Older age at the time of the Norwood procedure is a risk factor for early postoperative mortality^†

Abstract

OBJECTIVES

The Norwood procedure is commonly used as first-step palliation in children with hypoplastic left heart syndrome or related congenital malformations. In most cases, the operation is carried out during the first week of life. Excessive pulmonary blood flow in univentricular physiology can soon lead to compromised cardiac function and pulmonary hypertension and could have an influence on postoperative outcomes.

METHODS

Between 1997 and 2011, 257 Norwood procedures were performed at the Kinderherz-Zentrum Linz. Twenty-seven patients were older than 20 days at the time of the Norwood operation without previous pulmonary artery banding. A retrospective analysis of the two age groups was carried out concerning early mortality, interstage mortality and major events until the bidirectional Glenn operation.

RESULTS

Patients older than 20 days had a significantly higher mortality rate before the bidirectional Glenn operation than younger patients (44.4 vs 20.3%; P = 0.002). They also showed a significantly higher need for postoperative inhalative nitric oxide therapy (40.7 vs 14.5%, P = 0.002). Summarized risk for either successful postoperative cardiopulmonary resuscitation (CPR), extracorporeal membrane oxygenation therapy or mortality before the bidirectional Glenn operation was 59.3% in the older age group vs 23.8% in the younger age group (P < 0.001).

CONCLUSIONS

Age older than 20 days seems to be an independent risk factor for early postoperative mortality and fatal events after the Norwood procedure. Long-persisting excessive pulmonary blood flow and preoperative cardiac decompensation can be the reason for postoperative lethal pulmonary hypertensive crisis and compromised ventricular function.

INTRODUCTION

The Norwood procedure is commonly used as first-step palliation in patients with hypoplastic left heart syndrome and related congenital malformations. Although results improved over the last decade, mortality and major complications are still remarkable until the bidirectional Glenn operation is performed [1–5]. The Norwood operation is usually carried out during the first 7 days of life [1, 2, 4]. Preoperative excessive pulmonary blood flow can lead to congestive heart failure and compromised ventricular function. Nevertheless, in a certain number of patients, the Norwood procedure is done at an older age. Reasons can be late referral due to complicated transfer, treatment of extracardiac malformations and complications or previously failed biventricular treatment in borderline anatomical cases. Without pulmonary artery banding, these children might be at increased risk of postoperative potentially fatal pulmonary hypertensive crises or cardiac failure due to chronic overload of the pulmonary circulation. Previous reports stated an increased and continuing risk of death during all steps of single-ventricle palliation in older children [6] and proposed the use of large modified Blalock–Taussig shunts (RMBTSs) instead of right ventricle to pulmonary artery conduits (RVPACs) to increase postoperative pulmonary blood flow in the setting of increased pulmonary resistance [6, 7]. Because of the common practice of early treatment, only small numbers of older infants are reported [6–10].

Our study shall evaluate the influence of age older than 20 days at the time of the Norwood procedure on postoperative outcomes in a single-centre experience.

MATERIALS AND METHODS

Between 1997 and 2011, 257 consecutive Norwood procedures were performed at the Children's Heart Center Linz, Austria. Patients were retrospectively divided into two groups by their age at the time of the Norwood procedure. Three patients were excluded from this study because they had received a bilateral pulmonary artery banding prior to the Norwood procedure.

Group 1 includes 227 patients with an age younger than 20 days at the Norwood procedure. Median age was 7 days [quartiles: 5; 9]. Group 2 includes 27 patients with age 20 days and older (median age 35 days [quartiles: 25; 57]; ranging from 20 to 193 days). Six of the 27 patients of Group 2 underwent Stage I palliation after a failed biventricular repair. Initial procedures in these patients were aortic valve dilatation in 2 patients, surgery at the aortic isthmus in 3 patients and 1 Warden procedure, mostly at other centres. Twenty-one patients reached our centre after late referral due to transfer problems from foreign countries or prolonged conservative therapy there. Only 11 of the 27 Group 2 patients received long-lasting prostaglandine infusion before surgery. The duration ranged between 20 and 86 days (median 24 days). The 16 patients without prostaglandine infusion were older at the time of the Norwood (median 52 days, ranging from 31 to 193 days).

