THE ANATOMICAL RECORD 302:19–31 (2019)

Remodeling of the Embryonic

Interventricular Communication
in Regard to the Description
and Classification of Ventricular
Septal Defects
ROBERT H. ANDERSON ,1* DIANE E. SPICER,2 TIMOTHY J. MOHUN,3
JILL P.J.M. HIKSPOORS,4 AND WOUTER H. LAMERS4
1
Institute of Genetic Medicine, Newcastle University, Newcastle upon Tyne, UK
2
Department of Pediatric Cardiology, University of Florida, Gainesville, Florida
3
Francis Crick Institute, London, UK
4
Department of Anatomy, Maastricht University, Maastricht, The Netherlands

ABSTRACT
Ventricular septal defects are the commonest congenital cardiac mal-
formations. Appropriate knowledge of the steps involved in completion of
ventricular septation should provide clues as to the morphology of the dif-
ferent phenotypes. Currently, however, consensus is lacking regarding the
components of the developing ventricular septum, and how best to describe
the different phenotypes seen in postnatal life. We have reassessed the pre-
vious investigations devoted to closure of the embryonic interventricular
communication. On this basis, we discuss how studies in the early part of
the 20th century correctly identified the steps involved in the remodeling
of the embryonic interventricular foramen subsequent to the stage at which
the outflow tract arises entirely above the cavity of the developing right
ventricle. There has, however, already been remodeling of the foramen
from the stage at which the atrioventricular canal is supported exclusively
by the developing left ventricle. We show how these temporal changes in
morphology can provide explanations for the different ventricular septal
defects seen in the clinical setting. Thus, muscular defects represent inap-
propriate coalescence of muscular ventricular septum. The channels that
are perimembranous are due to failure of closure of the persisting embry-
onic interventricular foramen. Those that are doubly committed and juxta-
arterial reflect failure of formation of the free-standing subpulmonary
muscular infundibular sleeve. The findings also point to the importance of
appropriate alignment, during development, between the developing atrial
and ventricular septums, and between the apical component of the ventricular
septum and the ventricular outlet components. Anat Rec, 302:19–31, 2019.
© 2018 Wiley Periodicals, Inc.

*Correspondence to: Professor Robert H. Anderson, 60 Earlsfield DOI: 10.1002/ar.24020

Road, London SW18 3DN, UK. Telephone: 00-44-20-8870-4368. Published online 8 November 2018 in Wiley Online Library
E-mail: sejjran@ucl.ac.uk (wileyonlinelibrary.com).
Received 1 December 2017; Revised 7 January 2018; Accepted
19 January 2018.

© 2018 WILEY PERIODICALS, INC.

