Long-Term Outcomes After Interrupted Aortic Arch Repair

Abstract

BACKGROUND:

Interrupted aortic arch (IAA) is associated with left ventricular outflow tract obstruction (LVOTO) and DiGeorge syndrome. High-risk infantile surgery is required to address IAA, with limited data available on long-term outcomes. We used the Pediatric Cardiac Care Consortium, a multicenter US-based registry for pediatric cardiac interventions, to assess long-term outcomes after IAA repair by patient characteristics and surgical approach.

METHODS:

This is a retrospective cohort study of patients undergoing IAA repair between 1982 and 2003. Kaplan-Meier plots and Cox proportional hazards regression were used to examine associations with post-discharge deaths tracked by matching with the US National Death Index.

RESULTS:

Among 390 patients meeting inclusion criteria, 309 (79.2%) survived to discharge. Over a median follow-up of 23.6 years, 30-year survival reached 80.7% among patients surviving hospital discharge after initial IAA repair. Adjusted analysis revealed higher risk of death for type B vs. A (aHR 3.32, 95%CI: 1.48-7.44), staged repair (aHR 2.50; 95%CI: 1.14-5.50) and LVOTO interventions during initial hospitalization (aHR 4.12, 95%CI: 1.83-9.27), but not for LVOTO without need for interventions or presence of DiGeorge syndrome. There was a trend towards improved in-hospital and long-term survival over time during the study period.

CONCLUSIONS:

Staged repair, type B IAA and need for LVOTO intervention during initial hospitalization for repair are associated with high risk of death out to 30 years. Survival outcomes are improving, but further efforts need to minimize staged approach and risks associated with LVOTO relief procedures.

Interrupted aortic arch (IAA) is characterized by luminal discontinuity at three possible sites: type A (isthmus), B (between the carotid and subclavian arteries), and C (between the common carotid arteries). ¹ Each type can be subclassified based on the ipsilateral subclavian artery origin into subtype 1 (normal origin from the innominate artery) and 2 (aberrant origin). ² IAA is associated with ventricular septal defect (VSD) and posterior misalignment of the conal septum, causing left ventricular outflow tract obstruction (LVOTO). We use the term isolated IAA to distinguish IAA (with or without VSD) from IAA as part of complex congenital heart disease (CHD) with truncus arteriosus, aortopulmonary window, double-outlet right ventricle, transposition of great arteries, and single ventricle physiology. ³ IAA is associated with DiGeorge syndrome (DGS), a microdeletion syndrome with neural crest abnormalities. ⁴

Surgical approaches include primary repair (preferred approach if possible) and two-stage approach. ^3,5-8 Coexisting anomalies and variation in surgical approaches make it challenging to understand long-term outcomes of isolated IAA. We provide a nuanced report of long-term survival outcomes after isolated IAA repair using a large registry-based cohort tracking outcomes over 30 years.

PATIENTS AND METHODS:

This is a retrospective cohort study of patients with operated IAA enrolled in the PCCC, a US-based clinical registry of interventions for CHD, between 1982 and 2011. ^9,10 In-hospital survival outcomes were provided by the PCCC, and post-discharge events were identified by matching with the NDI. ^11,12 The study was approved by Emory University Institutional Review Board (IRB00080706) with consent waiver for patients enrolled up to April 15, 2003, date of implementation of the stricter HIPAA (Health Insurance Portability and Accountability Act) regulations.

We queried the PCCC for patients undergoing repair for isolated IAA. Inclusion criteria were repair at <1 year of age at a US PCCC center, US residence at time of repair, available direct identifiers for matching with the NDI registry and enrollment before April 15, 2003. Patients with missing surgical information or coexisting complex CHD were excluded.

Variables included patient sex, age at repair (neonates: 0-27 days, infants: 28 days-<1 year), weight at repair (<2.5 kg or >2.5 kg), era of surgery by tertiles (early: 1982-1992, middle: 1993-1997, late: 1998-2003), presence of DGS or other chromosomal abnormality, IAA type (A, B or C) and subtype (1 or 2), surgical repair (end-to-end anastomosis, subclavian and carotid flap repair, interposition graft repair), repair strategy as primary or staged, concomitant procedures during the same surgical admission such as pulmonary artery (PA) band or LVOTO relief with Damus-Kaye-Stansel (DKS) anastomosis or septal myectomy, and status of LVOTO as none, present not requiring intervention, and present requiring intervention. Variables were extracted from PCCC operative and catheterization notes.

