Featured Review

Central retinal artery occlusion

Sohan Singh Hayreh

The pathogeneses, clinical features, and management of central retinal artery occlusion  (CRAO) are Access this article online
discussed. CRAO consists of the following four distinct clinical entities: non‑arteritic CRAO  (NA‑CRAO), Website:
transient NA‑CRAO, NA‑CRAO with cilioretinal artery sparing, and arteritic CRAO. Clinical characteristics, www.ijo.in
visual outcome, and management very much depend upon the type of CRAO. Contrary to the prevalent DOI:
belief, spontaneous improvement in both visual acuity and visual fields does occur, mainly during the 10.4103/ijo.IJO_1446_18
first 7  days. The incidence of spontaneous visual acuity improvement during the first 7  days differs PMID:
*****
significantly (P < 0.001) among the four types of CRAO; among them, in eyes with initial visual acuity of
counting finger or worse, visual acuity improved, remained stable, or deteriorated in NA‑CRAO in 22%, 66%, Quick Response Code:
and 12%, respectively; in NA‑CRAO with cilioretinal artery sparing in 67%, 33%, and none, respectively; and
in transient NA‑CRAO in 82%, 18%, and none, respectively. Arteritic CRAO shows no change. Recent studies
have shown that administration of local intra‑arterial thrombolytic agent not only has no beneficial effect but
also can be harmful. Investigations to find the cause and to prevent or reduce the risk of any further visual
problems are discussed. Prevalent multiple misconceptions on CRAO are discussed.

Key words: Central retinal artery occlusion, retinal artery occlusion, retinal arteries, retinal vessels

Central retinal artery occlusion (CRAO) has been known as a CRAO  (NA‑CRAO), except for the presence of a patent
clinical entity since 1859, when von Graefe[1] first described CRAO cilioretinal artery [Fig.  1], which can have a marked
due to embolism. After that, Schweigger[2] in 1864 described it on influence on the visual outcome and retinal circulation
ophthalmoscopy. It is an ophthalmic emergency because of instant, (3) Transient NA‑CRAO: In this, occlusion of the CRA may vary
massive visual loss. A voluminous literature has accumulated from several minutes to many hours, depending upon the
on its various aspects, riddled with serious controversies and cause of occlusion [Fig. 2]. The visual outcome in this type of
misconceptions, particularly about its management. CRAO can be totally different from any of the other types, and
it depends entirely upon the duration of transient CRAO
Since 1955, I have investigated various aspects of CRAO
(4) Arteritic CRAO: In this condition, giant cell arteritis is
comprehensively, by anatomical studies on the central retinal
the cause of development of permanent CRAO  [Fig.  3].
artery (CRA),[3,4] experimental studies,[5‑10] and multiple clinical
My studies on giant cell arteritis have shown that these
studies[11‑23] on its different features. Recently, I discussed in
eyes almost invariably have associated arteritic anterior
detail my findings on CRAO, along with a detailed review of
ischemic optic neuropathy.[24,25] Clinically, these eyes have
the literature on CRAO, in my book, entitled “Ocular Vascular
the classical fundus findings of CRAO, except for associated
Occlusive Disorders.”[23] Following is a very brief account of
optic disk edema due to arteritic anterior ischemic optic
various clinical aspects of CRAO.
neuropathy.
Classification of CRAO In my study of 260 eyes with CRAO,[14] 67% had NA‑CRAO,
CRAO has always been considered as one disease. However, 16% transient NA‑CRAO, 14% NA‑CRAO with cilioretinal
my prospective study of 244 consecutive patients (260 eyes) artery sparing, and 4% arteritic CRAO.
with CRAO showed that in fact it consists of the following four It is important to note that clinical characteristics, visual
distinct clinical entities:[14] outcome, and management very much depend upon the type
(1) Non‑arteritic CRAO: This shows the classical clinical picture of CRAO.
of permanent CRAO with retinal infarction, cherry red spot
and retinal arterial changes [Fig. 1], absent or poor residual Demographic characteristics of various types of CRAO
retinal circulation on fluorescein fundus angiography, and In my study of 244  patients,[14] the mean (range) was: In
no evidence of giant cell arteritis NA‑CRAO mean 68 (26–90) years, in transient NA‑CRAO mean
(2) Non‑arteritic CRAO with cilioretinal artery sparing: This
has the classical clinical picture of permanent non‑arteritic
This is an open access journal, and articles are distributed under the terms of
the Creative Commons Attribution‑NonCommercial‑ShareAlike 4.0 License,
which allows others to remix, tweak, and build upon the work non‑commercially,
Department of Ophthalmology and Visual Sciences, College of
as long as appropriate credit is given and the new creations are licensed under
Medicine, University of Iowa, Iowa City, Iowa, USA the identical terms.
Correspondence to: Dr.  Sohan Singh Hayreh, Department of
Ophthalmology and Visual Sciences, University Hospitals and For reprints contact: [email protected]
Clinics, 200 Hawkins Drive, Iowa City, Iowa  ‑  52242‑1091, USA.
E‑mail: sohan‑[email protected] Cite this article as: Hayreh SS. Central retinal artery occlusion. Indian J
Ophthalmol 2018;66:1684-94.
Manuscript received: 27.08.18; Revision accepted: 27.09.18

© 2018 Indian Journal of Ophthalmology | Published by Wolters Kluwer - Medknow

 Hayreh: Central retinal artery occlusion
December 2018 1685

63 (20–89), in NA‑CRAO with cilioretinal artery sparing mean narrowest part of the CRA, where it enters the sheath of the
67 (39–87) years, and in arteritic CRAO mean 74 (62–87). This optic nerve [Figs. 4 and 5][3,4] (not at the lamina cribrosa, as
shows that although CRAO usually develops in older persons, is often erroneously described). The emboli originate from
it does occur in children and young persons also.[23] There is no plaques in the carotid artery or the heart[19,20] [Table 1]; rarely,
evidence that race or any other demographic parameter make CRAO is due to vasculitis, chronic systemic autoimmune
any difference in CRAO development. diseases, or thrombophilia.[23]
2. Transient NA‑CRAO is most often caused by a migrating
Causes of Various Types of CRAO embolus, and sometime by a transient marked fall of
1. Classical NA‑CRAO is most common due to permanent perfusion pressure in CRAO,[14] or high rise of intraocular
occlusion of the CRA, caused by an impacted embolus at the pressure.
3. Arteritic CRAO is due to thrombosis of the common trunk
of the posterior ciliary artery and CRA arising from the
ophthalmic artery[24] [Fig. 6], caused by giant cell arteritis,
not of CRA per se.