20 ANDERSON ET AL.
Key words: development; cardiac embryology; outflow tract;
atrioventricular cushions; outflow cushions
Ventricular septal defects are the commonest congeni-
tal cardiac malformations. Not only do they exist in isola-
tion, in which setting they show various phenotypic
variations, but they are an integral part of other lesions,
such as tetralogy of Fallot, common arterial trunk, and
double outlet right ventricle. As yet, however, there is no
agreed system for the description and categorization of
these defects, neither in their isolated form, nor when
they are part of more complex malformations. Despite the
comment made almost a century ago, namely that knowl-
edge of cardiac development could provide the basis for
understanding congenital cardiac malformations (Abbott,
1936), it is surprising that the necessary information
regarding the development of the normal ventricular sep-
tum remains contentious. Some investigators, for exam-
ple, suggest that, during its formation, the septum
possesses up to four developmental components (Van
Praagh, Geva, and Kreutzer, 1989; Geva, 2016). We
recently summarized our own interpretation of the steps
involved in the processes of normal development
(Anderson et al., 2014). We revisit that information here,
again taking advantage of the knowledge that has
Fig. 1. The image, taken from an episcopic data set prepared from a
mouse embryo sacrificed at embryonic day 10.5, shows how the atrial
appendages are ballooning in parallel (horizontal white arrows with red
borders) from the atrial component of the heart tube, which retains its
connection to the pharyngeal mesenchyme through the dorsal
mesocardium. The trabeculated components of the ventricles are
ballooning in series (oblique downward white arrows with red borders)
from the inlet and outlet parts of the ventricular loop. The primordium of
the muscular ventricular septum (white star with red borders) is
developing between the ballooning trabecular components.
accrued over the past decade using the technique of epi-
scopic microscopy (Mohun and Weninger, 2011). We place
these findings into the setting of previous descriptions.
We show how knowledge of the temporal changes in for-
mation of the ventricular septal components can provide
a platform for appreciating the phenotypic differences
between the channels, which provide shunting between
the ventricles in the postnatal heart.
FORMATION OF THE VENTRICULAR LOOP
When initially formed, the heart is a linear entity
encased within a developing pericardial cavity. Evidence
that has emerged over recent decades has shown that this
linear component of the tube, when first seen, subse-
quently forms little more than the definitive left ventricle.
Part of the work showing the importance of the so-called
second heart field in adding material to the developing
tube originated from the laboratory of Roger Markwald,
in whose honor this special issue is being collated
(Mjaadvedt et al., 2001). Along with others (Kelly et al.,
2001, Waldo et al., 2001), these investigators showed that
the new material entering the developing heart from the
heart-forming areas within the body of the embryo
formed the components of the atrial chambers, along with
much of the right ventricle and the outflow tract. It is
subsequent to the addition of this new material that it
becomes possible to recognize the ventricular loop. When
first formed, the walls of the loop, along with those of the
atrial component and the outflow tract, are made up of
myocardium that does not stain positively for either atrial
natriuretic factor or connexin 40. These walls were said
to be made up of primary myocardium (Moorman and
Christoffels, 2003). With subsequent development, the
atrial appendages expand from the atrial component of
the tube, while the trabecular ventricular components
grow from the inlet and outlet parts of the ventricular
loop (Fig. 1). The walls of these components, subsequent
to the process described as “ballooning,” were then noted
to react positively to both atrial natriuretic factor and
connexin 40. The walls were then considered to represent
chamber myocardium (Moorman and Christoffels, 2003).
This process of ballooning is also important for under-
standing development of the ventricular septum. This is
because the primordium of the muscular ventricular sep-
tum first becomes apparent concomitant with the expan-
sion of the ventricular trabecular components (Fig. 1). The
trabecular components balloon in series from the loop,
with the left ventricular component expanding from its
inlet, and the right ventricular part from the outlet of the
loop (Fig. 1). It is subsequent to the ballooning that it
becomes possible to recognize the embryonic interventricu-
lar communication. When first seen, it has exclusively
muscular borders, as at this stage the atrioventricular
canal opens exclusively into the cavity of the developing
left ventricle, while the cavity of the developing right ven-
tricle supports the entirety of the outflow tract. By virtue
of these arrangements, the entirety of the blood entering
 REMODELING OF EMBRYONIC INTERVENTRICULAR COMMUNICATION
the ventricular loop through the atrioventricular canal
must perforce pass through the embryonic interventricular
communication so as to reach the outflow tract (Fig. 2).
The morphology of the outlet component of the ventric-
ular loop at this early stage of development is itself perti-
nent to controversies regarding the structure of the
second chamber observed in the congenital cardiac lesions
known as double inlet left ventricle and tricuspid atresia.
For many years, hearts with double inlet left ventricle
were considered the exemplars of “single ventricle” (Van
Praagh, Ongley, and Swan, 1964; Edwards, 1977). This
was despite the fact that such lesions have two obvious
chambers within their ventricular mass. The logic
required to support their description as being “univentri-
cular hearts” was provided by suggesting that the second
chamber was no more than an infundibulum, thus dis-
qualifying it from ventricular status. Van Praagh himself
continues to espouse this notion (Van Praagh, 2015). In
this context, Geva also suggests that the “infundibulum”
is a discrete component of the developing right ventricle,
possessing a trabeculated component, which extends to
the ventricular apex (Geva, 2016). As can be seen from
Figure 2, however, the default option for the early embryo
is double inlet to the developing left ventricle. At this
early stage, there is no direct connection between the
right atrium and the developing right ventricle, with the
communication between the developing ventricles having
exclusively muscular borders (Fig. 3). It is this absence of
the right atrioventricular connection, as seen in the devel-
oping heart that is now the acknowledged phenotypic fea-
ture of the commonest examples of tricuspid atresia
(Anderson and Cook, 2004). Examination of the develop-
ing right ventricle at this early stage of development, fur-
thermore, shows that it is incomplete, lacking its inlet
Fig. 2. The image is from another dataset prepared from a developing
mouse embryo sacrificed at embryonic day 10.5. It is a frontal
section across the ventricular loop, showing how the entirety of the
atrioventricular canal is supported by the developing left ventricle, while
the outflow tract arises exclusively from the developing right ventricle.
All the blood entering the left ventricle through the atrioventricular canal
must pass through the initial, or primary, embryonic interventricular
communication, which is bounded caudally by the developing muscular
septum (white star with red borders), and cranially by the inner heart
curvature, so as to reach the outflow tract.
21
Fig. 3. The image comes from an episcopic data set prepared from a
human embryo at Carnegie stage 13. The data set is cut in a plane
replicating the subcostal oblique echocardiographic sections. The
image shows that, at this stage of development, there is no direct
communication between the atrial chambers and the developing right
ventricle (dashed white line). The developing right ventricle, although
lacking any inlet, and hence being incomplete, already possesses a well
formed apical trabecular component. It also supports the entirety of the
outflow tract, with the outflow cushions unfused at this early stage. As
yet, there has been no formation of the infundibulum. The similarity
between the incomplete right ventricle and the second chamber seen in
tricuspid atresia and double inlet left ventricle with concordant
ventriculo-arterial connections is striking.
component. It is, nonetheless, much more than a mere
infundibulum, as it already possesses a well-formed api-
cal trabecular component (Fig. 3). It is this trabeculated
component that forms the basis of the developing right
ventricle. Its resemblance to the second chamber in the
functionally univentricular heart with a dominant left
ventricle justifies the description of the smaller chamber
as an incomplete right ventricle (Jacobs and Anderson,
2006). It also follows that, as the entirety of the circum-
ference of the atrioventricular canal is initially supported
by the developing left ventricle, it is not possible at this
early stage for the developing embryo to close the inter-
ventricular communication. To achieve complete separa-
tion and subsequent septation of the ventricles, it is
necessary for the boundaries of the communication to be
remodeled such that the right ventricle achieves its direct
inlet component. It is then necessary to transform the ini-
tially solitary outflow tract such that discrete outlets are
provided for each of the developing ventricular chambers.
It is during this process of remodeling that the bound-
aries of the initial foramen, which is well-described as the
primary interventricular foramen, become converted into
secondary, and then tertiary, foramens (Fig. 4). It is the
remodeling of the primary to the secondary foramen that
provides the direct communication between the right
atrium and the right ventricle (Fig. 5, left-hand plot). The
ongoing changes from the secondary to the tertiary fora-
men then produce the outlet for the left ventricle (Fig. 5,
middle plot; Odgers, 1938; Frazer, 1916). Subsequent to
this remodeling, the developing embryo is then able to
22 ANDERSON ET AL.