The anatomical subgroup distribution was different between the two groups, showing relatively fewer cases of aortic atresia, but more cases of unbalanced atrioventricular canal and of single or dominant left ventricular anatomy in the older age group. The decision for surgery was made independently of anatomical subgroup.

Detailed patient characteristics of the two groups are summarized in Table 1. Data were gained retrospectively from patients’ medical records and statistically analysed. Study endpoint was the bidirectional Glenn operation. The study outcomes are the analysis of the postoperative course [primary chest closure, need for postoperative nitric oxide, successful cardiopulmonary resuscitation (CPR), extracorporeal membrane oxygenation (ECMO) use, length of ICU therapy and hospital stay], early and interstage mortality of the two age-dependent groups after the Norwood operation and the identification of independent risk factors for postoperative adverse events until the bidirectional Glenn operation. Three patients with a long period of dreadful congestive heart failure after the Norwood operation died soon after the bidirectional Glenn procedure and are included in the interstage mortality.

Operative technique

The Norwood procedure was performed in a standard manner irrespective of patient's age. Changes in surgical techniques occurred in historical episodes concerning perfusion technique and the method for restoring pulmonary blood flow. Both patient groups were affected in the same way by these changes.

Until October 1999, the Norwood procedure was carried out in deep hypothermia and circulatory arrest for the aortic arch reconstruction. During the next approximately 4 years, antegrade cerebral perfusion via the truncus brachiocephalicus was used at a flow rate of 30 ml/kg/min. Since July 2003, nearly all Norwood operations have been done using continuous whole body perfusion via a second arterial cannula in the descending aorta. The aortic arch reconstruction was done using homograft patch enlargement (177 cases), curved polytetrafluorethylene patch enlargement (49 cases) or by direct anastomosis (28 cases), depending on availability and anatomy. Pulmonary perfusion was re-established either with a right-sided modified Blalock–Taussig shunt (3.5 or 4 mm polytetrafluorethylene prosthesis) or predominantly with a right ventricle to PAC (5 or 6 mm) since 2001. The sternum was left open in the majority of cases depending on haemodynamic performance and risk factors for postoperative instability.

Median cross-clamp times were nearly similar with 69 min in Group 1 and 68 min in Group 2. Also, the median bypass time did not show significant difference (199 vs 208 min; P = 0.28).

Forty-one (18.1%) patients of the younger age group received a right-sided modified Blalock–Taussig shunt and 186 (81.9%) a right ventricle to PAC. In Group 2, there were 9 RMBTSs (33.3%) and 18 RVPACs (66.7%) (P = 0.069). See detailed information in Table 2.

Statistical analysis

Logistic regression analysis was carried out to detect independent risk factors for the composite endpoint CPR or ECMO or exitus. Age (</≥20 days), anatomy [aortic atresia/aortic stenosis/single ventricle (SV) or dominant left ventricle/unbalanced atrioventricular (AV) canal], perfusion technique [deep hypothermic circulatory arrest (DHCA)/antegrade cerebral perfusion (ACP)/whole body perfusion (DESC)], pulmonary blood flow (RMBTSs/RVPACs), sex (male/female), weight (kg), CPB time (min), cross-clamp time (min) and order of operation (–) were used as independent variables. For the comparisons of age subgroups (</≥ 20 days), all data sets of metric variables were checked for normal distribution (test of normality: Kolmogorov–Smirnov with Lilliefors significance correction, Type I error = 5%). Normally distributed data sets were compared by the t-test (test for variance homogeneity: Levene test, Type I error = 5%) for independent samples, metric variables without normally distributed data sets by the Mann–Whitney U-test and categorical variables either by Fisher's exact test or by the exact χ² test. The Type I error was not adjusted for multiple testing. Therefore, the concerning P-values are only descriptive. The open-source R statistical software package, versions 2.14.1 and 3.0.1, and BIAS 10.0 (Epsilon publishers) were used for statistical analysis.