Systemic Conditions Playing a Role in

Development of CRAO
These systemic conditions include the following:
1. H i g h c h o l e s t e r o l   ( p a r t i c u l a r l y l o w ‑ d e n s i t y
lipoproteins  –  “bad cholesterol”) leads to the build‑up
of plaques on the walls of carotid arteries, which are the
common source of embolism causing CRAO

Figure 1: Fundus photograph of the left eye with non‑arteritic

CRAO (NA‑CRAO), and two patent cilioretinal arteries (arrows). It
shows cherry‑red spot, retinal opacity of posterior fundus – most marked a b
in the macular region, cattle‑trucking in the retinal vessels
Figure 2: Fundus photograph and fluorescein fundus angiogram
of left eye with transient NA‑CRAO.(a) Fundus photograph shows
large number of cotton wool spots, maximum in the macular region.
(b) Fluorescein angiogram, during retinal arterial phase, shows almost
normal but slightly sluggish retinal circulation, except for the absence
of filling in the foveal region, and cotton wool spots at places mask the
background choroidal fluorescence

a b

c
Figure 3: Right eye with arteritic CRAO and arteritic anterior ischemic
optic neuropathy. (a) Fundus photograph shows chalky‑white optic disk
edema, some mild cattle‑trucking in the retinal vessels, and cherry red Figure 4: Schematic representation, showing the course of central
spot with perifoveolar retinal opacity.(b) Fluorescein fundus angiogram, retinal artery and its branches and anastomoses. A = Arachnoid;
during the early phase, shows posterior ciliary artery occlusion with C = choroid; CRA = central retinal artery; Col. Br. = Collateral branches;
a few small patches of choroidal filling, and early filling of the central CRV = central retinal vein; D = dura; LC = lamina cribrosa; ON = optic
retinal artery.(c) Fluorescein fundus angiogram during the late phase nerve; PCA = posterior ciliary artery; PR = prelaminar region; R = retina;
shows cattle‑trucking in the retinal vessels, with almost no disk staining S = sclera; SAS = subarachnoid space
1686 Indian Journal of Ophthalmology Volume 66 Issue 12

Table 1: Carotid Doppler/Angiography and

echocardiogram findings on the side with central retinal
artery occlusion[19]
Test finding Non‑arteritic central retinal
artery occlusion
Patent foramen ovale (n=234)
6 (2%)
Carotid Doppler/angiography: (n=199)
Occlusion
0‑15% 109 (55%)
16‑49% 23 (12%)
50‑79% 32 (16%)
80‑99% 13 (7%)
100% 22 (11%)
Carotid Doppler/Angiography (n=187) Figure 5: A transverse section of the human optic nerve, showing the
Plaque present 133 (71%) central retinal artery (CRA) inferiorly lying within the substance of dural
Echocardiogram (n=131) sheath of the optic nerve (ON)
Normal 39 (30%)
Abnormal, no embolic source 24 (18%)
6. There are several controversial reports suggesting that
Abnormal, with embolic source 68 (52%)
thrombophilia plays a role in the development of CRAO[23]
Echocardiogram: (of n=68 with embolic source)
7. Rarely, CRAO has been seen in association with hematological
Abnormal embolic source
Mitral valve 18 (26%) disorders, for example, sickle cell hemoglobinopathies,
Aortic valve 26 (38%) leukemia, systemic non‑Hodgkin’s lymphoma, and orbital
Mitral and aortic valves 24 (36%) lymphoma[23]
Echocardiogram: (of n=42 with mitral valve as 8. CRAO has also been reported perioperatively following
Mitral valve lesion types embolic source) a variety of surgical procedures, for example, following
orbital, eye or head injuries, facial and retrobulbar injections,
Calcified valve 24 (57%) peribulbar anesthesia, and intraocular gas injection used as
Mitral valve prolapse 7 (17%) tamponade for rhegmatogenous retinal detachment[23]
Other types of lesions 11 (26%) 9. CRAO following hemodialysis is known. It is well known
Echocardiogram: (of n=50 with aortic valve as that during hemodialysis, there is often a marked fall of
embolic source) blood pressure. Moreover, these patients almost invariably
Aortic valve lesion types
have marked vasculopathy[23]
Calcified valve 39 (78%)
Other types of lesions 11 (22%)
10. Other rare miscellaneous conditions associated with CRAO
include Fabry’s disease, incontinentia pigmenti, oral
contraceptives, usage of cocaine, snake bite,[18] migraine,
2. Carotid artery disease can produce CRAO by the following Marfan’s syndrome, nephrotic syndrome, and others.[23]
three mechanisms:
(a) Embolism, which is by far the most common cause of I have discussed all these conditions and reviewed literature
CRAO[19] at length, elsewhere.[23]
(b) A significant stenosis (about 70% or more) or complete
occlusion of the internal carotid artery, by markedly
Clinical Features
reducing the ocular blood flow, can result in development Symptoms
of CRAO. In my study,[19] >80% stenosis of the internal Typically, there is a sudden, massive loss of vision in the
carotid artery was seen in 18% of CRAO cases [Table 1] involved eye. It can occur at any time of the day. In my study
(c) Our study on atherosclerotic monkeys showed that of 260 eyes with CRAO,[14] visual loss was discovered on
serotonin, a powerful vasoconstrictor, released by waking up in the morning in 35% in NA‑CRAO, in 29% in
platelet aggregation on atherosclerotic plaques in the NA‑CRAO with cilioretinal artery sparing, in 39% in transient
carotid artery, produces a transient spasm, which can CRAO, and in 30% in arteritic CRAO. When the visual loss
cause transient, complete occlusion, or impaired blood is discovered on waking up in the morning, it may be due to
flow in the CRA[26] [Fig. 7] embolism, thrombosis, or due to transient CRAO from a fall of
3. The association of diabetes and arterial hypertension with perfusion pressure during sleep caused by nocturnal arterial
CRAO has been reported by a large number of studies[23] hypotension[14] [Fig.  8]. The visual loss may be preceded by
4. A variety of cardiac conditions can produce CRAO due history of amaurosis fugax, caused by transient migrating
to embolism from the heart[19] [Table 1]. In addition, there embolism or giant cell arteritis.[25] I found a history of amaurosis
are reports of development of CRAO following invasive fugax before development of CRAO in 12% in NA‑CRAO, in
cardiovascular procedures[23] 20% in NA‑CRAO with cilioretinal artery sparing, in 30% in
5. CRAO associated with vasculitis, giant cell arteritis, arteritic CRAO, and in 13% in transient CRAO.[14] In patients
and chronic systemic autoimmune diseases has been aged 50 years or older with a history of amaurosis fugax,[25]
reported[23] the first essential is to rule out giant cell arteritis, which is an
 Hayreh: Central retinal artery occlusion
December 2018 1687

