Childs Nerv Syst (2004) 20:607–617

DOI 10.1007/s00381-004-1011-6 SPECIAL ANNUAL ISSUE

Dale M. Swift
Bradley Weprin
Total vertex craniopagus
Fred Sklar with crossed venous drainage:
David Sacco
Kenneth Salyer case report of successful surgical separation
David Genecov
Raul Barcello
Mario Ortega
Ken Shapiro

Received: 26 April 2004
Published online: 29 July 2004

Springer-Verlag 2004
D. M. Swift · B. Weprin · F. Sklar ·
D. Sacco · K. Salyer · D. Genecov ·
R. Barcello · M. Ortega · K. Shapiro
Children’s Medical Center of Dallas,
North Texas Hospital for Children at
Medical City Dallas,
Dallas, TX, USA
D. M. Swift ())
Neurosurgeons for Children,
1935 Motor Street, Dallas, TX 75235, USA
e-mail: dswift@neurosurg4kids.com
Tel.: +1-214-4566660
Abstract Case report: Twin boys
joined at the head in a total vertex
configuration were born in Egypt in
June 2001. At 12 months, they were
transported to Dallas for evaluation
and eventual surgical separation. In
Dallas, a large multidisciplinary team
of health care providers from two
pediatric hospitals was assembled to
care for the boys. Extensive radio-
graphic evaluation revealed that the
twins had essentially separate, well-
formed brains, each with regions of
schizencephaly. Each child’s left ce-
rebral hemisphere drained most of the
venous blood to the right jugular
system of the other. A detailed as-
sessment of the foreseeable risks of
surgical separation was then estimat-
ed and presented to the parents, as
well as to the ethics committee of the
two institutions. The decision was
then made to proceed with separation.
Surgical planning included the con-
struction of multiple polymer models,
and the design and construction of
customized head holders and an op-
erating table. Prior to separation a
series of preparatory operations were
performed to expand the scalp, as
well as the fascia lata for dural
grafting. At the age of 28 months, the
twins were successfully separated
during a 33-h operation. No attempt
was made to reconstruct the dural
venous sinuses. Scalp closure was
adequate, requiring a split-thickness
skin graft on one boy. Outcome:
Postoperatively each child demon-
strated an incomplete right hemipar-
esis. One twin required cerebral spi-
nal fluid shunting. Neither child had a
CSF leak or a CSF infection. At
6 months follow-up, both boys are
rapidly acquiring speech in both En-
glish and Arabic, motor function is
improving, and both are progressing
toward independent ambulation.
Keywords Craniopagus · Conjoined
twins · Polymer models

