Customize your JAMA Network experience by selecting one or more topics from the list below.
Therapeutic embolization is an accepted component in the management
of vascular tumors, arteriovenous malformations, and active bleeding sites.1 Complications are rare, but include infarction
by unintended occlusion of vessels.1 This
occurs either when particles are injected from a proximally placed catheter
or when they reflux from distal vessels during flushing.
Inadvertent occlusion of the ophthalmic or central retinal artery during
external carotid artery embolization has been reported in 4 cases,2-4 with immediate
loss of vision. We report a case in which stepwise decline of vision related
to progressive swelling of polyvinyl alcohol (PVA) particles lodged in retinal
arterioles occurred throughout a period of 8 days.
Report of a Case
A 39-year-old woman underwent preresection embolization of a left-posterior
frontal convexity meningioma. A preembolization arteriogram visualized the
middle meningeal branches to the tumor, as well as an anastomosis to the ophthalmic
artery via the sphenoidal artery (Figure 1 and Figure 2A).
Preembolization left external
carotid arteriogram showing middle meningeal artery (asterisk), parietal/ branch
(solid arrow) to tumor (blush of small vessels), sphenoidal artery (open arrow),
and ophthalmic artery (arrowhead).
Schematic view of tumor and retinal
embolization. A, Left middle meningeal branches, including supply to convexity
meningioma, and anastomosis to ophthalmic artery. B, Catheter injecting polyvinyl
alcohol (PVA) particles to occlude vessels in the tumor bed. It is positioned
in a meningeal branch distal to the branch that leads to the sphenoidal artery.
C, Catheter has been pulled back to proximal meningeal artery in preparation
for postembolization arteriogram. The pull-back is done to avoid reflux of
loose PVA particles lying in tumor vessels. Those PVA particles residing in
the microcatheter are inadvertently flushed into meningeal branches that anastomose
to sphenoidal, ophthalmic, and retinal arteries. These particles occlude retinal
arterioles and, as they swell, sequentially infarct the retina.
A 5F catheter was positioned in the left external carotid artery, and
through this catheter, a 2F microcatheter was advanced into the middle meningeal
artery supplying the lesion. The patient was systemically heparinized during
the procedure. The catheter was positioned distal to the take-off of the sphenoidal
artery (Figure 2B). To confirm that
the catheter was beyond all dangerous anastomoses, provocative testing was
performed by injecting amobarbital and lidocaine through the microcatheter.5 When this testing produced novisual or other neurological
deficits, 45- to 150-µm PVA particles suspended in radiographic contrast
were slowly injected under fluoroscopic guidance into the tumor (Figure 2B). After slowed flow to the tumor
was seen, the microcatheter was flushed with heparinized saline to evacuate
any remaining particles in the catheter. To prevent reflux of emboli that
had been deposited in the distal vessels supplying the tumor, the catheter
was then pulled back proximal to the sphenoidal artery take-off (Figure 2C), and contrast was injected to
confirm devascularization of the meningioma. As the contrast was injected,
the patient reported a sudden flash of light in the left eye, and she soon
developed an inferior scotoma on that side. Polyvinyl alcohol particles retained
within the catheter had evidently been flushed into the ophthalmic artery
anastomosis and reached the retinal arterioles (Figure 2C).
Ophthalmic examination performed 3 hours after the procedure disclosed
visual acuities of 20/25 OD and 20/25 OS, and a left afferent pupil defect.
Ophthalmoscopy revealed ischemic clouding of the retina superotemporally.
Small, yellow-white particles were visible in retinal arterioles of the superior
and inferior arcades, including the vessel that was occluded in the superior
arcade (Figure 3A). A formal visual
field test (Humphrey Field Analyzer Program 24-2; Humphrey Instruments, San
Leandro, Calif) disclosed a corresponding inferonasal scotoma (Figure 3B). Seven days later, the patient reported a further decline
in vision in her left eye. Ophthalmoscopy revealed enlargement of the superior
retinal infarct (Figure 4A) and
of the corresponding scotoma (Figure 4B).
The particle in the proximal inferior temporal arteriole had enlarged. Despite
4 mg of oral dexamethasone taken 3 times daily, she developed loss of the
superior field one day later. Visual acuity had fallen to 20/400; ophthalmoscopy
showed that the particle in the inferior temporal artery had further enlarged,
that the inferior retina was now infarcted (Figure 5A), and that the scotoma involved most of the visual field
(Figure 5B). Follow-up examination
2 months later showed a pale left optic disc and no change in visual function.
