Approaches to Pulmonary Atresia With Major

Aortopulmonary Collateral Arteries

David J. Barron and Phil Botha

Pulmonary atresia with major aortopulmonary collateral arteries (MAPCAs) is one of the most chal-
lenging surgical conditions to manage—not only because of the technical complexity of the surgery
but also in terms of defining the anatomy of the pulmonary vasculature, the timing of surgery,
and decision making on staged vs complete repair. The importance of early definition of pulmo-
nary blood supply is paramount, establishing which areas of the lung are supplied by MAPCAs
alone and which have dual supply with the native system (noting that 20% of patients have absent
intrapericardial native vessels). Early unifocalization (3-6 months) is ideal, with closure of the ven-
tricular septal defect (VSD) performed if 15 or more out of 20 lung segments can be recruited.
Leaving the ventricular septal defect open with a limiting right ventricle-pulmonary artery conduit
can be a useful interim or even definitive circulation in patients with borderline vasculature. Re-
habilitation of small native vessels with central shunts can be very effective, but best outcomes
are achieved by a combination of unifocalization of MAPCAs together with the native vessels (if
present). A variety of reconstructive techniques are necessary to be able to effect these complex
repairs with careful choice of materials. Ideally, surgery can be completed through sternotomy
alone, but separate thoracotomies may be necessary to control and access some MAPCAs.
Semin Thorac Cardiovasc Surg Pediatr Card Surg Ann 21:64–74 Crown Copyright © 2017 Pub-
lished by Elsevier Inc. All rights reserved.
Keywords: pulmonary atresia, aortopulmonary collateral arteries, MAPCAs, unifocalization

Introduction
Within the spectrum of tetralogy of Fallot with pulmonary atresia,
20%-40% of pulmonary atresia cases will have multiple sources
of pulmonary blood flow from the systemic circulation (major
aortopulmonary collateral arteries [MAPCAs]). These are among
the most challenging of all congenital heart lesions to manage1
because they require detailed and exhaustive definition of the
anatomy and a recognition of the great heterogeneity of MAPCAs
from 1 patient to the next. Consequently, surgery needs careful Unifocalization of MAPCAs incorporating areas of dual supply and of sole
preoperative planning and involves the need to operate in the MAPCA supply.
posterior mediastinum among structures that are unfamiliar to
most surgery in congenital heart disease. A staged approach may Central Message
be necessary in more complex anatomies, and the timing and Pulmonary atresia with MAPCA management requires early and detailed def-
strategy for surgery is key to delivering the best outcomes for inition of pulmonary blood supply, with clear understanding of which areas
this complex and varied set of patients, with a combined ap- are supplied by MAPCA alone and which have dual supply with the native
system. Early unifocalization (3-6 months) is ideal, with closure of the VSD
proach from surgeons and interventional cardiologists. performed if 15 or more out of 20 lung segments can be recruited; leaving
The essential principles of management are as follows: the VSD open with a limiting right ventricle-pulmonary artery conduit can
be a useful interim or even definitive circulation in patients with borderline
a. Early and thorough assessment of the anatomy of the vasculature. Rehabilitation of small native vessels with central shunts can be
pulmonary blood supply. very effective, but best outcomes are achieved by a combination of unifocalization
of MAPCAs together with the native vessels (if present).
b. Focus on achieving early unifocalization (within the first
6 months of life).

Department Cardiac Surgery, Birmingham Children’s Hospital, UK. Surgery, Birmingham Children’s Hospital, Steelhouse Lane, Birmingham B4
Address correspondence to: David J. Barron, FRCS(CT), Department Cardiac 6NH, UK. E-mail: david.barron1@nhs.net

