Interactive CardioVascular and Thoracic Surgery 28 (2019) 291–300 STATE-OF-THE-ART

doi:10.1093/icvts/ivy224 Advance Access publication 28 July 2018

Cite this article as: Wu Y, Jin X, Kuang H, Lv T, Li Y, Zhou Y et al. Is balloon angioplasty superior to surgery in the treatment of paediatric native coarctation of the
aorta: a systematic review and meta-analysis. Interact CardioVasc Thorac Surg 2019;28:291–300.

Is balloon angioplasty superior to surgery in the treatment of

paediatric native coarctation of the aorta: a systematic review and
meta-analysis
Yuhao Wua,b,c,d,†, Xin Jina,b,c,d,†, Hongyu Kuangb,c,d,e, Tiewei Lvb,c,d,e, Yonggang Lia,b,c,d,

Downloaded from https://academic.oup.com/icvts/article-abstract/28/2/291/5061132 by guest on 21 November 2019

Yuehang Zhoua,b,c,d and Chun Wua,b,c,d,*
a
Department of Cardiothoracic Surgery, Children’s Hospital of Chongqing Medical University, Chongqing, China
b
Ministry of Education Key Laboratory of Child Development and Disorders, Chongqing, China
c
China International Science and Technology Cooperation Base of Child Development and Critical Disorders, Chongqing, China
d
Chongqing Key Laboratory of Pediatrics, Chongqing, China
e
Department of Cardiology, Children’s Hospital of Chongqing Medical University, Chongqing, China

* Corresponding author. Department of Cardiothoracic Surgery, Children’s Hospital of Chongqing Medical University, No. 136 Zhongshan Second Road, Yuzhong
District, Chongqing 400014, China. Tel: 8618883883009; e-mail: wuchun007@sina.com (C. Wu).

Received 12 February 2018; received in revised form 12 June 2018; accepted 12 June 2018

Summary CONGENITAL
A meta-analysis was performed to compare the outcomes between surgery and balloon angioplasty (BA) for native coarctation of the aorta
in paediatric patients. Electronic databases, including PubMed, EMbase, Medline and Cochrane Library were searched systematically for
literature aimed mainly at comparing the therapeutic effects for native coarctation of the aorta administered by surgery or BA.
Corresponding data sets were extracted and 2 reviewers independently assessed the methodological quality. Ten studies meeting the in-
clusive criteria were identified involving a total of 723 subjects. Eventually, it was observed that compared with BA, surgery was signifi-
cantly associated with a lower incidence of recoarctation, repeat intervention due to recoarctation and residual transcoarctation gradient
in mid- to long-term follow-up. However, BA was significantly associated with a shorter hospitalization time. Incidence of aneurysm for-
mation, perioperative mortality, complications and immediate transcoarctation residual gradient were not statistically different between
†The first two authors contributed equally to this work.