Fig. 4. The drawings show, in diagrammatic fashion, the steps involved in remodeling of the embryonic interventricular communication (double
headed white arrow in Figs. 2). The left-hand plot shows the foramen as the double headed green arrow, bounded cranially by the inner heart
curvature, and caudally by the crest of the muscular ventricular septum. Its dorsal boundary is marked by the blue line and the ventral boundary by
the red line (see Fig. 5, left-hand plot). The middle plot illustrates how remodeling of the dorsal boundary by expansion of the atrioventricular canal
produces the right ventricular inlet (see Fig. 5, middle plot). The remaining channel between the ventricles is now the secondary foramen.
Remodeling of its ventral border then produces the outlet for the left ventricle (right-hand plot, see also Fig. 5, right-hand plot). The embryo is then
able to close the persisting channel, which is between the aortic root and the right ventricle, and which can be described as the tertiary
interventricular foramen. The drawings do not, however, represent the three-dimensional changes that take place during the remodeling of the
boundaries of the primary foramen. These can better be appreciated when superimposed on reconstructed hearts, as shown in Figure 5.

close the tertiary foramen by formation of the so-called foramen are not shown to size in Figure 4. It is apprecia-
membranous septum (Fig. 5, right-hand plot). The dimen- tion of the processes involved in this remodeling, how-
sions of the various components of the remodeling ever, which we will describe in greater detail below, that

Fig. 5. The steps of remodeling shown diagrammatically in Figure 4 have been superimposed on hearts from developing mice in this figure. The
left-hand plot shows the starting point for remodeling, as already illustrated in Figure 2. The middle plot shows how, during E11.5, the dorsal part of
the primary foramen (blue line) becomes remodeled so as to form the boundaries of the right ventricular inlet. This produces the secondary
foramen, which remains roofed by the inner heart curvature. Further remodeling during E12.5 transforms the ventral part of the secondary foramen
into the boundaries of the left ventricular outflow tract (red line in right-hand plot). The developing embryo then closes the tertiary foramen, which is
the channel between the aortic root and the right ventricle. It is represented by the double headed green arrow in the right-hand plot, and is closed
by formation of the membranous septum. The evidence underscoring the remodeling is shown in Figure 6.
 REMODELING OF EMBRYONIC INTERVENTRICULAR COMMUNICATION
provides the key to understanding the phenotypic vari-
ability in the channels between the ventricles seen in the
setting of congenital cardiac disease.
EXPANSION OF THE ATRIOVENTRICULAR
CANAL
It is expansion of the atrioventricular canal that is the
first step in remodeling of the interventricular communi-
cation. As indicated above, this provides the right ventri-
cle with its inlet component. This step did not attract
attention in the otherwise authoritative account of remo-
deling of the interventricular communication provided by
Kramer (Kramer, 1942), nor did Odgers discuss these ini-
tial changes (Odgers, 1938). The mechanism of transfer of
the right ventricular inlet had remained controversial
prior to the investigation made using an antibody to the
no-dose ganglion of the chick, known as GlN2, which ser-
endipitously proved to mark the boundaries of the inter-
ventricular communication (Fig. 6; Lamers et al., 1992).
The antibody recognizes the same epitope, an N-linked
carbohydrate present in several adhesion molecules
(Mitsumoto et al., 2000) as the HNK-1, Leu7, and CD57
antisera. The patterns of GlN2-staining showed that,
although the atrioventricular canal was supported
23
exclusively by the developing left ventricle at the initial
stages of development (Fig. 2), the muscular floor of the
right atrium was already in continuity with the roof of
the developing right ventricle at the rightward margin of
the canal (Fig. 7, left-hand plot). All that was required to
bring the cavities of the developing right atrium and right
ventricle into direct communication was expansion of the
right margin of the atrioventricular canal. It is this
expansion that underscores the initial remodeling of the
dorsal part of the initial interventricular communication
(Figs. 4–7). The investigations performed using the GlN2
antiserum also showed that, subsequent to this expan-
sion, the initial atrioventricular canal myocardium
became incorporated into the right atrium as its vestibu-
lar component (Wessels et al., 1992; Lamers et al., 1992).
This permitted the inference to be made that the entirety
of the muscular walls of the developing right ventricle
were derived from the outlet component of the initial ven-
tricular loop. The findings also showed that it is inappro-
priate to argue that the so-called “infundibular”
component of the right ventricle has a different develop-
mental origin from its “sinus” (Geva, 2016).
Appreciation of the mechanism of expansion of this
rightward component of the embryonic atrioventricular
canal is pertinent to the understanding of one of the most

Fig. 6. Serial transverse sections through the heart of a CS14 (top row) and a CS23 (bottom row) human embryo. The sections are stained with
the GlN2 antiserum, which identifies the junction between the embryonic left and right ventricles and the future ventricular conduction system. Plots
A–C show 4-chamber views through a CS14 heart, with plot A as most superior and plot C as most inferior section. Note that the GlN2 “ring”
marks the top of the ventricular septum (B) and the roots of the bundle branches that straddle the interventricular septum (C). The stain further
marks the ventricular pole of the AV node (A) and the junction between the right atrium and the right ventricle (B, C). Plots D–G show near
transverse serial sections of the heart axis of a CS23 embryo, with interventricular septum, aorta, and right AV junction. Plot D shows the most
superior and G the most inferior section. GlN2-positive myocytes mark the ventricular pole of the AV node (E, F), the left bundle branch (D), and the
right AV junction (G). Plots F and G show that the GlN2-positive ring marks the ventricular border of the AV canal. Note that the GlN2 antiserum,
which recognizes a possibly sulfated carbohydrate epitope on proteins, also marks cells in the arterial walls and nerves.
24 ANDERSON ET AL.