64 PEDIATRIC CARDIAC SURGERY ANNUAL • 2018 https://doi.org/10.1053/j.pcsu.2017.11.001

1092-9126/Crown Copyright © 2017 Published by Elsevier Inc. All rights reserved.
Approaches to Pulmonary Atresia With Major Aortopulmonary Collateral Arteries
Figure 1 An example showing common relationships of MAPCAs to the trachea, main bronchi, and esophagus. (Color version
of figure is available online.)
c. Understanding the relationship between native pulmo-
nary arteries and MAPCAs to identify areas of dual supply
and sole supply.
d. Unifocalization is more important than complete repair,
and it can be valuable to leave the ventricular septal
defect (VSD) open initially in more difficult anatomy.
One of the characteristics of pulmonary atresia with ventricu-
lar septal defect and major aortopulmonary collateral arteries (PA/
VSD/MAPCAs) is the great heterogeneity of pulmonary blood
supply, which will dictate not only the clinical condition of the
patient but also the options for surgical management.2 The
MAPCAs themselves are variable in both size and number and,
furthermore, are variable in their course from site of origin to
their point of entry into the lung: MAPCAs most typically arise
from the descending thoracic aorta at about T4-T6 level, but the
next most common origin is from the underside of the aortic
arch. Other common sites of origin are from the subclavian or
brachiocephalic arteries and they can also arise from the distal
descending aorta and even from the abdominal aorta, rising up
behind the diaphragm. The extrapulmonary course of MAPCAs
can be direct entry into the lung, or may be a convoluted and
tortuous course, frequently with origin stenosis that can be present
from birth or may develop and progress with time (reflecting
the abnormal structure of the media of these vessels, which can
be thickened and muscular). Because MAPCAs usually arise very
posteriorly in the mediastinum, their relationship to the airways
and the esophagus is also highly variable. They can pass behind,
in front of, or even through the muscular wall of the esophagus
65
and tend to run along the main bronchi—but can be above,
below, anterior, or posterior to the bronchi before entering the
lung (Fig. 1). A common pattern for MAPCAs arising from the
mid-thoracic aorta is for them to run under the carina of the
trachea and follow the course under the main bronchi to enter
the lung.
The intrapulmonary course of the MAPCAs is equally vari-
able but we have noted that MAPCAs that come to enter the lung
at the hilum frequently adopt a branching pattern and appear-
ance much like a native pulmonary artery.3 Conversely, MAPCAs
running directly unto the lung, sometimes posterior to the bron-
chus, can also have abnormal intrapulmonary branching patterns
with stenotic areas within the lung that are not accessible to
surgery.
The final and most important variation in blood supply is the
relationship between native pulmonary arteries and MAPCAs.
Around 80%-85% of all the patients will have confluent
intrapericardial native pulmonary arteries, which are usually small
and underdeveloped to a variable extent. There is normally no
forward flow into these vessels (occasionally there can be a very
small communication providing a wisp of forward flow), but they
are filled by retrograde flow from one or more of the MAPCAs,
appearing on angiography to have the appearance of a seagull
in flight due to the midpoint of the vessels being connected to
the heart and so moving up and down with ventricular con-
traction (Fig. 2). Although the central (intrapericardial) component
of these native vessels is usually small, and even diminutive, the
branching pattern within the lungs can be much better but is,
again, variable. Regardless of the absolute size of these branches,
66
Figure 2 Angiogram showing a large MAPCA arising from the descend-
ing aorta, which is feeding the native pulmonary artery system. This system
fills the small intrapericardial native pulmonary arteries retrogradely—
which have the appearance of a seagull in flight. This is an example of
dual supply of the pulmonary vasculature by a MAPCA and the native
pulmonary artery system.
the key issue is whether each area of the lung is supplied by these
native vessels, or whether areas are supplied exclusively by
MAPCAs separate from the native system. Frequently, there is
a mixture of some areas of the lung supplied by native vessels
and some supplied by MAPCAs but it is also common to see
the MAPCA and native vessels clearly supplying the same distal
vasculature within the lung—so-called areas of dual supply.
The remaining 10%-15% of cases have complete absence of
central intrapericardial pulmonary arteries and are, by defini-
tion, dependent exclusively on MAPCA supply to the lungs.4,5
However, as described previously, frequently these MAPCAs may
enter the lung close to the hilum and have a distribution into
the lung similar to that of a native system (Fig. 3).
This whole picture of pulmonary blood supply is further com-
plicated by the natural history of the MAPCAs themselves, which
can develop progressive stenosis and even occlusion of their
origins or proximal course—which, in the setting of an area of
sole supply, can lead to failure of subsequent development of
the vasculature to that area of the lung. This is part of the reason
why early assessment of the pulmonary blood flow is so im-
portant, so as to avoid ‘losing’ areas of the pulmonary vasculature
that were supplied by stenosed MAPCAs.5-7
This variable nature of the pulmonary blood flow can mean
that some areas of the lung can be overperfused (with large, un-
obstructed MAPCAs, such as in Fig. 3) and so at risk of developing
pulmonary hypertensive changes, whereas other areas of the lung
may be at risk of losing their blood supply from stenosed
MAPCAs.
Patients and Methods
Patients can be classified according to their clinical presenta-
tion into:
D.J. Barron and P. Botha
Figure 3 Angiogram showing an example where the pulmonary blood flow