C The Author(s) 2018. Published by Oxford University Press on behalf of the European Association for Cardio-Thoracic Surgery. All rights reserved.
V
292 Y. Wu et al. / Interactive CardioVascular and Thoracic Surgery
surgery and BA. The overall level of evidence for our study was low and randomized controlled trials should be designed to evaluate and
compare the safety and effectiveness of both approaches for native coarctation of the aorta.
Keywords: Balloon angioplasty • Surgery • Paediatric • Treatment outcome
INTRODUCTION
Native coarctation of the aorta (NCA) is a common congenital heart
defect that accounts for 5–8% in all congenital heart disease patients
and with a frequency of 0.4 per 1000 live births [1]. The surgical re-
pair of coarctation of the aorta (CoA) was reported by Crafoord and
Nylin [2] in 1944 for the first time, involving resection and recon-
struction with an end-to-end anastomosis. Surgery was the only
treatment of CoA until balloon angioplasty (BA) was performed by
Singer et al. [3] in 1982. Since then, BA was widely adopted for CoA,
which was thought to be less invasive and safer. In the past 2 deca-
des, there has been an increasing number of studies comparing the
outcomes of BA versus surgery in the treatment of CoA. Despite
that, there is no clear evidence so far for the superiority of BA over
surgery. Since vascular growth in paediatric patients is fast, the com-
parative effectiveness of BA and surgery could be totally different
from those of adults. Therefore, this meta-analysis aimed to evaluate
the clinical outcomes of BA and surgery for NCA in paediatric
patients, as well as to provide explicit evidence to determine if BA in
the treatment of paediatric NCA is superior.
MATERIALS AND METHODS
Literature search
A systematic search of the PubMed, EMbase, Medline and
Cochrane Library for the relevant published studies comparing the
outcomes of the BA and surgery for NCA was conducted. The key
words used for searching were (surgery OR surgical treatment OR
conventional surgery) AND (balloon angioplasty OR interventional
therapy OR intervention OR transcatheter) AND (coarctation of the
aorta OR aortic coarctation OR CoA). References from all the in-
cluded studies and other relevant literature were also manually
reviewed to identify additional eligible studies. We contacted the
authors to obtain extra information through e-mail as necessary.
Study selection
A study was included in this systematic review when the follow-
ing criteria were met: (i) observational studies (cohort or case-
controlled studies) or randomized controlled trials (RCTs); (ii)
studies presenting head-to-head comparison of BA and surgery
for NCA; (iii) paediatric patients younger than 16 years old were
included. A study was excluded in this systematic review when
the following criteria were met: (i) multiple studies were based
on the same data; (ii) study included the treatment of recurrent
CoA patients; (iii) study included patients older than 16 years old.
Reviews, conference records, case reports and animal experi-
ments were also excluded. Only the study with biggest sample
size or with most complete set of recorded clinical outcomes was
included when several studies were based on the same sample.
Two reviewers (W.Y.H. and J.X.) screened all the studies inde-
pendently, and any disagreement on the eligibility of studies
were resolved by consensus with the third reviewer (K.H.Y.).
Data extraction
Data were extracted by both reviewers (W.Y.H. and J.X.) indepen-
dently, any disagreement was resolved by consensus with the
help of a third reviewer (K.H.Y.). A standardized extraction form
in an Excel spreadsheet was used. The following information was
extracted: (i) basic characteristics of included studies: first author,
Downloaded from https://academic.oup.com/icvts/article-abstract/28/2/291/5061132 by guest on 21 November 2019
publication year, study district, study design, surgical approach,
sample size, weight, age, follow-up time and surgical strategy; (ii)
outcomes of both surgical approaches for: residual transcoarcta-
tion gradient, hospitalization time, perioperative mortality,
incidence of repeat intervention due to recoarctation (re-CoA),
re-CoA and complications. The postoperative complications in-
cluded chylothorax, spinal cord injury, bleeding, etc. Since aneu-
rysm formation is commonly observed after intervention, we
independently analysed the incidence of postoperative aneurysm
formation. Re-CoA was defined as the transcoarctation gradient
after initial operation >20 mmHg (evaluated with echocardiogra-
phy or catheterization). Repeat intervention was defined as sur-
gery or BA performed only for re-CoA that occurred after
discharge following initial intervention. Perioperative mortality
was defined as death related to surgical treatment before dis-
charge. Short-, mid- and long-term follow-up were defined as
follow-up times of <1 year, 1–3 years and >3 years, respectively.
Patients with hypoplastic aortic arch were identified as having a
proximal or distal transverse arch diameter less than 50% of as-
cending aorta diameter as evaluated by cardiac echo or CT scan.
Quality assessment
Quality assessment was performed by both reviewers (W.Y.H. and
K.H.Y.) independently, and any disagreement on the assessed
quality was resolved by consensus with a third reviewer (J.X.). The
quality of the none-randomized controlled trials (NRCTs) was
evaluated by the Newcastle-Ottawa Scale (NOS) scores. A star sys-
tem was adopted in which a study was judged on 3 parts of the
NOS scale: the selection of study groups, the comparability of
study groups, and the ascertainment of exposure [4].The maxi-
