329

Open angle glaucoma and non-central retinal vein occlusion –the chicken or the egg?Bertil Lindblom

Department of Ophthalmology, Institute of Clinical Neuroscience, GoteborgUniversity, Goteborg, Sweden

ABSTRACT.Purpose: To determine whether non-central retinal vein occlusion (nCRVO) oropen angle glaucoma (OAG) is the primary event in eyes suffering from bothdiseases.Methods: The study has two parts. The first is a retrospective survey of a clin-ical glaucoma database. In eyes with OAG and nCRVO, the temporal relation-ship between the conditions was determined when possible. The second part isa ten-year follow-up study of eyes with isolated nCRVO.Results: Study 1: Among 576 OAG patients, 25 (4.3%) had nCRVO. In all eyesexcept one, OAG occurred before nCRVO. Almost without exception a vein onthe optic disk or at the disk margin was occluded. Study 2: Most eyes withisolated nCRVO had an occluded vein at a retinal artery-vein crossing. Onlyone of 34 patients had developed OAG after ten years.Conclusions: This study suggests that OAG in most eyes is the primary event.nCRVO had different characteristics in eyes with OAG compared to eyes with-out OAG.

Key words: branch retinal vein occlusion – central retinal vein occlusion – open angle glaucoma.

Acta Ophthalmol. Scand. 1998: 76: 329–333Copyright c Acta Ophthalmol Scand 1998. ISSN 1395-3907

Aclose association has been recog-nised between OAG and retinal

vein occlusion. CRVO or branch retinalvein occlusion are common in eyes withOAG (for a review see Sonnsjo andKrakau, 1993). These authors also con-sider the disk haemorrhages frequentlyseen in some patients with OAG to repre-sent small vein occlusions. The coexist-ence of these different conditions has ledto the suggestion that OAG, retinal veinocclusion and disk haemorrhage share acommon pathogenesis (Sonnsjo &Krakau, 1993). This hypothesis hasgained popularity because an increasingnumber of researchers favour some formof a ‘‘vascular’’ glaucoma model for thepathogenesis of OAG (for a review, seeHayreh, 1994). However, there have alsobeen advocates, primarily in the olderliterature, for the hypothesis that retinal

vein occlusion is a complication of OAG(Bertelsen, 1961; Dobree, 1957; Dryden,1965), occurring secondarily to structuraldisk changes (Kohner & Shilling, 1976).However, the results of many olderstudies are difficult to interpret becausetheir definition of OAG is often based onintraocular pressure alone without opticdisk or visual field criteria. Anotherproblem that is apt to colour the resultsis the difficulty of differentiating betweenprimary and secondary glaucoma in eyeswith CRVO. While the differentiationmay be simple in the full-blown case withneovascularizations of the iris, manycases are much less obvious.

This study explores the relation be-tween OAG and nCRVO (encompassingbranch retinal vein occlusion and hemi-spheric retinal vein occlusion) using cur-rent principles for glaucoma definition.

The reason for separating eyes withnCRVO from eyes with CRVO was thatthe former seldom develop secondaryglaucoma (Sanborn & Margargal, 1984),This condition could be difficult or im-possible to differentiate from primaryOAG.

The study is presented in two parts.The first part surveys the prevalence ofnCRVO in a group of patients with estab-lished OAG. The second part is a ten-yearfollow-up study of eyes with nCRVOwithout OAG.

Study 1: Prevalence ofnCRVO in Eyes withestablished OAGMaterial and Methods

Beginning in 1992, all patients with glau-coma or ocular hypertension examinedby the author at the University Eye Clinicin Goteborg have been registered in acomputerised database. Each record con-tains information about the glaucoma di-agnosis, other eye diseases, systemic dis-eases, systemic medications, and heredity.In addition, results from the most recenteye examination are recorded. Each timea patient is examined, old data is cor-rected if necessary, and any new relevantinformation is added. Only patientsexamined by the author are included tokeep the database as homogeneous aspossible, with identical inclusion criteriaapplied to patients in each diagnosticgroup. For a subject to be registered withthe diagnosis of OAG, both of the follow-ing criteria have to be met:1. Sector-shaped defects in the retinal

nerve fibre layer and/or optic diskchanges compatible with glaucoma.