is derived entirely from MAPCAs with no intrapericardial native pulmo-
nary arteries. There are 4 large vessels arising from the descending aorta
and the brachiocephalic artery. Note that the aorta is right sided, running
down the right side of the vertebral column.
a. Cyanosis—moderate to severe cyanosis with no signs
of heart failure
b. Cardiac failure—usually not clinically cyanosed because
of high pulmonary blood flow
c. Balanced circulation—usually mildly cyanosed but not
in congestive failure
Clinical findings usually reflect the adequacy of pulmonary
blood flow and so range from congestive cardiac failure (large,
unobstructed MAPCAs) through to profound cyanosis (small
numbers of poorly developed MAPCAs or severely stenotic
MAPCAs). Cases that are ‘balanced’ are not in high output failure
and are typically moderately cyanosed—however, the clinical
picture of a balanced circulation could still mean that some areas
of the lung are overperfused and others are underperfused, and
so still need early investigation.
Investigation and Assessment
Early and thorough definition of pulmonary blood supply is es-
sential. Detailed imaging is paramount, but assessment must
include pressure and flow data as well as an appreciation of the
distribution and runoff of each vessel within the lung. Cardiac
catheterization is the cornerstone of this assessment and needs
a thorough and exhaustive delineation of pulmonary blood flow.
Individual angiograms of each MAPCA in 2 planes are essen-
tial, demonstrating the runoff and distribution of each vessel
within the lung. Injections in the brachiocephalic and subcla-
vian arteries should be done to identify any MAPCA origins from
these vessels, and use of balloon occlusion of the descending aorta
while dye is injected proximal to it can be a useful way of de-
fining vessels arising from the upper thoracic aorta. Great
efforts should be made to confirm or exclude central
intrapericardial native pulmonary arteries, which may only appear
Approaches to Pulmonary Atresia With Major Aortopulmonary Collateral Arteries
Figure 4 The ‘occult’ pulmonary artery: Angiogram showing a wedge in-
jection into the left lower pulmonary vein which reveals a previously hidden
left pulmonary artery filling retrogradely. This vessel would have been
originally supplied by a MAPCA which has become occluded.
on follow-through or on injection directly into an individual
MAPCA, looking for the characteristic ‘seagull’ sign. If there appear
to be areas of the lung with absent blood supply, then it is es-
sential to perform pulmonary vein wedge injections, which may
identify patent arterial vessels that have lost their antegrade flow
but may yet be recruitable—the so-called occult pulmonary artery
(Fig. 4). The surgeon needs to study the angiograms in great detail,
establishing the course of each MAPCA in relation to the airways
and to appreciate their AP relationship because many MAPCAs
are very posteriorly positioned within the mediastinum. Careful
review of each angiogram may show areas of contrast washout
as a tell-tale sign of areas with dual supply.
Approach to Surgery
The primary aim in management of this condition is to achieve
unifocalization of as much of the pulmonary vasculature as pos-
sible. Ideally, we aim for surgery between 3 and 9 months of
age to avoid the development of pulmonary vascular disease in
overperfused regions and to prevent stenosed vessels from being
lost. Furthermore, the vessels themselves remain relatively elastic
and are easier to manipulate at this age than in older children.
Vessels may need to be rotated and moved within the chest to
achieve unifocalization, and working within this age group also
reduces the distances across which vessels need to be manipulated.
Unifocalization
Careful preoperative planning with detailed review of all imaging
is essential. The aims can be summarized as follows:
a. Gain control of all MAPCAs before going onto bypass,
preferably before giving heparin. Running onto bypass
with large open MAPCAs can be extremely hazardous
as all flow is lost into the lungs and it is difficult to main-
tain adequate perfusion pressure. Trying to identify
MAPCAs when fully heparinized can be difficult and
67
cause bleeding. The exact approach has to be individu-
alized to the patient as it depends on the pattern of
MAPCAs and their points of origin.
i. An initial thoracotomy, usually on the side of the
descending aorta (25%-30% of cases have right-
sided aorta that descends within the right side of
the chest), may be necessary. The origins of the
MAPCAs can all be accessed and vessel loops are
placed around them. An opening is made into the
pericardium and the vessel loops fed loosely inside
for ready access at the next stage. The thora-
cotomy has the added advantage of being able to
mobilize the ipsilateral vessels and native pulmo-
nary artery (if present) to facilitate the subsequent
repair, and can even be used to perform individu-
al anastomoses within the chest if it is felt that this
will be easier access than from the midline. Finally,
the approach can also be used to work within the
fissures of the lung as, occasionally, vessels that might
otherwise be too posterior or too small external to
the lung develop into larger more accessible vessels
within the fissures. The thoracotomy is then closed,
the patient is turned supine, and sternotomy is
performed.
ii. Very occasionally, the anatomy is such that the
MAPCAs on the opposite side of the descending aorta
are those that need to be mobilized and devel-
oped. In this situation, the surgical strategy needs
to be confident that vessels supplying the contra-
lateral lung will be readily accessible from the
midline.
iii. MAPCAs from the underside of the arch or the sub-
clavian vessels can be readily accessed from the
midline.
iv. Midline sternotomy alone may be all that is re-
quired to access all the native vessels and
MAPCAs—an approach that has been very success-
fully used by the Stanford group.8,9 However, the