contentious types of ventricular septal defect, namely, the
defect of so-called “atrioventricular canal type”. It is now
well-recognized that some patients with atrioventricular
septal defect and common atrioventricular junction can
have shunting between the chambers confined at ventric-
ular level. This is because the bridging leaflets of the com-
mon atrioventricular valve are attached to the leading
edge of the atrial septum. It is these lesions that are the
true ventricular defect of atrioventricular canal type. The
hearts exhibit commonality of the atrioventricular canal,
and have a trifoliate left atrioventricular valve. The
lesion initially described as being of “atrioventricular
canal type” was designated as such because the muscular
ventricular septum extended across the full width of the
overriding orifice of the tricuspid valve (LaCorte, Fellows
and Williams, 1976). Others continue to designate these
lesions as being of “atrioventricular canal type,” arguing
that they exist because of failure of formation of an
alleged septum of the atrioventricular canal (Van Praagh,
Geva, and Kreutzer, 1989). The investigations using the
GlN2 antibody, however, had shown that the developing
atrioventricular bundle was positioned on the crest of the
muscular ventricular septum (Lamers et al., 1992; Fig. 6).
And, subsequent to the completion of ventricular septa-
tion, the atrioventricular bundle is sandwiched between
the crest of the muscular ventricular septum and the
insulating tissues of the atrioventricular junctions. This
means that there cannot have been formation of a new
muscular “septum of the atrioventricular canal,” as had
been suggested by Van Praagh and his colleagues (Van
Praagh, Geva, and Kreutzer, 1989), and as Geva con-
tinues to claim (Geva, 2016). It has now been shown that
a structure deserving of the title “septum of the atrioven-
tricular canal” is formed during this stage of develop-
ment. It is derived by muscularization of the vestibular
spine, also known as the “dorsal mesenchymal protru-
sion” (Briggs, Karkala, & Wessels, 2012). Muscularization
of this component, which occurs only after the rightward
expansion of the right atrioventricular junction is com-
plete, produces the antero-inferior buttress of the atrial
septum. This structure is formed on the atrial side of the
insulating tissues of the atrioventricular junctions
(Anderson et al., 2014). The phenotypic feature of the so-
called ventricular septal defect of “atrioventricular canal
type” as described by Lacorte and colleagues is malalign-
ment between the atrial septum and the muscular ven-
tricular septum. The hearts with these features do not
have a common atrioventricular junction, but rather pos-
sess a tricuspid valve overriding the right atrioventricu-
lar junction. They are perimembranous defects with
atrioventricular septal malalignment (Fig. 8). As we will
discuss below, others subsequently identified perimem-
branous defects with inlet extension as being of “atrioven-
tricular canal type” (Wells and Lindesmith, 1985). These
defects also have separate atrioventricular junctions for
the mitral and tricuspid valves. They, too, are not appro-
priately described as possessing “atrioventricular canals.”
FURTHER REMODELING OF THE EMBRYONIC
INTERVENTRICULAR COMMUNICATION
Expansion of the right margin of the atrioventricular
canal is sufficient in itself to provide the right ventricle
with its inlet component. In the developing mouse, this
process is completed during embryonic day 11.5. At embry-
onic day 12.5, the entirety of the developing outflow tract
remains supported by the right ventricle. This is compara-
ble to the situation seen in the developing human heart at
Carnegie stage 16. This arrangement was well-illustrated
by Kramer in embryos of 12 to 14 mm crown-rump length

Fig. 7. The images are taken from episcopic data sets from mouse embryos sacrificed early (top plot) and later (bottom plot) during embryonic
(E) day 10.5. The left-hand plot shows that, although at the early stage the developing right atrioventricular orifice opens to the cavity of the
developing left ventricle (white arrow), the floor of the right atrium is in continuity with the roof of the developing right ventricle (red dotted line). All
that is required to produce direct continuity between the cavities of the right atrium and right ventricle is to expand the atrioventricular canal
rightward (white arrow with red borders). As is shown in the right-hand plot, this has been achieved by the end of E10.5, with the right
atrioventricular orifice (white arrow) now opening to the right of the developing muscular ventricular septum (white star with red borders in both
plots). The initial embryonic interventricular communication has been remodeled (see Figs. 4 and 5) so as to surround the newly developed right
ventricular inlet (blue dotted line).
 REMODELING OF EMBRYONIC INTERVENTRICULAR COMMUNICATION
Fig. 8. The image shows a “four chamber” section from a human
heart with a ventricular septal defect produced by malalignment
between the atrial septum and the muscular ventricular septum. Some
authorities suggested that the lesion is an “atrioventricular canal
defect”. As is shown, nonetheless, there are separate atrioventricular
junctions, with the left atrioventricular junction guarded by a
morphologically mitral valve. The lesion is a perimembranous inlet
ventricular septal defect with atrioventricular septal malalignment. This
type of defect is best understood as the consequence of incomplete
rightward expansion of the embryonic atrioventricular canal.
(Kramer, 1942; Fig. 9). So as to complete ventricular septa-
tion, therefore, it is necessary to continue the remodeling
of the interventricular communication. This is achieved by
transferring the aortic root such that it arises above the
cavity of the left, rather than the right, ventricle (Figs. 4
and 5). This process occurs in the developing mouse during
embryonic days 12.5 and 13.0, equivalent to human devel-
opment through Carnegie stages 16 through 18. The proxi-
mal outflow cushions complete their fusion during this
period, creating a shelf in the roof the right ventricle. Con-
comitant with this process, the margins of the embryonic
interventricular communication provide the outlet for the
blood entering the left ventricle from the left atrium to
reach the aortic root (Fig. 5, central plot; Fig. 10, left-hand
plot). A new communication between the ventricles, how-
ever, is then created between the rightward margin of the
aortic root and the cavity of the right ventricle. This area,
which can be considered to represent the tertiary foramen,
is then readily identified as being discrete from the second-
ary interventricular communication (Fig. 10, central plot).
Then, by the end of embryonic day 14.5, the aortic root has
been realigned to achieve its definitive position above the
cavity of the left ventricle. This means that the boundaries
of the secondary interventricular communication (Fig. 4)
have been remodeled to form the left ventricular outflow
tract (Fig. 10, right-hand plot, Fig. 5, right-hand plot).
The changes occurring during remodeling of the second-
ary interventricular communication (Figs. 4–5, and 10) are
recapitulated in the features seen in the morphology of the
25
channels between the ventricles found in the settings of
double outlet right ventricle with subaortic defect, tetral-
ogy of Fallot, and the commonest type of ventricular septal
defect coming to surgical correction. In the setting of dou-
ble outlet right ventricle, the channel between the ventri-
cles is equivalent to the secondary interventricular
communication (Fig. 11, left-hand plot). In the setting of
tetralogy of Fallot, where there is overriding of the aortic
root, the part of the channel between the ventricles now
representing the secondary foramen has already been
remodeled to become the outflow tract of the left ventricle.
It is the persisting communication between the overriding
aortic root and the right ventricle that is equivalent to the
tertiary foramen. This is the area identified by most inves-
tigators as the “ventricular septal defect” (Fig. 11, middle
plot). It is this channel that also represents the commonest
type of ventricular septal defect encountered in patients
referred for surgical correction. In these latter patients,
the secondary embryonic interventricular communication
has again remodeled to become the outflow tract for the
left ventricle (Fig. 11, right-hand plot). In the setting of
double outlet right ventricle, paradoxically, most pediatric
cardiologists still describe the channel equivalent to the
secondary interventricular communication as the “ventric-
ular septal defect.” This channel between the ventricles,
however, which represented the outflow tract for the left
ventricle, cannot be closed during surgical correction.
Instead, during most surgical corrections, an area within
the right ventricle, roofed by the outlet septum, is patched
by the surgeon so as to redirect the flow from the left ven-
tricle into the aortic or pulmonary root. If the area repre-
senting the secondary foramen is directed to the
pulmonary root, the operative procedure also includes an
arterial switch. It is the area patched by the surgeon,
therefore, that logically should be described as the ventric-
ular septal defect. The channel between the ventricles is
more accurately described simply as the interventricular
communication (Ebadi et al., 2017; Fig. 11).
CLOSURE OF THE PERSISTING
INTERVENTRICULAR COMMUNICATION
Transfer of the aortic root to the left ventricle, there-
fore, serves to reorient the secondary interventricular
communication to become the left ventricular outflow
tract. The changes involved in remodeling the interventri-
cular foramen were well described by Frazer and Odgers
in seminal investigations published in the first part of the
20th century (Frazer, 1916; Odgers, 1938). It was they
who pointed out how, after transfer of the aortic root to
the left ventricle, the persisting interventricular foramen
was a new entity (Figs. 4 and 5). It is this tertiary fora-
men that is closed by formation of the membranous part
of the ventricular septum. Odgers showed that the new
component of the ventricular septum was formed by the
rightward margins of the atrioventricular cushions, which
he called the tubercles (Odgers, 1938). Kramer, while
endorsing the larger part of the account provided by
Odgers, argued that his material suggested that the prox-
imal outflow cushions, or ridges as they were termed by
him, also contributed to the membranous septum,
describing the formation of a “mesenchymal mass.” Our
episcopic images suggest that both Odgers and Kramer
could be correct in their descriptions. The proximal out-
flow cushions, subsequent to their fusion, create a shelf in
26 ANDERSON ET AL.

