Author + information
- Received June 2, 2010
- Revision received November 15, 2010
- Accepted December 23, 2010
- Published online August 16, 2011.
- Diego Porras, MD⁎,
- David W. Brown, MD⁎,
- Audrey C. Marshall, MD⁎,
- Pedro del Nido, MD†,
- Emile A. Bacha, MD† and
- Doff B. McElhinney, MD⁎,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Doff B. McElhinney, Children's Hospital, Department of Cardiology, 300 Longwood Avenue, Boston, Massachusetts 02115
Objectives The aim of this study was to determine whether there are identifiable factors associated with increased risk for aortic arch reintervention in patients who have undergone balloon aortoplasty (BD) for aortic arch obstruction (COA) after the Norwood procedure (NP).
Background BD has been shown to be an effective acute therapy for COA after the NP. However, recurrent obstruction requiring repeat intervention is not uncommon.
Methods All patients who underwent BD as the initial intervention for COA after the NP from to January 1993 to May 2009 were retrospectively analyzed (n = 116).
Results The median age at initial BD was 4.5 months. The median follow-up period was 3.4 years. Procedures were considered acutely successful in 92% of patients, with a median gradient reduction overall from 24 to 3 mm Hg (p < 0.0001) and a COA diameter increase of 52% (p < 0.0001). By Kaplan-Meier analysis, freedom from reintervention was 69% at 1 year and 58% at 5 years, and freedom from reoperation was 82% at 1 year and 79% at 5 years. By Cox regression analysis, proximal arch obstruction, age <3 months at BD, moderate or greater ventricular dysfunction, moderate or greater atrioventricular valve regurgitation on pre-catheterization echocardiography, and higher post-BD coarctation gradient were associated with shorter freedom from reoperation.
Conclusions Despite a high acute success rate, a significant proportion of patients treated with BD for post-NP COA underwent reintervention during follow-up. The risk for arch reintervention is highest in patients with proximal arch obstruction, those age <3 months at the time of BD, and those with less successful acute results.
Aortic arch obstruction (COA) remains an important problem in patients who have undergone the Norwood procedure (NP). It has been reported to occur in 9% to 37% of patients, and is a significant cause of morbidity (1–7). COA has been associated with deterioration of ventricular function in this patient population, as well as atrioventricular valve regurgitation (AVVR) and imbalance of systemic and pulmonary blood flow (8). It has also been associated with increased risk for death (9,10). Balloon aortoplasty (BD) has become the standard first-line therapy for post-NP COA and is generally effective, with reported acute success rates ranging from 89% to 100% (3–7). In fact, Zeltser et al. (7) did not find a difference in survival between patients treated with BD for COA after the NP and those who did not develop arch obstruction. Despite favorable acute results, little is known about the intermediate and long-term outcomes after BD of the reconstructed aortic arch. Factors associated with reintervention after BD have not been characterized and may have implications for decision making around methods of aortic reconstruction, as well as the timing and method of reintervention. We evaluated all patients who underwent BD as the primary modality of treatment for post-NP COA over the past 16 years, with the objective of identifying factors associated with reintervention on the aortic arch in this patient population.
We reviewed all patients who underwent BD for post-NP COA at Children's Hospital Boston from January 1993 to May 2009. Patients were included if the NP had been performed for hypoplastic left-heart syndrome or related single-ventricle lesions and no prior arch interventions had been performed. Only patients who underwent the NP and BD at Children's Hospital Boston were included. Patients who underwent primary stenting were excluded. The diagnosis of hypoplastic left-heart syndrome was based on echocardiographic evidence of diminutive ascending aorta, aortic atresia or stenosis, mitral atresia or stenosis, and hypoplastic or dysfunctional left ventricle. The NP was defined as a surgical procedure that included anastomosis of the native pulmonary trunk and aorta, excision of ductal tissue, reconstruction of the aortic arch, and creation of a systemic–to–pulmonary artery shunt or right ventricle–to–pulmonary artery conduit, with or without an open atrial septectomy. Patients were grouped by era on the basis of whether they were among the first or last 50% of patients. Cross-sectional follow-up was obtained at the time of most recent clinical evaluation, death, or transplantation. The institutional review board of Children's Hospital Boston approved this study.