surgeon must be confident that all large vessels can
be safely and adequately controlled through midline
sternotomy if this is to be the preferred approach.
We utilize sternotomy alone if the anatomy of the
MAPCAs is suitable but, if any doubt, will first
perform a thoracotomy as described previously.
b. The aim is to create a platform of native tissue across
the midline that will form the basis of the reconstruc-
tion. This is most commonly achieved by utilizing the
native pulmonary artery system but may created from
MAPCAs that have been divided, mobilized, and brought
forward.
c. Always utilize the native pulmonary artery system if
present. Even if small, these can be laid open and create
a central platform onto which to build the unifocalization.
d. In areas of dual supply, the feeding MAPCA can usually
be ligated. If in doubt, it is best to recruit the MAPCA
into the central confluence and, in cases where the
feeding MAPCA is large and readily accessible, then we
would often incorporate the MAPCA regardless because
68
Figure 5 The ‘Melbourne shunt’: There is pulmonary atresia with diminutive native pulmonary arteries. The arteries are con-
nected directly into the facing side of the aorta to create a central shunt or aortopulmonary window. This drives antegrade
flow into the native system to stimulate growth. An alternative is to place a small interposition shunt between the aorta and
the pulmonary arteries. (Color version of figure is available online.)
the more native tissue than can be used in the conflu-
ence, the better.
e. It is not essential to close the VSD as part of the initial
procedure. If the overall pulmonary vasculature looks
poor or if insufficient vessels can be unifocalized (vessels
may be too posteriorly located or too distant from the
hilum to be recruited), it may be safer to leave the VSD
open and place a deliberately restrictive right ventricle-
pulmonary artery (RV-PA) conduit into the unifocalized
vessels. The restrictive conduit will prevent overcirculation
while still delivering pulsatile, antegrade flow into the
pulmonary vessels at a controlled pressure. As these pa-
tients have normal AV connections and well-developed
ventricles, the streaming effect is such that predomi-
nantly desaturated blood will be directed into the lungs,
gaining maximum oxygen extraction for the Qp:Qs. An
alternative would be to create a fenestrated VSD but we
prefer the restrictive conduit as this avoids having to
undertake any intracardiac procedure at this stage. A
further alternative would be to place a central shunt into
the unifocalized vessels, which avoids the need for
ventriculotomy,9 but does not have such advantageous
hemodynamics as the RV-PA conduit, which delivers better
Sata O2 for the same Qp:Qs and better diastolic pressure.
Shunt Procedures
Although unifocalization can be achieved in the majority of pa-
tients (80%-90% in larger series), it may not be achievable in
patients with poorly developed vasculature or small and scanty
MAPCAs. The most effective shunt is in the setting of small native
pulmonary arteries forming the characteristic ‘seagull’ shape on
angiography, which are of small caliber centrally but which have
relatively good distribution within the lungs themselves. In this
situation, a direct aortopulmonary connection can be made (the
so-called Melbourne Shunt),10 connecting the diminutive PAs di-
rectly into the back of the aorta with or without a small
interposition shunt (Fig. 5). The vessels are usually so small that
there is little risk of overcirculation.
D.J. Barron and P. Botha
More classical shunts may be necessary into isolated MAPCAs
or unilateral native PAs, ideally with origin as centrally as pos-
sible so that they are readily accessible at subsequent operations.
Choice of shunt and access (ie, thoracotomy or sternotomy) will
be dependent on the individual anatomy—in some cases, it may
be possible to anastomose 2 ipsilateral vessels and place a shunt
into the conjoined vessels (Fig. 6). Shunt procedures do not pre-
clude subsequent unifocalization, and the Melbourne shunt can
be particularly successful in promoting growth and develop-
ment of the native PA system such that full unifocalization and
even complete repair can be considered in the future. Occa-
sionally, sequential shunt procedures may be necessary to
rehabilitate different areas of the lung vasculature before
unifocalization can be considered.
Surgical Techniques and Materials
Thoracotomies are usually performed on the ipsilateral side to
the descending aorta in the fourth intercostal space. Note that
in PA/VAS/MAPCAs, 25%-30% of patients have a right-sided aortic
arch with right-sided descending aorta. This will provide access
to the MAPCA origins and the vessels can be mobilized and fol-
lowed out into the lung on the ipsilateral side. MAPCAs running
into the contralateral side can be mobilized a limited distance
beyond the midline but this can be very useful as it creates a
space around them that is readily found when approached
through subsequent midline sternotomy. It may be possible to
join neighboring MAPCAs (or MAPCA to native PA) together if
it felt that access is easier from thoracotomy—especially for pos-
teriorly placed MAPCAs running into the posterior aspect of the
lung. Test occlusion with side-biting clamps should be done before
proceeding to ensure the patient does not desaturate signifi-
cantly while these vessels are temporarily occluded.
At sternotomy, it is important to mobilize the aorta as much
as possible so that it is possible to retract it from side to side to
access the transverse sinus and work behind it. It may help to
avoid cannulating the SVC initially so that there is no cannula
in the way when working around and behind the SVC on the
right-sided vessels. Working in the space between the aorta and
the SVC, it is possible to dissect out through the posterior
Approaches to Pulmonary Atresia With Major Aortopulmonary Collateral Arteries 69