Introduction of the condition. In recent years, however, technological

Twins conjoined at the head or craniopagus1 is a rare and
dramatic condition [1]. Surgical separation is generally
deemed desirable, but success in this undertaking is lim-
ited. This is due to the rarity, variability, and complexity
1 Although the neurosurgical portion of the operation
The word craniopagus is singular and refers to a set of twins
conjoined at the head. Since the term refers to both children as one
entity its use is limited in this article
advances have led to improved preoperative understand-
ing of the involved anatomy and presumably better out-
comes.
The separation of total vertex-conjoined twins requires
collaboration among three medical specialties: neurosur-
gery, craniofacial plastic surgery, and anesthesiology.
constitutes the “separation,” a successful outcome re-
quires that each specialty has carefully considered its
608
contribution and that the individual specialists function
together as a coherent team. Inevitably each specialty will
have to change its traditional approach to optimize the
ultimate result. We report a case of successful separation
of total vertex-conjoined twins operated on in Dallas at
the age of 28 months. Close multidisciplinary coopera-
tion, extensive preoperative planning, and innovative
equipment design are all thought to have contributed to
the successful outcome.
Case report
Twin boys joined at the head were born in Egypt in June 2001. The
twins were joined at the vertex of each in a pattern termed “total
vertex craniopagus” (Fig. 1). The interaxis angle was nearly 180.
Interaxis rotation was about 90 (i.e., their faces were oriented
about 90 from one another). At 12 months of age they were
transported to Dallas for evaluation. The twins were alert, active,
and demonstrated no focal neurological deficits. Neither boy
seemed significantly impaired in comparison to the other. General
pediatric assessment revealed no abnormalities of other organ
systems. Developmental and cognitive assessment showed both
boys to have mild to moderate developmental delays, even con-
sidering the physical constraints imposed by their attachment.
Nonetheless, the twins appeared to be thriving and demonstrated
positive responses to physical and occupational therapies.
Cerebral anatomy
The cerebral anatomy was delineated by a series of neuroradio-
graphic studies including computerized tomography (CT), com-
puterized tomographic angiography (CTA), magnetic resonance
imaging (MRI), magnetic resonance angiography (MRA), magnetic
resonance venography (MRV), and cerebral angiography. The CT
and MRI data were reformatted to provide 3D representations. Each
child demonstrated fully developed cerebral hemispheres, basal
ganglia, and cerebelli. Each child also had a limited area of schi-
zencephaly in the medial occipito-parietal region extending to a
cystic dilatation of the third ventricle (Fig. 2). The left hemisphere
of each twin projected into the cranium of the other twin. There was
marked positional flattening of the skull posteriorly, which was
Fig. 1 The patients at 17 months of age. The interaxis angle is
about 180, and the rotation about 90
transmitted to the shape of the brain. The brains appeared nearly
completely separate, with a small shelf of dura in limited locations.
At the apex of the parietal aspect of one left hemisphere, the margin
on MRI was indistinct. Indeed, at surgery this area was found to be
adherent to the other twin’s left hemisphere.
Conventional cerebral angiography demonstrated relatively
normal arterial supply to the brains with distortion consistent with
the cerebral anatomy. The venous drainage, however, was markedly
abnormal, demonstrating “crossed venous drainage.” The left ca-
rotid injection on each child filled the left hemisphere of the same
child, but the majority of venous drainage proceeded though cor-
tical draining veins into a superficial dural venous sinus, which
became the right transverse sinus and ultimately the right jugular
vein of the other child (Fig. 3). Right hemispheric drainage stayed
within the same child via the right jugular system. Thus, the left
jugular veins were rudimentary.
The anatomy of the brains and the dural venous sinuses was
remarkably symmetric (Figs. 4, 7). Neither child demonstrated a
more exuberant venous system than the other, and equal numbers of
left-sided major cortical veins drained into the “shared” right jug-
ular systems of each. Thus, the major dural venous sinus—analo-
gous to the sagittal sinus in its drainage distribution—did not,
however, course between the two hemispheres of each twin. This
sinus system ran superficially along the junction of the two brains
and was continuous from the right jugular vein of one twin to the
right jugular vein of the other with a small diagonally oriented
connection posteriorly. Blood flow in this anastomosis was pre-
sumed to be bidirectional. Although this pattern has been referred
to as a “shared sagittal” or “circumferential” sinus in other cranially
conjoined twins, in this case the sinus was more aptly described as
“spiral”—implying a midpoint appropriate for division. This mid-
point was located just to the right of each child’s glabella. The dural
sinus in this region was presumed to be bidirectional and extended
deep between the two brains as a highly vascular “shelf” of dura.
The MRV revealed small veins running through the inter-
hemispheric fissure of each twin in the expected location of the
sagittal sinus (Fig. 5). Because of the apposition of the twins’
brains, these vessels coursed deeply rather than superficially.
Defining flow through these vessels on the angiogram was not