In-hospital deaths were ascertained from the PCCC, and post-discharge events were identified by matching identifiers with events in the NDI up to December 31, 2021. ^11,12 Causes of death were obtained from NDI-Plus and categorized by International Classification codes (ICD-9 and 10). They were grouped CHD, CVD, or Non-CHD/Non-CVD. ¹³

Cohort characteristics were summarized by descriptive statistics using median and 25^th-75^th percentile (IQR) for non-normally distributed variables and counts and percentages for categorical variables. Continuous and categorical variables were compared by Wilcoxon Rank Sum and Chi-Square.

We used Kaplan-Meier survival plots with log-rank to compare unadjusted long-term survival for the total cohort and subgroups of interest. Proportional hazard was evaluated using log-log curves and by including a Heaviside function. To examine associations affecting in-hospital and post-discharge mortality, we used logistic regression and Cox proportional hazards regression and reported odds (OR) and hazard ratios (HR) with 95% confidence intervals (CI). We considered age, weight at repair, sex, era of repair, IAA type, presence of DGS or other chromosomal abnormality, absence or presence of LVOTO, surgical strategy, type of IAA repair, and concomitant procedures including PA banding, DKS, or septal myectomy. In adjusted logistic regression, we included variables of predetermined interest (sex, age group at repair, DGS, and surgical method) and those with p-value of <0.2 in univariable analysis. Due to strong correlation between staged approach and certain concomitant procedures, we constructed two multivariable models: I for repair strategy and II for concomitant procedures.

RESULTS

Between 1982 and April 15, 2003, 464 patients met inclusion criteria with 390 having direct identifiers (Figure 1 and Table 1). The most common type was B. Aberrant subclavian occurred frequently, especially with type B IAA (Supplemental Table 1). LVOTO was more common with type B but without reaching not statistical significance. Given the scarcity of type C, it was excluded from mortality analysis.

Figure 1.

Flowchart of patient selection for IAA cohort and their outcomes

Table 1.

Characteristics of patients undergoing IAA repair

Clinical characteristics	Overall N=390 (%)	Discharged alive N=309 (%)	In-hospital deaths N=81 (%)	Post-discharge deaths N=55 (%)
Median age at repair in days (IQR)	7 (4-12)	7 (4-12)	6 (4-13)	6 (3-9)
Age at repair
Newborns (< 28 days)	33 (8.5)	28 (9.1)	5 (6.2)	4 (7.3)
Infants (28 days-<1 year)	357 (91.5)	281 (90.9)	76 (93.8)	51 (92.73)
Sex
Female	186 (47.7)	135 (43.7)	51 (63.0)	27 (49.1)
Male	204 (52.3)	174 (56.3)	30 (37.0)	28 (50.9)
Median weight at repair in kg (IQR)	3.2 (2.8-3.7)	3.28 (2.9-3.8)	2.9 (2.3-3.2)	3.1 (2.8-3.7)
Weight at repair
≥2.5 kg	319 (84.4)	263 (88)	56 (70.9)	46 (86.8)
<2.5 kg	59 (15.6)	36 (12)	23 (29.1)	7 (13.2)
Type of IAA
A1	98 (25.1)	88 (28.5)	10 (12.4)	8 (14.6)
A2	9 (2.3)	6 (1.9)	3 (3.7)	0
B1	181 (46.4)	139 (45)	42 (51.9)	29 (52.7)
B2	96 (24.6)	72 (23.3)	24 (29.6)	17 (30.9)
C	6 (1.5)	4 (1.3)	2 (2.5%)	1 (1.8)
Subclavian aberrancy
Non-aberrant	279 (71.5)	227 (73.5)	52 (64.2)	37 (67.3)
Aberrant	105 (26.9)	78 (25.2)	27 (33.3)	17 (30.9)
Type of IAA repair
End-to-end	252 (64.6)	203 (65.7)	49 (60.5)	29 (52.7)
Interposition graft	89 (22.8)	65 (21.0)	24 (29.6)	18 (32.7)
Vascular flap	49 (12.6)	41 (13.3)	8 (9.9)	8 (14.6)
Repair strategy
Primary	287 (73.6)	229 (74.1)	58 (71.6)	31 (56.4)
Staged	103 (26.4)	80 (25.89)	23 (28.40)	24 (43.6)
Concomitant procedures
None	54 (13.8)	40 (12.94)	14 (17.28)	2 (3.64)
Yesa	336 (86.2)	269 (87.1)	67 (82.7)	53 (96.4)
VSD closure	220	179	41	22
PA band	103	80	23	24
Septal myectomy	19	19	0	3
DKS	14	11	3	8
LVOTO
None	187 (47.9)	144 (46.6)	43 (53.1)	18 (32.7)
LVOTO w/o intervention	170 (43.6)	135 (43.7)	35 (43.2)	26 (47.3)
LVOTO w/ interventionb	33 (8.5)	30 (9.7%)	3 (3.7)	11 (20.0)
Chromosomal abnormalities
None	258 (66.2)	201 (65.1)	57 (70.4)	37 (67.27)
DGS	126 (32.5)	104 (33.7)	22 (27.2)	18 (32.7)
Other	6 (1.54)	4 (1.3)	2 (2.5)	0
Turner Syndrome	1 (0.3)	1 (0.3)	0	0
Trisomy 21	5 (1.3)	3 (1.0)	2 (2.5)	0
Surgical era
Early (1982-1992)	134 (34.4)	101 (32.7)	33 (40.7)	24 (43.6)
Mid (1993-1997)	121 (31.0)	93 (30.1)	28 (34.6)	20 (36.4)
Late (1998-2003)	135 (34.6)	115 (37.2)	20 (24.7)	11 (20.0)