RESULTS

The chest could be closed primarily more often in the younger patient group (34.4% in Group 1 vs 14.8% in Group 2; P = 0.049). Postoperative need for inhaled nitric oxide was lower in Group 1 (14.5% in Group 1 vs 40.7% in Group 2; P = 0.002).

In total, 7.9% of our patients survived at least 1 episode of CPR including chest compression (CPR) in the interstage period. Thirteen of these patients belonged to Group 1 and 7 patients to Group 2. So significantly fewer younger patients (5.7 vs 25.9%; P = 0.002) suffered successful CPR.

In our centre, ECMO therapy was used for the first time in 2008 in Norwood patients. Since then, ECMO use has been significantly higher in the older patient group (9.6% in Group 1 vs 40.0% in Group 2; P = 0.008). ECMO was usually installed either after CPR on the intensive care unit or under deleterious haemodynamic instability in the operating room or later on. One patient of Group 2 had ECMO therapy already preoperatively because of chronic congestive heart failure after failed biventricular repair. This patient needed ECMO therapy also postoperatively after a resuscitation event on the ICU. Seven of 13 ECMO patients of our series could be weaned off bypass primarily, but only 4 underwent a successful bidirectional Glenn operation.

Thirty-one of the 227 patients of Group 1 died within 30 days of the Norwood operation (early mortality rate 13.7%). The rate decreased over time and with every change in our surgical technique. We found the best outcome in the subgroup of our patients, who early received a right-sided RV-PA conduit under whole body perfusion (early mortality rate 8.1%; 9/111 patients). Reasons for early death in Group 1 were cardiac failure in 22 cases, sepsis in 3, early BTS thrombosis in 3, 1 case of sinus thrombosis and 2 cases of untreatable pulmonary failure.

Nevertheless, significant interstage mortality raised the death rate in the whole group 1 up to 20.3% until Stage II palliation. Six patients of Group 1 died in the interstage period due to chronic congestive heart failure, 2 because of sepsis and 7 died of unknown reasons mostly unexpected at home after discharge.

In comparison, we already lost 33.3% (9 of 27) of the Group 2 patients early and in total 44.4% until the bidirectional Glenn operation (P = 0.008). Eight of the 9 early deceased patients of Group 2 died soon after the operation or could not be weaned off postoperative ECMO. Reasons for death were either sudden or persisting cardiopulmonary failure. One patient died due to lethal bleeding on postoperative day 11. The 3 interstage mortality cases of Group 2 occurred on postoperative days 40, 84 and 140; 2 of them due to chronic congestive heart failure. Six of these 12 deceased patients had a right-sided Blalock–Taussig shunt (2 of them 4 mm in diameter) and 6 had a right-sided RV-PA conduit (2 of them 6 mm in diameter).

Six patients of Group 2 had interventional or surgical procedures prior to the Norwood operation to address several problems of their Shone complex. After the Norwood procedure, 3 of them died within 30 days and 1 soon after a failed bidirectional Glenn operation (66.7% mortality).

Summarized risk for either successful postoperative CPR, ECMO therapy or mortality before the bidirectional Glenn operation was 23.8% in the younger age group vs 59.3% in the older age group (P < 0.001).

Logistic regression analysis for the composite endpoint CPR or ECMO or exitus showed only age (</≥ 20 days) and anatomical subgroup (aortic atresia/aortic stenosis/SV or dominant left ventricle/unbalanced AV canal) as independent risk factors (each with P < 0.001). Age ≥20 days and the diagnosis unbalanced AV canal influence outcomes in a negative way.

There were no significant differences between the two groups concerning length of ICU therapy and length of stay when children with postoperative exitus were excluded. Median time between the Norwood and bidirectional Glenn operation was 115 days in Group 1 and 100 days in Group 2 (P = 0.073). Detailed data are given in Table 3.