mum scores of NOS was 9, and each study was graded as either
low (0–5) or high quality (6–9). The quality of the RCTs was evalu-
ated with the Cochrane collaboration tool [5], which provided the
assessment of the risk of bias. This tool determined the risk of bias
based on the following 6 dimensions: (i) details of randomization
method; (ii) allocation concealment; (iii) blinding of participants,
personnel and outcome assessment; (iv) incomplete outcome
data; (v) selective reporting; and (vi) other sources of bias.
We assessed the level of evidence with the Grades of
Recommendation, Assessment, Development and Evaluation sys-
tem (GRADE) [6]. The GRADE system included: (i) high quality:
further research is very unlikely to change the confidence in the
estimate of effect; (ii) Moderate quality: further research is likely
to have an crucial impact on the confidence in the estimate of ef-
fect and may change the estimate; (iii) Low quality: further re-
search is extremely likely to have an crucial impact on the
confidence in the estimate of effect and is likely to change the
 Y. Wu et al. / Interactive CardioVascular and Thoracic Surgery
Figure 1: Flowchart of literature search in the meta-analysis. BA: balloon angioplasty; CoA: coarctation of the aorta.
estimate; (iv) Very low quality: any estimate of effect is very
uncertain.
Statistical analysis
All statistical analyses were conducted by using Stata 12.0 (Stata
Corp, TX, USA) and RevMan software (version 5.3; The Nordic
Cochrane Centre, Copenhagen, Denmark), and P-value <0.05 was
considered statistically significant. Odds ratio (OR) or relative risk,
weighted mean difference (WMD) were employed for dichoto-
mous and continuous data, respectively. The Cochrane Q test
and the I2 statistics were used to assess the heterogeneity, with I2
>50% indicating heterogeneity. If the I2 statistic >50%, a random-
effects model was employed. Subgroup analyses and sensitivity
analyses were used to explore the sources of heterogeneity; oth-
erwise, a fixed-effects model of analysis was employed. If only
the median value and range were available, then the formulas
provided by Hozo et al. [7] were used to estimate the mean val-
ues and standard differences. Sensitivity analyses were performed
using the leave-one-out method in which the outcomes of meta-
analysis were conducted with each study removed in turn until
significant reduction in heterogeneity was observed. If the direc-
tion and magnitude of pooling estimates with respect to the out-
comes did not vary markedly with the removal of any given
study, it indicated that the meta-analysis was confirmed and the
data were not overly influenced by any study. In the meta-
analysis of the immediate residual transcoarctation gradient, a
subgroup analysis was conducted separately for the peak or
mean gradient.
RESULTS
Characteristics of included studies and quality
assessments
A total of 3716 studies were obtained initially. After screening for
duplicates and relevance in titles and abstracts, only 33 studies
were available for the full-text evaluation of eligibility. Eventually,
293
Downloaded from https://academic.oup.com/icvts/article-abstract/28/2/291/5061132 by guest on 21 November 2019
this meta-analysis was based on 9 NRCTs and 1 RCT [8–17]. A
flowchart depicting the search strategy is shown in the Fig. 1.
Of the 10 included studies, 9 studies were retrospective NRCTs
and 1 study was RCT. A total of 723 patients were involved in this
study, of which 279 patients were in the BA group and 444
patients in the surgery group. Only patients under the age of 16
years were included (Table 1).
Quality assessments of included studies were in accordance
with the NOS and the Cochrane collaboration tool. All NRCTs
were retrospective and deemed to be of high quality on the basis
of the NOS score for retrospective cohort study; 1 study was
assessed with 7 stars, and 8 studies with 6 stars. All 9 NRCTs clari-
CONGENITAL
fied their inclusive and exclusive criteria, however, all 9 studies
had flaws in comparability (Table 2) or adequacy of follow-up.
Pairs match or propensity score matching was not adopted in
any of our included NRCTs. No statistical difference was found
between surgery and BA groups in baseline characteristics such
as age, weight, gender and preoperative gradient among all stud-
ies except one [13]. However, in 2 of our included studies [9, 14],
the real P-value was not reported. Unfortunately, the details of
blinding, randomization method and allocation concealment
were absent in the RCT by Cowley et al. [17], therefore, its risk of
bias was critical. The characteristics and quality assessments of
the included studies are shown in details in the Table 1.
Associated anomalies and the percentage of aortic arch hypo-
plasia in CoA patients were presented in the Table 3. In our in-
cluded studies, patent ductus arteriosus, ventricular septal defect
and atrial septal defect were the leading associated cardiac
anomalies of CoA. Four studies [8, 10, 12, 13] which did not re-
port isthmus hypoplasia or arch hypoplasia were included in our
studies. Two out of 10 studies reported the incidence of hypopla-
sia of aortic arch as 39.6% and 37.9%. However, no statistical dif-
ference was found between both approaches with regard to
hypoplastic aortic arch [9, 15]. Two studies [11, 14] did not report
the percentage of hypoplastic aortic arch but had insignificant
differences between the 2 approaches. No data was available re-
garding hypoplastic aortic arch in the last 2 studies [16, 17].
Incidence of re-CoA was provided by all 9 NRCTs and 1 RCT,
therefore, we only conducted the funnel plot of the incidence of
 294

