Since the first U.S. endovascular aortic graft was performed, radiologists and physicians have learned a great deal about abdominal aortic aneurysms (AAA), endografts, and patient outcomes. This article presents critical information regarding aneurysm measurement, patient and device selection, device placement, management of endoleaks, and long-term patient outcome.
is a Professor and Director, and
is a Research Assistant, Department of Radiology, The Alfred,
is a Vascular Surgeon, Vascular Surgery Unit, The Royal Melbourne
Hospital, Victoria, Australia.
is a Regional Sales Manager for William A. Cook Australia Pty.
Ltd., Brisbane, Australia.
Dr. Thomson is a shareholder in Angiodynamics Inc. and has
received research and travel grants from Angiodynamics Inc., Cook
Inc., Boston Scientific Corp., and Terumo Corp. Dr. Milne has
received research and travel grants from Cook Inc. and
TriVascular Inc. Dr. Tan has received a research grant from
William A. Cook Australia Pty. Ltd. Mr. McIntyre is an employee
of William A. Cook Australia Pty. Ltd.
The exclusion of an abdominal aortic aneurysm (AAA) by placement
of an intraluminal, stent-anchored, Dacron prosthetic graft
retrograde from the femoral artery is vastly different from the
open conventional Dacron repair, and this approach has galvanized
vascular surgeons worldwide. Although stent grafts were performed
in Russia in the 1980s, Parodi
first described his technique in 1991. The first endovascular
aortic graft performed in the United States was reported in May
Eleven years later, we have learned a great deal about the
aneurysm, the endograft, and the outcome of the new procedure. The
key lessons are in accurate measurement of the aneurysm, careful
patient selection, selection of an appropriate device, device
placement, management of endoleaks, and long-term outcome.
The risk for AAA rupture is related to its maximum diameter.
Rupture occurs when the wall stress exceeds the tensile strength of
the aortic wall. Maximum diameters >5 cm, recent rapid growth,
or pain are indications of a need for repair. Rupture of the aorta
is associated with a high mortality. A longitudinal population
screening study of 41,000 males in West Australia showed no
definite benefit from ultrasound screening to detect abdominal
aortic aneurysm. There was, however, some benefit in selected
groups of men suitable for endografts between the ages of 65 and
75, provided that the rate of open operation was low.
In contrast, in the United Kingdom, screening once on 62-year-old
men was shown to detect the 15% of men with disease, which enabled
longitudinal follow-up of this group and reduced the rupture rate.
An AAA is often detected as a result of abdominal ultrasound or
computed to-mographic (CT) examinations for other conditions.
Angiography with a graduated catheter was once a prerequisite
for assessment for endografting, but, as a result of improvements
in CT, preoperative planning is now largely performed by multislice
CT with 3-dimensional and maximum-intensity projection
reconstructions. A typical CT protocol is listed in Table 1 and
The most critical measurement is the length of the upper
aneurysm neck. Ideally, the neck should be ≥15 mm in length, <30
mm in diameter, cylindrical in shape, and free of plaque without
excessive tortuosity. The aim of measurement is to determine the
size and length of endograft that is required. Modular bifurcated
endografts have made the selection of the correct overall length
much less of a problem, as they can be overlapped to create an
adjustable overall length in situ.
Currently, angiography has a more limited role in patients with
tortuous aortas or in those in whom other procedures, such as
lumbar or iliac embolization, are required. Embolization of large
lumbar arteries is often considered to reduce the opportunity for
type II endoleaks, and there is evidence that embolization of a
patent inferior mesenteric artery prior to placement of an
endograft is associated with a lower incidence of endoleak and a
more rapid decrease in aneurysm size.
When the aneurysm sac has a low thrombus load, there appears to be
more opportunity for connections to persist across the excluded sac
between lumbar arteries and the inferior mesenteric artery.
When the internal iliac origin is involved in the aneurysm, the
endograft is usually extended into the external iliac artery and
the internal iliac artery is embolized to prevent back filling of
Patients selected for endograft repair should have the same
indications as for a surgical repair. Although generally the
patients undergoing endograft repair have smaller maximum aneurysm
dimensions, there has been no benefit shown by earlier treatment.