Figure 6 Example of a shunt procedure performed through a thoracotomy. Two interconnecting MAPCAs are shown arising
from the aorta with a stenosed segment between them. The vessels are disconnected from the aorta, enlarged with a patch of
pulmonary homograft, and then connected to the ascending aorta with an interposition shunt. This provides controlled flow
into the vessels and is readily accessible from midline sternotomy for future recruitment. (Color version of figure is available
online.)

pericardium, working over the top of the left atrium and into
the subcarinal space. There are vascular lymph nodes here that
can be carefully resected and this provides access to the poste-
rior mediastinum where many of the MAPCAs can be found.
Creating space around these vessels at preceding thoracotomy
can be very helpful in finding the vessels at this stage. Ideally,
the origins of all important MAPCAs should be identified and
controlled before running onto cardiopulmonary bypass, after
which each MAPCA is doubly ligated at their origin to stabilize
bypass.
Most of the reconstruction can usually be achieved on bypass
with the heart beating. Vessels are controlled as distally as pos-
sible using either fine spring-loaded vascular clips (such as Yasargil,
Aesculap Inc, PA) or with fine vascular clamps for larger vessels.
Laying open the native PAs from hilum to hilum creates a plat-
form onto which additional MAPCAs can be joined into (Fig. 7),
even when they are very small. If they are absent, then the aim
is to bring together the largest MAPCAs from either side and anas-
tomose their posterior walls to create a new platform (Fig. 8).
If no central platform can be created, then ipsilateral vessels can
be joined together, sometimes with a homograft patch augmen-
tation to create a ‘sump’ on each side. These 2 sumps can then
be connected together with a tube graft (usually Gore-Tex, WL
Gore, Newark, DE) and an RV-PA conduit is then connected to
this graft (Fig. 9).
Pulmonary homograft is our preferred material for patch re-
construction of the unifocalized vessels. Glutaraldehyde-
treated native pericardium can be used as an alternative. Xenograft
patches can be used but they tend to be stiffer and liable to
kinking or creasing rather than taking on the natural curvature
of the vessels, and are more prone to degeneration and calcifi-
cation in the long-term. Conduits need to be strong enough to
withstand high pressure, particularly if the VSD is to be left open.
Even in the setting of a full repair, RV pressures may be rela-
tively high postoperatively because of the nature of the vasculature,
so strong conduits may still be preferable. We prefer a Dacron
conduit such as Hancock (Medtronic Inc, Minneapolis, MN) or
an aortic homograft.

Figure 7 Unifocalization. Example 1: (A) Native pulmonary arteries are fed by a large MAPCA arising from the descending aorta
(dual supply). A second MAPCA supplies the left upper lobe (LUL) territory, which does not connect with the native system
(lone supply). (B) Both MAPCAs have been ligated at their origins. The native vessels have been laid open widely out into
their branches. The MAPCA to the LUL has been mobilized, laid open, and connected into the native system. (C) The unifocalized
vessels have been reconstructed with a patch of pulmonary homograft. A defect is cut into this patch to receive a valved RV-
PA conduit. (Color version of figure is available online.)
70 D.J. Barron and P. Botha

Figure 8 Unifocalization. Example 2: (A) Three large MAPCAs supply the pulmonary vasculature. The lower right MAPCA has
dual supply with the diminutive native pulmonary arteries. (B) The diminutive intrapericardial vessels are too small to be of
use. The MAPCAs on both sides are disconnected from their origins, mobilized, and laid open. (C) Unifocalization is per-
formed on either side, enlarging the vessels with patches of pulmonary homograft. A Gore-Tex tube graft is then placed from
hilum-to-hilum, connecting the reconstructed vessels. A valved RV-PA conduit is then placed into the Gore-Tex tube. (Color
version of figure is available online.)

Results awaiting further recruitment (n = 6) or are unlikely to ever achieve

A total of 275 consecutive cases of pulmonary atresia with
MAPCAs presenting to Birmingham Children’s Hospital between
1988 and 2016 have been reviewed. Unifocalization was achieved
in 249 of the 275 patients (88%). The remaining group of 26
patients received only shunt-type procedures and are either
unifocalization because of poor vasculature (n = 19).
Within the group of 249 patients undergoing unifocalization,
the total in-hospital early mortality was 2.8%. Ten-year actuar-
ial survival for this whole group was 93 ± 4% and 15-year actuarial
survival was 88 ± 9%. Combined thoracotomy and sternotomy
approach was used in 63% of cases (158 of 249) and

Figure 9 Unifocalization. Example 3: (A) Four large MAPCAs supply the entire pulmonary vasculature. There are absent native
vessels. (B) The 4 MAPCAs are ligated at their origins, mobilized, and laid open. (C) The mobilized MAPCAs are then brought
together to create a central confluence. (D) The confluence is patched over with a pulmonary homograft and a valved RV-PA
conduit is then placed into the patch to complete the repair. (Color version of figure is available online.)
Approaches to Pulmonary Atresia With Major Aortopulmonary Collateral Arteries 71