IAA=interrupted aortic arch, LVOTO=left ventricular outflow tract obstruction, DGS=DiGeorge syndrome, VSD=ventricular septal defect, PA= pulmonary artery, DKS=Damus-Kaye-Stansel. Numbers in parenthesis represent percentiles by columns.

^a20 patients had >1 procedure.

^bLVOTO interventions include DKS and septal myectomy.

Most patients underwent end-to-end anastomosis (64.6%), staged repair with PA band occurred in 103, while interventions for LVOTO were needed in 33. One third of the cohort had DGS, with almost all having type B IAA (Table 1, Supplemental Table 2).

In-Hospital Mortality

In-hospital death occurred in 81 patients. Higher risk for in-hospital death was associated with females (aOR 1.83; 1.06-3.15), weight of <2.5 kg (aOR 2.60; 95% CI 1.34-5.04), type B (aOR 3.13; 1.53-6.41), and early surgical era (aOR 2.16; 1.02-4.56) (Table 2). DGS was associated with lower in-hospital mortality (aOR 0.44; 0.24-0.82), while age, type of repair, repair strategy, concomitant procedure, aberrant subclavian and LVOTO with or without intervention did not affect in-hospital mortality.

Table 2:

Variables associated with in-hospital mortality after IAA repair

Clinical Characteristics	Unadjusted OR (95% CI)	P-value	Adjusted ORa (95% CI)	P-value
Age at repair
Infant vs. Neonate	0.67 (0.25-1.79)	0.42	0.62 (0.22-1.76)	0.37
Sex
Female vs. Male	2.14 (1.29-3.56)	0.003	1.83 (1.06-3.15))	0.03
Weight at repair
<2.5 kg vs ≥2.5 kg	2.92 (1.59-5.35)	0.0005	2.60 (1.34-5.04)	0.005
Type of IAA
B vs A	2.26 (1.19-4.30)	0.013	3.13 (1.53-6.41)	0.002
Subclavian aberrancy
Aberrant vs non-aberrant	1.51 (0.89-2.57)	0.13	-	-
Type of IAA repair
Flap/graft vs end-to-end	1.30 (0.78-2.15)	0.32	1.33 (0.66-2.67)	0.43
Repair Strategy
Staged vs Primary	1.02 (0.59-1.78)	0.92	0.73 (0.34-1.59)	0.43
Concomitant Procedure
Present vs Absent	0.81 (0.47-1.38)	0.43	-	-
LVOTO
None	Ref	Ref	Ref	Ref
LVOTO w/o intervention	0.92 (0.55-1.53)	0.74	1.01 (0.59-1.75)	0.96
LVOTO w/ intervention	0.36 (0.11-1.25)	0.11	0.28 (0.08-1.03)	0.06
DGS
Yes vs No	0.72 (0.42-1.25)	0.25	0.44 (0.24-0.82)	0.01
Era of Surgery
Early (1982-1992)	1.81 (0.97-3.36)	0.06	2.16 (1.02-4.56)	0.04
Mid (1993-1997)	1.66 (0.87-3.15)	0.12	1.74 (0.88-3.44)	0.11
Late (1998-2003)	Ref	Ref	Ref	Ref

