due to these changes in four cases. After median sternotomy, surgical right atrial cannulation, and aortic cannulation, all patients were placed on hypothermic cardiopulmonary bypass. Myocardial protection was obtained with cold antegrade crystalloid (St. Thomas solution) repeated at 20–30 min intervals and topical hypothermia; short periods of circulatory arrest were also used when necessary to improve visualization. After transverse aortotomy, a careful commissurotomy just short of the aortic annulus was performed in order to avoid development of aortic regurgitation. Obstructive myxomatous and fibrous nodules on the leaflets were removed. This implied careful and meticulous thinning/shaving of leaflets. Modified ultrafiltration at the end of CPB was applied in all patients since 1992. Mean CPB time was 50 ± 43 min (range 12–273 min), median aortic cross-clamp time was 24 min, whereas total circulatory arrest in eight patients lasted 10 min. Hospital survivors were seen in outpatient clinics at regular intervals by pediatric cardiologists. Two-dimensional echocardiography and Doppler studies were routinely performed before discharge from the hospital and during each outpatient visit. The transvalvular aortic peak pressure gradient was calculated using the simplified Bernoulli equation: gradient = 4V2, where V is the peak instantaneous transvalvular Doppler velocity. Aortic valve incompetence was graded in our unit by mapping the dimensions of the regurgitation jet with color flow Doppler echocardiography [4]. A peak gradient of at least 50 mmHg and progression to Grade 3 aortic regurgitation were indication for evaluation by heart catheterization to discuss reintervention. All hospital survivors, except one patient, have been followed up between 16 and 261 months (mean 115 ± 67 months) postoperatively. Data collected retrospectively included patients’ demographics, anatomical diagnoses, clinical status at presentation, echocardiography and cardiac catheterization, operations, and follow-up clinical status. They are expressed as median or mean ± standard deviation and/or percentages with 95% confidence limits (CL). Patients were divided into three groups according to mortality and morbidity as follows: survivors without reintervention (Group A), survivors with reintervention (Group B) and non-survivors (Group C). The three groups were compared to identify the changing patterns in characteristics, management, and outcomes using χ2 tests and Krunskal Wallis analysis of variance. Risks for death were assessed by univariate analysis. Kaplan–Meier curves for actuarial survival, freedom from reintervention including re-operation or interventional catheterization were calculated using the GraphPad Prism (San Diego, CA, USA). The level of statistical significance was set at a p value of less than 0.05. Two neonates died early on the first operative day from heart failure: operative mortality of 6% (2/34), 0.7–19.6% (95% CL). Significant risk factors for early mortality as displayed in Table 1 were leaflet morphology (monocuspid valve, p = 0.005), presence of endocardial fibroelastosis (p = 0.013) and impaired left ventricular function (shortening fraction < 28%, p = 0.026). Comparison of preoperative and operative variables of the three groups Mean duration of endotracheal intubation for survivors was 116 ± 127 h. Hemodynamic instability required delayed sternal closure in eight patients. Surgical re-exploration was necessary in three patients because of sustained bleeding. Low cardiac output with renal insufficiency led to peritoneal dialysis in five patients. One patient developed thrombosis of the subclavian vein requiring embolectomy by interventional catheterization. In the same patient a permanent pacemaker had to be implanted after resection of a subaortic stenosis and VSD closure during the primary repair. Twenty-five patients (74%) made an uneventful recovery. No patient died late. One patient was lost to follow-up. Seven out of 31 patients (23%) needed a re-operation (n = 6) or reintervention by catheterization (n = 1). Concerned were five previous bicuspid valves, one monocuspid and one tricuspid valves. The indication for the second surgical intervention was recurrent aortic stenosis in two patients, Grade 3 aortic insufficiency in two patients, combined aortic regurgitation and stenosis with subvalvular stenosis in one patient, and combined valvular and subvalvular aortic stenosis in the remaining patients. The mean interval between the first and the second procedure was 79 ± 62 months (range 6–168 months). One patient underwent percutaneous balloon valvuloplasty 61 months after initial repair. Re-aortic stenosis was surgically addressed