Data were obtained by review of medical records, pre-operative echocardiograms, and angiograms from the initial BD procedure, which were reviewed by a single reviewer in blinded fashion. Pre-operative patient factors included diagnosis, diameter of the ascending aorta, sex, age, and weight at the time of the NP. Operative parameters included date of operation, type of pulmonary artery–to–aortic anastomosis (proximal side-to-side anastomosis vs. pulmonary trunk to underside of the aortic arch), whether a coarctectomy was performed, material used to augment the reconstructed aortic arch, and the source of pulmonary blood flow used (modified Blalock-Taussig shunt vs. right ventricle–to–pulmonary artery shunt). Coarctectomy was defined as resection of the aorta at the level of the coarctation, including wide resection of all ductal tissue and the posterior shelf and direct end to end anastomosis between the greater curvature of the native aortic arch and the posterior wall of the descending aorta. Pre-catheterization data included weight and age at the time of BD, as well as the degree of ventricular dysfunction and AVVR on pre-catheterization echocardiography.
Patients presented for cardiac catheterization based on either clinical suspicion of arch obstruction or an unexplained deterioration in clinical status (unplanned catheterization group) or in anticipation of their next surgical procedure (planned catheterization group). The decision to perform BD was made on a case-by-case basis; no standardized criteria were in place. During the study period, BD was generally considered the first-line therapy for post-NP COA at our institution. Procedural factors obtained from the catheterization report included peak gradient across the obstruction before and after balloon dilation (pre- and post-BD gradients), diameters of the first and largest balloons used for BD, and number of balloon inflations. The following measurements were made using the lateral or left anterior oblique projection of the aortic arch angiogram (Fig. 1): neo-ascending aorta and transverse arch diameter, narrowest diameter at the site of obstruction (COA diameter), and diameter of the descending thoracic aorta at the level of the diaphragm. When the descending thoracic aorta at the level of the diaphragm was not visible in the angiograms, the diameter measurement was made as distal as possible in the descending thoracic aorta, beyond the area of post-stenotic dilation that is commonly present. The COA index was calculated by dividing the COA diameter by the descending thoracic aortic diameter. The arch geometry was also assessed and classified as having Romanesque (semicircular rounded form), Gothic (triangular form with acute angulation between the ascending and descending aorta), or Crenel geometry (rectangular form) (10). We also found that some patients had tortuous or S-shaped arches.
The strength of indication for BD was retrospectively assessed based on the gradient across the area of obstruction and the ventricular function. A strong indication was defined a priori as a COA index <50% and/or a peak gradient ≥15 mm Hg in the presence of normal ventricular function or a gradient ≥10 mm Hg in the presence of significant ventricular dysfunction. Significant ventricular dysfunction was defined a priori as moderate or worse ventricular dysfunction on pre-catheterization echocardiography or evidence of hemodynamic compromise (ventricular end-diastolic pressure ≥14 mm Hg and/or calculated cardiac index <2.6 l/min/m2).
Patients were divided into 2 groups based on the anatomic region of the arch involved in the obstruction: distal COA was defined as obstruction immediately distal to the takeoff of the left subclavian artery (LSCA) (Fig. 2A); proximal COA occurred at or proximal to the takeoff of the LSCA (Fig. 2B).
The primary outcome variables were freedom from any arch reintervention and freedom from any arch reoperation after BD. Throughout this report, “reintervention” and “reoperation” refer to any aortic arch reintervention (catheter based or surgical) or reoperation after BD, respectively. Another outcome variable analyzed was transplantation-free survival. Survival data are presented as percent survival (95% confidence interval) at defined intervals. Predictor variables included the demographic, surgical, pre-catheterization, and BD-related variables listed previously.