OR=odds ratio, IAA=interrupted aortic arch, LVOTO=left ventricular outflow tract obstruction, DGS=DiGeorge syndrome.

^acalculated after controlling for sex, IAA type, presence of DGS, surgical era, age at repair, low surgical weight <2.5kg and LVOTO status

Long-term Survival After Repair

Among 309 patients discharged, 55 deaths occurred over a median follow-up of 23.6 years (IQR: 20.04-28.05 years, max 38.2 years). Survival after discharge declined rapidly within 3 years after repair followed by a steady low attrition to reach a 30-year post-discharge survival of 80.7% among patients survived to hospital discharge after their initial surgical repair (Figure 2A). Late surgical era was associated with improved 20-year-survival but without reaching statistical significance (Figure 2B). Patients with type B had significantly lower survival (p=0.03), as well as patients with LVOTO interventions or staged repair (Figure 3A-C). DGS (Figure 3D), sex, age and weight at repair, type of repair and presence of aberrant subclavian artery were not associated with post-discharge death.

Figure 2.

Post-discharge Kaplan-Meier survival plot with 95% CI following IAA repair for the overall cohort (A) and by surgical era (B).

Figure 3:

Post-discharge Kaplan-Meier survival plot with 95% CI following IAA repair by (A) IAA type, B) LVOTO status, (C) repair strategy and (D) presence of DGS

Adjusted analysis for associations with higher 30-year risk of death (Model I solved for repair strategy) revealed higher risk for type B (aHR 3.32, 95% CI: 1.48-7.44), patients undergoing staged repair (aHR=2.50; 95% CI 1.14-5.50), and for those requiring intervention for LVOTO (aHR 4.12, 95% CI: 1.83-9.27) (Table 3). In this model, there was no statistical difference in outcomes by era. Model II (solved for concomitant procedures) revealed a trend towards decreased risk of death with more recent era (p=0.05) (Supplemental Table 3).

Table 3:

Variables associated with post-discharge mortality after IAA repair

	Unadjusted HR (95% CI)	P-value	Adjusted HRa (95% CI)	P-value
Age at repair
Infant vs. Neonate	0.76 (0.28-2.11)	0.60	1.45 (0.50-4.18)	0.50
Sex
Female vs. Male	1.21 (0.71-2.07)	0.48	1.06 (0.61-1.86)	0.84
Weight at repair
< 2.5 kg vs. ≥ 2.5 Kg	1.27 (0.57-2.83)	0.55	1.29 (0.57-2.92)	0.54
Type of IAA
B vs. A	2.96 (1.39-6.29)	0.005	3.32 (1.48-7.44)	0.004
Subclavian aberrancy
Aberrant vs non-aberrant	1.44 (0.81-2.55)	0.22	-	-
Type of IAA repair
Flap/graft vs. End-to-end	1.65 (0.96-2.83)	0.07	1.16 (0.56-2.37)	0.69
Repair Strategy
Staged vs. Primary	2.59 (1.51-4.46)	<.001	2.50 (1.14-5.50)	0.02
LVOTO
None	Reference		Reference
LVOTO w/o intervention	1.58 (0.86-2.91)	0.14	1.43 (0.77-2.68)	0.26
LVOTO w/ intervention	3.89 (1.83-8.27)	<.001	4.12 (1.83-9.27)	<.001
DGS
Yes vs No	1.08 (0.61-1.90)	0.80	1.10 (0.57-2.12)	0.78
Era of Surgery
Early (1982-1992)	2.09 (1.00-4.36)	0.05	1.70 (0.72-4.02)	0.23
Mid (1993-1997)	2.07 (0.98-4.36)	0.06	2.03 (0.94-4.36)	0.07
Late (1998-2003)	Reference		Reference