a b

c d
Figure 7: Fluorescein fundus angiograms (a, c, d) and a fundus
photograph (b) of right eye of an atherosclerotic monkey before and
after intravenous infusion of serotonin.(a) Baseline angiogram during
retinal arteriovenous phase, showing normal filling of the central retinal
artery, the retinal vasculature, and the choroid, one week before (b, c).
(b, c) During infusion of serotonin (b) is a black and white fundus
photograph about 2 min after the start of serotonin infusion, showing
evidence of CRAO, including pale optic disk, retinal opacity, and
cherry red spot at fovea. (c) An angiogram about 4 min, after the start
of serotonin infusion, shows normal filling of the choroidal circulation
but complete occlusion of the central retinal artery. (d) An angiogram
during regression phase shows once again normal filling of the central
retinal artery and the choroid

Initial visual acuity

Initially, visual acuity in NA‑CRAO was 20/200–20/400 in 7%
and counting fingers or worse in 93%. In transient NA‑CRAO,
it depended upon the duration of CRAO, and in NA‑CRAO
Figure 6: An example of a common trunk of origin of central artery with cilioretinal artery sparing upon the area of distribution of
of the retina and posterior ciliary arteries arising from the ophthalmic the patient cilioretinal artery – in both, it varied from 20/40 or
artery, as seen from below. CAR = Central artery of the retina, better to counting fingers or worse; in arteritic CRAO, it varied
LPCA = lateral posterior ciliary artery, MPCA = medical posterior from 20/200 to counting fingers or worse.[14]
ciliary artery, OA = ophthalmic artery, ON = optic nerve. PPS = point
of penetration into the sheath by the central retinal artery.*Common Initial visual field defects
trunk of origin of CAR and MPCA Visual acuity essentially represents the function of the foveal
region and not of the rest of the retina; only visual fields plotted
ophthalmic emergency, because if it is not treated immediately, with a Goldmann perimeter provide full information about the
it can result in massive visual loss which can involve both eyes. function of the entire retina – after all, CRAO involves the entire
Some patients complain of a purplish hue to the blur, which retina, not only the fovea. Unfortunately, there has been little
is highly suggestive of retinal ischemia. Hallucinations have information on the visual field defects in CRAO in the literature.
been reported with severe vision loss associated with CRAO.[23] My study[14] was the first to evaluate visual fields in 145 CRAO
Simultaneous onset of CRAO in both eyes does not normally eyes, where it was possible to plot them. It provided invaluable
occur; it happens rarely, when there is compression of both eyes information on visual function in CRAO. Initially, there was
during a prolonged surgical procedure or during hemodialysis. a complete loss of the central 30° visual field  (producing
central scotoma) in NA‑CRAO in 48%, in NA‑CRAO with
Visual Function in CRAO cilioretinal sparing in 22%, and with transient NA‑CRAO in
4% only. Initially, peripheral fields were normal in 22% with
The devastating visual loss and its management is a major NA‑CRAO [Fig. 9] and in 63% with transient CRAO.
topic in CRAO. Therefore, from the functional point of view,
the most important consideration, both for the patient and Natural history of visual loss in CRAO
the ophthalmologist, is the visual outcome and natural I investigated the natural history of visual outcome in CRAO in
history. In view of that, it requires a detailed discussion of this group of patients.[14] It is widely believed that CRAO results
various aspects of visual function in CRAO. My prospective in permanent visual loss, with no chance of any spontaneous
study[14] in 260 CRAO eyes provided this information for visual improvement. But my study showed that not to be true,
the first time. as is evident from the following.
1688 Indian Journal of Ophthalmology Volume 66 Issue 12

Visual Acuity Improvement: Table 2 gives detailed good peripheral visual fields [Fig. 9], as in age‑related macular
information about the initial and final visual acuities in various degeneration, can lead a reasonably satisfactory functional
types of CRAO. In eyes seen within 7 days and with counting life although unable to read or do fine work. Unfortunately,
fingers visual acuity, it improved by 22% in NA‑CRAO, by 67% the common use of automated static threshold perimetry
in NA‑CRAO with cilioretinal artery sparing, and by 82% in nowadays, rather than manual kinetic perimetry performed
transient NA‑CRAO, but in none in arteritic CRAO. with a Goldmann perimeter, does not provide full information
about the peripheral visual fields, because automated perimetry
Visual Field Improvement: Central visual field improved in
does not provide information beyond the central 30° or less
39% of transient NA‑CRAO, 25% of NA‑CRAO with cilioretinal
of the field, whereas kinetic perimetry covers the peripheral
artery sparing, and 21% with NA‑CRAO. The peripheral visual
visual field up to about 90°.
field improved in NA‑CRAO in 39%; in transient NA‑CRAO in
39%, and in NA‑CRAO with cilioretinal artery sparing in 25%. In conclusion,  (i) the initial and final visual acuities and
Interestingly, in transient NA‑CRAO with initial central and visual field defects differ significantly (P  <  0.001) among
peripheral visual fields field defects, those recovered to normal the four types of CRAO,  (ii) significant improvement can
in 26% and 30%, respectively. In addition, in NA‑CRAO, occur in visual acuity and visual field without treatment,
peripheral visual fields initially were normal in 22%, and on and (iii) classification of CRAO is crucial for understanding
follow‑up, these improved in 39% of the remaining eyes. In differences in visual outcome.
other words, 61% of eyes with NA‑CRAO finally had normal
Factors play crucial roles in determining the visual outcome
to good peripheral visual fields, which is of great importance
in visual function, as discussed below. in CRAO
Understanding the fundamental facts behind the factors that
Importance of peripheral visual fields for visual function play a crucial role in determining the visual outcome in CRAO
Central visual field loss results in impaired central vision, while is essential. These include the following:
the peripheral visual field is essential for day‑to‑day living and 1. Duration of acute retinal ischemia: This is by far the
navigation; therefore, a person with central visual loss but with most important factor. My experimental study [9] of