Figure 10 Kaplan-Meier plot showing actuarial survival after unifocalization procedure for PA/VSD/MAPCAs. Survival is ana-
lyzed according to whether unifocalization was based on native pulmonary arteries only (green curve), a combination of native
pulmonary arteries and MAPCAs (orange line), or on exclusively MAPCAs (blue line). (Color version of figure is available online.)

sternotomy alone in the remaining 37% cases. An initial shunt
procedure was performed in 129 cases (47%) before
unifocalization, of which 56 (22%) were a central AP-window
type of shunt to the native pulmonary arteries (Fig. 5).
Patients were categorized in 2 ways:
1. According to their pattern of pulmonary blood supply:
those in whom unifocalization was achieved using native
pulmonary artery system alone, those who had a mixture
of MAPCAs and native PA recruited to achieve
unifocalization, and those who had unifocalization based
exclusively on MAPCAs.
2. According to whether the VSD was closed at the initial
procedure, closed at a subsequent procedure, or never
closed.
The outcomes in relation to the pattern of pulmonary blood
supply is shown in Figure 10. The most common groups were
those with combined native PA and MAPCA unifocalization (119
of 249, 48%) followed by those whose repair was based entire-
ly on native PAs, ligating the feeding MAPCAs (72 of 249, 29%).
The least common groups were those with absent native PAs,
in whom unifocalization was based entirely on the MAPCAs (58
of 249, 24%). There was no statistical difference in early or late
survival between any of these groups. Patients with exclusively
native PA reconstruction and those with combination of native
PAs and MAPCAs had very similar survival curves. The actuar-
ial survival of those patients with absent native pulmonary arteries
was lower than that of the other groups (80 ± 8% at 10 years)
but this did not reach statistical significance.
The VSD was left open at the initial procedure in 97 of 249
patients (39%) and was subsequently closed in 46 of these pa-
tients, leaving a group of 51 patients who have never had the
VSD closed (6 of these patients are still under review and are
predicted to achieve VSD closure in the future). Survival ac-
cording to VSD status is shown in Figure 11. There was no
difference in early mortality in relation to whether the VSD was
closed at the time of unifocalization or not. Actuarial survival
was similar in those patients who had either complete initial repair
or subsequent VSD closure. Those patients who never achieved
VSD closure have a significantly worse survival, with 10-year
survival of 78% compared with 92% in those who have achieved
VSD closure (P = 0.003).
In terms of surgical techniques, a total of 474 MAPCAs were
recruited with a further 301 MAPCAs ligated (because of dual
supply) within the entire patient group. Pulmonary homograft
patching was used exclusively as patch material for reconstruc-
tion. A hilum-to-hilum Gore-Tex tube graft was used to achieve
unifocalization in 63 patients (as in Fig. 9), usually after patch
augmentation of the ipsilateral vessels as described previously.
72 D.J. Barron and P. Botha

Figure 11 Kaplan-Meier plot showing actuarial survival following unifocalization procedure for PA/VSD/MAPCAs. Survival is
analyzed according to whether the VSD was closed, either at the initial or subsequent procedures (green line)—or whether it
has been left open with a limiting RV-PA conduit (orange line). (Color version of figure is available online.)