HR=hazard ratio, IAA=interrupted aortic arch, LVOTO=left ventricular outflow tract obstruction, DGS=DiGeorge syndrome

^acalculated after controlling for sex, IAA type, presence of DGS, surgical era, age group at repair, repair strategy, low surgical weight <2.5kg and LVOTO status.

Among 55 post-discharge deaths, 55% were attributed to the IAA itself. Additional cardiovascular conditions were identified in 47%, most in association with type B. Nine deaths were not attributed to either CHD or CVD (Supplemental Table 4).

COMMENT

IAA features challenging surgical management due to frequent LVOTO. ^3,14,15 In this large multi-center cohort study in-hospital mortality was 20.8%, with higher risk in females, low weight (<2.5 kg) at surgery and early surgical era (1982-1992), consistent with contemporary reports.^3,5,7 Counterintuitively, DGS was associated with a protective effect after adjustment for other risk factors, possibly due to unmeasured variables not accounted in our study. The 30-year post-discharge survival reached 80.7% and was lower with type B, staged approach, and interventions for LVOTO. Mortality in those with LVOTO without intervention was no different than absence of LVOTO. Age, weight, sex, and DGS were not associated with long-term survival. There was a trend towards improved in-hospital and post-discharge survival in the most recent cohort.

To our knowledge, this is the largest IAA cohort with longest follow-up. Several single centers assessed post-surgical mortality in IAA with <10 years of follow-up. ^6,8,15,16 Fewer are multi-institutional studies with longer follow-up, usually describing mixed cohorts including IAA with complex CHD. ^3,17 Complex CHD can severely affect long-term outcomes and do not allow assessment of the salient features of isolated IAA, namely the LVOTO. The previous longest follow-up, reported by the Congenital Heart Surgeons Society (CHSS), ^3,17 included 472 neonates undergoing IAA repair between 1987 and 1997 and showed a 16-year survival of 59%, comparable with our 20-year survival (approximately 60%). Due to the cohort differences, direct comparisons of LVOTO associations with outcomes is difficult. In our cohort, we found no association of LVOTO without intervention with long-term survival, possibly due to limited granularity on LVOTO degree (measurements of aortic valve and LVOT). Grading LVOTO adequacy to maintain systemic outflow remains challenging prior to VSD closure and there are only scarce data on LVOT and aortic valve sizes and Z-scores. ^18-20 Even when available, there is debate on their significance for risk of residual obstruction post IAA repair. Previous reports have suggested that small aortic valve annulus (z-score<−5) predicted development of LVOTO after arch repair, ^21,22 while others suggested indexed LVOT cross-sectional area of <0.7 mm/m², not the valve, predicted post-repair LVOTO. ^18,20 In addition, catch-up growth of the aortic valve and subaortic area have been reported and may at least partially explain the lack of LVOTO association with long-term survival. ^14,23-26 Precise and standardized LVOT measurements are evolving, offering potential new insights into the relationship between LVOTO and late outcomes. The CHSS group also reported lower birth weight, younger age at surgery, and female sex to be associated with higher early mortality risk after IAA repair. While we noticed similar trends for in-hospital outcomes, none of this affected long-term survival.

The most relevant report for comparison is from an Australian group. ⁵ This group reported 30-day postsurgical mortality and 10-year survival for a single center cohort of patients undergoing IAA repair, including 122 patients with simple IAA mirroring our study’s characteristics. This cohort had similar distribution of surgical approaches with slightly lower use of staged palliation. Both cohorts show similar outcomes when converted to 30-day mortality, with comparable 10-year survival rates for all eras, except for slightly lower but non-statistically significant survival in the PCCC cohort during the later era (74.8% vs. 92%, p=0.53). Neither study found association of LVOTO with early or late outcomes, whereas our study found a trend towards improved early outcomes after interventions for LVOTO but higher risk of death over the 25 years of post-discharge follow-up. This difference between the two studies may be explained by higher statistical power in our study with larger patient numbers (390 vs 122) and longer follow-up (23.6 vs 11.5 years). The differential effect of LVOTO intervention on early vs late outcomes may be explained by the likely protective effect the relief of obstruction offers in the immediate postoperative period, while in the longer term these patients remain at risk secondary to incomplete relief/recurrence of the LVOTO in the patients with septal myectomy or to DKS-related morbidity in the rest of them.