Table 1: Characteristics and quality assessment of included studies

Authors Study Study Surgical Patients, Age Weight (kg) Follow-up Surgical Quality
district design approach male/female time strategy assessment
(gender)

Alaei et al. [8] Iran Retrospective SU group 112 132 ± 89.4 days 4.7 ± 1.3 6–24 months NA 6 (high)
BA group 55 153 ± 95.6 days 6.8 ± 9.0
Chiu et al. [9] Taiwan Retrospective SU group 128 3.3 ± 7.2 years NA 10.0 ± 5.3 years 70 RETE 6 (high)
BA group 41 3.0 ± 4.9 years 11.0 ± 5.3 years 25 PPA
7 SFA
14 IG
Zhang et al. [10] China Retrospective SU group 38/15 59.6 (15–190) daysa 4.0 (2.1–6) 33 (6–63) month 13 RETE 6 (high)
BA group 29/10 52.3 (9–184) days 4.0 (1.9–7) 62 (9–119) month 40 EETE
Lin et al. [11] Taiwan Retrospective SU group 9/3 26 (9–94) daysa 3.4 (2–5.5) 557 (56–1560) days NA 6 (high)
BA group 7/2 27 (7–104) days 3.7 (2.3–6.2) 1179 (232–1689) days
Dijkema et al. [12] Netherland Retrospective SU group 13/16 4.9 ± 5.2 years NA 28.8 ± 9.7 months 20 RETE 6 (high)
BA group 18/1 5.4 ± 4.7 years 17.1 ± 2.9 months 15 SFA
1 IG
3 unknown
Walhout et al. [13] Netherland Retrospective SU group 18 0.6 (0.4–14) yearsa NA 7.2 ± 2.4 years 16 RETE 6 (high)
BA group 28 5.8 (0.3–15) years 5.4 ± 2.8 years 1 SFA
1 RETE and SFA
Fiore et al. [14] USA Retrospective SU group 20/14 7.7 daysb 3.4 38 months 26 EETE 7 (high)
BA group 15/8 8.8 days 3.6 36 months 8 SFA
Hernández- Mexico Retrospective SU group 20/8 7.0 ± 4.1 years 26.6 ± 15.1 221 daysb NA 6 (high)
González
et al. [15]
BA group 22/8 6.6 ± 4.4 years 25.4 ± 12.5 196 days
Rao et al. [16] USA Retrospective SU group 8/6 27 ± 35 days 3.5 ± 0.9 4.5 years 10 RETE 6 (high)
(3 months–9 years)a
BA group 9/6 29 ± 27 days 3.8 ± 1 4.0 years 2 SFA
(4 months–8 years)
2 IG
Cowley et al. [17] USA RCT SU group 16 5.7 ± 2.1 years 19.9 ± 5.2 11.0 ± 4.9 years NA Low quality
BA group 20 6.3 ± 2.0 years 21.7 ± 8.1 9.9 ± 5.2 years
Y. Wu et al. / Interactive CardioVascular and Thoracic Surgery

a
Median value and range.
b
Mean value.
BA: balloon angioplasty; EETE: extended end-to-end anastomosis; IG: interposition graft; NA: not available; PPA: pericardium patch aortoplasty; RCT: randomized controlled trial; RETE: resection and end to end anasto-
mosis; SFA: subclavian flap angioplasty; SU: surgery.

Downloaded from https://academic.oup.com/icvts/article-abstract/28/2/291/5061132 by guest on 21 November 2019

 Y. Wu et al. / Interactive CardioVascular and Thoracic Surgery 295

Table 2: P-value in the comparability of included non-randomized controlled trials

Authors Age (years) Weight Gender Preoperative Associated BSA Percentage of Coarctation/AsAo Pairs match/ Include/
(kg) gradient anomalies hypoplastic dimensiona propensity exclude
aortic arch score matching criteria

Alaei et al. [8] 0.17 0.10 0.40 0.57 SDb NA NA Clarified

Chiu et al. [9] NSDc NA NSDc NA NSDd NA 0.58 NA NA Clarified
Zhang et al. [10] 0.46 0.82 0.78 0.72 NSDe 0.85 NA Clarified
Lin et al. [11] 0.72 0.78 1.00 0.23 NSDd NA 0.32 0.32 NA Clarified
Dijkema et al. [12] 0.69 NA <0.001 NA NSDd NA NA Clarified
Walhout et al. [13] 0.001 NA NA 0.50 NSDd NA NA Clarified
Fiore et al. [14] NSDc NSDc NSDc NSDc NA NA NSDc NSDc NA Clarified

