Abstract
Objective: A small sub-group of patients with hypoplastic left heart syndrome (HLHS) have normal-sized ascending aorta and arch. An alternative to the Norwood I procedure in these patients is the creation of an aorto-pulmonary (AP) window with a distal pulmonary artery band (PAB). We reviewed our experience with this technique and compared outcomes to the Norwood procedure for HLHS. Methods: All patients treated for HLHS in a single institution between 1992 and 2005 were analysed. This identified 13 patients treated with AP window and PAB compared to 333 patients undergoing stage I Norwood procedure. An unrestrictive AP window was created and the main PA was banded. Patient records and echocardiograms were analysed. Median follow-up was 10 (IQR 0–655) days and 100% complete. Results: There were seven early deaths (54%) in the AP window group and two conversions to Norwood circulation. This was a significantly worse outcome than for the Norwood procedure over the same period, which had an early mortality of 29% (p = 0.03). Kaplan–Meier actuarial analysis demonstrated a continued survival benefit of the Norwood group at 6 months (p = 0.0005). Deaths were due to either low cardiac output syndrome (n = 4) or sudden unheralded arrest (n = 3). This occurred despite aortic cross-clamp and circulatory arrest times being significantly lower in the AP window group compared to the Norwood group (35 ± 27 vs 55 ± 16 min, p < 0.01 and 16 ± 29 vs 55 ± 20 min, p < 0.01, respectively). No differences in arterial saturations or systolic blood pressure existed between the groups, but diastolic blood pressure was significantly lower in the AP window group at 27 ± 10 mmHg compared to 42 ± 8 mmHg in the Norwood group (p = 0.01) with evidence of flow reversal in the descending aorta. Differences in diastolic blood pressure between groups were abolished after conversion to stage II. Conclusions: Despite favourable anatomy and shorter ischaemic times, the AP window/PAB technique has a poor outcome compared to the Norwood procedure for HLHS. Low diastolic blood pressure with reversal of descending aortic flow in diastole was a feature of the AP window/PAB circulation. We recommend the Norwood procedure for these sub-types. This may have implications for newer ‘hybrid’ procedures for HLHS which create a similar palliative circulation.
1 Introduction
Hypoplastic left heart syndrome (HLHS) includes a variety of anatomical sub-types that vary from classical aortic and mitral atresia through to unbalanced AVSDs and borderline ‘small’ left ventricles (LVs) with forward flow across the aortic valve. The Norwood procedure and its modifications have emerged as a successful palliative approach that can be applied to all variants of HLHS with improving results since its first application in the early 1980s.
HLHS is characterised by a varying degree of hypoplasia of the ascending aorta and transverse arch with aortic coarctation. The success of the Norwood procedure is partly due to the fact that it addresses these components whilst providing adequate coronary blood flow. However, within the spectrum of HLHS there is a small sub-group of patients who have a normal-sized ascending aorta and arch with no coarctation but an unequivocally hypoplastic LV which is inadequate to support the systemic circulation. We proposed that the Norwood physiology could be reproduced in these patients through the creation of an unrestrictive aorto-pulmonary (AP) window and distal pulmonary artery band (PAB). This approach had the attraction of avoiding extensive aortic arch surgery, and achieving a similar palliative circulation with a simpler and shorter procedure. The right ventricle would support the systemic circulation through the AP window whilst the PAB protected the pulmonary circulation from excessive flow.
This study set out to compare the results of this procedure with the standard Norwood procedure carried out at a single institution since the development of the HLHS programme here in 1992.
The results have particular relevance to newer ‘hybrid’ techniques of palliation for HLHS that propose ductal stenting with bilateral PA banding [1]. These techniques propose a very similar physiology to that created by the AP window/PAB approach.
2 Patients and methods
The Norwood programme for HLHS began in 1992 at the Diana, Princess of Wales, Birmingham Children’s Hospital, UK. Between 1992 and August 2005, a total of 333 children with HLHS have been treated. A retrospective review of the departmental database was undertaken to identify these patients, and their hospital records were reviewed to identify those patients receiving palliation with an AP window and PAB during this period.
During this period the techniques for the Norwood procedure have evolved in line with those recorded in the literature and the classical technique employing a systemic–pulmonary artery shunt has been replaced by the RV–PA conduit. These results have been published previously and this database was used as the reference group for evaluation of the AP window/PAB technique [2,3].
Suitability for the AP window approach was defined as a normal-sized ascending and transverse aorta (greater than 4.5 mm ascending aorta and transverse arch ≥weight +1 mm) with no evidence of coarctation. These criteria were all based on preoperative trans-thoracic echocardiography. The presence of forward flow in the ascending aorta and the presence or absence of aortic atresia were not considered in decision-making but were recorded in the analysis. Patient characteristics are given in Table 1 .