possible. These sagittal veins were the only obvious aspect of the
twins’ venous drainage that was not symmetric as it appeared as
though both drained into twin M’s jugular system.
Assessment of risk
Upon arrival in Dallas there had been no definitive decision made
as to whether or not to proceed with surgical separation. The twins
were in no apparent distress and grossly neurologically intact. A
reasonable assessment of risk had to be determined and then pre-
sented to the parents as well as the ethics committees of the two
institutions involved. This assessment was based upon the anatomy
demonstrated by CT, MRI, and cerebral angiography, as well as
consultation with other surgical teams who had experience in
similar cases.
Fig. 2a–e MRI obtained at 13 months of age. Axial T1 MRIs with
contrast. a Scan at the level of the lateral ventricles of twin M. Note
the cystic dilatation of the third ventricle and limited schizen-
cephaly. b Scan roughly at the junction of the twins. Three hemi-
spheres are visible, two from twin M and the left of twin A. c Scan
at level of the lateral ventricles of twin A. Note the rotation from a
and similar cystic dilatation of the third ventricle with limited
schizencephaly. d Sagittal image of twin M (bottom). e Coronal
image with twin M at the bottom. Note the indistinct margin be-
tween the two brains (circle). This region proved more difficult to
dissect at surgery
609
610
Fig. 3a, b Cerebral angiogram at 13 months of age. a Left internal
carotid injection of twin A results in the majority of venous
drainage to the right jugular vein of twin M. b Right internal carotid
injection of twin M similarly empties into twin A’s right jugular
system
In addition to the general risks of surgery of this magnitude, the
patients’ individual anatomy presaged two major concerns. First,
the presence of crossed venous drainage appeared to carry a
significant risk of venous infarction. Second, the lack of tissue
available for closure of the anticipated calvarial defect carried
significant risk of central nervous system (CNS) infection.
Once sufficient details of the patients’ anatomy had been elu-
cidated, a diligent effort was made in order to identify similar
patients and discuss their management with their respective surgical
teams and care providers. The authors gratefully acknowledge and
are particularly indebted to the teams at UCLA and Singapore
General Hospital, both of whom previously separated total vertex-
conjoined twins. In addition, a number of other neurosurgeons with
similar experience graciously reviewed this case and rendered
valuable opinions.2
Upon review of the available information, a reasonable as-
sessment of risk seemed to predict that the twins had a good chance
of survival, but that the left cerebral hemispheres were at significant
risk of venous infarction. Numerous possible options for managing
the dural sinuses were proposed. The final surgical plan utilized the
exquisite symmetry apparent in the patients and disallowed any
“sacrifice” of one twin’s venous drainage system for the benefit of
the other. The surgical plan also emphasized numerous measures to
prevent CNS infection, which had proven disabling in other cases.
Review of the information by the two ethics committees resulted in
recommendations that separation surgery be offered to the parents,
who in turn accepted and gave informed consent.
Preoperative preparation
Multidisciplinary team
In Dallas, a multidisciplinary team was assembled with pediatric
neurosurgery, craniofacial plastic surgery, and pediatric anesthesia
at its core. Further support was provided by other pediatric sub-
specialists including neuroradiology, critical care, general surgery,
neurology, ENT, physical therapy, occupational therapy, social
work, chaplaincy, and others. None of the members of the team had
previous experience with separating cranially conjoined twins.
Some members of the neurosurgery team (K.S., F.S.) had separated
twins conjoined at the spine, and members of both the craniofacial
and neurosurgery teams (K.S., K.S.) had performed repair of cranial
defects resultant from separation of cranially conjoined twins. The
leader of the anesthesia team (M.O.) had previously directed the
anesthetic management of six sets of somatically conjoined twins.
Tissue expansion
Upon review of similar cases, it became clear that complete cov-
erage of the surgical defect with healthy autologous tissue and
without foreign materials would be crucial in the prevention of
cerebrospinal fluid (CSF) leaks and resultant CNS infection. In-
fections of these types are known to cause significant neurologic
damage and thought to be responsible for poor neurocognitive
outcome observed in some separated twins. Two separate flaps
were planned, scalp and pericranium (Fig. 5). The roughly trian-
gular scalp flaps were frontal and occipital in each twin. Frontal
flaps were based on supratroclear, supraorbital, and superficial
temporal blood supply, whereas the occipital flaps were fed by the
posterior auricular and occipital arteries. The pericranial flaps were
based temporally on the deep temporal arteries. Thus, a series of
operations was performed to provide appropriate tissue expansion.
Tissue expansion progressed over a period of nearly 6 months. A
total of six inflatable expanders with varying configurations and
volumes was placed. Aggregate volume was 2,150 cc. One ex-
pander was removed due to an infection, which had completely
resolved by the time of separation.
2
The authors would like to acknowledge the generous expert ad-
vice provided by Drs’ Keith Goh, John Frazee, Jorge Lazareff,
Barbara Van de Wiele, Duke Samson, Marion Walker, Jonathan
Peter, Dachling Pang, Ken Winston, and others
 611