The Hancock valved conduit was used in 90% of cases, includ-
ing all those in which the VSD was left open.
Discussion
Surgical management of pulmonary atresia with MAPCAs has
transformed the prognosis for this complex condition, but the
timing and approach to surgery are still in evolution. There is
undoubtedly a move toward earlier surgery and the impor-
tance on aiming to achieve unifocalization early, so as to secure
as much pulmonary vasculature as possible. The relationship of
the MAPCAs to the native pulmonary artery system is funda-
mental to the success of the surgical approach and has generated
much debate, with some authors suggesting that there is no
benefit in recruiting MAPCAs.11 There is no doubt that the native
pulmonary artery system provides the best substrate for
unifocalization and reconstruction and should always be used
if possible. However, our results demonstrate that it is essential
to recruit MAPCAs that do not collateralize with the native system
to achieve best results. Furthermore, up to a fifth of these pa-
tients have complete absence of intrapericardial native pulmonary
arteries, and unifocalization is dependent on MAPCA recruit-
ment in these cases.4,5 Using these strategies has achieved complete
repair in 80%-90% of cases both in these results and in those
of the Stanford group.3,5,9
The Melbourne group has proposed a fundamentally differ-
ent approach to management described as a “rehabilitation”
strategy based entirely on the native pulmonary artery system.
This was based on their findings that, in a series of 60 unifocalized
MAPCAS,12 26 vessels had thrombosed and 12 had stenoses of
>50% at a mean of 3.4 years. Only in patients presenting heart
failure (approximately 11% of their cohort) is unifocalization and
single stage complete repair are undertaken. This strategy has
resulted in complete repair in 73% of patients, with median PA
pressure of 0.52 (2). Mortality was 10% with no cardiac-
related deaths. Other centers have reported achieving complete
repair in 60% of patients with this strategy, although RV/LV pres-
sure ratio was greater than 0.5 in more than half the patients,
which may result in long-term detriment to right ventricular
function.13
The nature of the unifocalization procedure has varied between
centers, but in our view should be taken to mean simultane-
ous amalgamation of all MAPCAs (to areas without dual supply)
together with the native pulmonary arteries plus the relief of ste-
noses in all surgically accessible vessels with patch enlargement
using pulmonary artery homograft. This is very much in line
with the management strategies promoted by the Stanford group,
Approaches to Pulmonary Atresia With Major Aortopulmonary Collateral Arteries
who have the largest experience in the world (over 450 cases)
and outstanding outcomes, with 57% of patients undergoing
unifocalization and complete repair in a single procedure, and
90% of patients ultimately achieving VSD closure.5,9 This has been
achieved with mean RV/LV pressure ratio of 0.41 ± 0.12 through
meticulous incorporation of all useful native tissue to the pul-
monary vascular bed through unifocalization, typically undertaken
between 3 and 6 months of age. With either the rehabilitation
or unifocalization strategies, it would therefore seem that early
intervention is key, in the former, by early central shunt to maxi-
mize native PA growth, or in the latter, to remove MAPCAs from
the systemic circulation before the onset of pulmonary vascu-
lar disease in these vessels.14
We have found the concept of leaving the VSD open to be
very helpful in managing patients with uncertain pulmonary vas-
culature. Use of a limiting conduit should ensure that
overcirculation should not occur and we usually choose a conduit
around half of the predicted full size for the body surface area
of the patient. Using 1 or 2 sizes larger than this and then clip-
ping down the conduit externally is another option as these clips
can be ballooned off in the catheter laboratory in the future to
allow for growth. It is important that a stiff conduit is used that
will maintain a reliable cross-sectional diameter under pres-
sure, and not dilate. This policy does not preclude from closure
of the VSD at a later date, which was achieved in almost half of
this group of patients. It was encouraging to note that early and
late survival were not influenced by staged closure of the VSD
as the late survival was similar in both groups, so long as the
VSD was ultimately closed. This inevitably leaves a group in whom
it is never possible to close the VSD (19% of all unifocalized pa-
tients in this series) whose late survival is not as good, but this
is the group of patients at the end of the spectrum with poor
pulmonary vasculature, and this approach has still provided this
difficult group with good quality of life,3 which is substantially
better than the natural history.1,6
We have retained a degree of flexibility in terms of decision
making for VSD closure but use the principle that at least 15 of
20 lung segments should be recruited to safely close the VSD. In
borderline cases with concerns over the quality of the vessels within
the lung or areas of the lung with hypertensive segments, we would
still err on the side of caution and leave the VSD open. Alterna-
tive strategies in these situations have been to perform a fenestrated
VSD closure or to use a central shunt into the unifocalized
confluence.9,15 We feel that the limiting RV-PA conduit provides
optimal hemodynamics while avoiding having to operate on the
VSD at all, simplifying the surgery. Other methods for assessing
whether or not the VSD can be closed include more detailed mea-
surement total pulmonary artery size, such as Nakata index,16 or
performing flow studies intraoperatively. There is evidence that
flow studies may be a better predictor of VSD closure than pul-
monary artery measurements and provide a more objective