Patient characteristics
Patient characteristics
2.1 Operative technique
The great vessels were fully mobilised and the duct and branch PAs controlled with slings. Cardiopulmonary bypass was established between high in the ascending aorta and a single atrial cannula. The duct was ligated and deep hypothermia established at full flows using an α-stat strategy. The aorta was then cross-clamped and the heart arrested with cold crystalloid cardioplegia (St Thomas’ solution at 30 ml kg−1). A longitudinal incision was made into the facing sides of the ascending aorta and the main PA (5–6 mm in length). The openings were enlarged anteriorly and a small amount of tissue was excised as necessary such that an unrestrictive opening existed (accepting a size 6 dilator). The posterior walls of the facing aorta and PA were then joined together with a running suture and, after placing the band (see below), this conjoined area was roofed over with a patch of bovine pericardium to complete the window.
According to the patients’ weight a 4 mm (weight > 2.5 kg) or 3 mm (weight ≤ 2.5 kg) dilator was then passed though the window across the PA bifurcation. A braided 3/0 nylon suture was then placed within the adventitia of the distal PA and was tied down over the dilator. The window was then completed by roofing over the AP connection with a patch of bovine pericardium (Fig. 1 ). Short periods of circulatory arrest were used to provide a bloodless field if necessary and to allow for the atrial septectomy which was performed through the venous cannulation site. Postoperative echo is shown in Fig. 2 with the Doppler recording showing continuous flow in Fig. 3 .
Schematic diagram showing the AP window with pulmonary artery band.
Schematic diagram showing the AP window with pulmonary artery band.
Colour flow Doppler imaging showing the AP window with pulmonary artery band.
Colour flow Doppler imaging showing the AP window with pulmonary artery band.
Continuous Doppler ultrasound showing the continuous flow signal across the AP window.
Continuous Doppler ultrasound showing the continuous flow signal across the AP window.
The techniques used for the Norwood procedure have been outlined elsewhere [2,3]. The exact technique has evolved over the period of this study from the classical technique with a systemic–pulmonary shunt (n = 260, 78%) to the RV–PA conduit (n = 73, 22%). The Norwood procedure was performed at a median age of 4 days (IQR 3–7). The majority (n = 265, 80%) were operated within the first 7 days and only 13 patients (3.9%) were operated at >30 days of age. The median weight at operation was 3.1 kg (IQR 2.8–3.5). The weight at operation Z-score was 1.17–1.22, based on the British 1990 reference population [4]. The median body surface area (BSA) at operation was 0.20 m2 (range 0.14–0.33 m2).
Delayed sternal closure was utilised routinely in all patients and inotropic support was provided with milrinone 0.5 μg kg−1 min−1 or dobutamine 10 μg kg−1 min−1 plus adrenaline 0–0.3 5 μg kg−1 min−1 to maintain mean arterial pressure >40 mmHg. We have previously described a risk model for the Norwood I procedure for HLHS based on the results from this institution. This uses the most significant factors from a multivariate analysis and produces a predicated risk of death based on patients’ ventricular function, size of ascending aorta and type of Norwood operation [2]. This model was used to analyse the AP window group to establish their predicted risk if they had gone through a Norwood pathway.
Follow-up information was obtained from the patients’ records and through contact with the referring cardiologists if necessary; median follow-up was 62.1 months and was 100% complete.
2.2 Statistical methods
Comparisons between groups were performed using the unpaired Student’s t-test. Survival was calculated using the Kaplan–Meier method and comparison between groups performed using the log-rank test. Statistics were performed using the SPSS for Windows software package (version 12, SPSS Inc., Chicago, Ill.) and probabilities of p < 0.05 were accepted as showing statistical significance.
3 Results
The AP window group was significantly older than the Norwood group with a median age of 19 days (IQR 8–24 days) versus median age of 4 days (IQR 3–7 days). This reflects the fact that three patients with a borderline LV had survived an initial trial of conservative therapy, but remained duct dependent and were referred to this centre for further management. Two patients had undergone attempts at balloon aortic valvuloplasty at another centre prior to referral; both had resulted in moderate aortic incompetence.
There were seven early deaths (54%) in the AP window group compared to an early mortality of 29% in the entire Norwood group (p = 0.03). This was clinically significant. Four of the deaths occurred from low cardiac output state: two died, earlier in the series, on the operating table unable to wean from cardiopulmonary bypass. The first patient was profoundly acidotic preoperatively and was only stabilised following a balloon septostomy. On discontinuing bypass the patient became severely desaturated. The band was loosened and, despite two further attempts to wean off bypass, the patient died of right ventricular dysfunction, desaturations and low cardiac output. The second patient also became profoundly desaturated with low cardiac output following bypass discontinuation. The remaining two patients died within the first 18 h after returning from theatre in a low cardiac output state unresponsive to increasing doses of adrenaline, in established renal failure on peritoneal dialysis despite good fractional shortening on echocardiography. The diastolic blood pressures in these patients were low, with a mean of 20 mmHg for the first two patients and 25 and 45 mmHg for the patients who returned from theatre. Extracorporeal support was not available in this institution for postoperative salvage in palliative circulations at that time.