Downloaded from https://academic.oup.com/icvts/article-abstract/28/2/291/5061132 by guest on 21 November 2019

Hernández- 0.97 0.90 0.65 0.26 NA NA 0.79 NA NA Clarified
González et al. [15]
Rao et al. [16] 0.86 0.40 0.88 NA 0.73 NA NA NA NA Clarified

P-value <0.05 considered statistically different.

a
The diameter of the narrowest aorta to ascending aorta diameter ratio.
b
Several types of associated anomalies were reported and most of them were statistically different.
c
The real P-value was not reported.
d
Several types of associated anomalies were reported and most of them were not statistically different.
e
All the patients were associated with VSD.
BSA: body surface area; NA: not available; NSD: no statistically different; SD: statistically different; VSD: ventricular septal defect.

Table 3: The associated anomalies of CoA and percentage of hypoplastic aortic arch

Authors Associated cardiac anomalies (n) Hypoplastic aortic arch

(%)

Alaei et al. [8]a 15 MS; 18 AS; 122 PDA; 35 ASD; 77 VSD; 13 TGA 0
Chiu et al. [9] 107 PDA; 90 VSD; 13 bicuspid aortic valve; 21 LVOTB 39.6b
Zhang et al. [10]c 92 VSD 0
Lin et al. [11] 12 VSD; 2 AS; 1 AVSD; 3 ASD; 1 cor triatriatum NAb
Dijkema et al. [12]a 31 bicuspid aortic valve; 2 AS; 2 ASD; 5 VSD 0
Walhout et al. [13]a 13 bicuspid aortic valve; 13 PDA; 8 AS; 3 ASD; 8 VSD; 1 TGA 0

CONGENITAL
Fiore et al. [14] NA NAb
Gonzalez et al. [15] NA 37.9b
Rao et al. [16] 6 VSD; 3 AS; 1 TGA; 2 double-inlet left ventricle; 1 AVSD NA
Cowley et al. [17] NA NA
a
Neither isthmus hypoplasia nor arch hypoplasia were included in this study.
b
No statistical difference was found regarding to hypoplastic aortic arch between surgery and BA groups.
c
Only the short-segment CoA patients with the transverse aortic arch-to-ascending aorta diameter ratio more than 0.3 were included.
AS: aortic stenosis; ASD: atrial septal defect; AVSD: atrioventricular septal defect; BA: balloon angioplasty; CoA: coarctation of the aorta; LVOTB: left ventricular out-
flow tract obstruction; MS: mitral stenosis; NA: not available; PDA: patent ductus arteriosus; TGA: transposition of great arteries; VSD: ventricular septal defect.

re-CoA and no significant publication bias was found (Begg’s test

P = 0.59, Egger’s test P = 0.91) (Fig. 2).

Repeat intervention due to recoarctation

A total of 8 NRCTs [8–14, 16] and 1 RCT [17] documented repeat
intervention due to re-CoA, and the strategy in the treatment of
re-CoA of both approaches is shown in detail in the
Supplementary Material, Table S1. A heterogeneity test revealed
that the heterogeneity across NRCTs was 48% (v2 = 13.34,
P = 0.06), and the overall heterogeneity was 40% (v2 = 13.40,
P = 0.10), therefore a fixed effect model was employed (Fig. 3).
The result of this meta-analysis indicated that compared with BA,
surgery was significantly associated with a lower incidence of re-
peat intervention due to re-CoA [OR 0.40, 95% confidence inter-
val (CI) 0.27–0.60; P < 0.001]. Figure 2: Publication bias analysis by funnel plot. OR: odds ratio.
296 Y. Wu et al. / Interactive CardioVascular and Thoracic Surgery
Figure 3: Forest plot of meta-analysis of repeat intervention due to recoarctation. BA: balloon angioplasty; CI: confidence interval; df: degree of freedom; NRCT: none-
randomized controlled trial; RCT: randomized controlled trial.
Recoarctation
A total of 10 studies [8–17] analysed the incidence of re-CoA of
both surgical approaches (Supplementary Material, Table S1) and
all of them were included in this meta-analysis. A heterogeneity
test revealed that the heterogeneity across NRCTs was 48% (v2 =
15.28, P = 0.05), and the overall heterogeneity was 42% (v2 =
15.39, P = 0.08), therefore a fixed effect model was employed
(Fig. 4). The result indicated that compared with BA, surgery was
significantly associated with a lower incidence of re-CoA (OR
0.43, 95% CI 0.30–0.61; P < 0.001).
Complications
A total of 7 NRCTs [9–15] analysed the incidence of complica-
tion of both surgical approaches and the complications of
both approaches were shown in detail in Supplementary
Material, Table S2. A heterogeneity test revealed that v2 =
9.76, P-value = 0.08, I2 = 49%, therefore a fixed effect model
was employed (Fig. 5). The result of this meta-analysis indi-
cated that no statistical difference was observed between sur-
gery and BA with regard to incidence of complications (OR
1.77, 95% CI 0.95–3.28; P = 0.07).
Postoperative aneurysm formation
A total of 8 NRCTs [9–16] and 1 RCT analysed the incidence of
postoperative aneurysm formation (Supplementary Material,
Downloaded from https://academic.oup.com/icvts/article-abstract/28/2/291/5061132 by guest on 21 November 2019
Table S2). A heterogeneity test revealed that heterogeneity across
NRCTs was 42% (v2 = 5.19, P = 0.16), and the overall heterogene-
ity was 50% (v2 = 8.02, P = 0.09), and as such, a randomized effect
model was employed (Supplementary Material S1). Meta-analysis
of postoperative aneurysm formation indicated that no statistical
difference was observed between surgery and BA (OR 0.32, 95%
CI 0.07–1.57; P = 0.09).
Perioperative mortality
A total of 7 NRCTs [8, 10, 11, 13–16] analysed the perioperative
mortality of both surgical approaches (Supplementary Material,
Table S2) and they were included in this meta-analysis. A hetero-
geneity test revealed that v2 = 0.61, P-value = 0.89, I2 <0.001,
hence a fixed effect model was employed (Supplementary
Material S2). Meta-analysis of perioperative mortality indicated
that no statistical difference was observed between surgery and
BA (OR 2.57, 95% CI 0.87–7.61, P = 0.09).
Hospitalization time
A total of 5 NRCTs [8, 11, 13, 15, 16] analysed the hospitalization
time of both surgical approaches, however, only 3 of them [8, 11,
16] were available for data pooling. Therefore, 3 NRCTs were in-
cluded in this meta-analysis. A heterogeneity test revealed that v2
= 0.8, P-value = 0.67, I2 <0.001, thus a fixed effect model was
employed (Supplementary Material S3). The result of this meta-
analysis indicated that compared with surgery, BA was
 Y. Wu et al. / Interactive CardioVascular and Thoracic Surgery 297