Fig. 4a–d Diagram of the ve-

nous system. a, b Anterior and
posterior views depicting the
“spiral” sinus continuous from
one right jugular vein to the
other. Note the two potential
division points where blood
flow is presumed to be bidirec-
tional. c, d Broken lines indicate
points where the cortical veins
and dural sinuses were divided

Calculations of the dural defects were 165 cm2 and 192 cm2.
Typically, pericranium would be considered for such a closure, but
in this case pericranium was an essential element of the scalp
closure and another source of autologous membrane was sought.
This amount of fascia was not present in an acceptable donor site
elsewhere in the bodies so it was decided to expand the fascia lata.
In both boys 280 cc tissue expanders were inserted under the fascia
lata of each thigh and gradually inflated to 150% of their intended
volume.
The final operative date was determined by the adequacy of the
tissue expansion. Models of the head with the scalp expanders in
place were created and the flap design tested with latex “skin.” This
step proved indispensable as the initial design of the flap had to be
modified to adjust for the effects of growth and tissue expansion.
The simulated flaps identified areas where coverage would be in-
complete. These areas were then repositioned over skull and peri-
cranium, enabling the plastic surgeons to employ a vacuum-assisted
closure system (VAC; KCI, San Antonio, TX, USA) without fear of
injuring the brain.
612
Fig. 5a–d Flap concept; ideal-
ized diagram demonstrating
two-layer scalp closure with
pericranium and full thickness
skin flaps. Note that the flap
bases are offset by 90. At sur-
gery modifications were neces-
sary
Models
The use of physical models proved invaluable in this case (Fig. 6)
[2]. Models (Medical Modeling, Golden, CO, USA) were created to
simulate multiple aspects of the patients’ anatomy. A full-scale
total body model of both twins measuring 6 feet (1.83 m) in length
and weighing 100 pounds (45.4 kg) was used to design and test the
head holder and operating table, as well as during multiple “dress
rehearsals.” Multiple separate models of the brains, intracranial
vasculature, skulls, and scalp were studied and manipulated, re-
sulting in a more complete preoperative understanding of the
anatomy than could have been achieved by studying the radio-
graphic images alone. The model incorporating the venous struc-
tures was particularly useful when correlated with blood flow
demonstrated by cerebral angiography. Updated models were also
created after tissue expansion and then used to test the scalp flap
design. A latex “skin” covering the model was created and then cut
and rotated, revealing necessary modification to flap design.
Innovative surgical equipment
Such a novel and complex case as this required innovation in the
design of equipment that is otherwise in routine use. Our team was
fortunate to have the cooperation of imaginative engineers and
representatives from companies with expertise in surgical tables
(KCI), operative head holders (Schaerer-Mayfield, Munsingen,
Switzerland), and frameless stereotaxy (Cbyon, Palo Alto, CA,
USA). The willingness and efficiency with which these manufac-
turers responded was without precedent and greatly appreciated.
Surgical table. Intuitively separation of vertex-conjoined twins
requires a 360 surgical field. Conventional operating tables do not
provide this flexibility and dictate that multiple changes of position
will be required as the operation proceeds. Since each major change
in positioning requires repeat prepping and draping of the surgical
field, the likelihood of infection multiplies and an already lengthy
procedure becomes longer. In addition, operating tables that adjust