Fig. 9. The left-hand plot is modified from Figure 8 in the publication of Kramer from 1942. It depicts reconstructions made from human embryos
of 12 to 14 mm crown-rump length. The arrangement is directly comparable to the episcopic image shown in the right-hand plot, which is taken
from a dataset prepared from a mouse embryo sacrificed at embryonic day 12.5. The interventricular foramen has already been remodeled to
produce the inlet of the right ventricle (See Fig. 5, left-hand plot). At this stage, the developing right ventricle continues to support the entirety of the
outflow tract, which is being divided into the aortic and pulmonary outflow tracts by fusion of the proximal outflow cushions. AV – atrioventricular.

the roof of the developing right ventricle. This process in
itself reduces the dimensions of the persisting communi-
cation between the aortic root and the right ventricle
(Fig. 10, right-hand plot). By the time the cushions have
created the shelf in the right ventricle, however, their
surface has muscularized (van den Hoff et al., 1999), pro-
ducing initially an extensive wedge-shaped structure. The
initial work regarding the myocardialization of the out-
flow tract was also performed, in part, in the laboratory of
Roger Markwald. As the cushion mass muscularizes, its
central part shows evidence of attenuation. It is the right-
ward surface of the muscularized component, subsequent
to extensive remodeling, which becomes the free-standing
subpulmonary infundibular sleeve. This supports the leaf-
lets of the pulmonary valve in the base of the ventricular
mass (Fig. 12, left-hand plot). Kramer did not take
account of the attenuation of the core of cushion mass,
arguing instead that the supraventricular crest formed by

Fig. 10. The episcopic images are taken from data sets prepared from mouse embryos sacrificed during embryonic day 12.5 and early 13.5. They
show how the secondary interventricular communication (red double headed arrow, see Figs. 4 and 5) becomes remodeled to produce the outflow
tract of the left ventricle. It is the persisting interventricular communication between the aortic root and the right ventricle, or the tertiary
interventricular foramen (green double headed arrow), as seen in the right-hand plot that is closed by formation of the membranous septum (see
below).
 REMODELING OF EMBRYONIC INTERVENTRICULAR COMMUNICATION 27

Fig. 11. The images are from congenitally malformed human hearts with, to the left hand, double outlet right ventricle with subaortic defect, in the
middle tetralogy of Fallot, and to the right hand, the commonest type of ventricular septal defect coming to surgical correction. Each heart has
been sectioned to replicate the echocardiographic four-chamber cut incorporating the aortic root. The images show the channel representing the
secondary interventricular foramen as the red double headed arrow (see also Figs. 4 and 5), and the area for eventual closure of the ventricular
septum as the green double headed arrow. They illustrate a paradoxical situation. It is the boundaries of the locus providing the outflow tract for the
left ventricle as seen in the setting of tetralogy of Fallot (middle plot) that is currently named by most as the “ventricular septal defect” when both
arterial trunks arise from the right ventricle (left-hand plot).