Time-to-event outcomes were evaluated using Kaplan-Meier analysis with log-rank tests or Cox regression analysis. The proportionality assumption of Cox regression analysis was checked by plotting cumulative hazards for the independent variables that were analyzed. Multivariate freedom-from-event analysis was not performed, because of the relatively small number of outcome events. Hazard ratios are presented with 95% confidence intervals. Comparisons of paired pre- and post-BD pressure gradients and COA indexes were performed using Wilcoxon signed rank tests. Comparisons between groups were performed using Fisher exact tests, Student t tests, or Wilcoxon rank sum tests as appropriate. Logistic regression was used to adjust for the strength of indication for BD when comparing patient-related factors (e.g., age, indication for catheterization) between patients who had residual post-BD gradients ≤10 mm Hg and those with gradients >10 mm Hg. Data are presented as frequency, median (minimum to maximum), or mean ± SD as appropriate. SPSS version 16.0 (SPSS, Inc., Chicago, Illinois) was used for statistical analysis.
Between January 1993 and May 2009, 556 patients underwent the NP at Children's Hospital Boston; follow-up data at least 1 month after the NP were available to us for 462 of these patients (19 patients had no follow-up information after discharge from the hospital, and 75 died within 1 month of the NP). Of these 462 patients, 133 had interventions for arch obstruction (29%). BD was the first intervention in 120 (90%) of these patients, but 4 were excluded from this study: 1 had BD previously performed at another institution, and 3 underwent primary stent placement to relieve COA (a 7-year-old with a tortuous arch and 2 infants in the early post-operative period). The other 116 patients were included in the study.
Table 1 summarizes the demographic and operative characteristics at the time of NP for the study cohort.
Table 2 summarizes the demographic, procedural, and follow-up characteristics of patients undergoing BD. BD was acutely effective at reducing the peak gradient across the COA (Fig. 3A), which fell from a median of 24 mm Hg (0 to 65 mm Hg) to 3 mm Hg (0 to 17 mm Hg) (p < 0.001). Of the 116 patients, 107 (92%) had post-BD peak gradients ≤10 mm Hg. There was no difference in the proportion of patients with post-BD peak gradients ≤10 mm Hg according to age at BD (p = 0.21) or indication for catheterization (p = 0.50). This held true even after correcting for the strength of indication for BD. The COA index after BD also increased significantly (p < 0.001) (Fig. 3B).
All patients who did not undergo reintervention after BD (n = 69) were followed for at least 6 months. During follow-up, 105 patients (91%) proceeded with staged single-ventricle palliation: 27 patients (23%) underwent bidirectional Glenn procedures only, and 78 patients (67%) underwent Fontan procedures. Six patients (5%) underwent conversion to biventricular circulation, and 3 (3%) died before any subsequent procedures.
All reoperations (n = 24) consisted of patch arch augmentation (homografts in 12 patients, autologous pericardium in 5, bovine pericardium in 3, and synthetic grafts in 4). All but 3 of these patients underwent reoperation at a subsequent surgical stage (not the sole indication for surgery). The indication given for reoperation was inadequate result of BD in 17 patients (71% of 24) and recurrence of obstruction in 7 (29%). All but 1 patient survived to discharge after reoperation.
Survival analyses for each outcome are presented in Figure 4. Most of the reinterventions occurred in the first 6 months after BD (n = 31 [66%]). The primary indication given for reintervention in these patients was an inadequate result of the initial BD in 19 (61%) and recurrence of COA despite acutely successful results in 12 patients (39%) (Fig. 5). Freedom from reintervention was 69% at 1 year (60% to 76%) and 58% at 5 years (48% to 70%). Freedom from reoperation was 82% at 1 year (73% to 88%) and 79% at 5 years (71% to 85%). Transplantation-free survival for the cohort was 94% at 1 year (88% to 97%), 83% at 5 years (67% to 87%), and 74% at 10 years (59% to 85%).