Fig. 6a–d Models. a, b The skull model has been cut, simulating the amount of cranial bone removed at surgery. c The brain models have
been placed in the skull models. d Venous model
independently may result in torque being applied to the brains and
cerebral vasculature, once the rigid support of the skull is removed.
These factors influence surgical planning and could easily lead to
operative steps that would not be ideal. To obviate these restric-
tions, we worked with KCI to develop a table that allowed both
children to be rotated 360 together initially and which would then
ultimately separate into two operating tables. The table suspended
each child in a “clamshell”-style body jacket so that the conjoined
heads were in the center of rotation. A retractable, removable
bracket supported each head and provided stabilization during the
application of the head holders. The table was constructed to allow
the anesthesiologists to coordinate and control rotation from either
end of the table. Rotation of each end was yoked to the other side of
the table, preventing torque on the heads of the children. The table
was tiltable to either side to assist in the distribution of blood
volume if necessary. Conversion to two separate tables was
achieved by disconnecting a rod joining the two drive shafts, thus
enabling two teams to work simultaneously after separation. A wire
frame was constructed to provide support for surgical drapes and
allow the anesthesiologists access as needed. The beds rotated
beneath the frame preventing the drapes from becoming entangled
with changes in position. Special care was taken in padding the
body jackets and lining them with materials that would prevent
pressure sores. The prolonged duration of the procedure dictated
that special temperature control is used, as the body jackets would
limit temperature regulation. This factor was of great importance
for the team of anesthesiologists.
Head holders. Conventional skull pin fixation would not provide
the access necessary to perform the operation. Furthermore, the
array of pins would have interfered with the scalp flaps and their
613
blood supply. Close cooperation with Shaerer-Mayfield allowed us
to develop two customized head holders that allowed pin fixation to
be applied at sites that did not interfere with the scalp flaps or
encroach upon the surgical field. The head holder consisted of a
customized ring that attached to traditional skull pins or threaded
pins and allowed a self-retaining retractor system to be incorpo-
rated. The deformity of each twin’s head dictated that each ring
conform to each child’s head shape. A circular ring would end up
being too far from the head and interfere with surgical access.
However, an exact duplicate of each child’s head shape would have
narrowed portions of the arc necessary for mounting and sliding
fixation devices and retractor arms. The craniofacial surgeons
provided transmalar pins so that the head holder would lie below
the equator of the skull. The ring of each child was linked to the
other by a crossbar that provided further stabilization. The position
of this linkage could be changed as the operation progressed and
was ultimately removed when the separation was complete. The
need for access to the endotracheal tubes was answered by a re-
movable element of the ring.
Close cooperation between KCI and Shaerer-Mayfield facili-
tated the design for mounting the two head holders to the table. The
models proved invaluable during the process of designing and
testing these devices. The customized design of the table and head
holders for a single application meant that the implementation and
integration of the devices needed to be carefully rehearsed to ensure
their usefulness during the actual operation. Each trial of the sys-
tems identified necessary modifications.
Frameless stereotaxy. At first we were unsure as to whether
frameless stereotaxy would be feasible in an operation of this type
due to multiple position changes. However, with the development
614
of the customized head holder and operating table, frameless ste-
reotaxy became a viable option. We employed the Cbyon system
and received close support from the company. A dynamic reference
frame (DRF) was developed that could be mounted on the rotating
operating table without intruding on the available surgical space.
The attachment to the table was rigid enough to prevent accidental
movement of the reference frame. Two DRFs were used, one on
each half of the body jackets holding the children. This ensured that
the camera could “see” the patients’ heads throughout the 360 of
rotation planned during the operation.
Operative technique and events
On the day prior to the separation, the twins were taken to the
operating room where they were anesthetized and endotracheal
tubes wired in position. They then had venous, central venous,
arterial, and urinary catheters placed. An MRI with fiducial markers
was performed. Lumbar spinal drains were placed (unsuccessfully
in twin A), and malar pins were applied for cranial fixation. At this
time final adjustments were made to the positioning plans and the
patients were kept intubated in anticipation of the next day’s sur-
gery. The twins were 28 months of age and weighed 12 kg and
14 kg.
On the day of separation the patients were re-anesthetized and
positioned on the customized operating table. Prior to placing the
anterior portion of the body jacket, the tissue expanders were re-
moved from the thighs of both boys bilaterally. The expanded
fascia latae were maintained as capsules around the expanders to
prevent contraction prior to use. The custom head holders were then
applied using the malar pins anteriorly and occipital skull pin fix-
ation posteriorly. Posterior pin placement was complicated by the
presence of the scalp expanders, which had distorted and thinned
the regional skull. After final positioning and application of the