the process of muscularization was a “conal septum”
(Kramer, 1942). Van Praagh and his colleagues continue
to hold this opinion (Van Praagh, Geva and Kreutzer,
1989). Geva, as we have discussed, suggests that an even
greater part of the septum belongs to the “conus” (Geva,
2016). We now know that, subsequent to eventual closure
of the persisting communication between the aortic root
and the right ventricle, and formation of the membranous
septum from the tubercles of the atrioventricular cush-
ions as described by Odgers (Odgers, 1938), the central
part of the proximal cushion mass becomes transformed
into extracavitary fibroadipose tissue. There is no rem-
nant of a septum in the postnatal heart between the free-
standing subpulmonary infundibular sleeve and the aor-
tic root (Anderson et al., 2014).
FAILURE TO CLOSE THE TERTIARY
INTERVENTRICULAR COMMUNICATION
Prior to its closure, the persisting channel between the
ventricles is positioned directly beneath the aortic root. It
can now be considered to represent the tertiary interven-
tricular communication. It opens to the right ventricle
centrally within its base. This is the precise location of
the commonest type of ventricular septal defect found in
patients coming forward for surgical correction of their
defect. For many years, such defects were described as
being “membranous.” Becu and his colleagues, however,
had already noted that the dimensions of such defects
were appreciably larger than the size of the membranous
septum itself (Becu et al., 1956). They commented that
the defects likely represented deficiencies of the muscular
septum in the environs of the interventricular communi-
cation. This concept was subsequently endorsed by Sher-
man (Sherman, 1963). It was this concept that was
embraced by ourselves when, in collaboration with
European colleagues, we proposed that the defects were
better considered as being perimembranous (Soto et al.,
1980). We pointed out that, depending on the extent of
the deficiency of the muscular septum, the perimembra-
nous defects could open centrally (Fig. 13, central plot), or
could extend so as to open primarily to the inlet or outlet
of the right ventricle. The defects opening to the inlet of
the right ventricle would be shielded by the septal leaflet
of the tricuspid valve (Fig. 13, left-hand plot). Wells and
Lindesmith had interpreted such perimembranous
defects opening to the right ventricular inlet as represent-
ing the “atrioventricular canal type” of defect (Wells and
Lindesmith, 1985). These defects are different from
the malalignment defects accorded this description by
LaCorte et al. (1976). Sherman had already pointed out
(Sherman, 1963) that the defects lacked a common atrio-
ventricular junction, with the left-sided junction guarded
by a morphologically mitral valve. They are no more than
perimembranous defects opening predominantly to the
right ventricular inlet (Fig. 13, left-hand plot). They differ
markedly from the perimembranous inlet defects with
atrioventricular septal malalignment (Fig. 7). The defects
opening to the ventricular outlet, in contrast, could do so
either because the supraventricular crest, derived from
the proximal outlet cushions, was itself hypoplastic
(Fig. 13, right-hand plot), or because the crest was devi-
ated in antero-cephalad or postero-caudal fashion relative
to the muscular ventricular septum itself. Perimembra-
nous defects, nonetheless, can also be confluent, opening
to both the right ventricular inlet and outlet, while
retaining their central location at the ventricular base
(Fig. 14, left plot).
It also follows that, should the proximal outflow cush-
ions themselves have failed to muscularize, then the defi-
ciency of the supraventricular crest can be sufficiently
extreme that the resulting defect is roofed by fibrous con-
tinuity between the leaflets of the aortic and pulmonary
valves. It is for this reason that, on occasion, it is possible
to identify a fibrous rather than a muscular outlet sep-
tum. It is the presence of such fibrous continuity between
28 ANDERSON ET AL.

Fig. 12. The episcopic images are from data sets prepared from mouse embryos sacrificed on embryonic day 14.5, at the time of closure of the
tertiary interventricular (IV) communication. The left-hand plot, which is cut to reveal the septal surface of the right ventricle, shows how
the proximal outflow cushions have formed the supraventricular crest, with part of the trabecular layer of the right ventricle compacting to form the
septomarginal trabeculation and its septoparietal extensions, labeled as “compacting trabeculations” in the image. The tertiary interventricular
communication, or the channel between the aortic root and the right ventricle, itself is being closed by the tubercles of the atrioventricular
(AV) cushions. As is shown in the right-hand plot, a four-chamber section, these form the nonmyocardial part of the ventricular septum. A remnant
of the septal cushion can still be seen within the developing supraventricular crest, but this does not contribute to the definitive membranous
septum. Note also that, as shown in the right-hand plot, there remains a substantial subaortic infundibulum subsequent to closure of the persisting
interventricular communication. This will not become fibrous in the mouse until embryonic day 16.5.

the leaflets of the arterial valves that is the phenotypic Such defects often have a muscular postero-inferior rim,
feature of the type of ventricular septal defect that is now produced by fusion of the trabecular layer of the right
described as being doubly committed and juxta-arterial. ventricular wall with the inner heart curvature. Being

Fig. 13. The images show how defects that share the phenotypic feature of fibrous continuity in their roof between the leaflets of the aortic and
tricuspid valves, making them perimembranous, can open to the inlet of the right ventricle (left-hand plot), centrally at the ventricular base (central
plot), or to the outlet of the right ventricle (right-hand plot) depending on the extent of deficiency of the muscular ventricular septum. In the case of
the outlet defect, the deficiency is due to hypoplasia of the supraventricular crest, such that the defect opens between the limbs of the
septomarginal trabeculation.
 REMODELING OF EMBRYONIC INTERVENTRICULAR COMMUNICATION 29

Fig. 14. The images show defects that open to the right ventricle because of hypoplasia of the supraventricular crest, which is formed by
muscularization of the proximal outflow cushions. The left-hand plot shows a defect that is perimembranous, as revealed by the fibrous continuity
between the aortic and tricuspid valvar leaflets. Clearly opening to the right ventricular outlet, it also extends to open to the right ventricular inlet.
This makes it a confluent defect. In the right plot, the heart shows no muscularization of the proximal outflow cushions at all, so that the defect is
roofed by fibrous continuity between the leaflets of the aortic and pulmonary valves. This makes the defect doubly committed and juxta-arterial.
Both defects open to the right ventricle between the limbs of the septomarginal trabeculation, but the doubly committed defect has a muscular
postero-inferior rim, which presumably is produced by compaction of the right ventricular trabeculations fusing with the inner heart curvature.