measurement—the Stanford group have popularized this method
using flow rates of 3 l·min−1·m−2 and accepting pressures of
≤25 mm Hg as acceptable for VSD closure.9,17 The Toronto group
report using a protocol of 2.5 l·min−1·m−2 and accept pressures of
≤30 mm Hg18 and both groups have reported excellent results. Nev-
ertheless, other authors have found flow studies less predictive of
midterm survival in patients after VSD closure.19,20
73
One major aspect of this condition that is not covered in this
study is the burden of reintervention. Even accounting for the
inevitable conduit revisions in growing children, these vessels
frequently have abnormal branching patterns within the lung
that may need serial dilatations or even stenting. The artificial
materials used and the multiple suture lines are prone to fibro-
sis and distortion that may also require interventional cardiology
support and further reoperations.21 Future management needs
to examine whether these reinterventions can be minimized and
it would be hoped that, as surgical strategies evolve to earlier
and more complete unifocalization, the pulmonary vasculature
can be better preserved and protected at a younger age.
References
1. Bull K, Somerville J, Ty E, et al: Presentation and attrition in complex
pulmonary atresia. J Am Coll Cardiol 1995;25:491-499
2. Liao PK, Edwards WD, Julsrud PR, et al: Pulmonary blood supply
in patients with pulmonary atresia and ventricular septal defect.
J Am Coll Cardiol 1985;6:1343-1350
3. Davies B, Mussa S, Davies P, et al: Unifocalization of major
aortopulmonary collateral arteries in pulmonary atresia with ven-
tricular septal defect is essential to achieve excellent outcomes
irrespective of native pulmonary artery morphology. J Thorac
Cardiovasc Surg 2009;138:1269-1275
4. Griselli M, McGuirk SP, Winlaw DS, et al: The influence
of pulmonary artery morphology on the results of operations for
major aortopulmonary collateral arteries and complex congeni-
tal heart defects. J Thorac Cardiovasc Surg 2004;127:251-
258
5. Carrillo SA, Mainwaring RD, Patrick WL, et al: Surgical repair of
pulmonary atresia with ventricular septal defect and major
aortopulmonary collaterals with absent intrapericardial pulmo-
nary arteries. Ann Thorac Surg 2015;100:606-614
6. Leonard H, Derrick G, O’Sullivan J, et al: Natural and unnatural
history of pulmonary atresia. Heart 2000;84:499-503
7. Puga FJ, Leoni FE, Julsrud PR, et al: Complete repair of pulmo-
nary atresia, ventricular septal defect, and severe peripheral
arborization abnormalities of the central pulmonary arteries. Ex-
perience with preliminary unifocalization procedures in 38 patients.
J Thorac Cardiovasc Surg 1989;98:1018-1029
8. Reddy VM, McElhinney DB, Amin Z, et al: Early and intermedi-
ate outcomes after repair of pulmonary atresia with ventricular septal
defect and major aortopulmonary collateral arteries: experience
with 85 patients. Circulation 2000;101:1826-1832
9. Malhotra SP, Hanley FL. Surgical management of pulmonary atresia
with ventricular septal defect and major aortopulmonary collaterals:
a protocol-based approach. Semin Thorac Cardiovasc Surg Pediatr
Card Surg Annu 2009;145-151
10. Watterson KG, Wilkinson JL, Karl TR, et al: Very small pulmo-
nary arteries: central end-to-side shunt. Ann Thorac Surg 1991;52:
1132-1137
11. Nørgaard MA, Alphonso N, Cochrane AD, et al: Major aorto-
pulmonary collateral arteries of patients with pulmonary atresia
and ventricular septal defect are dilated bronchial arteries. Eur J
Cardiothorac Surg 2006;29:653-658
12. d’Udekem Y, Alphonso N, Nørgaard MA, et al: Pulmonary atresia
with ventricular septal defects and major aortopulmonary collat-
eral arteries: unifocalization brings no long-term benefits. J Thorac
Cardiovasc Surg 2005;130:1496-1502
13. Chen Q, Ma K, Hua Z, et al: Multistage pulmonary artery reha-
bilitation in patients with pulmonary atresia, ventricular septal defect
and hypoplastic pulmonary artery. Eur J Cardiothorac Surg 2016;50:
160-166
14. Watanabe N, Mainwaring RD, Reddy VM, et al: Early complete
repair of pulmonary atresia with ventricular septal defect and major
74
aortopulmonary collaterals. Ann Thorac Surg 2014;97:909-
915
15. Marshall AC, Love BA, Lang P, et al: Staged repair of tetralogy of
Fallot and diminutive pulmonary arteries with a fenestrated ven-
tricular septal defect patch. J Thorac Cardiovasc Surg 2003;126:
1427-1433
16. Amark KM, Karamlou T, O’Carroll A, et al: Independent factors
associated with mortality, reintervention, and achievement of com-
plete repair in children with pulmonary atresia with ventricular
septal defect. J Am Coll Cardiol 2006;47:1448-1456
17. Mainwaring RD, Sheikh AY, Vadiyala RP, et al: Surgical out-
comes for patients with pulmonary atresia/major aortopulmonary
collaterals and Alagille syndrome. Eur J Cardiothorac Surg 2012;42:
235-240
18. Honjo O, Al-Radi OO, MacDonald C, et al: The functional intra-
operative pulmonary blood flow study is a more sensitive predictor
D.J. Barron and P. Botha
than preoperative anatomy for right ventricular pressure and phys-
iologic tolerance of ventricular septal defect closure after complete
unifocalization in patients with pulmonary atresia, ventricular septal
defect and major aortopulmonary collaterals. Circulation 2009;120:
S46-S52
19. Trezzi M, Albanese SB, Albano A, et al: Impact of pulmonary flow
study pressure on outcomes after one-stage unifocalization. Ann
Thorac Surg 2017;doi:10.1016/j.athoracsur.2017.05.020.
20. Zhu J, Meza J, Kato A, et al: Pulmonary flow study predicts sur-
vival in pulmonary atresia with ventricular septal defect and major
aortopulmonary collateral arteries. J Thorac Cardiovasc Surg
2016;152:1494-1503, e1
21. Carotti A, Albanese SB, Fillipelli S, et al: Determinants of outcome
after surgical treatment of pulmonary atresia with ventricular septal
defect and major aortopulmonary collaterals. J Thorac Cardiovasc
Surg 2010;140:1092-1103