330

Fig. 1. Flow charts illustrating the design of the two studies. A: Study .1. nCRVO in eyes with established OAG. B: Study .2. Incidence of OAG ineyes with nCRVO. Figures in parentheses show number of patients; figures in italics between brackets show number of eyes. Abbreviations: RVO –retinal vein occlusion. CRVO – central retinal vein occlusion, nCRVO – non-central retinal vein occlusion, OAG – open angle glaucoma, OH –ocular hypertension.

2. Visual field changes consistent withglaucoma and not explained by anyother condition. Computer-assistedperimetry is performed with theOphthimus high-pass resolution tech-nique (HighTech Vision, Molndal,Sweden) or with Humphrey Field Ana-lyzer (Humphrey Instruments, San Le-andro, CA), program 24–2 or 30–2.

For the present study, the OAG groupcomprised patients with primary OAG(exfoliation and simple glaucoma) andOAG considered secondary to CRVO. Adiagnosis of exfoliation glaucoma wasmade if pseudo-exfoliations were presentin at least one eye. Both normal-tensionand high-tension glaucoma were in-cluded.All OAG patients with retinal vein oc-clusion were identified from the database.Fundus photographs and fluorescein an-giograms were examined when available.

Results

A flow-chart summarising the study de-sign and results is shown in Fig. 1A. Fivehundred seventy-six patients with OAGwere registered in the glaucoma database(295 women and 281 men). In patientsstill alive in 1995, mean age was 80 years(82 for women and 79 for men). Agerange was 32 to 102 years (51 to 100 forwomen, 32 to 102 for men). Mean age at

which OAG was diagnosed was 69 years.Two hundred ninety-two subjects sufferedfrom simple glaucoma while 284 (49.3%)had exfoliation glaucoma. Twelve pa-tients had normal-tension glaucoma;none of them had pseudo-exfoliations.

Among all OAG subjects, 64 (10.6 percent) had some form of retinal vein oc-clusion: 4.3 per cent had nCRVO and 6.3per cent CRVO. Thirty-one were femalesand 33 males. One patient had nCRVO inone eye and CRVO in the other. Meanage in the nCRVO group was 79 years(range 32–92) and in the CRVO group 81years (range 61–94). In three cases fileswere lost and the exact nature of the oc-clusion could not be established. Fifteen(24%) of the 64 subjects with OAG andretinal vein occlusion had systemic hyper-tension, another three (5%) had cerebro-vascular disease, and seven (11%) haddiabetes mellitus. Corresponding figuresin OAG patients without retinal vein oc-clusion were 15%, 3%, and 8%, respec-tively. The differences were not statisti-cally significant.

For 23 of the 25 eyes with nCRVO andOAG, OAG was obvious already at thetime of the occlusion. In the majority ofeyes, OAG had been diagnosed and glau-coma treatment instituted prior to thevein occlusion while in the remaining eyesthe optic disk showed obvious glaucoma-

tous changes at presentation of nCRVO(Fig. 2). In only one eye was nCRVO theinitial event, occurring 15 years prior toOAG. For one eye the time sequencecould not be established.

Fundus photographs were available for18 eyes with OAG and nCRVO. Fluor-escein angiography had been performedin only a few eyes. A majority of the pa-tients had lens opacities, reducing the im-age quality. The location of the vein oc-clusion was on the optic disk surface orat the disk border in 16 eyes (88.8%). Inone eye the occlusion site was 0.5 disk di-ameters from the border (Fig. 3). Thiswas the only eye in which occlusion of avein peripheral to the optic disk borderhad occurred remote from an artery-veincrossing. The other eye with nCRVO out-side the optic disk belonged to the patientin which nCRVO was diagnosed long be-fore OAG. Of the 16 eyes with occlusionof a vein on the optic disk surface, sevenwere hemispheric vein occlusions, i.e., oc-clusion of a first-order branch of the cen-tral retinal vein. In nine eyes, the oc-clusion engaged a second-order branchon the optic disk. In seven eyes the oc-clusion site was in an area where glau-comatous changes were obvious (e.g., un-dermining of the disk border or an exca-vation notch).