Figure 8: A 24‑h ambulatory blood pressure recording in a woman,

starting at about 11:00 a.m. and ending at about 10:00 a.m. next day.
Note that during the waking hours the blood pressure is perfectly Figure 9: Visual fields of right eye with NA‑CRAO, plotted with a
normal but shows a marked drop during the sleeping hours (nocturnal Goldmann perimeter, show a large absolute central scotoma, with
arterial hypotension) almost normal peripheral visual field with I‑4e and V‑4e

Table 2: Initial and final visual acuity by CRAO type

Type of CRAO Visual 20/15 20/20 20/25 20/30 20/40 20/60 20/70 20/80 20/100 20/200 20/400 CF HM LP NLP Total*
acuity
Non‑arteritic CRAO Initial 1 1 6 13 66 44 28 12 171
Final 1 1 9 14 43 23 19 12 122
Non‑arteritic CRAO with Initial 1 4 2 3 3 1 1 2 9 2 35
cilioretinal artery sparing Final 2 2 3 3 6 2 4 6 1 29
Arteritic CRAO Initial 1 1 2 2 7 13
Final 1 1 1 8 11
Transient Non‑ arteritic Initial 2 9 4 2 2 2 1 3 2 10 4 41
CRAO Final 3 8 4 5 3 2 1 1 3 3 5 38
*The differences in the total number of eyes for initial and final visual acuity are due to some patients having only one visit
 Hayreh: Central retinal artery occlusion
December 2018 1689

CRAO in 38 elderly, atherosclerotic, hypertensive rhesus In conclusion, all these factors play important roles in the
monkeys  (similar to most patients with CRAO) showed visual outcome of CRAO and require evaluation. Therefore, it
that the retina suffers no detectable damage with temporary is essential to classify CRAO into its different types to make
CRAO if the circulation is restored within 97 min, but, after scientifically accurate observations about the visual outcome,
that, the longer the ischemia lasts, the more extensive is the which, as shown above, can be very different in the various
permanent damage. CRAO lasting for about 240 min results types.
in massive, irreversible retinal damage. In eyes where the
Misconceptions prevalent about the visual outcome in CRAO
retinal circulation was restored to normal after CRAO of
more than 2 but less than 4 h duration, retinal function did There are several misconceptions about it; they are discussed
not show signs of major improvement until many hours or at length elsewhere.[15] These include the following.
even a day after restoration of circulation – the longer the 1. There is an almost universal impression that CRAO causes
ischemia, the longer the lag before any improvement of marked, generalized loss of vision and that CRAO cannot
function started. This is a key finding clinically, because it produce central scotoma only, with intact peripheral visual
has unfortunately fostered the mistaken notion that visual fields. As discussed above, although central scotoma is
recovery can occur in eyes with CRAO lasting for much the most common visual field defect in CRAO, the eye
longer than 4 h or even for days with CRAO can still have fairly normal peripheral visual
2. Type of CRAO: As shown above, this plays an important role fields [Fig. 9].
3. The cause of CRAO: This can also play a role. Although 2. There is no chance of an eye with CRAO having spontaneous
embolism is a far more common cause of CRAO than visual improvement. My study findings  (above) showed
thrombosis,[19,20] rarely, when CRAO is caused by giant cell that this is not true
arteritis, vasculitis, trauma, or other conditions causing 3. That visual acuity testing gives required information about
thrombosis of CRA, there is poor visual outcome the visual outcome and function in CRAO. As mentioned
4. Site of occlusion in the CRA: As mentioned above, it is above, visual acuity essentially represents the function of
generally believed that the site of occlusion in CRAO is at the the foveal region and not the rest of the retina. By contrast,
level of the lamina cribrosa (as revealed by histopathologic visual fields plotted with a Goldmann perimeter provide
studies of enucleated eyes). Since the narrowest lumen of the information about the function of the entire retina. For
artery is where it pierces the dura mater of the optic nerve day‑to‑day living and navigation, peripheral visual fields
sheath[3,4] [Figs. 4 and 5], in cases of CRAO due to embolism, provide the essential information
the chances of an embolus getting impacted at this site are 4. When evaluating visual acuity improvement, one has to
much higher than at any other site in the artery. In contrast to keep in mind that apparent visual acuity improvement
that, the site of occlusion in thrombosis of the CRA is at the may simply be due to eccentric fixation, because patients
lamina cribrosa, as shown by the histopathological studies. with a central scotoma often learn with experience to
The site of occlusion is an important factor in determining fixate eccentrically, resulting in an apparent visual
the amount of available anastomoses by the CAR [Fig. 4] acuity improvement, which does not represent a genuine
and residual retinal circulation in CRAO improvement in the retinal function. Obviously, this
5. Residual retinal circulation: Fluorescein angiography almost important artifact can lead to a mistaken claim of visual
always shows this in fresh CRAO.[5,14,20] That has erroneously improvement following a therapy. Therefore, for genuine
been claimed as a sign of only partial occlusion of the CRA visual acuity improvement, there must be simultaneous
and used as a justification for claims of visual improvement improvement in both the central scotoma and visual
for many advocated therapies, even when therapy is acuity.
instituted many hours or even days after the onset of CRAO.
I specifically investigated this issue experimentally in 101 Anterior Segment of the Eye
eyes of rhesus monkeys.[5,9] In these experimental studies, the
In the anterior segment of the eye, there is usually no specific
CRA was cut, but there was still residual retinal circulation,
abnormality related to the CRAO. Initially, the intraocular
which clearly showed that there was no partial occlusion of
pressure often tends to be slightly lower than in the fellow
CRA; there was no correlation between the residual retinal
normal eye, similar to what is seen with retinal vein occlusion,[27]
circulation and recovery of visual function.[20] My studies
but, unlike in the latter, the lower pressure does not last long.
showed that the duration of CRAO is almost always the
When CRAO is due to ocular ischemic syndrome, the eye may
principal determining factor in the production of irreversible
develop iris and angle neovascularization;[12,13] Otherwise,
retinal damage, that is, the longer the ischemia, the more
CRAO eyes do not develop any kind of neovascularization.
marked the retinal damage.[9] The only exception is when
there is a large patent cilioretinal artery, which seems to have There is a prevalent belief that CRAO can cause anterior
some protective effect, particularly in the macular retina segment neovascularization and neovascular glaucoma,
6. Presence and area of supply by a patent cilioretinal artery: This similar to that seen following ischemic central retinal vein
can have a major impact on the visual outcome in CRAO, occlusion.[15] It is wrong to consider ischemic central retinal
as shown above. Importantly, in CRAO when the supply vein occlusion and CRAO as being similar in nature, as far as
by the cilioretinal artery just touches the foveola [Fig. 10], the development of neovascular glaucoma is concerned. My
initially there is a marked decline in visual acuity, but large studies[12,13] on CRAO have shown no cause‑and‑effect
within 2–3  weeks, the visual acuity improves markedly, relationship between CRA and neovascular glaucoma, thereby
spontaneously (to almost 6/6) which may erroneously be contradicting claims that CRAO results in development of
attributed to whatever treatment is being administered.[15] neovascular glaucoma.
1690 Indian Journal of Ophthalmology Volume 66 Issue 12