The results of univariate analysis for the primary outcomes are summarized in Table 3. Of note, there was no difference in the statistically significant variables when the analysis included only patients with stage I circulation at the time of BD (n = 97) and no difference in freedom from reintervention or reoperation between patients who underwent BD in the first and second halves of the study period. Factors associated with shorter transplantation-free survival were moderate or worse RV dysfunction at the time of pre-catheterization echocardiography (hazard ratio: 5.3; 95% confidence interval: 2.0 to 14.5; p = 0.001) and proximal COA (hazard ratio: 7.4; 95% confidence interval: 2.9 to 18.9; p < 0.001).
Risk factors for arch reintervention
Site of Obstruction
The most common site of COA was distal to the LSCA (n = 102 [88%]). Of the 14 patients with proximal COA, the obstruction was in the transverse arch (between the origins of the innominate artery and the LSCA) in 10 patients and immediately proximal to the origin of the brachiocephalic vessels in 4 patients.
Proximal COA was significantly associated with shorter freedom from reintervention, reoperation, and transplantation-free survival (Fig. 4D). Patients with proximal COA underwent initial BD during an unplanned catheterization more often than patients with distal COA (86% vs. 38%, p = 0.001). There were no other differences in demographic, surgical, or functional variables between the 2 groups. Although there was no difference between the pre-BD pressure gradients in the 2 groups (median 25 mm Hg vs. 23 mm Hg, p = 0.99), patients with proximal COA had significantly higher post-BD gradients (median 8 mm Hg vs. 2 mm Hg, p = 0.05).
Age at Time of BD
Age <3 months at the time of BD was associated with shorter freedom from reintervention and reoperation. As would be expected based on our management of single-ventricle patients, all patients who underwent BD at <3 months of age did so during an unplanned catheterization. Unplanned catheterization, regardless of age, was also associated with shorter freedom from reintervention and reoperation.
Pre-BD Catheterization Data
The pre-BD gradient was not associated with any of the outcomes. A lower pre-BD COA index was associated with shorter freedom from reintervention but not with freedom from reoperation.
Acute Response to BD
Post-BD COA index was associated with shorter freedom from reintervention, and the post-BD gradient was associated with shorter freedom from reoperation and reintervention.
We did not find any surgical factors that were associated with freedom from reintervention or reoperation after the initial BD. There was also no associations between individual surgeons and patient outcomes. However, it should be mentioned that during the study period, there were more than 8 different surgeons performing this procedure. Analysis limited to the surgeons with the most cases showed no statistically significant difference.
Ventricular and Atrioventricular Valve Function
Both moderate or worse right ventricular dysfunction and moderate or worse AVVR on pre-catheterization echocardiography were associated with shorter freedom from reoperation. Moderate or worse right ventricular dysfunction was also associated with shorter freedom from reintervention.
The most common arch geometry was Gothic (50% or 56 of the 111 patients with angiograms available for review). Romanesque arches were seen in 35 (32%), Crenel arches in 17 (15%), and S-shaped arches in 3 (3%). There was no statistically significant association between the presence of a Gothic arch and freedom from reintervention, freedom from reoperation, or transplantation-free survival.
We performed a separate analysis on the group of patients with distal arch obstruction (n = 102). The predictive variables of significance were the same. However, 2 additional variables were significant in this subgroup by univariate analysis: 1) a higher pre-BD gradient was associated with shorter freedom from reintervention (p = 0.03); and 2) a lower pre-BD coarctation index was associated with shorter freedom from reoperation (p = 0.002).
Follow-up echocardiography and hemodynamic evaluation
Follow-up echocardiograms were available in all but 1 patient. The median time from BD to the last available echocardiogram was 37.4 months (0.3 to 192.8 months). Significant ventricular dysfunction (qualitatively moderate or worse) was present in 36 patients (31%) and did not differ in frequency between patients who had reintervention and those who did not (32% vs. 31%, p = 1.00). Significant AVVR (qualitatively moderate or worse) was present in 18 patients (16%) and did not differ in frequency between patients who had reintervention and those who did not (22% vs. 12%, p = 0.20).