The remaining three early deaths occurred in patients who had not exhibited evidence of low output state. These patients had been making steady progress, were weaning from inotropic support and one was weaning from ventilation. Two suffered cardiac arrest following a short episode of bradycardia that was unaccounted for and the third suffered a VF arrest.
There were an additional three late deaths in the AP window group, all from congestive cardiac failure. One patient with a complex problem of non-compacted RV and HLHS had remained on the PICU for 23 days with evidence of congestive cardiac failure and failure to wean from mechanical ventilation. He had persistently low diastolic blood pressure and so underwent conversion to a standard Norwood physiology with an RV–PA conduit. Postoperatively he developed necrotising enterocolitis, recurrent sepsis, focal seizures and became ventilator dependent. He developed signs of restrictive cardiomyopathy with high PA and atrial pressures and was therefore not suitable for conventional stage II palliation. However, his trans-pulmonary gradient was low enough to consider cardiac transplantation. He was extubated and returned to the ward but died from congestive cardiac failure awaiting the transplant decision. In the second patient, ventricular function remained poor. He had persistent low diastolic pressure and developed thrombus in the left atrium. He underwent a high-risk stage II procedure with excision of thrombus at the age of 2 months but remained in low cardiac output postoperatively. The final patient died at the age of 22 months having presented with a chest infection. She had initially progressed well through stage II with moderately impaired ventricular function but was found to have deterioration of function and severe tricuspid regurgitation. Subsequently, she began having a series of cardiac arrests requiring increasing inotropic support and further resuscitation failed.
Overall, four patients underwent stage II procedure with ligation of the main PA. None required any PA reconstruction.
Fig. 4 compares the Kaplan–Meier actuarial survival between the Norwood group and the AP window group. Log-rank test demonstrated a significantly worse survival with the AP window group (p < 0.001).
Kaplan–Meier actuarial survival between the Norwood group and the AP window group.
Kaplan–Meier actuarial survival between the Norwood group and the AP window group.
The CPB times for the AP window group was 56 ± 30 min (mean ± SD) compared to 71 ± 39 min in the Norwood group (p = 0.04). Aortic cross-clamp time was 35 ± 27 min versus 55 ± 16 min (p < 0.01) and deep hypothermic arrest time 16 ± 29 min versus 55 ± 20 min (p < 0.01).
There was no difference in postoperative arterial saturations between the two groups with 75 ± 9% and 74 ± 7% in the AP window and Norwood groups, respectively.
On analysis of the first 24 h blood pressure recordings on return to the PICU there was no difference in the systolic pressures that were 57 ± 11 mmHg and 55 ± 10 mmHg in each group. However, the diastolic pressures were 27 ± 10 mmHg in the AP window group compared to 42 ± 8 mmHg in the Norwood group (p = 0.01) with no evidence of pulmonary incompetence.
Examination of the early postoperative echo recordings revealed evidence of flow reversal in the descending aorta during diastole in all of the AP window patients (Figs. 5 and 6 ). Using the Birmingham preoperative risk model for the Norwood I procedure [2], the number of predicted deaths in this group of patients would have been 2.88 (22% early mortality) assuming a classical Norwood operation and 1.05 (8% early mortality) assuming a RV–PA conduit technique (Table 2 ).
Colour flow Doppler image demonstrating flow reversal in the aortic arch in a patient with AP window and PA band.
Colour flow Doppler image demonstrating flow reversal in the aortic arch in a patient with AP window and PA band.
Continuous Doppler ultrasound in a patient with AP window and PA band showing the diastolic run-off in the aortic arch.
Continuous Doppler ultrasound in a patient with AP window and PA band showing the diastolic run-off in the aortic arch.
Standardised mortality ratio for the AP window group
Standardised mortality ratio for the AP window group
The model accounts for ventricular function. Thus, although this group of patients had a high incidence of impaired ventricular function, the outcome was still significantly worse than if they had undergone a standard Norwood approach. For example, the specific risk for a patient with impaired RV function but with a normal-sized aorta (6 mm) in the risk model is 0.11 for a Sano Norwood procedure (i.e. 11%) and is 0.32 for a classical Norwood procedure (i.e. 32%). Compare this to a mortality of 54% in the observed population.