Posterior Segment of the Eye

Ophthalmoscopic findings in CRAO
Nettleship[28] in 1891 published a detailed and comprehensive
description of the acute ophthalmoscopic appearance of CRAO
in the English literature. He stated: “The classical dense, white
haze of the central region of the retina with a well‑marked
clear patch (‘cherry red spot’) at the yellow‑spot was very
well shown; there were no hemorrhages; the arteries and
veins were of about the normal size, but no pulsation could
be produced in any of them by pressure with the finger upon
the globe.” He went on to describe the appearance of the blood
vessels: “The second feature of interest … is the stagnation of
the arterial blood stream without diminution in the size of the
arterial column … the large size of the vessels appeared to be
inconsistent with the conclusion that their supply was cut off.
This phenomenon is by no means uncommon in recent cases
of embolism or thrombosis of the arteria centralis, and,… is
quite compatible with almost, if not quite complete extinction
of sight; on the other hand in some cases the arteries are, as is
Figure 10: Fundus photograph of the right eye with NA‑CRAO, and well known, found to be extremely shrunken from the first.”
a patent cilioretinal artery in the macular region. The retinal supply by He also gave an excellent diagram of cilioretinal collaterals on
the patent cilioretinal artery passes through the foveal region (arrow) the surface of the optic disk seen in a patient 12 months after
development of CRAO.
I investigated prospectively the incidence, pattern, and
evolution of fundus changes in 248 consecutive eyes with
CRAO,[16] as well as experimentally induced CRAO in 101 eyes
in rhesus monkeys.[5,6,9]
There is a misunderstanding that retinal whitening, due to
infraction of the inner retinal layers, must involve the entire
retina to justify the diagnosis of CRAO. In fact, it is mostly
a b
located in the posterior pole, maximum within the temporal
arcades, with normal‑looking peripheral fundus [Fig.  11a].
Figure 11: Fundus photograph and fluorescein fundus angiogram of
There is a widespread impression that the “cherry red spot” is
left eye four day after the onset of transient NA‑CRAO. (a) Fundus
due to the red color of the normal underlying choroidal blood.
photograph shows retinal opacity of posterior fundus – most marked in
the macular region, a small cherry red spot, a few retinal hemorrhages That is not true. The color of the normal fundus is due to the
a small patent cilioretinal artery (arrow), and normal‑looking peripheral color of the retinal pigment epithelium, not the blood in the
retina.(b) Fluorescein fundus angiogram, during the retinal arteriovenous choroid. Fundus in black people is gray, but they do not have
phase, shows slightly delayed filling of the retinal arteries and veins, black blood! The retinal opacity usually starts to resolve in
with central macular region showing no retinal capillary filling due to about a week and generally resolves completely within about
no‑reflow phenomenon caused by thick ischemic retina in that region a month or so, and the retina regains its transparency.
There is no universally consistent pattern of retinal vessels
changes. During the initial stages of CRAO, their appearance
varies markedly, being extremely attenuated to normal‑looking.
Cattle‑trucking is seen in both retinal arteries and veins during
the first week after onset of CRAO [Fig. 1].
Initially, the optic disk may show pale or hyperemic edema,
and later on, there is optic atrophy. Cilioretinal optic disk
collaterals [Fig. 12] may develop in 18% within 3 months or
later, and they may be misdiagnosed as disk neovascularization.
Because of this erroneous impression, some authors have
claimed that CRAO results in optic disk neovascularization,
a b but my detailed studies[12,16] clearly demonstrated that there is
Figure 12: (a) Fundus photograph of the left eye with multiple cilioretinal no such cause‑and‑effect relationship.
collaterals on and around the optic disk in an eye with old NA‑CRAO.
Fluorescein fundus angiographic findings in CRAO
(b) Fluorescein fundus angiogram of the right eye, during retinal arterial
phase, with multiple cilioretinal collaterals on and around the optic disk It must be performed in all CRAO eyes, because it helps to
in an eye with old NA‑CRAO. Note that, unlike neovascularization, the distinguish various types of CRAO and determine the extent
collateral shows no fluorescein leak of fundus circulatory abnormalities. NA‑CRAO eyes almost
 Hayreh: Central retinal artery occlusion
December 2018 1691

Figure 13: Fluorescein fundus angiogram of the left eye of a rhesus Figure 14: Fluorescein fundus angiogram of the left eye of a patient
monkey, immediately after experimentally cutting of the central retinal with NA‑CRAO, 29 s after injection of the dye, shows normal filling of
artery at its site of entry into the optic nerve. It shows slow filling of the the choroid and optic disk vessels (supplied by the posterior ciliary
central retinal artery and the retinal vascular bed, in spite of cutting of artery circulation), but no filling of the central retinal artery at all as yet.
the central retinal artery The retinal vessels show typical “cattle‑trucking”