Downloaded from https://academic.oup.com/icvts/article-abstract/28/2/291/5061132 by guest on 21 November 2019

Figure 4: Forest plot of meta-analysis of incidence of recoarctation. BA: balloon angioplasty; CI: confidence interval; df: •••; NRCT: none-randomized controlled trial;
RCT: randomized controlled trial.

CONGENITAL
Figure 5: Forest plot of meta-analysis of incidence of complication. BA: balloon angioplasty; CI: confidence interval; df: •••.

significantly associated with a shorter hospitalization time (WMD
19.40, 95% CI 15.82–22.99; P < 0.001).
Residual transcoarctation gradient
In summary, a total of 6 NRCTs [8, 10, 11, 13–15] recorded the
preoperative transcoarctation gradient, while only 4 NRCTs [8,
11, 13, 15] and 2 NRCTs [8, 11] were available for data pooling
for the meta-analysis of immediate residual transcoarctation
gradient and the residual transcoarctation gradient in the mid- to
long-term follow-up, respectively (Supplementary Material, Table
S3). The immediate transcoarctation residual gradient of the sur-
gery group and residual transcoarctation gradient in mid- to
long-term follow-ups of both groups were evaluated with trans-
thoracic echocardiography. The immediate transcoarctation re-
sidual gradient of BA group was evaluated with catheterization.
Peak gradient was employed in 4 of our included studies [8, 10,
11, 14], while 2 studies employed the mean gradient [13, 15].
Catheterization was only employed for immediate
298 Y. Wu et al. / Interactive CardioVascular and Thoracic Surgery
transcoarctation residual gradient measurements in the BA
group. The meta-analysis of residual transcoarctation gradient in
the short-term follow-up was absent due to lack of relevant data.
In the meta-analysis of the immediate residual transcoarctation
gradient, 2 studies recorded the peak gradient, whereas, the
other 2 recorded the mean gradient. Therefore, we conducted a
subgroup analysis separately for the peak and mean gradient
studies. The heterogeneity test revealed that the overall v2 = 3.58,
P-value = 0.31, I2 = 16%, whereas for the subgroup of peak gradi-
ent, v2 = 0.05, P-value = 0.82, I2 <0.001, and for the subgroup of
mean gradient, v2 = 1.63, P-value = 0.2, I2 = 39% (Supplementary
Material S4). Additionally, no statistical difference was observed
between surgery and BA in the meta-analysis in either of the 2
the subgroup analyses (WMD -3.37, 95% CI -6.91 to 0.16;
P = 0.06; WMD 0.24, 95% CI -3.49 to 3.97; P = 0.9) nor in the over-
all group analysis (WMD -1.66, 95% CI -4.23 to 0.90; P = 0.2).
Only 2 NRCTs [8, 11] analysed the residual transcoarctation
gradient in mid- to long-term follow-up of both surgical
approaches and they were included in this meta-analysis. A het-
erogeneity test showed the overall v2 = 0.19, P-value = 0.67, I2
<0.001 (Supplementary Material S5). The meta-analysis result in-
dicated that compared with BA, surgery significantly resulted in a
lower gradient in the mid- to long-term follow-up (WMD -0.85,
95% CI -12.34 to -3.76; P < 0.001).
Sensitivity analyses
Due to significant heterogeneity, sensitivity analyses using the
leave-one-out method were performed in the meta-analysis of
repeat intervention due to re-CoA, incidence of re-CoA, aneu-
rysm formation, and complications. In the meta-analysis of re-
peat intervention due to re-CoA and incidence of re-CoA, the I2
was reduced to lower than 0.001 when we removed the Fiore
et al. [14] study, and the pooling estimate still stayed statistically
significant. In the meta-analysis of incidence of complications,
the I2 decreased to 29% when the Lin et al. [11] was removed and
the pooling estimate still stayed statistically insignificant. In the
meta-analysis of incidence of aneurysm formation, the I2 fell to
lower than 0.001 when we removed the Hernández-González
et al. [15] study, however, the pooling estimate showed a statisti-
cally significant change.
GRADE evaluation for the level of evidence
The included NRCTs and RCT had the same 3 outcomes: inci-
dence of re-CoA, aneurysm formation and repeat intervention
(Supplementary Material, Table S4). The recommended level of
this study was defined as low due to a low level of evidence in all
3 included outcomes.
DISCUSSION
BA is considered to be an acceptable option for NCA patients
with discrete and mild coarctation. However, the application of
BA on young infants and neonates is controversial because of
higher incidences of re-CoA and aneurysm formation reported
compared with surgery [18–20]. In contrast, a case series reported
by Patel et al. [21] demonstrated a safe outcome for young
infants and neonates who underwent BA with low mortality and
morbidity. In our study, consistent with previous published
studies, BA entailed a higher incidence of re-CoA and repeat in-
tervention compared with surgery in paediatric NCA. Young age
at CoA intervention, high gradient before dilation and hypoplasia
of transverse arch were associated with repeat intervention [10].
Furthermore, Alaei et al. [8] suggested BA could be a palliative
procedure for small infants to survive under critical situations
due to the immediate relief of gradient. Compared with native
CoA, BA is usually the preferred treatment for recurrent CoA [22].
Both surgery and BA are proven to be efficient and safe in the
treatment of CoA. However, there might be centre preferences
and era effects for each of the treatment. In our included studies,
only 2 reported the relevant data on centre preference and era
Downloaded from https://academic.oup.com/icvts/article-abstract/28/2/291/5061132 by guest on 21 November 2019
effects: (i) Alaei et al. [8] reported the treatment choice was made
according to the paediatric cardiology interventionist’s discretion;
(ii) Dijkema et al. [12] reported era effects on treatment choice.
Unfortunately, no conclusion could be drawn due to limited data
regarding preferred practice outcomes. In our study, no signifi-
cant difference was observed regarding complications, perioper-
ative mortality, and immediate residual transcoarctation gradient
between BA and surgery. However, literature reviews with regard
to postoperative complications of both surgical approaches sug-
gested that complications from surgery seemed to be more seri-
ous [11, 15, 16]. Longer hospitalization time and more serious
complications possibly led to a higher medical cost with respect
to surgery. Aneurysm formation was commonly detected in the
follow-up after BA. The meta-analysis of aneurysm formation
also indicated that no significant difference was observed be-
tween 2 approaches. The mechanism causing postoperative an-
eurysm formation possibly involved the disruption of aortic
intima and media followed by dilation [23].
BA prompted a smoother recovery with less pain and shorter
hospitalization times [8]. In our study, BA was associated with a
shorter hospitalization time compared with that of surgery.
However, other relevant data on postoperative recovery such as
pain control, intensive care unit stay and ventilation times were
not available for data pooling with our included studies. It should
be noted that femoral vessel injury was critical in the BA group.
Thrombosis and haematoma were commonly observed as a ma-
jor complication of BA. Moreover, long-term follow-up should
be conducted to explore whether injury of femoral vessels would
lead to retardation of lower extremities.