by repeat open valvotomy/commissurotomy (n = 3) with subvalvular myomectomy (n = 2) and Ross procedure (n = 1). The Ross operation was for re-aortic obstruction of a monocuspid valve 93 months after primary repair. In the patient with combined aortic regurgitation, a hole in one leaflet was found and closed by a pericardial patch. The other two patients with aortic incompetence underwent aortic valve replacement with a mechanical prosthesis size 21 mm, 149 and 168 months after open valvotomy. One of these patients required pacemaker implantation because of a postoperative complete heart block. There was no additional mortality with any of these reinterventions. Overall outcome is schematically summarized in Fig. 1. Kaplan–Meier estimate for time-related freedom from reintervention for aortic valve dysfunction after initial valvotomy is shown in Fig. 2. It is 85.1 ± 6.9%, 78.0 ± 9.3%, and 53.5 ± 15.9% at 5, 10, and 15 years, respectively. Thirty-four consecutive patients, 26 neonates and 8 infants younger than 3 months, underwent surgical valvotomy for aortic stenosis. Seven patients needed re-operation (n = 6) or catheter reintervention (n = 1). Indication for the second intervention was recurrent aortic stenosis in two patients, severe aortic insufficiency in two patients, a combined aortic regurgitation and stenosis with left ventricular outflow tract obstruction (LVOTO) in one patient and recurrent aortic stenosis with LVOTO in another patient. The procedure was percutaneous balloon valvuloplasty in one patient, repeat open valvotomy in one, repeat open valvotomy and LVOTO resection in one, aortic valve plasty and LVOTO resection in one, ROSS operation in one, and aortic valve replacement in two patients. AR: aortic valve regurgitation, AS: aortic valve stenosis, AVP: aortic valvuloplasty, AVR: aortic valve replacement, BD: balloon dilatation, ED: early death, RP: repeat procedure. Kaplan–Meier estimate of freedom from reintervention for aortic stenosis or regurgitation after surgical aortic valvotomy (n = 34). Vertical bars represent standard error of the mean (SEM). Echocardiographical examination in all but two survivors before discharge from hospital demonstrated a good left ventricular function. In five patients, the aortic valve function was normal. In the remaining 27 survivors, regurgitation and/or residual stenosis were present. Incompetence was Grade 1 in 18 patients and Grade 2 in 4. The mean peak systolic pressure gradient across the aortic valve was 21.3 ± 13.7 mmHg. The gradient was less than 30 mmHg in 21 patients and between 30 and 50 mmHg in 6 patients. There were no significant differences in the immediately postoperative aortic pressure gradient between Group A and Group B patients (Table 2). Early and late functional status in survivors After repeat open valvotomy/commissurotomy (n = 3), echocardiography before hospital discharge showed Grade 1 aortic insufficiency in all patients. The peak systolic transaortic gradient was between 25 and 35 mmHg. Regurgitation was Grade 2 in the patient who secondarily underwent pericardial patch repair of the aortic valve. For the patient whose recurrent aortic stenosis was managed by percutaneous balloon valvuloplasty, peak systolic gradient dropped from 100 to 30 mmHg. One patient was lost to follow-up. There was no late death. Twenty-six patients (84%) were in New York Heart Association (NYHA) functional class I. Five patients (16%), NYHA class II, either presented with mild symptoms or on medical treatment especially captopril (n = 2) because of aortic regurgitation or digoxin (n = 2). All patients, except two children with pacemaker implantation, have retained normal sinus node function. At the latest echocardiography examination for the 28 survivors with native aortic valve, aortic regurgitation was Grade 1 in 10 patients and Grade 2 in 14 patients. Sixteen patients had a systolic ejection gradient across the aortic valve less than 30 mmHg, 7 patients had a peak gradient between 30 and 50 mmHg, and 5 patients had more than 50 mmHg. The mean gradient was 29.8 ± 18.2 mmHg. There were no significant differences in the latest postoperative aortic pressure gradient between Group A and Group B (Table 2). This study bears the usual drawbacks of retrospective series over a long period. In particular, there were variations with regard to patient selection for percutaneous balloon valvuloplasty and to criteria for reintervention. Nevertheless, this series is relatively homogeneous as primary surgical valvotomy was performed by the same surgeon. Critical aortic stenosis commonly presents in the neonatal period with a poor clinical condition. In this series, 14 patients (41%) had to be