Type B is associated with higher risk for long-term mortality following IAA repair even after adjustment for associated risk factors such as concomitant LVOTO, presence of aberrant subclavian and coexistence with DGS. ^{3,16,17,27,28} The worse outcomes with type B may be explained by interaction between the anatomic characteristics associated with this anatomy, although this could not be tested with the available number of cases and events. The differential outcome may be from unmeasured variables such as distance of interruption in type B, as type A is anatomically akin to coarctation of the aorta.

DGS is not associated with increased long-term mortality, aligning with previous reports. ²⁹ A similar observation was noted in a contemporary report from the PCCC on outcomes for truncus arteriosus communis (TAC) with type B IAA. ³⁰ Arch repair within the context of TAC had a 30-year survival of 90%, which is comparable to the survival for isolated IAA without LVOTO (88.1%) in the current study, but higher than the 77.7% survival with IAA and LVOTO without intervention and 54.6% for patients requiring LVOTO intervention. We suspect this is related to the less frequent LVOTO in IAA with TAC. Mortality after IAA repair modestly improved over time . Surgical and medical advancements have contributed to reduced long-term mortality over the length of our study. Multiple factors are likely responsible including less reliance on artificial tissue and frequent one-stage repair with primary anastomosis and VSD closure. Post-discharge mortality due to cardiovascular causes of death suggests significant residual cardiac pathology including heart failure and arrhythmias, highlighting the importance of surveillance in clinics for adults with CHD.

Our study’s strengths include the multi-center representation, extended follow-up and focus on isolated IAA, thus providing a pure assessment of this group’s salient features. Limitations are the retrospective registry-based nature including less granularity on features like length of the interrupted segment, degree of LVOTO or nature of the aortic valve. Thus, we used a qualitative description of the LVOTO. There are also differences between subgroups with and without identifiers that may contribute to higher in-hospital mortality in the latter. In the PCCC cohort, patients lacking identifiers are typically younger with more severe disease and increased in-hospital mortality. ^11,12,30 Sensitivity analysis using inverse probability weighting in the overall cohort revealed a minimal impact on long-term survival (1.1%), ¹¹ but potential bias related to unaccounted characteristics remains. An additional limitation is the use of ICD-9 codes for causes of death as we have reported previously. ¹³ Nevertheless, the NDI-Plus is the “gold standard” for reported causes of death from public datasets and has been found to provide meaningful results. Lastly, there may be unmeasured variables that explain the unlikely benefit of DGS on in-hospital mortality. Further, we cannot provide comprehensive assessment of residual lesions (such as LVOTO) nor reoperations as some may undergo subsequent procedures in non-PCCC centers. Although technical innovation for IAA and LVOTO has not evolved much over the last two decades, advancements in management have occurred including more precise diagnostics for assessing LVOT adequacy.

Despite these limitations, our report extends previously reported outcomes on IAA by 10-15 years and up to an age that captures the impact of reintervention and interaction with adult cardiovascular conditions. Surveillance of this population is paramount for early detection and treatment of severe, but modifiable morbidity from heart failure and arrhythmias. Future studies incorporating standardized assessment of the LVOTO will provide better selection of patients to benefit from judicious use of LVOTO interventions.

ABBREVIATIONS

CHD

Congenital Heart Disease

CVD

Cardiovascular Disease

DGS

DiGeorge Syndrome

DKS

Damus-Kaye-Stansel

IAA

Interrupted Aortic Arch

LVOTO

Left Ventricular Outflow Tract Obstruction

NDI

National Death Index

PA

Pulmonary Artery

VSD

Ventricular Septal Defect