Fundus photographs and visual
fields of the left eye on the day of procedure. A, superior branch retinal
artery occlusion as ischemic whitening. Several yellow-white polyvinyl alcohol
particles are impacted within proximal retinal arterioles. B, Corresponding
inferior nerve fiber bundle defect.
Fundus photographs and visual
fields of the left eye 7 days after the procedure. A, Retinal infarction has
extended nasally. B, Enlargement of visual field defect, corresponding to
enlarged area of infarction.
Fundus photographs and visual
fields of the left eye 8 days after the procedure. A, Retinal infarction now
also involves the inferior branch artery. The particle lodged in the proximal
inferior temporal retinal arteriole has enlarged. B, Visual field defect now
involves the upper field, corresponding to new infarction of the inferior
Our patient had a rare complication of therapeutic PVA embolization,
namely, inadvertent occlusion of branches of the central retinal artery. This
complication is based on known anastomoses between the external carotid circulation
and the ophthalmic artery.6-8
External carotid branches may replace the internal carotid artery as the predominant
supply of the ophthalmic artery in 5% of patients with otherwise normal cephalic
vessels7 and in a much higher proportion
in those with arteriosclerosis, which preferentially affects the internal
Aware that this anastomosis might cause a serious problem, our interventional
neuroradiologist (J.P.D.) positioned the catheter far distal to it, performed
preembolization pharmacologic provocative testing, embolized gently without
repositioning the catheter, and flushed the catheter after embolization. However,
these techniques failed to prevent errant emboli from reaching the retinal
vessels. Some particles were undoubtedly present in the catheter even after
careful flushing with saline, suggesting that it may not be possible to completely
remove all particles from the microcatheter.
Four additional measures might have prevented this complication. First,
embolization could have been performed with particulate agents too large to
enter the anastomotic pathway, but there are 2 drawbacks to this maneuver.
The first is that the size of anastomoses and particles vary, making it impossible
to be sure that some particles will not enter the anastomosis. The second
drawback is that large particles are less effective in devascularizing the
tumor bed.9 The second measure would have
been to selectively occlude the anastomosis with a microcoil prior to embolization
of the tumor bed. This would have prevented small particles from entering
the anastomosis, but this is not feasible if the anastomosis is too small
or tortuous to be selectively catheterized. A third preventive measure would
have been to remove the microcatheter used for injecting emboli, and to then
introduce a new microcatheter to perform the postembolization arteriographic
studies. The drawback to this measure is that if further embolization of the
tumor through a distal site is found to be necessary, additional time and
risk would be added to the procedure. Fourth, we could have injected contrast
dye during the follow-up arteriogram through the guiding catheter positioned
in the external carotid artery, rather than through the microcatheter. This
would have avoided the need to manipulate the microcatheter and would have
virtually eliminated the risk of injecting retained PVA emboli. The main disadvantage
of this otherwise practical solution is that a guiding catheter must have
a large enough inner lumen to allow passage of the inner microcatheter and
leave enough room around it to inject viscous contrast material. Perhaps a
larger 6F guiding catheter would have been preferable in this case.
An unusual aspect of our case is the sequential nature of the branch
retinal artery occlusions occurring throughout a period of 8 days. We attribute
this phenomenon, (not previously reported) to progressive swelling and/or
thrombus formation in these vessels. Swelling is a feature that makes PVA
a useful agent for therapeutic embolization, but that increases the possibility
that errant PVA particles will cause neurological deficits. Unfortunately,
there is no antidote to reverse the swelling of the particles.
Although our case has unusual features, we doubt that inadvertent occlusion
of the retinal artery during preoperative embolization via the external carotid
system is as rare as the published literature would suggest. An authoritative
textbook10 describes 2 unpublished cases.
In presenting this case to gatherings of ophthalmologists and neuroradiologists,
many members of our audiences have told us of similar experiences.
Corresponding author: Jonathan D. Trobe, MD, Kellogg Eye Center,
1000 Wall St, Ann Arbor, MI 48105 (e-mail: firstname.lastname@example.org).
Turner T, Trobe JD, Deveikis JP. Sequential Branch Retinal Artery Occlusions Following Embolization of an Intracranial Meningioma. Arch Ophthalmol. 2002;120(6):857–860. doi:
Artificial Intelligence Resource Center