Of the 69 patients who did not have reintervention, 52 (75%) underwent follow-up catheterization. The median time from BD to the last available catheterization in these patients was 32.4 months (2.1 to 126.7 months). The median gradient across the aortic arch recorded at this catheterization was 0 mm Hg (0 to 15 mm Hg), and all but 1 patient had a gradient <10 mm Hg.
Despite significant advances in the surgical management of patients with hypoplastic left-heart syndrome and related single-ventricle defects, COA after the NP continues to be a significant problem (1–8,11,12). When it occurs, BD has been established as the first-line therapy and is generally effective acutely (3–7), as our study confirms. However, despite a high acute success rate, our data demonstrate that reintervention and reoperation on the aortic arch are common, especially in the 6 months after the initial BD. This finding is consistent with smaller series reported by other groups (4,6,7).
As we expected, patients who underwent BD at a younger age and had symptomatic COA were at higher risk for reoperation and reintervention. Importantly, however, reduction in the COA gradient to levels ≤10 mm Hg and increase in the COA index was achieved in the great majority of the patients, regardless of the indication for catheterization, the strength of the indication for BD, or the age of the patient.
In our study, the site of obstruction relative to the takeoff of the LSCA was strongly associated with outcome after BD for post-NP COA. The most common site of obstruction was distal to the LSCA, which accounted for 89% of the cases in this series. Patients with proximal COA were found to be less responsive to BD and to have worse outcomes. Although some reports have found an association between arch obstruction and mortality after the NP, more recent experience suggests that in the balloon angioplasty era, mortality is similar for patients who do and do not undergo BD for arch obstruction (7). Here, we identify a subgroup of patients in whom obstruction was less responsive to BD and was associated not only with increased reintervention rates but also with decreased survival. Of the 14 patients with proximal arch obstruction, only 3 patients survived without reintervention. Symptomatic patients appear to be at particularly high risk: of the 12 patients with symptomatic proximal COA, 11 underwent reintervention, died, or underwent transplantation during the follow-up period. Based on these findings, it would be reasonable to advocate for a more aggressive approach to the management of patients with proximal COA, especially those with symptomatic obstruction. It will also be important to determine risk factors for the development of proximal arch obstruction, although that is beyond the scope of this project.
Several studies have shown that surgical technique, patch materials, and baseline anatomy influence the risk for arch obstruction after the NP (1,2,12–15). We did not find any surgical factors that were associated with freedom from reoperation or reintervention after BD. Coarctectomy has been shown in some series to protect against late COA after the NP. In this study, patients with histories of coarctectomy who had developed COA and underwent BD had no difference in freedom from subsequent reintervention compared with patients who did not have coarctectomy. We take this to mean that coarctectomy does not provide an advantage in the response to BD once COA develops. The presence of an aberrant right subclavian artery may alter the technique of NP (12); therefore, this subpopulation may respond differently to BD for arch obstruction. In our cohort, there were only 5 patients with this diagnosis, limiting our ability to analyze it as a risk factor for arch reintervention after BD.
The currently preferred surgical technique for arch reconstruction by most surgeons at our institution consists of posterior shelf resection (coarctectomy) and subsequent reconstruction of the greater curvature of the distal arch, followed by patch augmentation of the lesser curvature of the neoaorta, preferably using glutaraldehyde-treated autologous pericardium. However, there continues to be some variation in practice among surgeons.
Pre-BD gradient was not associated with any of the outcomes in the group as a whole. However, among patients with distal COA, pre-BD gradient and pre-BD COA index were associated with shorter freedom from reintervention and reoperation, respectively. These data suggest that for patients with distal COA, more severe obstruction (as judged by the catheterization data) is a risk factor for reintervention on the aortic arch.