anterior body jacket, the scalp flaps were marked, fiducials regis-
tered, and the table rotated to the prone position. The operation was
then divided into three predefined stages.
Stage one. In the prone position, the posterior aspects of the scalp
flaps were created. The tissue expanders were removed, retaining
the outer capsule with the scalp. The inner capsule remained as the
pericranium. The layers of the two flaps were continually moist-
ened and kept on slight traction throughout the lengthy procedure.
Craniotomies exposing the posterior dural sinuses were then per-
formed. The bone was markedly thinned in some areas beneath the
tissue expanders. The dura was then opened, isolating the diago-
nally oriented, posterior anastomotic sinus, which was then divided
without incident (Fig. 4d). The dural openings then proceeded
outward, but only on the side covering the left hemispheres of each
child. Cortical draining veins of the left hemispheres were tem-
porarily occluded, then cauterized and divided. The arrangement of
the draining veins was relatively symmetric and an attempt was
made to divide corresponding veins simultaneously to minimize the
possibility of blood volume shift between the twins. Predivision
temporary occlusion was performed in order to observe any pos-
sible effect upon the brain.
Stage two. The entire operating table was then rotated 180 so that
the midpoint between each twin’s face was uppermost. To prevent
continued blood loss and maintain tension, the posterior scalp flaps
were approximated prior to turning. This also served as a support
for the exposed brain while in a dependent position. The anterior
incisions of the scalp flaps were now made and tissue expanders
removed. A craniotomy was then created over the midpoint ex-
tending past each child’s anterior midline. The dura was opened on
both sides of the midposition sinus, which extended deep as a dural
“shelf” (Fig. 7). The exact vascularity of this sinus was not known
preoperatively, but blood flow was presumed to be bidirectional,
indicating a physiologic as well as anatomic dividing point. The
sinus was occluded with a combination of clips and then oversewn
without incident.
Separation of the left hemispheres of each twin from the sinus
was then accomplished. Again, temporary occlusion was per-
formed, followed by coagulation and division. Using the rotation of
the table, veins were alternately divided in each patient to minimize
blood volume shift. Neither visible cerebral edema nor significant
blood volume shift occurred.
Separation of the brains then proceeded in a circumferential
fashion. This was relatively straightforward in the limited regions
where the brains were separated by dura. In other regions the pial
layers were approximated and separation was more difficult. This
was especially problematic at the apex of the left hemisphere of
twin M, where it abutted the left hemisphere of twin A. The situ-
ation was analogous to a difficult dissection of the interhemispheric
fissure and at times the plane was lost and dissection subpial.
Frameless stereotaxy was particularly helpful during this portion of
the operation.
As the dissection proceeded centripetally, some previously un-
recognized veins were divided without obvious cerebral or hemo-
dynamic consequence. The interhemispheric “sagittal” vein of twin
A was divided, but that of twin M remained. When it appeared as
though the separation was complete, the rigid “yoke” connecting
the head holders was carefully removed and the table disconnected
and slowly distracted. The exposed brains at this point required
continuous support to minimize gravitational distortion.
Stage three. After complete separation was achieved, the operating
table was converted to two separate tables. The heads of each table
were then turned 90 away from each other and two combined
teams of neurosurgeons and craniofacial surgeons worked simul-
taneously on the closures. The cranium was not reconstructed and
the cranial bone was placed into the pocket created by the fascia
lata tissue expanders. With continuous manual support of the
brains, the dura was closed with the fascia lata and the fibrous
tissue capsules harvested from the tissue expanders. The amount of
this tissue proved insufficient and therefore free grafts of pericra-
nium were utilized, maintaining the flap base. Scalp closure was
then accomplished using the pericranial and scalp flaps. Scalp
closure was complete in twin A, but twin M required skin grafting
over the temporal regions. On the left a 34-cm split-thickness skin
graft was used and on the right a 35-cm full thickness graft har-
vested from the expanded skin of the thigh was used. Drains were
placed at the base of the flaps in each child and a vacuum-assisted
closure system (VAC KCI) was used to bed down the skin grafts in
twin M.
The procedure lasted approximately 33 h. Estimated blood loss
was 2,800 ml in twin A and 3,200 ml in twin M. Anticipated
problems such as cerebral edema and blood volume shifts between
the twins did not occur to any significant degree. During stage two
of the procedure the patients received loading doses of pentobar-
bital according to a predetermined plan. The patients exhibited
stable vital signs and were transported to the ICU without incident.
Postoperative course
Postoperatively the patients were maintained on pentobarbital in-
fusions for 72 h in anticipation of potential cerebral edema. Since
the bony calvaria had not been reconstructed, this was undertaken
more to protect the scalp closure and minimize the risk of CSF
leakage. Twin A had a lumbar spinal drain placed on postoperative
day (POD) 1. Both lumbar drains were drained at head level for
5 days then removed on POD 7. The suture lines remained intact
without CSF leakage. CTs of both boys obtained on POD 1 showed
no significant cerebral edema or hemorrhage.