Case report. A 33-year-old male ac-

331

Fig. 2. A: Right eye of a 33-year-old male (see case report) showing inferior hemispheric retinal vein occlusion. Note the thin collateral vessel alongthe temporal disk border connecting superior and inferior retinal veins. The disk has excavation extending to the superior rim. B: In the left eyethere is large excavation showing vertical asymmetry. The nerve fibre layer is more visible superiorly.

countant complained of slowly decreas-ing vision in the right eye for about sixmonths. He was in perfectly good generalhealth. Visual acuity was 20/50 in theright eye, 20/15 in the left. The right vis-ual field showed defects, most pro-nounced inferiorly. The left eye had minorfield changes. At the first examination,intraocular pressures were 19 and 32 mm,respectively. On follow-up examination,pressures were 25 and 27, respectively.Both eyes had open angles with densepre-trabecular uveal meshworks as signsof goniodysgenesis. Media were clear.Right optic disk showed a large exca-vation extending to the superior rim.There was an inferior hemispheric retinalvein occlusion with haemorrhages in theright eye (Fig. 2A). The left optic diskshowed a vertically asymmetric exca-vation and loss of retinal nerve fibre layerin the inferior arcuate bundle (Fig 2B).Because of extensive glaucoma damageand fresh haemorrhages in the right eyeit was concluded that OAG was the pri-mary event and nCRVO a secondary ef-fect. Two years later, he was still in goodhealth. Medical examinations includinglaboratory work-up remained normal.

Study 2: Incidence ofOAG in Eyes withnCRVOMaterial and Methods

Patients with nCRVO but without anysigns of glaucoma had been enrolled in aprospective study of laser photocoagu-

lation of nCRVO during the years 1984 to1987. Consecutive patients with nCRVOseen at the University Eye Clinic in Gote-borg were registered by that time. Theirfiles, fundus photographs, and fluoresceinangiograms were examined in retrospect.All patients that could be reached in 1995were invited to a follow-up examination.All were examined by the author. Per-imetry was performed with HumphreyField Analyzer program 24–2 and fundusphotographs were obtained.

The occlusion site was identified in thefundus photographs or angiograms. Thegeneration of the occluded retinal vesselbranch was established and the distance

Fig. 3. Right eye from a 64-year-old male with glaucomatous disk changes. The superior temporalvein is occluded about 0.5 disk diameters above the disk border (arrow). The occlusion site isremote from the artery-vein crossing.

between the occlusion site and the opticdisk border was measured. Since photo-graphs were obtained with different fun-dus cameras, all distances were expressedin optic disk diameters. The vessel anat-omy at the occlusion site was alsostudied. Special attention was paid towhether the vein was crossed by an arteryat the point of occlusion.

Results

A flow-chart summarising the study de-sign and results is shown in Fig. 1B. Sixty-one patients had been examined fornCRVO during 1984–1987. Mean age atexamination was 66 years (range 37–84).

332

Twenty-four were women and 37 men. Infive instances the patients’ files were lost.Fundus photographs and fluorescein an-giograms were examined for the remain-ing 56 patients (58 eyes). In 1995, aboutten years after the initial examination, 16of the patients were dead. The files fromthese patients were analysed. Mean fol-low-up time was 25 months (11–72). Noneof the deceased patients had developedOAG during the follow-up period.

The remaining 40 patients were invitedfor a follow-up examination. Three hadmoved from the area and another threewere unable to come due to infirmity.Thus, 34 patients were examined. Meanfollow-up time was 121 months (SD 13.7,range 92–152). Twenty eyes (20 patients)had been treated with argon laser photo-coagulation: twelve because of macularoedema and another eight because of neo-vascularization in the retina or on the op-tic disc. One patient had developed OAGin the eye contralateral to the nCRVO. An-other patient showed ocular hypertensionin both eyes but no glaucomatous fieldchanges. The remaining 32 patientsshowed no sign of OAG. Twelve patientshad systemic hypertension.

The pattern of nCRVO was different innon-glaucoma eyes compared to eyeswith OAG. Only 16 of the 58 eyes (28%)in the nCRVO group had occluded a veinon the optic disk surface. Ten were hemi-spheric vein occlusions and six were oc-clusions of a second-order branch. Theremaining 42 eyes (72%) had occlusion ofa retinal vein peripheral to the optic disk.Without exception, the occlusion site ineyes with nCRVO peripheral to the opticdisk was at an artery-vein crossing (Fig.4). Mean distance between the occlusionsite and optic disk border in the wholegroup of nCRVO eyes was 0.74 disk di-ameters (SD 0.72). Corresponding figuresin eyes with nCRVO and OAG were 0.06disk diameters (SD 0.17) (p∞0.0001).