An aneurysm >5 cm in maximum dimension or one that is rapidly
enlarging or tender are the usual indications for repair. Those
patients who are unable to tolerate conventional open surgery have
a high mortality if an endovascular repair fails during the
procedure. Some causes of such failure include an inability to
deploy the endograft because of tortuosity of the iliac vessels or
the aneurysm itself, failure to obtain exclusion of the aneurysm
with a large endoleak, and rupture of the aneurysm or the iliac
arteries during deployment.
Modular devices such as the Cook Zenith Flex AAA Endovascular
graft (Cook Inc., Bloomington, IN) are 3-part bifurcated endografts
designed for extreme tortuosity and for patients with small-caliber
iliac arteries. Small iliac arteries are common in Asian
populations (Figure 2).
In the nature of new devices, guidelines for patient suitability
for endografts have been tested. In the Australian cohort study
with the Zenith endograft, 68% of the patients did not meet the
manufacturer's recommended guidelines.
The most ignored guideline was the neck length, and breaching this
guideline resulted in a 4-fold increase in endoleak; however,
primary and secondary aneurysm exclusion rates were still 87% and
Selection of an appropriate device
In the abdominal aorta, a bifurcated endograft that extends from
the renal arteries into the iliac arteries is almost always
required. Early models of endografts were straight aortic tubes or
aortouni-iliac devices and required occlusion of the contralateral
common iliac artery and a femorofemoral crossover Dacron graft
(Figure 3). However, bifurcated endografts are now considered
The proximal fixation of an endograft is one of the most
critical considerations in the repair. While all endografts employ
graft expansion as a fixation method, some devices provide
additional bare stents that extend proximally across the renal
arteries with or without hooks and barbs to provide additional
resistance to migration. These cross-renal devices have not been
associated with significant increase in renal dysfunction, and most
of the observed renal dysfunction after endograft insertion for AAA
is probably due to iodinated contrast.
Endograft systems with fixation that is not solely dependent on
expansile force have proved to have a lower rate of migration and
type I endoleak. The closer to the renal arteries an endograft seal
is placed, the less likely a type I endoleak is to occur.
With most devices, the first deployment step is the release of
the cranial portion of the endograft at the level of the renal
arteries. The Zenith and TriVascular (Boston Scientific, Natick,
MA) endografts, however, have a restraining device at the cranial
end that allows a measure of adjustment after the endograft is
released. Excessive movement of the partially deployed endograft
may increase the possibility of emboli from the luminal thrombus
and atheromatous plaque within the aneurysm (Figure 4). Adjustment
of the endograft is necessary if a fenestrated endograft is
selected. A fenestrated endograft is one in which there are
apertures in the Dacron fabric of the endograft to allow access to
major branches, such as renal or superior mesenteric arteries
Other design features associated with better outcomes are fully
stented endografts with higher columnar support and those with a
high resistance to kinking or limb extension dislocation. Limbs of
bifurcated grafts without stent support required further
intervention to prevent thrombosis or kinking in 40% of cases in
the EVT phase II trial
with the need for placement of additional self-expanding stents in
the limbs. Excessive tortuosity of the iliac arteries is associated
with limb occlusion from kinking and with embolism from excessive
Selection of an endograft is a current problem in the United
States, where the number of grafts approved by the Food and Drug
Administration is still limited compared with the number available
in Europe and Australia. In 2004, more than three fourths of the
patients with an AAA ≥5.0 cm in size who were seen in a tertiary
referral center were morphologically not suitable for endograft
repair using 2 currently approved bifurcated endografts.
The main reasons for exclusion were aneurysm angulation and
diameters outside the range of manufactured and commercially
available endografts. The authors, however, found that >90% of
patients referred to them were suitable for endografting, as they
have access to custom-made endografts that overcome the usual
contraindications for an "off-the-shelf" endograft. The exclusion
rate in our personal series of 350 procedures is >90% (PY Milne,
KR Thomson, unpublished data, July 1994 through 2004).
In Australia, where there is ready access to locally
manufactured Zenith grafts in both ready-made and customized
configurations, it is not surprising that Cook Inc. has, by far,
the largest share of the market.