Surgery is the definitive therapy for native CoA, and it is pre-
ferred over BA in infants and young children as well as in non-
discrete multilevel and complicated coarctation [24].The surgical
repair of native CoA included end-to-end anastomosis, subcla-
vian flap angioplasty, patch aortoplasty, and interposition graft.
In our included studies, the resection and end-to-end anastomo-
sis was the most commonly used surgical approach in the surgery
group. The surgery for native CoA is conducted via lateral thora-
cotomy which is associated with chest wall deformity, scoliosis,
and mammary maldevelopment [25]. One major advantage of
surgical repair is 1-stage repair via sternotomy in CoA patients
with other concomitant cardiac defects and complex coarctation
anatomy. In our study, other advantages for surgery over BA
have also been noted. The meta-analysis of residual gradient in
the mid- to long-term follow-up revealed a lower residual gradi-
ent in the surgery group compared with BA. However, no signifi-
cant difference was found in the meta-analysis of immediate
residual gradient. Absence of radiation should also be empha-
sized as another major advantage of surgery in the treatment of
paediatric NCA.
 Y. Wu et al. / Interactive CardioVascular and Thoracic Surgery
We conducted a subgroup analysis in the meta-analysis of im-
mediate residual transcoarctation gradient. The heterogeneity
test revealed an overall I2 of 16%, however, we observed that I2
<0.001 and I2 = 39% in the subgroup analyses of peak gradient
and mean gradient, respectively. The overall heterogeneity might
be related to the different types of observation values with re-
spect to residual gradient. The sensitivity analysis of incidence of
re-CoA and repeat intervention suggested a significant reduction
in heterogeneity when we removed the Fiore et al. [14] study and
the pooling estimates were confirmed and robust. In the Fiore
et al. [14] study, significantly higher incidences of re-CoA and re-
peat intervention were observed in the BA group. This observa-
tion could possibly be correlated with the younger age, lower
weight and the different size of balloons used when compared
with the other studies. The skill, experience and learning curve of
interventional physicians could also contribute to the above ob-
servation. After excluding the Lin et al. [11] study, the heterogene-
ity dropped to 29% in the meta-analysis of incidence of
complications. We noticed that thoracic complications were criti-
cal in the surgery group. This was probably related to the young
age and narrow operative field leading to overexerted traction.
As a result, it was noted that the meta-analyses of incidence of
re-CoA, repeat intervention and complications were not overly
affected by a single study. After removing the Hernández-
González et al. [15] study, the heterogeneity dropped to lower
than 0.001 in the meta-analysis of aneurysm formation; however,
the pooling estimate showed a statistically significant change.
The Hernández-González et al. [15] study reported that the num-
ber of patients with aneurysm after surgical repair was more than
that from BA, and this was possibly related to the small sample
size of the Hernández-González study [15].
We noticed that there was a previous meta-analysis published
in 2014 which also compared BA versus surgery for NCA [26].
However, in that particular study, not only children, but adults
were also included. Therefore, we are the first study to date to
compare the clinical outcomes between BA and surgery in the
treatment of paediatric NCA. We observed that the previous
meta-analysis suffered from the following limitations: firstly, the
studies included in that meta-analyses enrolled patients with an
unacceptable variability in age. It must be pointed out that the
comparative effectiveness of BA versus surgery of NCA is very dif-
ferent between children and adults, therefore, pooling these data
together represented a major methodological flaw. Secondly, the
previous meta-analysis, which mistakenly regarded Cowley et al.
[17] study as a retrospective NRCT, pooled data with other stud-
ies directly, without subgroup analysis. To our knowledge, the
Cowley et al. [17] study was a RCT, though with a low level of evi-
dence. Thirdly, 2 studies included in that meta-analysis were
based on the same sample [17, 27], therefore, the results could
be significantly contaminated due to the data overlap. Fourthly,
that previous analysis failed to categorize the gradient into peak
or mean gradient, consequently, the pooling of these data could
possibly lead to significant heterogeneity. Fifthly, the number of
included studies in the funnel plots were less than 9 in general,
hence, the publication bias assessment could be inaccurate.
Limitations
Although we have attempted to overcome all the shortcomings
of previous meta-analyses and updated the latest relevant litera-
ture, there were still a few limitations in our study. Firstly, only 1
299
RCT was identified with our search criteria, hence, our meta-
analysis was based on 9 NRCTs and 1 RCT, even though the only
included RCT had a high risk of bias. Consequently, the level of
evidence of our study remained low. Similarly, based on the
GRADE system, the recommended level was also low. The RCT
[17] offered the data on aneurysm formation, re-CoA and repeat
intervention due to re-CoA, so we conducted meta-analysis of
these 3. The heterogeneity was moderate but at an acceptable
level. Therefore, we believe our results were not overly influenced
by the NRCTs and that the results are clinically applicable.
Secondly, BA is rarely used in patients with hypoplastic aortic
arch and these patients generally require surgery instead of BA,
therefore, BA groups do not have these high-risk patients. We
Downloaded from https://academic.oup.com/icvts/article-abstract/28/2/291/5061132 by guest on 21 November 2019
have summarized the presence of hypoplastic aortic arch for a
baseline comparison and no statistical difference was found be-
tween surgery and BA groups regarding hypoplastic aortic arch
in most of our included studies (8 out of 10) [8–15]. However, rel-
evant data regarding hypoplastic aortic arch was not available in
2 of our included studies [16, 17], therefore, potential selection
bias might still exist in our study. Thirdly, common baseline data
significantly affecting clinical outcomes such as age, weight, gen-
der, preoperative gradient and associated anomalies were not
statistically different across most of the NRCTs. The literature
search, data extraction and quality assessment were all evaluated
by 2 reviewers independently to reduce selection bias. We also
made strict inclusive and exclusive criteria. Therefore, we believe
the results of our meta-analysis are trustworthy, although the se-
lection bias of the included NRCTs did lack matched pairs and
propensity score matching. Finally, we initially planned to include
studies comparing surgery versus stent implantation which was
regarded as a more modern approach, however, only 1 study
[28] was identified, and they only compared the hospitalization
time and cost fee between stent implantation and surgery, there-
fore, the meta-analysis of stent versus surgery could not be con-
CONGENITAL
ducted. Stent implantation was reported to lower risk of