Several criteria have been applied to evaluate the acute results of balloon angioplasty for COA after the NP, mostly on the basis of the post-BD gradient and/or the COA index (3,6,7,16). However, it is unclear whether these measures correlate with better long-term outcomes. We found that both a higher post-BD gradient and a lower COA index after BD were associated with shorter freedom from reintervention and that a higher post-BD gradient was also associated with shorter freedom from reoperation. However, these acute outcomes taken in isolation are unlikely to be reliable predictors of long-term success. Many factors, including the arch anatomy, the patient's presentation, ventricular function, and AVVR, likely contribute.
If untreated, COA can result in deleterious effects on ventricular function, which is a negative prognostic indicator (17). In this study, moderate or worse ventricular dysfunction at the time of pre-catheterization echocardiography was associated with significantly shorter transplantation-free survival. Although studies have shown that ventricular function does recover if the obstruction is relieved in a timely manner (5,8), it is still unclear whether there are long-term implications of the duration and magnitude of increased afterload to which these patients are exposed.
Patients with moderate or worse ventricular dysfunction or moderate or worse AVVR on pre-catheterization echocardiography had shorter freedom from reoperation. These findings may reflect increased ventricular function and AVVR effects in patients with higher COA severity or a lower threshold for arch reoperation in patients with higher grade ventricular dysfunction or AVVR.
In the bigger picture, the hydrodynamic efficiency of the reconstructed aortic arch of the NP, with geometric variability, compliance mismatch due to rapid tapering, heterogeneous aortic wall properties due to patch material and possibly scar, and other factors, is not well understood. Patients who have undergone the NP have decreased elasticity within the reconstructed aortic arch (18), and elastic properties of the central arterial tree are known to have a profound influence on ventricular function (19–21). In patients who have undergone anatomically successful repair of COA not associated with NP, abnormal arch geometry has been associated with abnormalities of central aortic fluid dynamics and biomechanics that may predispose to adverse cardiovascular outcomes (10). We did not find an association between arch geometry and outcome after BD for COA after the NP. However, discrete COA is but 1 form of increased ventricular afterload, and further study is necessary to characterize ventricular-arterial interaction and to understand more fully the anatomic, hydrodynamic, and mechanical load imposed by a reconstructed aorta after the NP.
Interpretation of our results is subject to certain limitations. Although we found that among the patients who underwent NP and had follow-up for longer than 1 month, 29% had interventions for COA, this may not accurately reflect the incidence of COA after the NP. The primary goal of this study was not to determine the incidence of, or risk factors for, arch reintervention after the NP. The study population represents a selected group, because patients who died with COA, those who underwent arch surgery without prior BD, and those with obstruction who did not undergo any arch reintervention were not included. Also, because of the retrospective nature of this study, there were no pre-established criteria for diagnosing recurrent COA; therefore, the indications for repeat intervention may not have been uniform across the study population.
BD was acutely successful in the large majority of patients with COA after the NP. However, both catheter-based and surgical reintervention were common in this patient population, especially in the first 6 months after BD. We did not find any association between surgical factors and acute or intermediate outcomes after BD. Obstruction that involved the neoaortic arch proximal to the takeoff of the LSCA was less responsive to BD and was associated with worse outcomes. Finally, younger patients and those with less successful acute results were more likely to undergo reintervention.
The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- atrioventricular valve regurgitation
- balloon aortoplasty
- aortic arch obstruction
- left subclavian artery
- Norwood procedure
- Received June 2, 2010.
- Revision received November 15, 2010.
- Accepted December 23, 2010.
- American College of Cardiology Foundation
- Zeltser I.,
- Menteer J.,
- Gaynor J.W.,
- et al.
- Ou P.,
- Celermajer D.S.,
- Raisky O.,
- et al.
- Hosein R.B.,
- Clarke A.J.,
- McGuirk S.P.,
- et al.
- Eren M.,
- Gorgulu S.,
- Uslu N.,
- et al.
- Hundley W.G.,
- Kitzman D.W.,
- Morgan T.M.,
- et al.