The recognition that occlusion of the internal iliac artery is a
cause of significant morbidity from buttock claudication has led to
placement of endografts in ectatic iliac arteries with satisfactory
even though there is a theoretical risk of rupture of the ectatic
iliac segment. Coil occlusion of the internal iliac artery is also
associated with buttock claudication, particularly if the coils are
placed distally in the internal iliac artery.
A recent development is the iliac bifurcated endograft that
provides a branch of the iliac limb to supply the internal iliac
artery. This device is a forerunner of similar branched endografts
for use in the arch of the aorta and to treat thoracoabdominal
Preservation of the internal iliac artery removes the problem of
The need for a percutaneous device is less important now that it
has been recognized how diseased the common femoral artery is in
patients with AAA and how low the morbidity of a femoral cutdown
procedure is, particularly if an oblique incision is used.
Percutaneous placement of the contralateral limb is commonly
performed, as the insertion sheath for the contralateral limb is
usually ≤18F in size. A variety of suture closure devices have been
used to obtain hemostasis.
Good results of endografts for aortic transection
have been cited to extend the use of endografts for leaking AAA.
This has the potential to significantly reduce the mortality from
ruptured AAA but requires an "off-the-shelf" endograft system. Such
a system is used in Melbourne, Australia at the two Level I trauma
hospitals (The Alfred and The Royal Melbourne Hospitals) for
thoracic aortic injury, but the range of sizes required for AAA
makes this area more problematic.
In spite of some manufacturer's claims to the contrary, no
radiologist believes that a mobile image intensifier is as good as
a fixed angiographic unit, in the abdomen or other thick body
parts. However, usually for reasons of control and the remote
possibility of a need for open conversion, some vascular surgeons
still perform endografting of AAA in an operating room environment.
In our series, no acute conversions occurred from 1995 to 2003. One
acute conversion in 1994 was required for rupture during the
procedure, and a second rupture occurred at the common iliac artery
origin during a procedure in 2005 (PY Milne, KR Thomson,
unpublished data, 2005). Both conversions were performed in the
Apart from the imaging quality, the next most critical issues
are familiarity with the device and technical skill in angiographic
procedures. Care must be taken to prevent guidewires from reaching
the aortic arch during abdominal endovascular procedures for AAA,
as there is a risk of stroke, guidewire dissection, or perforation
by the tip of the guidewires.
If the device is measured incorrectly and is too short, an
extension piece may be added; but if the device is too long, it may
not be possible to engage the contralateral limb or the distal ends
may inadvertently cover the internal iliac artery on one or both
sides. When the neck is short and severely angulated, the choice of
the correct device is best determined by the patient's anatomy
(Figure 6). In patients in whom open surgery is impossible and in
whom the aneurysm sac is large enough, it is possible to "dump" the
endograft in the aneurysm sac and perform a second procedure, at a
later date, with the correct-sized graft (Figure 7). Once inserted
and deployed, it is almost impossible to remove an endograft by
interventional means. A type I endoleak at the conclusion of an
endograft placement is considered a failure, as the aneurysm is
still pressurized and has not been excluded.
If the internal iliac artery is to be covered by one of the
limbs of the endograft, internal iliac embolization is usually
performed. In some centers, this is done as a separate procedure to
reduce anesthetic time and iodinated contrast load.
Complex endograft procedures with significant manipulation of
the device and the use of multiple access catheters have a high
morbidity and, sometimes, a lethal rate of trash embolism.
Suprarenal loose atheromatous material is dangerous and should be
noted if it is present on the planning CT scan. Impotence is a rare
complication of endograft repair in males but is a good reason to
attempt to preserve both internal iliac arteries, if possible.
Management of endoleaks
An endoleak is defined as a communication from the arterial
circulation with the aneurysm sac (Table 2). Endoleaks may occur at
the time of deployment of the graft or may develop at a later date,
up to months or years following deployment. The need for
surveillance and management of endoleaks is the principal cost of
endografts compared with conventional open surgical repair.
However, not all sac enlargement or failure of shrinkage after
treatment is due to an endoleak; in such cases, a low-pressure
hygroma or seroma probably exists.
Imaging is used extensively to identify the presence or absence of
an endoleak and to confirm the shrinkage of the aneurysm sac. Color
Duplex ultrasound (CDU) and CT are the imaging modalities used for
The CT protocol should allow time for a small endoleak to fill with
contrast, so arterial and delayed imaging is required. Angiography
is usually indicated only when a leak has been identified by less
invasive means or when there is an unexplained enlargement of the
Type I endoleaks are usually large and obvious (Figure 8).