outlet defects, such channels continue to open to the right
ventricle between the limbs of the septomarginal trabecu-
lation (Fig. 14, right-hand plot). Our inferences made
regarding the morphogenesis of the different defects
depending on failure to close the tertiary interventricular
foramen, or hypoplasia or malalignment of the supraven-
tricular crest, have now been confirmed by observations
made in a colony of mice in which the Furin enzyme was
perturbed. Several of these mice showed persistence of
the tertiary interventricular foramen at embryonic day
15.5, producing a ventricular septal defect bordered by
continuity between the developing leaflets of the mitral
and tricuspid valves, in other words a perimembranous
defect. The defects opened centrally into the base of the
right ventricle (Fig. 15, left-hand plot). Other mice
showed hypoplasia and failure of muscularization of the
proximal outflow cushions, leaving ventricular septal
defects that were doubly committed and juxta-arterial
(Fig. 15, right-hand plot).
COMPACTION OF THE MUSCULAR
VENTRICULAR SEPTUM
It is often presumed that the individual trabeculations
making up the trabecular layer of the developing ventric-
ular walls, which form the greater part of the mural
thickness until embryonic day 14.5, they coalesce to pro-
duce the compact component of the walls. This is not true,
as the compact part of the ventricular walls is formed, in
its larger part, by hyperplasia of the initially thin com-
pact layer. This thin compact layer is present from the
outset of formation of the chamber myocardium
(Anderson et al., 2017). It is the initial trabeculations,
nonetheless, which do coalesce to form the muscular ven-
tricular septum. The initial trabeculations also coalesce
to form the papillary muscles of the atrioventricular
valves, along with the septoparietal and septomarginal
trabeculations of the right ventricle. Failure of complete
coalescence of the septal trabeculations, therefore, can
account for the presence of muscular defects within any
part of the muscular septum. It is widespread failure of
such coalescence that produces the so-called “Swiss-
cheese septum.” Such defects can open to the inlet of the
right ventricle, but more frequently they are found within
the apical part of the septum, either in mid-septal posi-
tion, or at the ventricular apex. They can also be found
anterior to the body of the septomarginal trabeculation.
The muscular defects can also open to the outlet of the
right ventricle when there is hypoplasia of the supraven-
tricular crest. Muscular defects, however, cannot open
centrally within the ventricular base. This is the area
occupied by membranous septum. An intact membranous
30 ANDERSON ET AL.

Fig. 15. The images are prepared from episcopic data sets obtained from mouse embryos sacrificed at embryonic day 15.5 subsequent to
perturbation of the Furin enzyme. By this stage of development, the channel between the aortic root, and the right ventricle would normally have
closed. The images show a central perimembranous defect in the left-hand plot, and a doubly committed and juxta-arterial defect in the right-hand
plot, that latter being found in the presence of double outlet from the morphologically right ventricle.