always show sluggish filling of the retinal vasculature, and a and consequently in permanent ganglion cell death in the
variable amount of residual retinal circulation [Fig. 13], and it nonperfused retina; the area of central retinal capillary
is rare to see complete absence of retinal vascular filling in these non‑filing may vary from eye to eye [Figs.  2b and 11b],
eyes [Fig. 14]. I have discussed the mechanism of that residual depending upon the severity of retinal swelling in the macular
filling elsewhere.[15,20] The presence of a variable amount of region. This results in the variable size of the permanent central
residual retinal circulation has resulted in a mistaken belief scotoma.
among a majority of ophthalmologists that the CRA is not
completely occluded in CRAO. Transient CRAO shows normal There is marked improvement in retinal circulation filling
or almost normal retinal circulation [Figs.  2b and 11b]. In in CRAO eyes with time  –  the time taken and the extent of
CRAO with cilioretinal artery sparing, fluorescein angiography this restoration vary markedly from eye to eye, depending
is the only way to definitely outline the region supplied by upon the site of occlusion and availability and number of CRA
the patent cilioretinal artery [Fig. 15], and that is important anastomoses [Fig. 4].[3,4]
for determining the visual outcome. In arteritic CRAO, the
key diagnostic feature is the presence of associated posterior
Combined CRAO and Central Retinal Vein
ciliary artery occlusion[24] [Fig. 3b] because, as discussed above, Occlusion
giant cell arteritis causes thrombosis of the common trunk of There are several reports of this in the literature. The basic
posterior ciliary artery and CRA arising from the ophthalmic question to ask is: Is the simultaneous development of these
artery [Fig.  6]. Missing a diagnosis of arteritic CRAO and two very different clinical entities purely coincidental or do
consequently of giant cell arteritis can result in catastrophic
they represent one disease? CRAO is usually embolic in origin,
visual loss in both eyes; thus, fluorescein angiography plays
while central retinal vein occlusion is always a thrombotic
a crucial role in the diagnosis of arteritic CRAO and giant cell
disorder; therefore, there is no reason why two very different
arteritis. It must, therefore, be performed in all CRAO patients
processes should occur simultaneously. My clinical, basic,
50 years and older.
and experimental studies on central retinal vein occlusion
During the acute phase of CRAO, my experimental[5,9] and have provided an explanation for this clinical entity. They
clinical CRAO studies have shown that ischemic retinal whitish have shown that central retinal vein occlusion is usually due
opacity and swelling in CRAO are essentially located in the to occlusion of the central retinal vein in the optic nerve, at
macular region – maximum in the perifoveolar region [Figs. 1, variable distance posterior to the lamina cribrosa. However,
3a, and 11a]. If there is restoration of circulation in the CRA, when the central retinal vein is occluded at the lamina cribrosa,
the retinal capillaries in the central, thickest part of the macular it completely blocks retinal circulation, because the central
region do not refill [Figs. 2b and 11b], because of compression retinal vein has no means to establish collateral circulation
of retinal capillaries by the surrounding swollen superficial soon [Fig.  4] and that converts the retinal circulation into a
retinal tissue, resulting in the “no re‑flow phenomenon”[9] closed loop – if the blood cannot get out, the blood cannot
1692 Indian Journal of Ophthalmology Volume 66 Issue 12

of treatment advocated for CRAO would involve writing

along, detailed description, which is not desirable in such a
review. The two cited references[20,23] give detailed discussion
about the various advocated treatments; anyone interested to
get more detailed information can consult those references.
I have been told that in India, most vitreoretinal surgeons are
advocating immediate vitrectomy with low pressure settings
in eyes with CRAO irrespective of length of CRAO. Any claim
that vitrectomy dislodges emboli has no scientific validity at
all, and I have discussed that in my publications on that topic.
I am aware that a paper on this method was published from
China,[29] based on 10 patients where this procedure was done
24 h to 6 days after the onset of CRAO. There are three major
problems with their conclusion. (i) Retinal circulation was said
to be restored to normal in four cases, but visual acuity was
said to have improved in six cases; that means visual acuity
improved in two eyes even when there was no restoration of
retinal circulation. (ii) Most importantly, as discussed above, the
retina suffers irreversible ischemic damage within 4 h of the onset
of CRAO.[9] Therefore, performing this surgical procedure 24 h
to 6 days after the onset of CRAO cannot restore function in the
irreversibly damaged retina. (iii) As shown above, spontaneous
visual improvement does occur as a part of natural history that
Figure 15: Fluorescein fundus angiogram of the left eye with NA‑CRAO then is used as evidence for advocating a mode of treatment.
with cilioretinal retinal sparing. It shows the area supplied by the These problems invalidate the claim that their surgical procedure
cilioretinal artery and retrograde filling the adjacent retinal circulation helped to improve visual acuity. I consider such practice
based on ignorance. It is unfortunate that such scientifically
get in. This must result in complete stoppage of the retinal misleading papers get published; it is even more unfortunate
circulation; consequently, that results in secondary CRAO. that such financially rewarding procedures get perpetuated by
Therefore, when CRAO is associated with central retinal vitreoretinal surgeons, at no benefit at all to the poor patient.
vein occlusion, it constitutes a hemodynamic blockage in the Of these therapies, local intra‑arterial thrombolysis is
CRA and is not due to actual thrombosis or embolism in the currently the most popular. In this mode of treatment,
CRA. Thus, simultaneous development of central retinal vein intra‑arterial fibrinolysis is delivered directly into the
occlusion and CRAO represents one disease entity. ophthalmic artery by super‑selective administration of
thrombolytic agent. Almost all the studies are based on
Management of CRAO retrospective review of charts. Reputable studies by Beatty and
There has been a great interest and controversy in its management Au Eong,[30] Fraser and Siriwardena,[31] Noble et al.,[32] Framme
ever since CRAO has been known. Usually, one or a combination et al.,[33] and Schumacher et al.[34] conclusively showed that local
of the following conventional modes of treatments has been intra‑arterial fibrinolysis not only has no beneficial effect, but
advocated in acute CRAO and has claimed success. These[20,23] also can be associated with serious systemic complications. For
include (i) ocular massage, in an effort to dislodge the embolus in example, Framme et al.[33] in a study of 62 patients compared
the CRA; (ii) a reduction of IOP by paracentesis, massage of the visual recovery after intra‑arterial fibrinolysis to conventional
eyeball, administration of acetazolamide, and so on to improve treatment. They concluded that there was no difference
blood flow; (iii) vasodilation of the CRA; (iv) inhalations of 95% between the two modes of treatment in the improvement of
oxygen and 5% carbon dioxide; (v) rebreathing of expired CO2 visual acuity; additionally, there is an increased risk of a stroke
in a bag; and  (vi) retrobulbar vasodilators. Except for ocular from intra‑arterial fibrinolysis. Schumacher et al.[34] conducted a
massage, which occasionally dislodging the embolus, there is prospective, randomized, multicenter clinical trial study in 84
no evidence that the rest show any benefit. CRAO patients, of whom 40 received conventional treatment
and 44 local intra‑arterial fibrinolysis. The mean best corrected
Other advocated treatments included (i) thrombolysis
visual acuity showed no significant differences  (P  =  0.69)
by administering a thrombolytic agent, (ii) isovolumic
between the two groups. Adverse reactions of therapy were
hemodilution, (iii) hyperbaric oxygen, (iv) reduction of red
seen in 37.1% in the local intra‑arterial fibrinolysis group. They
blood cell rigidity by giving pentoxifylline, (v) systemic
concluded that in light of these two therapies’ similar outcomes
steroids intravenously to reduce vascular endothelial edema
and the higher rate of adverse reactions associated with local
following CRAO, (vi) neodymium: yttrium aluminum garnet
intra‑arterial fibrinolysis, they could not recommend local
laser arteriotomy and embolectomy, (vii) cannulation of the
intra‑arterial fibrinolysis for the management of acute CRAO.
supraorbital artery and retrograde injection of antispasmodic
I feel that should settle the issue of the role of thrombolytic
papaverine, (viii) direct massage of the CRA, and (ix) glutamate
therapy. This is not surprising, because, as discussed
receptor antagonists.
elsewhere,[14,15,23] thrombolysis therapy has little scientific
I have discussed in detail each of these treatments rationale in CRAO and is not entirely safe. However, more
elsewhere.[20,23] To discuss at length, each and every mode recently, Johns Hopkins Hospital study[35] in 42 patients, half
 Hayreh: Central retinal artery occlusion
December 2018 1693