Dilatation of the aneurysm neck and migration of the endograft
distally or separation of the endograft limbs from the body of the
endograft can produce a large type I leak that may proceed rapidly
to rupture of the sac. Type II leaks, on the other hand, are more
subtle and harder to find. With CDU, the flow pattern in the
endoleak may indicate which endoleaks will seal spontaneously.
Endoleaks that are effectively a channel from one vessel to another
are less likely to close than those channels that are
end-connections to the aneurysm sac (Figure 9).
Carbon dioxide angiography is particularly sensitive for
detecting a small type II endoleak. More recently, it has been
found that MR angiography is 50% more sensitive than CTA for
detecting a small type II endoleak.
Unfortunately, this technique cannot be used for stainless steel
stented endografts, more because of the susceptibility artifact
than because of any real risk of displacement of the endograft
Persistent type II endoleaks result in pressurization of the
sac, and the contents remain liable to thrombolysis and variation
in the size of the channel within the thrombus. Type II endoleaks
can rupture if left untreated, and failure of shrinkage or
enlargement of the aneurysm sac is worthy of further investigation.
Where there is no easy transluminal access through lumbar or
inferior mesenteric arteries, a direct translumbar approach is
performed. It is then possible to catheterize the inlet and outlet
vessels selectively from the translumbar sheath (Figure 11).
Complete occlusion of the pathway is required, as a few coils in
the aneurysm sac will not promote permanent occlusion. Our
experience of thrombin injection into the aneurysm sac to close an
endoleak was unsatisfactory. Not all endoleaks need intervention,
as aneurysm shrinkage can occur even in the presence of an
endoleak. Such endoleaks are usually small and type II in
When the aneurysm sac does not decrease in size or even enlarges
slightly and there is no radiologic evidence of endoleak, a state
of endotension is said to occur. In cases in which the sac has been
punctured, clear or serous fluid has been found. A similar finding
has been reported after open repair using polytetrafluoroethylene
In our own series of >350 endograft repairs using Dacron-covered
stents, we have not found a case of endotension (PY Milne, KR
Thomson, unpublished data, July 1994 through 2004). However, others
are convinced of the existence of this condition, and it may be
related to the type of graft fabric used.
Studies comparing open surgical repair to endograft repair of
AAA have shown that both methods are effective in preventing
aneurysm rupture and that the late risk of rupture following
endografting is <1% per year. The 5-year survival after both
methods is approximately 70%.
Provided patient anatomy is suitable, the aneurysm-related death
rate appears to be lower following endovascular aneurysm repair
(EVAR); this is because of a lower perioperative mortality rate.
The major limitation of EVAR remains the need for ongoing
surveillance and the increased incidence of intervention in the
years following a successful treatment. Studies of the devices used
show fairly similar results for similar devices.
Failed grafts are most often managed successfully by further
The results of the Eurostar Registry
have resulted in a much better appreciation of the forces affecting
an endograft as the aneurysm thromboses and shrinks. Early devices
produced an excellent early result but poor long-term protection
because of endoleaks.
In addition to the forces exerted by the blood flow on the
endograft, the shrinkage of the sac after endografting causes quite
marked torsion force on the endograft. This appears to be the cause
of some limb separations seen in the Eurostar Registry, and the
lessons learned have influenced endograft design. The Eurostar
Registry results are available online and are updated regularly.
Following device deployment in short-necked aneurysms, we have
observed late device failure with detachment of the uncovered
suprarenal fixation from the covered trunk of the endograft. There
is no room for complacency in late follow-up of endografts, as the
Eurostar data shows a higher failure rate after 4 years
In the next few years, we shall see further developments,
including devices with biomaterials and devices that may not
require the extended and expensive imaging follow-up associated
with current endograft systems. However, because an endoleak can
cause pressurization and enlargement of the aneurysm sac and then
eventually result in an aneurysm rupture, an alternative follow-up
method is still required. The cost of current follow-up protocols
and the significant incidence of type II endoleaks are the most
significant challenges to be overcome. Otherwise, endografts for
AAA may become another discarded procedure.