DISCUSSION

The Norwood procedure is usually electively planned during the first week of life. Several current studies have shown the importance of optimal multiorgan status for best outcomes of this difficult operation. Several prospective multicentre investigations could identify risk factors for postoperative mortality such as lower birth weight, smaller ascending aorta, significant AV-valve insufficiency, lower centre volume or ECMO use [2, 11]. These investigations commonly did not include the patient group of a significantly older age without protection of the pulmonary circulation because of their prospective character. Chronic volume overload of the lungs in single-ventricle physiology or chronic congestion in Shone complex patients can lead to potentially negative sequelae. Consequence can be a hyper-reactive pulmonary vasculature which predisposes these patients to pulmonary hypertensive crises postoperatively. An additional risk can be damaged function of the systemic ventricle.

In our study, patients with age ≥20 days showed a significant higher need for postoperative inhaled nitric oxide, a strong pulmonary vasodilator, which was administered only in severe postoperative cyanosis due to pulmonary vasoconstriction. The rate of primary chest closure was also significantly lower in the older age group. Reasons for an open sternum were either haemodynamic instability or postoperative cyanosis. These findings might have their reasons in a hyper-reactive pulmonary vascular bed.

Although operative data such as total bypass time, median cross-clamp time and the kind of surgical procedure were comparable in both groups, older patients showed a significantly higher incidence of fatal events postoperatively. Not only postoperative ECMO need, but also successful CPR and postoperative mortality, were significantly more frequent in patient Group 2. As most of those fatal events occurred early after the operation and sometimes unexpectedly during an uneventful postoperative course, they might be explained by a sudden rise in pulmonary vascular resistance and subsequent cardiac failure.

The fact that our older patients had a probability of 59.3% (vs 23.8% in the younger patients) to suffer CPR, ECMO therapy or death until the bidirectional Glenn operation (BDG) identifies this group as a high-risk patient subpopulation.

It was recommended by Duncan et al. and Alsoufi et al. [6, 7] that these problems be encountered with a higher pulmonary blood flow (BT-shunts instead of RV-PA conduits or larger BT-shunts). This approach did not show a positive effect on the outcomes of our older patients. No noticeable advantage of the BT-shunt against the RV-PA conduit was demonstrated in Group 2. On the other hand, 6 of the 9 BT-shunt patients died before the bidirectional Glenn operation, whereas 6 of 18 RV-PA conduit patients did not survive until Stage II. This reflects the general advantage of the RV-PA conduit patients in our centre. Maybe the worse coronary blood flow in the BT-shunt group had an additional deleterious effect on cardiac function.

Patients who underwent the Norwood procedure late after failed biventricular treatment had a remarkably bad outcome (mortality 66.7% until BDG). In these cases, the decision for the change in strategy was usually made after a frustrating long therapy on the intensive care unit. Those patients can be divided into two groups. We saw on the one hand historical cases that survived biventricular therapy without chance for univentricular palliation at the primary care centre. On the other hand, ambitious approach of already foetal intervention nowadays brings more children towards borderline anatomy and biventricular treatment than natural foetal course. Although only few patients are in this special subgroup, the prognosis of these “failing” patients seems to be by far worse than after primary single-ventricle palliation. Compromised ventricular function and stressed pulmonary vasculature can be the reason for the poor results. This confirms the need for an early correct decision for borderline anatomical cases between univentricular palliation and biventricular repair. A step back to univentricular circulation seems to be hardly possible.

The most common reason for late operation in our centre was late referral from foreign countries. Whenever possible, immediate transfer to the surgical centre and early Norwood operation seems to be of great importance regarding postoperative complications. Facing the results, an alternative strategy for these patients has to be reviewed. All measures of medication against postoperative pulmonary hypertensive crises have to be undertaken to prevent lethal events. Maybe a temporary pulmonary artery banding prior to the Norwood procedure could relieve pulmonary circulation stress.

A limitation of the study is, of course, the small number of older infants. Additionally, the long investigation period, where several changes in technique and therapy have occurred, makes the interpretation of results difficult. The influence of preoperative status was not taken into account because of the retrospective character of the study.

ACKNOWLEDGEMENTS

The authors thank Wolfgang Schimetta and the Working Group for Systemic Optimization of Clinical Research Projects (ASOKLIF) for data analysis and statistics.

Conflict of interest: none declared.