potential aneurysm formation and play a crucial role in the treat-
ment of aortic rupture or dissection due to damage of BA [29,
30]. Therefore, stent implantation could be a great complement
to the interventional treatment of NCA and more studies to
compare the effectiveness of stents versus surgery should be
designed.
CONCLUSIONS
In conclusion, compared with BA, surgery was significantly asso-
ciated with a lower incidence of re-CoA, fewer repeat interven-
tions due to re-CoA and lower residual transcoarctation gradient
in the mid- to long-term follow-up. Conversely, BA was signifi-
cantly associated with a shorter hospitalization time. Incidence of
aneurysm formation, perioperative mortality, complications and
immediate transcoarctation residual gradient were not statisti-
cally different between surgery and BA. However, the results of
the meta-analysis for incidence of aneurysm formation was in-
conclusive due to fluctuating results in the sensitivity analysis. In
terms of the incidence of re-CoA, repeat intervention due to re-
CoA and residual transcoarctation gradient in the mid- to long-
term follow-up, our study showed the superiority of surgery over
BA in the treatment of paediatric NCA. The level of evidence of
our study was low and RCTs should be designed to evaluate and
compare the safety and effectiveness of both approaches for
NCA.
300 Y. Wu et al. / Interactive CardioVascular and Thoracic Surgery
SUPPLEMENTARY MATERIAL
Supplementary material is available at ICVTS online.
Conflict of interest: none declared.
REFERENCES
[1] Yang L, Chua X, Rajgor DD, Tai BC, Quek SC. A systematic review and
meta-analysis of outcomes of transcatheter stent implantation for the
primary treatment of native coarctation. Int J Cardiol 2016;223:1025.
[2] Crafoord C, Nylin G. Congenital coarctation of the aorta and its surgical
treatment. J Thorac Surg 1946;14:347–61.
[3] Singer MI, Rowen M, Dorsey TJ. Transluminal aortic balloon angioplasty
for coarctation of the aorta in the newborn. Am Heart J 1982;103:131–2.
[4] Wells GA, Shea BJ, O’Connell D, Robertson J, Peterson J, Welch J et al.
The Newcastle–Ottawa Scale (NOS) for assessing the quality of non-
randomized studies in meta-analysis. Appl Eng Agric 2014;18:727–34.
[5] Higgins J, Green S. Cochrane handbook for systematic reviews of inter-
ventions version 5.1.0. Naunyn Schmiedebergs Arch Pharmacol
2011;5:S38.
[6] Guyatt GH, Oxman AD, Vist GE, Kunz R, Falck-Ytter Y, Alonso-Coello P
et al. GRADE: an emerging consensus on rating quality of evidence and
strength of recommendations. BMJ 2008;336:924–6.
[7] Hozo SP, Djulbegovic B, Hozo I. Estimating the mean and variance from
the median, range, and the size of a sample. BMC Med Res Methodol
2005;5:13.
[8] Alaei F, Moghadam MY, Mortaezaian H, Alaei M, Bakhshandeh H.
Balloon angioplasty versus surgical repair of coarctation of aorta in
infants. J Tehran Heart Cent 2011;6:134–7.
[9] Chiu HH, Chiu SN, Hu FC, Chen CA, Lin MT, Wang JK et al. Late cardio-
vascular complications after surgical or balloon angioplasty of coarcta-
tion of aorta in an Asian cohort. Am J Cardiol 2009;104:1139.
[10] Zhang H, Ye M, Chen G, Liu F, Wu L, Jia B. A comparison of balloon an-
gioplasty of native coarctation versus surgical repair for short segment
coarctation associated with ventricular septal defect—a single-center ret-
rospective review of 92 cases. J Thorac Dis 2016;8:2046–52.
[11] Lin SM, Hwang HK, Wu SJ, Chen MR. Comparison between balloon an-
gioplasty and surgery for native coarctation of the aorta in neonates and
young infants. Acta Cardiol Sin 2008;24:204–8.
[12] Dijkema EJ, Sieswerda GT, Takken T, Leiner T, Schoof PH, Haas F et al.
Long-term results of balloon angioplasty for native coarctation of the
aorta in childhood in comparison with surgery. Eur J Cardiothorac Surg
2018;53:262–8.
[13] Walhout RJ, Lekkerkerker JC, Oron GH, Bennink GB, Meijboom EJ.
Comparison of surgical repair with balloon angioplasty for native coarc-
tation in patients from 3 months to 16 years of age. Eur J Cardiothorac
Surg 2004;25:722–7.
[14] Fiore AC, Fischer LK, Schwartz T, Jureidini S, Balfour I, Carpenter D et al.
Comparison of angioplasty and surgery for neonatal aortic coarctation.
Ann Thorac Surg 2005;80:1659–65.
[15] Hernández-González M, Solorio M, Conde-Carmona I, Rangel-Abundis
A, Ledesma M, Munayer J et al. Intraluminal aortoplasty vs. surgical aor-
tic resection in congenital aortic coarctation. A clinical random study in
pediatric patients. Arch Med Res 2003;34:305–10.
[16] Rao PS, Chopra PS, Koscik R, Smith PA, Wilson AD. Surgical versus bal-
loon therapy for aortic coarctation in infants <_ 3 months old. J Am Coll
Cardiol 1994;23:1479–83.
[17] Cowley CG, Orsmond GS, Feola P, McQuillan L, Shaddy RE. Long-term,
randomized comparison of balloon angioplasty and surgery for native
coarctation of the aorta in childhood. Circulation 2005;111:3453–6.
[18] Suárez de Lezo J, Pan M, Romero M, Segura J, Pavlovic D, Ojeda S et al.
Percutaneous interventions on severe coarctation of the aorta: a 21-year
Downloaded from https://academic.oup.com/icvts/article-abstract/28/2/291/5061132 by guest on 21 November 2019
experience. Pediatr Cardiol 2005;26:176–89.
[19] Lee CL, Lin JF, Hsieh KS, Lin CC, Huang TC. Balloon angioplasty of native
coarctation and comparison of patients younger and older than 3
months. Circ J 2007;71:1781.
[20] Wong D, Benson LN, Van Arsdell GS, Karamlou T, McCrindle BW.
Balloon angioplasty is preferred to surgery for aortic coarctation. Cardiol
Young 2008;18:79.
[21] Patel HT, Madani A, Paris YM, Warner KG, Hijazi ZM. Balloon angio-
plasty of native coarctation of the aorta in infants and neonates: is it
worth the hassle? Pediatr Cardiol 2001;22:53–7.
[22] Feltes TF, Bacha E, Beekman RH, Cheatham JP, Feinstein JA, Gomes AS
et al. Indications for cardiac catheterization and intervention in pediatric
cardiac disease: a scientific statement from the American Heart
Association. Circulation 2011;123:2607–52.
[23] Karl TR. Surgery is the best treatment for primary coarctation in the ma-
jority of cases. J Cardiovasc Med 2007;8:50–6.
[24] Torok RD, Campbell MJ, Fleming GA, Hill KD. Coarctation of the aorta:
management from infancy to adulthood. World J Cardiol 2015;7:765–75.
[25] Jaureguizar E, Vazquez J, Murcia J, Diez Pardo JA. Morbid musculoskele-
tal sequelae of thoracotomy for tracheoesophageal fistula. J Pediatr Surg
1985;20:511–4.
[26] Hu ZP, Wang ZW, Dai XF, Zhan BT, Ren W, Li LC et al. Outcomes of sur-
gical versus balloon angioplasty treatment for native coarctation of the
aorta: a meta-analysis. Ann Vasc Surg 2014;28:394.
[27] Shaddy RE, Boucek MM, Sturtevant JE, Ruttenberg HD, Jaffe RB, Tani LY
et al. Comparison of angioplasty and surgery for unoperated coarctation
of the aorta. Circulation 1993;87:793–9.
[28] George JC, Shim D, Bucuvalas JC, Immerman E, Manning PB, Pearl JM
et al. Cost effectiveness of coarctation strategies: endovascular stent ver-
sus surgery. J Am Coll Cardiol 2002;39:410.
[29] Baykan A, Narin N, Ozyurt A, Argun M, Pamukcu O, Mavili E et al.
Cheatham platinum stent implantation in children with coarctation of
the aorta: single-centre short-term, intermediate-term, and long-term
results from Turkey. Cardiol Young 2014;24:675–84.
[30] Ang HL, Lim CW, Hia C, Yip J, Quek SC. Coarctation of the aorta: nonsur-
gical treatment using stent implantation. Singapore Med J 2014;55:
302–4.