admitted into intensive care unit because of cardiorespiratory failure. Decision for further management was quickly taken to enhance chance for survival. In the period under review, HLHS cases were rapidly detected in our unit and neonates with isolated aortic stenosis were primary assigned to aortic valvotomy, unless CPB was contraindicated. At surgery, it is of paramount importance to thin/shave aortic leaflets that are usually thickened by myxomatous tissue. Indeed, obstruction can mainly be caused by nodular excrescences and protruding immature tissue on the cups, as seen in four of our cases. This policy resulted in a low early and late mortality, a low occurrence of significant aortic regurgitation and a low early recurrence of aortic stenosis, with freedom from reintervention for aortic valve dysfunction of 78.0 ± 9.3% ten years after initial valvotomy. Early mortality after neonatal surgical aortic valvotomy, which was high in the past (as high as 52.1% in the series of Gaynor et al. [5]), has been reduced in recent publications to 0–18% [6–10]. Impaired left ventricle function and endocardial fibroelastosis are well-recognized risk factors for early mortality [11,12]. Monocuspid valve, per se, constitutes a surgical risk, as demonstrated by Bhabra et al. [7] and as shown by this study: two among the three patients with monocuspid valve died and the survivor had to undergo Ross procedure 93 months later. Although surgical or catheter primary management of critical aortic stenosis is still debated, there is agreement that, in trained hands, both methods carry, actually, similar early mortality. It appears that the critical factors predicting outcome are systolic function and morphology of the left ventricle (LV) and the aortic valve. When the LV is small, with endocardial fibroelastosis, and dysfunctions, neither surgical nor balloon valvotomy gives good results. Monocuspid aortic valves are poor candidates for both methods, albeit theoretically worse for catheter procedure, as obstruction relief will inevitably result in aortic regurgitation. Fortunately the great majority of stenosed aortic valves are bicuspid or tricuspid, so that both managements may be discussed. The problem really concerns the residual or recurrent valve dysfunction after the primary procedure. As both procedures are palliative procedures, the question is freedom from reintervention over time for aortic valve stenosis or regurgitation. In a recent paper, Balmer et al. [13] showed that aortic regurgitation was frequently observed among neonates and infants with previous balloon valvotomy; freedom rate from reintervention was only 35% at 3 years. At this age, if the aortic valve cannot be repaired, there is no other choice than Ross procedure or aortic valve replacement by allograft. Survivors after primary valvotomy in most surgical series, like ours, enjoy much longer interval without reintervention, 10-year freedom rates from reintervention varying between 55 and 90% [6,10,14,15]. At this interval, the aortic valve annulus is usually big enough to accommodate an adult-size mechanical valve, if needed. It is not surprising that neonatal aortic balloon valvotomy frequently ends up with significant aortic regurgitation. Valve stenosis is due, in great majority of cases, not only to the fusion of rudimentary commissures but also (sometimes mainly) to irregular cups thickening. These excrescences protrude into valve orifice and hamper its opening. At surgery, it is possible to directly inspect the aortic valve and to carry out valve debridement. This is not achievable with the ‘blind’ balloon dilatation. Instead, cups are likely to get torn or perforated while trying to relief obstruction. By comparing survivors with and without reintervention, we could not find any parameter that might predict earlier postoperative failure of the aortic valve, most probably because of the small number of patients. It is to be presumed that trileaflet valves behave better than bileaflet ones [7]. It is noteworthy that reintervention was rarely due to significant aortic regurgitation. Overall long-term prognosis for hospital survivors was very good with regard to late survival, freedom from reintervention for aortic valve dysfunction and heart functional status, all being in NYHA class I or II. In our view, surgical aortic valvotomy under hypothermic cardiopulmonary bypass and cardioplegic myocardial protection remains the most favorable approach in the treatment of critical aortic stenosis in neonates and young infants. In centers where surgical expertise is readily available, this management will continue to challenge the balloon valvuloplasty in terms of better long-term outcomes.