4 Discussion
It is well recognised that there are a variety of anatomical sub-types within the spectrum of HLHS. Since most are characterised by varying degrees of hypoplasia of the ascending aorta and arch, surgical palliation has always addressed these lesions with the Damus-Kaye-Stansel type reconstruction with or without patch augmentation. However, as highlighted in this paper, there is a small sub-group (3.7% in this series) with HLHS who have a normal-sized ascending aorta and arch and so do not necessarily require such radical aortic augmentation. The hypothesis of the AP window/PAB approach was that the Norwood physiology could be reproduced with a less complex procedure which would reduce the potentially deleterious effects of bypass.
The results of this study would suggest that this was not the case. Despite what appears to be a favourable anatomy, the group of patients undergoing AP window/PAB suffered a significantly higher mortality than the comparative Norwood group. The striking clinical finding was the low diastolic pressure and evidence of reversed diastolic flow in the descending aorta (Fig. 5) and diastolic run-off (Fig. 6). This suggests that the physiologic effect of the PAB without the protection of a semilunar valve between it and the circulation results in a diastolic ‘steal’ phenomenon at the expense of systemic perfusion. This would also put the patients at risk of a coronary steal phenomenon, which would be in keeping with the common incidences of a sudden unheralded arrest. Coronary flow, particularly to the sub-endocardium, is dependent on diastolic flow, maintenance of diastolic pressure and length of diastole [5]. This theory would have been supported by evidence of myocardial ischaemia on the postoperative ECG in these patients, but there was no clear evidence of this. However, neonatal ECGs do not always show classical signs of ischaemia, as evidenced by findings in patients with anomalous origin of the left coronary artery [6]. Diastolic hypotension has been documented to cause irregular trans-mural myocardial perfusion and results in global sub-endocardial ischaemia [7,8] to which the heart can be particularly sensitive in the early postoperative period.
The findings strongly suggest that, despite a shorter and less complex procedure, the AP window/PAB is fundamentally flawed through the problem of diastolic run-off. This is supported by the finding that when these same patient characteristics were run through a risk model for the Norwood procedure (with RV–PA conduit), the predicted early mortality would have been 8% compared to the observed figure of 54%. This model was based on a detailed analysis of the Norwood procedure for HLHS in which multivariate analysis had shown ventricular function, ascending aortic diameter and source of pulmonary blood flow to be the strongest preoperative predictors of outcome. This would suggest that, in our hands, these patients would have had a better outcome with a standard Norwood approach.
The classical Norwood procedure, using a systemic–pulmonary shunt, has been criticised for similar reasons in terms of the continuous diastolic flow [9], but has not been associated with such profound diastolic hypotension or reversed diastolic flow. This may be because the haemodynamic effects of an unrestrictive AP window are more dramatic than those of a Gore-Tex shunt (which is both more peripheral in the circulation, narrower and longer, resulting in an inbuilt degree of fixed resistance which helps to dampen the degree of diastolic run-off).
This study is weakened by the differences that exist between the patient groups. The study group tended to be slightly older and, as might be expected from the normal aortic size, may have undergone a period of ‘trial’ of biventricular circulation in the setting of some forward flow across the aortic valve. This delay may have introduced an element of pulmonary hypertension and of volume-loaded heart failure in the older patients, although this cannot be established from the data available. Older age at time of Norwood palliation has certainly been cited as a risk factor in previous studies [10]. Also, two patients had undergone balloon aortic valvuloplasty at referring institutions, which had resulted in aortic regurgitation and may have had additional impact on the postoperative physiology.
These findings have particular relevance to the innovative techniques of ‘hybrid’ procedures for palliation in HLHS. These involve placement of bilateral PAB and direct placement of a stent in the ductus as an alternative to the Norwood procedure [1]. The technique has many attractions, particularly the avoidance of the need for cardiopulmonary bypass. However, the findings in this paper have highlighted the potential risks of this physiology and, as experience develops, it will be interesting to see whether similar problems with low diastolic pressure are encountered. Certainly, the avoidance of open-heart surgery might offset these concerns since there would be greater myocardial reserve in the early postoperative period. Conversely, if the hybrid procedure leaves the patient with a degree of hypoplasia of the ascending and transverse aorta, then the effects of diastolic coronary steal may be exacerbated. However, the avoidance of cardiopulmonary bypass may outweigh any such potential risk and there has been no mention to date of low diastolic pressures in such cases, and the published results are too early to establish whether there is any significant inter-stage attrition.
In summary, we feel that the AP window and PAB technique do not provide good palliation for these patients and we would recommend a standard Norwood procedure as offering the best outcome without the need of augmentation of the ascending aorta. The physiology of this form of palliation has relevance to newer ‘hybrid’ approaches to HLHS.