treated by local intra‑arterial t‑PA study (LIF) and the rest no most common source of emboli in the heart is the valves,
therapy, claimed that LIF administered in aliquots is associated although other lesions can also be responsible [Table  1].
with an improvement in visual acuity compared with standard Transesophageal echocardiography is the best way to
therapy and has few side effects. But that study presents evaluate heart lesions
problems: (a) the treatment was given 15 h after the onset of 4. Atherosclerosis is a common cause of lesions in the carotid
CRAO, but my experimental study[9] showed that CRAO lasting artery, so it is essential to do their lipid evaluation. I have
for 4 h results in irreversible ischemic retinal damage. (b) My found that invariably physicians tend to look only at the total
natural history study[14] in 260 eyes showed that the visual cholesterol level. I have seen many patients with perfectly
outcome very much depends upon the type of CRAO, and their normal total cholesterol levels but with high low‑density
sample size of 21 was far too small to rule that out. lipoproteins, which is the most important parameter. The
current guideline is to have low‑density lipoproteins not
The groundless claims of benefits for various modes of more than 70 mg/dL. Similarly, triglycerides should also
treatments are due to the following two factors: be in normal limits
1. The visual improvement as a part of the natural history of 5. As discussed above, there are many rare causes of CRAO. In
visual outcome[14] has been mistaken for a beneficial effect difficult cases, one has to keep those in mind. In my studies,
of treatment I have not found thrombophilia evaluation fruitful in CRAO
2. My study[9] showed that if the retinal circulation is restored 6. In eyes with CRAO, optical coherence tomography (OCT)
within an hour, there is no irreversible retinal damage with and OCT angiography do not provide any more worthwhile
return of normal vision. information than that provided by fundus ophthalmoscopic
Most importantly, when discussing any therapy for CRAO, evaluation, color fundus photography, and ordinary
the first essential is, does a therapy have a scientific rationale? fluorescein fundus angiography; other than it unnecessarily
Without it, any treatment must eventually prove useless or adds to the cost of medical care, which is unfair to the poor
even harmful. patient. I have found that in CRAO, these two new methods
of testing are not necessary.
Investigations in Patients with CRAO Financial support and sponsorship
From the discussion above, it is evident that patients with Nil.
CRAO should have the following investigations to find the
Conflicts of interest
cause and to prevent or reduce the risk of any further visual
problems or stroke. Contrary to the practice often advocated, There are no conflicts of interest.
they do not need urgent neurological evaluation, unless they
have neurological symptoms, as discussed elsewhere.[22]
References
1. In all patients 50  years and older with CRAO, it is 1. Von Graefe  A. Über Embolie der Arteria centralis retinae
absolutely critical to rule out giant cell arteritis because als Ursache plötzlicher Erblindung. Albrecht v Graefes Arch
that is an important ophthalmic emergency; if it is Ophthalmol. 1859;5:136‑85.
missed, that can result in catastrophic visual loss in both 2. Schweigger K. Vorlesungen über den Gebrauch des Augenspiegels.
eyes, which is preventable by immediate diagnosis and Berlin: Mylius’sche Verlags‑Buchhandlung; 1864.
aggressive management with corticosteroid therapy.[36‑38] 3. Hayreh SS. A study of the central artery of the retina in human
This fact has medicolegal implications, apart from the beings [Thesis for the Master of Surgery]. Panjab University (India);
1958.
human tragedy involved. All these patients must have
immediate evaluation of erythrocyte sedimentation rate 4. Singh S, Dass R. The central artery of the retina. Br J Ophthalmol
and C‑reactive protein levels estimated – of the two, the 1960;44:193‑212; 280‑299.
latter is more reliable.[38] Systemic symptoms of giant cell 5. Hayreh SS, Weingeist TA. Experimental occlusion of the central
arteritis,[38] when present, can be helpful. All these patients artery of the retina l. Ophthalmoscopic and fluorescein fundus
angiographic studies. Br J Ophthalmol 1980;64:896‑912.
must also have fluorescein angiography because the
presence of associated posterior ciliary artery occlusion 6. Hayreh  SS, Kolder  HE, Weingeist  TA. Central retinal artery
occlusion and retinal tolerance time. Ophthalmology 1980;87:75‑8.
is diagnostic in these cases [Figs. 3 and 6]
2. Since embolism is the most common cause of CRAO, 7. Jonas JB, Hayreh SS. Optic disc morphology in experimental central
retinal artery occlusion in rhesus monkeys. Am J Ophthalmol
all patients with CRAO must have urgently a complete
1999;127:523‑30.
evaluation to find the source of embolism. As discussed
above, the most common sources of emboli are the carotid 8. Hayreh SS, Jonas JB. Optic disk and retinal nerve fiber layer damage
after transient central retinal artery occlusion: An experimental
artery and the heart [Table 1]. In the carotid artery, the most study in rhesus monkeys. Am J Ophthalmol 2000;129:786‑95.
common source of embolus is a plaque, so it is essential to
9. Hayreh SS, Zimmerman MB, Kimura A, Sanon A. Central retinal
find out if there are any plaques; unfortunately, I have found
artery occlusion. Retinal survival time. Exp Eye Res 2004;78:723‑36.
that evaluation is usually primarily focused on whether
10. Kwon  YH, Rickman  DW, Baruah  S, Zimmerman  MB, Kim CS,
there is hemodynamically significant carotid artery stenosis,
Boldt  HC, et al. Vitreous and retinal amino acid concentrations
without paying attention to the presence of plaques. That in experimental central retinal artery occlusion in the
can miss the actual source of emboli primate. Eye 2005;9:455‑63.
3. Cardiac evaluation is equally essential, even if the carotid 11. Hayreh SS. Central retinal vascular occlusion‑ morphology and
artery shows lesions, because I have found that at times a pathogenesis. In: Brancato  R, Pratesi  F, editors. International
patient can have lesions in both places, and either of them Symposium on Ocular Angiology (1974: Firenze) Rome: Minerva
can be the source of emboli. My study[19] showed that the medica; 1977. p. 15‑40.
1694 Indian Journal of Ophthalmology Volume 66 Issue 12