REFERENCES

APPENDIX. CONFERENCE DISCUSSION

Dr T. Jones(Birmingham, UK): First of all, achieving an 8% mortality using the right-sided conduit in all-comers is an excellent achievement. To go into the paper, I think what you've highlighted to us, an experience that we all share, is that late age at presentation going into a Norwood is certainly a high-risk strategy. What I think is quite interesting is that in your group of late presentations, you've got two sets of patients. You've got one set of patients who, for whatever reason, delayed referral or other problems, have gone into a late Norwood. And then you've got the second set that you mentioned, who have gone down a biventricular strategy and then been converted. And I think these two pathways represent very different patients and different management decisions.

And I've got a question for both. The patients who presented late you've taken through a Norwood procedure. In the light of the data and the evidence now, have you changed practice and would you think about going through a hybrid strategy for these patients and seeing if that stabilizes their pulmonary vascular resistance and their blood flow changes?

And then the second comment, and you've already alluded to it partly, is for the patients with a failed biventricular repair, I mean you've shown in the data (that Ed Hickey has published as well) that if we make the wrong decision, it is a catastrophically bad outcome. So again, in the light of this, do you now go more to a Norwood rather than trying to push the biventricular repair?

Dr Sames-Dolzer: Regarding the late referral patients, the first aim has to be to get them to our centres earlier. We were lucky to have had no such older patients since the analysis, but what we discuss is to do a preventive bilateral pulmonary artery banding prior to the Norwood procedure and then wait for a week or ten days. We can then do the Norwood operation when we have a volume unload of the pulmonary vessels and of the ventricle.

We are not convinced of the original hybrid procedure (waiting two or three months and doing the Norwood procedure together with the Glenn) in these cases. It is more difficult to do the Norwood in a redo setting with the stent in the duct. Also, arch geometry changes a lot in the first months. I think a week or ten days should be enough time for protection of the pulmonary vascular bed.

Regarding the biventricular patients, no, we have not changed the strategy. We still have an aggressive trend towards biventricular repair, but it's important that we always keep in mind to protect the pulmonary vasculature. So we just had one case of borderline anatomy where we made a kind of hybrid with a pulmonary artery banding. By this method, we can gain some weeks waiting until the left ventricle grows.

Dr S. Sano(Okayama, Japan): In the case of the older patients with high pulmonary vascular resistance, we perform bilateral banding and wait for two or three weeks and then do a Norwood. When you do a bilateral band, initially the Doppler gradient might be 2 to 3 mmHg then gradually increase to 3.5 and 4, which means the pulmonary vascular resistance has already dropped. Then we do the Norwood rather than a Norwood-Glenn. The end result in our unit is much better.

Dr Sames-Dolzer: Yes, I think this would also be our strategy in the future.

Dr D. Anderson(London, UK): Your conclusions, to my thinking, are slightly contradictory. You say that in a situation of high pulmonary blood flow, that's a risk factor. And then your next conclusion was that pulmonary hypertension, and I presume by that you mean increased pulmonary vascular resistance, may be a problem in the recovery phase. So what's happening? If they've got high pulmonary blood flow, they must by definition have a low resistance, and yet at the end of the operation, you may have problems with pulmonary hypertension. So do you have a strategy to manage that or is there something that you're doing?

Dr Sames-Dolzer: What we saw is that a lot of the patients had sudden pulmonary hypertensive crisis. So they were initially doing very well and then they had a sudden onset of a life-threatening event and CPR became necessary. So I think that the pulmonary hypertension may show up in the form of a sudden onset crisis during the first days on the intensive care unit, so that is the real danger for those patients.

It was proposed in the literature to use larger shunts or to use B-T shunts instead of RV-PA conduits to counter high pulmonary resistance, but we could not find any advantage of this strategy.

For us, older patients also benefit from the Sano conduit and suffer from the adverse effects of the diastolic run-off of the B-T shunt. The majority of the older patients did well after the operation, but then had a severe problem with rapid onset during the intensive care therapy, so it was just that we came too late with our operation. So we have to do something preventive for these patients. It's not that the strategy is not working at all.

We also had cases that were among the oldest in this group which went very well, who already underwent the Fontan procedure without further problems. But most of the patients had, at least once, a problem requiring CPR.

Author notes