septum is always to be found in central position when
defects have exclusively muscular borders. The only
exception to this rule is when muscular defects coexists
with a second defect that is perimembranous.
CONCLUSIONS
According to Wells and Lindesmith, it had been “tradi-
tional” to categorize ventricular septal defects into four
categories according to their developmental heritage
(Wells and Lindesmith, 1985). These four types of defects,
labeled in numerical fashion, were said to have been out-
let, membranous, of “atrioventricular canal type”, or mus-
cular. We have been unable to trace the “tradition” on
which this categorization was allegedly based. It does
now prove to be the case, nonetheless, that according to
the deficiencies in formation of the normal ventricular
septum, it is possible to recognize the entities as
described. In our opinion, they can better be grouped into
three categories. This is because the so-called “atrioven-
tricular canal defect,” described by Wells and Lindesmith
them as the “Type 3” defect, is no more than a perimem-
branous defect with malalignment between the atrial and
muscular ventricular septal components. The so-called
“Type 1” defects in the categorization popularized by
Wells and Lindesmith are those that are doubly commit-
ted and juxta-arterial. These defects are well explained
on the basis of hypoplasia or malalignment of the supra-
ventricular crest, with this structure being formed by
muscularization of the proximal outflow cushions. The so-
called “Type 2” defect, that is, the consequence of failure
of closure of the tertiary interventricular communication,
this being the space remaining between the right ventri-
cle and the aortic root subsequent to fusion of the proxi-
mal outflow cushions. These defects are now usually
described as being perimembranous (Soto et al., 1980).
Depending on the deficiency of the muscular components
surrounding the interventricular communication, the
defects can open centrally, can extend toward the inlet or
outlet of the right ventricle, or can be confluent, opening
to both inlet and outlet. As emphasized, this perimembra-
nous category includes the so-called “Type 3” defect. The
“Type 4” defects, as defined by Wells and Lindesmith,
make up the muscular group. These channels can be pro-
duced either by lack of coalescence of the apical muscular
septum, or failure of fusion of the supraventricular crest
with the apical muscular septum. All defects, therefore,
can be categorized into the doubly committed and juxta-
arterial, perimembranous, and muscular groups. To
provide complete descriptions, nonetheless, it is also nec-
essary to describe their geography relative to the land-
marks of the right ventricle, and to account for the
presence or absence of septal malalignment. The mala-
lignment most frequently involves the outlet septum.
Atrioventricular septal malalignment as associated with
the perimembranous inlet defect is also of particular
importance, as this variant is associated with a grossly
abnormal disposition of the atrioventricular conduction
 REMODELING OF EMBRYONIC INTERVENTRICULAR COMMUNICATION
axis (Milo et al., 1979). It is knowledge of the phenotypic
variability based on borders that permits the location of
the conduction axis to be predicted with accuracy in all
the remaining defects. These details are all included in
the categorization that will shortly be used in the 11th
iteration of the International Classification of Disease,
albeit that the proposed system uses geography as the
primary criterion for categorization. Our experience,
nonetheless, in which we have compared the evidence
now available with regard to development with the find-
ings from examination of specimens, indicates that it is
emphasis on borders, rather than geography, which will
provide the most robust system of categorization.
LITERATURE CITED
Abbott ME. 1936. Atlas of congenital cardiac disease. New York:
American Heart Association. p 36–37.
Anderson RH, Cook AC. 2004. Morphology of the functionally univen-
tricular heart. Cardiol Young 14(Suppl. 1):3–12.
Anderson RH, Spicer DE, Brown NA, Mohun TJ. 2014. The develop-
ment of septation in the four-chambered heart. Anat Rec
(Hoboken) 297:1414–1429.
Anderson RH, Jensen B, Mohun TJ, Petersen SE, Aung N,
Zemrak F, Planken RN, MacIver DH. 2017. Key questions relating
to left ventricular non-compaction cardiomyopathy: is the Emperor
still wearing any clothes. Can J Cardiol 33:747–757.
Becu LM, Burchell HB, Dushane JW, Edwards JE, Fontana RS,
Kirklin JW. 1956. Anatomic and pathologic studies in ventricular
septal defect. Circulation 14:349–364.
Briggs LE, Kakarla J, Wessels A. 2012. The pathogenesis of atrial
and atrioventricular septal defects with special emphasis on the
role of the dorsal mesenchymal protrusion. Differentiation 84:
117–130.
Ebadi A, Spicer DE, Backer CL, Fricker FJ, Anderson RH. 2017.
Double outlet right ventricle revisited. J Thorac Cardiovasc Surg
154:598–604.
Edwards JE. 1977. Discussion. In: Davila JC, editor. 2nd Henry Ford
Hospital International Symposium on Cardiac Surgery. New York:
Appleton-Century-Crofts. p 242.
Frazer JE. 1916. The formation of the pars membranacea septi.
J Anat 51:19–29.
Geva T. 2016. Segmental approach to congenital heart disease. In:
Lai WW, Mertens LL, Cohen MS, Geva T, editors. Echocardiogra-
phy in pediatric and congenital heart disease: from fetus to adult.
2nd ed. Hoboken, New Jersey: JohnWiley & Sons, Ltd. p 31–43.
Jacobs ML, Anderson RH. 2006. Nomenclature of the functionally
univentricular heart. Cardiol Young 16(Suppl 1):3–8.
Kelly RG, Brown NA, Buckingham ME. 2001. The arterial pole of the
mouse heart forms from Fgf10-expressing cells in pharyngeal
mesoderm. Dev Cell 1:435–440.
Kramer TC. 1942. The partitioning of the truncus and conus and the
formation of the membranous portion of the interventricular sep-
tum in the human heart. Am J Anat 71:343–370.
31
LaCorte MA, Fellows KE, Williams RG. 1976. Overriding tricuspid
valve: Echocardiographic and angiocardiographic features.
Am J Cardiol 37:911–919.
Lamers WH, Wessels A, Verbeek FJ, Moorman AFM, Virágh S,
Wenink ACG, Gittenberger-de Groot AC, Anderson RH. 1992. New
findings concerning ventricular septation in the human heart.
Implications for maldevelopment. Circulation 86:1194–1205.
Milo S, Ho SY, Macartney FJ, Wilkinson JL, Becker AE,
Wenink ACG, Gittenberger-de Groot AC, Anderson RH. 1979.
Straddling and overriding atrioventricular valves morphology and
classification. Am J Cardiol 44:1122–1134.
Mitsumoto Y, Oka S, Sakuma H, Inazawa J, Kawasaki T. 2000. Clon-
ing and chromosomal mapping of human glucuronyltransferase
involved in biosynthesis of the HNK-1 carbohydrate epitope. Geno-
mics 65:166–173.
Mjaatvedt CH, Nakaoka T, Moreno-Rodriguez R, Norris RA,
Kern MJ, Eisenberg CA, Turner D, Markwald RR. 2001. The out-
flow tract of the heart is recruited from a novel heart-forming field.
Dev Biol 238:97–109.
Mohun TJ, Weninger WJ. 2011. Imaging heart development using
high resolution episcopic microscopy. Curr Opin Genet Dev 21:
573–578.
Moorman AFM, Christoffels VM. 2003. Cardiac chamber formation:
development, genes, and evolution. Physiol Rev 83:1223–1267.
Odgers PNB. 1938. The development of the pars membranacea septi
in the human heart. J Anat 72:247–259.
Sherman FE. 1963. An Atlas of Congenital Heart Disease. London:
Henry Kimpton. p 166–186.
Soto B, Becker AE, Moulaert AJ, Lie JT, Anderson RH. 1980. Classi-
fication of ventricular septal defects. Br Heart J 43:332–343.
Van den Hoff MJB, Moorman AFM, Ruijter JM, Lamers WH,
Bennington RW, Markwald RR, Wessels A. 1999. Myocardializa-
tion of the cardiac outflow tract. Dev Biol 212:477–490.
Van Praagh R, Ongley PA, Swan HJC. 1964. Anatomic type of single
or common ventricle in man. Morphologic and geometric aspects of
60 necropsied cases. Am J Cardiol 13:367–386.
Van Praagh R, Geva T, Kreutzer J. 1989. Ventricular septal defects:
How shall we describe, name and classify them? J Am Coll Cardiol
14:1298–1299.
Van Praagh R. 2015. Bulboventricular foramen. In: Ezon DS,
Goldberg JF, Kyle WB, editors. Atlas of Congenital Heart Disease
Nomenclature. Houston: Baylor College of Medicine, Texas Chil-
dren’s Hospital. p 45.
Waldo KL, Kumiski DH, Wallis KT, Stadt HA, Hutson MR,
Platt DH, Kirby ML. 2001. Conotruncal myocardium arises from a
secondary heart field. Development 128:3179–3188.
Wells WJ, Lindesmith GG. 1985. Ventricular septal defect. In:
Arciniegas E, editor. Pediatric cardiac surgery. Chicago: Year Book
Medical.
Wessels A, Vermeulen JLM, Verbeek FJ, Viragh SZ, K6lmin F,
Lamers WH, Moorman AFM. 1992. Spatial distribution of "tissue-
specific" antigens in the developing human heart and skeletal mus-
cle: III. An immunohistochemical analysis of the distribution of the
neural tissue antigen GlN2 in the embryonic heart; implications
for the development of the atrioventricular conduction system.
Anat Rec 232:97–111.