12. Hayreh SS, Podhajsky P. Ocular neovascularization with retinal pressure abnormalities associated with central and hemicentral
vascular occlusion. II. Occurrence in central and branch retinal retinal vein occlusion. Ophthalmology 2004;111:133‑41.
artery occlusion. Arch Ophthalmol 1982;100:1585‑96. 28. Nettleship E. Unusual appearance in a case of retinal embolism
13. Mizener JB, Podhajsky P, Hayreh SS. Ocular ischemic syndrome. about 30 hours after its occurrence. Festschrift zur Feier des
Ophthalmology 1997;104:859‑64. Siebzigsten Geburtstages von Herman‑von Helmholtz 1891;7:7‑8.
14. Hayreh SS, Zimmerman MB. Central retinal artery occlusion: Visual 29. Lu  N, Wang  NL, Wang  GL, Li  XW, Wang  Y. Vitreous surgery
outcome. Am J Ophthalmol 2005;140:376‑91. with direct central retinal artery massage for central retinal artery
15. Hayreh SS. Prevalent misconceptions about acute retinal vascular occlusion. Eye (Lond) 2009;23:867‑72.
occlusive disorders. Prog Retin Eye Res 2005;24:493‑519. 30. Beatty  S, Au Eong  KG. Local intra‑arterial fibrinolysis for acute
16. Hayreh  SS, Zimmerman  MB. Fundus changes in central retinal occlusion of the central retinal artery: A meta‑analysis of the
artery occlusion. Retina 2007;27:276‑89. published data. Br J Ophthalmol 2000;84:914‑6.
17. Hayreh SS. Intra‑Arterial thrombolysis for central retinal artery 31. Fraser S, Siriwardena D. Interventions for acute non‑arteritic central
occlusion. Br J Ophthalmol 2008;92:585‑7. retinal artery occlusion. Cochrane Database Syst Rev 2009. doi:
18. Hayreh  SS. Transient central retinal artery occlusion following 10.1002/14651858.CD001989.pub2.
viperine snake bite. Arch Ophthalmol 2008;126:870‑1. 32. Noble  J, Weizblit  N, Baerlocher  MO, Eng  KT. Intra‑arterial
19. Hayreh  SS, Podhajsky  PA, Zimmerman  MB. Retinal artery thrombolysis for central retinal artery occlusion: A systematic
occlusion: Associated systemic and ophthalmic abnormalities. review. Br J Ophthalmol 2008;92:588‑593.
Ophthalmology 2009;116:1928‑36. 33. Framme C, Spiegel D, Roider J, Sachs HG, Lohmann CP, Butz B,
20. Hayreh SS. Acute retinal arterial occlusive disorders. Prog Retin et al. Central retinal artery occlusion. Importance of selective
Eye Res 2011;30:359‑94. intra‑arterial fibrinolysis. Ophthalmologe 2001;98:725‑30.
21. Hayreh SS, Zimmerman MB. Ocular arterial occlusive disorders 34. Schumacher M, Schmidt D, Jurklies B, Gall C, Wanke I, Schmoor C,
and carotid artery disease. Ophthalmology Retina 2017;1:12‑18. et al. Central retinal artery occlusion: Local intra‑arterial fibrinolysis
22. Hayreh SS. Do patients with retinal artery occlusion need urgent versus conservative treatment, a multicenter randomized trial.
neurological evaluation? Am J Ophthalmol 2018:30457‑4. doi: Ophthalmology 2010;117:1367‑75.
10.1016/j.ajo. 2018.08.015. 35. Aldrich  EM, Lee  AW, Chen  CS, Gottesman  RF, Bahouth  MN,
23. Hayreh  SS. Ocular Vascular Occlusive Disorders. Heidelberg: Gailloud  P, et al. Local intraarterial fibrinolysis administered in
Springer‑Verlag; 2015. p. 31‑53, 239‑305. aliquots for the treatment of central retinal artery occlusion: The
24. Hayreh  SS. Anterior ischaemic optic neuropathy. II. Fundus on Johns Hopkins Hospital experience. Stroke 2008;39:1746‑50.
ophthalmoscopy and fluorescein angiography. Br J Ophthalmol 36. Hayreh  SS, Zimmerman  B. Management of giant cell arteritis:
1974;58:964‑80. Our 27‑year clinical study; new light on old controversies.
25. Hayreh SS, Podhajsky PA, Zimmerman B. Ocular manifestations Ophthalmologica 2003;217:239‑59.
of giant cell arteritis. Am J Ophthalmol 1998;125:509‑20. 37. Hayreh SS. Treatment of acute visual loss in giant cell arteritis:
26. Hayreh SS, Piegors DJ, Heistad DD. Serotonin‑induced constriction Should we prescribe high‑dose intravenous steroids or just oral
of ocular arteries in atherosclerotic monkeys: Implications for steroids? J Neuroophthalmol 2012;32:278‑87.
ischemic disorders of the retina and optic nerve head. Arch 38. Hayreh  SS, Podhajsky  PA, Raman  R, Zimmerman  B. Giant cell
Ophthalmol 1997;115:220‑8. arteritis: Validity and reliability of various diagnostic criteria. Am
27. Hayreh  SS, Zimmerman  MB, Beri  M, Podhajsky  P. Intraocular J Ophthalmol 1997;123:285‑96.