Fig. 7a–d Magnetic resonance venography (MRV). a, b Sagittal and axial MRVs at 13 months. c, d Three-dimensional reconstruction of
MRI data. c Anterior view of brains and dural sinus. d MRV with ventricles
Upon awakening, both patients had brief focal seizures that
were easily controlled with phenobarbital. Twin A developed hy-
drocephalus and a lumbo-peritoneal (LP) shunt was placed to avoid
further scalp incisions. The shunt failed with a distal infection.
Cultures of the CSF, however, remained sterile. The LP shunt was
then removed and a new ventricular shunt placed without incident.
Symptomatic subdural collections developed, requiring a subduro-
peritoneal shunt about 10 weeks postoperatively.
At 6 months postoperatively both twins have required gastros-
tomy tubes for persistent vomiting. Gastrointestinal evaluations
have been unrevealing and the origin is presumed to be cerebral.
Twin M demonstrates an improving right hemiparesis and twin A
an improving diparesis. Both children are alert, interactive, and are
gaining speech function. They are responding well to therapy and
independent ambulation is expected.
615
Discussion
Significant technological advances have been made since
Winston’s comprehensive review of craniopagus in 1987
[1, 3]. Advances in neuroimaging with improved preop-
erative delineation of anatomical details have been par-
ticularly crucial to surgical planning and presumably to
recent successes. In our case, it is certainly true that ex-
tensive neuroradiologic evaluation with 3D reconstruc-
tions and the creation of physical models helped the entire
team understand the complex anatomic concerns.
The magnitude and complexity of these types of cases
commands cooperation among the major involved spe-
cialties: neurosurgery, craniofacial plastic surgery, and
anesthesiology. Although discussion of the anesthetic
616
management of this case is outside the scope of this re-
port, a few items bear mention. Previous experience has
shown that conjoined twins display variable responses to
anesthesia. In addition, separation surgery has demon-
strated the potential for catastrophic blood loss as well as
blood volume shifts between conjoined twins. In our case,
by the time of the separation surgery, the twins had been
anesthetized numerous times and their responses to an-
esthesia known. Although the estimated blood loss ap-
proximated three blood volumes there were no hemody-
namic complications. Without the support of an experi-
enced and competent anesthesia team, assisted by well-
trained and rehearsed operating room personnel, the
success of this undertaking would have been in jeopardy.
The decision to proceed with separation surgery was
not inevitable. The twins were not experiencing any
systemic or neurologic deterioration, as has been reported
preceding some other separations [3, 4]. The fact that they
were stable may support a decision not to operate, but
patients who are deteriorating are more likely to be at
higher surgical risk [3]. As with most surgery, an estimate
of risk must be weighed against an estimate of potential
benefit. In this case both estimates, especially the latter,
are difficult and inexact at best. There is no a priori reason
why conjoined twins must be separated, although some
claim that cognitive decline is likely [3]. It is known that
there are examples of adult patients who decline separa-
tion [5, 6]. It may be likely that as craniopagus patients
grow older they would be less likely to choose separation,
but the recent decision of two highly educated adult
women to undergo separation is enlightening [8].
Our decision to proceed was based on the most likely
postsurgical outcome that could be anticipated. In this
regard, we estimated a likely result of separation to be left
hemispheric damage of an unknown severity in both pa-
tients. The participation of the ethics committees was
valuable in this regard. In the end it was thought that,
even with the sequelae of dominant hemispheric damage,
the boys could interact better in society if separated. We
anticipated survival in both patients and were fortunate
not to be faced with a situation where sacrifice of one
would lead to better outcome for the other.
Management of the cerebral venous system
The case presented in this article is best designated “total
vertex craniopagus.” Recent experience with this variant
has been without mortality in three occurrences (Dallas—
reported herein; Los Angeles—personal communication
and lay press; Singapore—[7]). When classified by the
deepest shared structure, the present case is Type C,
characterized by discrete brains, incomplete dural sepa-
ration, and shared arachnoid [1]. This classification does
not take into account the important aspect of venous
drainage. In the setting of total vertex craniopagus, it
appears that crossed venous drainage is frequently, if
not invariably, encountered. Angiography consistently
reveals reciprocal drainage of one hemisphere of one twin
into the other’s dural venous system. Frequently, the most
complicated aspect of surgical separation concerns this
crossed venous drainage with tentative risks of venous
infarction. The twins described in this report demonstra-
ted significant crossed venous drainage. Each twin’s left
cerebral hemisphere drained most of its angiographically
demonstrated venous blood into the other twin’s right-
sided jugular system.
Numerous approaches were considered to deal with
this problem. One of the most complex, but potentially
most beneficial would be to completely reconstruct the
dural sinus, giving each twin a complete drainage system.
Given the complexity of such a procedure and the history
of failure of cerebral venous grafts, we rejected this op-
tion. Failure in this setting might lead to venous infarction
of all hemispheres instead of just the two left ones. En-
dovascular balloon test occlusion (BTO) may predict the
need for revascularization. Unfortunately, in this case
there was no anatomic location suitable for BTO. Test
occlusion of a shared sinus would impair drainage of one
right and one left hemisphere of different twins.
Another option for managing the venous drainage is
staged occlusion. Staged occlusion of the draining veins
may in theory allow time for venous collaterals to de-
velop. While this is an attractive idea, and widely advo-
cated, there are a number of reasons why we chose not to
perform staged occlusion in this case. First of all, al-
though staged occlusion is thought to “redirect” venous
blood to deep veins, there is no guarantee that blood flow
will not simply increase to the remaining cortical veins,
which must ultimately be divided. Second, should cere-
bral edema occur following an initial staged division, it
may be poorly tolerated in a craniopagus patient with a
closed skull. While we were unsure whether or not ce-
rebral edema would develop in our patients, the lack of
bony skull postoperatively might have mitigated the ef-
fects of raised intracranial pressure. An additional con-
cern regarding staged craniotomies was only appreciated
after extracting the tissue expanders in our case. Inward
pressure against the skull exerted by the tissue expanders
resulted in significant deformity of the cranium. This ef-
fect might be even more pronounced in younger children
and in patients with healing craniotomies.
The most convincing reason to proceed in a single
operation came from review of the most recent similar
experiences. We extensively reviewed the cases from
Singapore (2001) and Los Angeles (2002). Neither of
these teams elected to perform staged occlusion and none
of the twins appeared to suffer significant venous in-
farction (personal communication, 7). In Dr Goh’s opin-
ion the length of the surgery is sufficient time in which to
open venous collaterals. Indeed, little is known about the
development of cerebral venous collaterals and many
 617

believe that existing veins may rapidly accommodate new
blood flow demands. Fortunately, in our case neither
significant cerebral edema nor venous infarction ensued.
Observed neurologic injury could be directly related to
dissection of the areas where the two brains were most
adherent.
essary for dural closure, in conjunction with judicious use
of pericranium the result was effective. Further efforts to
prevent CSF leaks included lumbar spinal drains and
postoperative barbiturates to minimize potential brain
swelling. These factors, plus the avoidance of foreign
materials and the careful placement of skin grafts all
contributed to the absence of any CSF infection.

Management of the calvarial defect

One of the most difficult aspects of this case was the
design and execution of the tissue expansion required to
cover the calvarial defects. These implants are prone to
infection and require many months to accomplish their
goals. Gradual expansion is preferable to rapid expansion,
which creates thinner skin at greater risk of breakdown.
The flap design was continually reassessed as the twins
grew and the expanders were inflated. Indeed, the flaps
were redesigned immediately before surgery due to
unanticipated growth patterns. Detection of this change
was significantly aided by working with physical models.
Both the neurosurgical and plastic surgical teams
recognized the need for adequate dural and scalp cover-
age without CSF leakage. This was thought to be crucial
to the prevention of CNS infection, which could ulti-
mately limit cognitive outcome. To this end, the novel
concept of fascia lata expansion was applied. Although
this technique did not supply all of the membrane nec-
Innovative devices
This case provided the impetus for the development of a
number of innovative devices and techniques. The con-
vertible table, with 360 rotation, coupled with the cus-
tomized head holder utilizing malar pins, provided supe-
rior function and safety by maintaining the twins’ position
relative to one another until separated and allowing
complete access to all aspects of the operative field.
Similarly, the uses of physical models and frameless
stereotaxy were extended beyond the usual indications.
Acknowledgments Although the success of this case is attributed
to the aforementioned concepts and techniques, in the end the un-
dertaking would have been impossible without the massive team
effort incorporating the combined resources of Dallas’ two pedi-
atric hospitals. The authors would like to acknowledge the support
of the Children’s Medical Center of Dallas, and the North Texas
Hospital for Children at Medical City Dallas in this endeavor.

References
1. Winston KR (1987) Craniopagi: ana-
tomical characteristics and classifica-
tion. Neurosurgery 21:769–781
2. Christensen AM, Humphries SM, Goh
KYC, Swift DM (2004) Advanced
“tactile” medical imaging for separation
surgeries of conjoined twins. Childs
Nerv Syst http://dx.doi.org/10.1007/
s00381-004-0982-7
3. Winston KR, Rockoff MA, Mulliken
JB, Strand RD, Murray JE (1987) Sur-
gical division of craniopagi. Neurosur-
gery 21:782–791
4. Khan ZH, Tabatabai SA, Saberi H
(1999) Anesthetic and surgical experi-
ence in a case of total vertical craniop-
agus. Surg Neurol 52:62–66
5. Todorov AG, Cohen KL, Spilotro V,
Landau E (1974) Craniopagus twins.
J Neurol Neurosurg Psychiatry
37:1291–1298
6. Dallas Morning News, Sisters were
joyful in 43 years together, 8/24/2003
Page 8W
7. Goh KYC (2004) Separation surgery for
total vertex craniopagus twins. Childs
Nerv Syst http://dx.doi.org/10.1007/
s00381-004-0978-3
8. Krauthammer C (2003) A doctor’s duty.
Time 162:80