General DiscussionIn this study eyes with nCRVO and OAGshowed differences compared to eyes withisolated nCRVO. In eyes with OAG, theocclusion site was in most cases on theoptic disk or at its border. There was atendency for occlusions to occur in anarea of the disk showing distinct glau-comatous changes. It is plausible that thisfinding relates to the peripapillary vaso-constriction reported in the sector withthe greatest cupping in glaucoma eyes

Fig. 4. Superior temporal branch retinal vein occlusion in an eye without glaucoma in a 52-year-old male. The occlusion site is at an artery-vein crossing.

(Rader et al., 1994). In contrast, mosteyes with nCRVO without OAG had pe-ripheral occlusions, always occurring at alocation where an artery crossed a vein.This is in keeping with earlier studies(Zhao et al., 1993). Only one eye in thenCRVO-OAG group had an occludedvein distal to the optic disk and this wasthe only eye in the study in which nCRVOpreceded OAG with several years. It wasalso the only eye in the study in which anCRVO distal to the disk occurred at alocation remote from an artery-veincrossing. In almost all eyes with nCRVOand OAG, OAG was diagnosed first and/or obvious disk and field changes consist-ent with glaucoma were present at the in-itial visit for nCRVO. These findings sug-gest that in eyes with both nCRVO andOAG, OAG is in most instances the pri-mary event.

The prevalence of all retinal vein oc-clusions was 11 per cent in our glaucomapopulation. This figure may be falsely highbecause of selection bias. It is plausiblethat OAG patients with additional ocularpathology will be more readily referred tothe university clinic than patients with iso-lated OAG. Whether the high prevalenceof exfoliation glaucoma in our region isimportant for the incidence of nCRVO isnot known. The number of eyes with nor-mal-tension glaucoma, on the other hand,was low. One reason could be that patientsoften are selected on the grounds of an el-evated intraocular pressure. The trueprevalence of normal-tension glaucoma inthe region is not known.

Only one case of OAG developed in 58patients with nCRVO during a ten-yearfollow-up. A large number of patients

were lost, the majority because of death.However, none of the patients lost to fol-low-up was known to have OAG. Becauseof the eye care organisation in the Gote-borg region, any nCRVO patient with adiagnosed OAG most probably wouldhave come to our attention. The low inci-dence of OAG in eyes with nCRVO is inagreement with an earlier study in whichonly two new cases of OAG were foundduring a ten-year follow-up study of 143patients with retinal vein occlusion (Rub-instein & Jones, 1976).

Recently, Sonnsjo and Krakau (1993)presented a vascular hypothesis of glau-coma. These authors found that in somecases OAG preceded the vascular oc-clusion while in other cases the oppositewas true. However, in essential parts oftheir study nCRVO and CRVO weregrouped together and treated synony-mously. A possible disadvantage of thisapproach is the difficulty of differentiat-ing between primary and secondary glau-coma in eyes with CRVO. Another expla-nation for the discrepancies between theirfindings and the those of the presentstudy could be differences in the defi-nition of OAG. Visual field examinationsin an eye with retinal vein occlusion aredifficult to interpret. Optic disc evalu-ation after an occlusion is also difficultbecause the atrophy that occurs after reti-nal vein occlusion can sometimes be in-distinguishable from that of OAG. It istherefore impossible to eliminate a com-ponent of subjectivity in the diagnosis ofOAG in these patients.

No attempt was made to differentiatehemispheric retinal vein occlusions intohemi-central retinal vein occlusion and

333

hemispheric retinal vein occlusion ac-cording to Hayreh and Hayreh (1980). Inthe former the central retinal vein has adual trunk of which one part is occluded.As remarked by Sanborn and Magargal(1984), the distinction between the twotypes is not always easy to make. In thepresent study the most important aspectwas to treat separately conditions afterwhich development of secondary glau-coma is uncommon.

The relation between OAG and retinalvein occlusion needs to be evaluatedfurther. In future studies it will be necess-ary to separate CRVO from nCRVO be-cause the relation to OAG may differ be-tween the two groups. For the same rea-son it may be sound to separate nCRVOon the optic disk from peripheralnCRVO. Recognition of these differencesmay help to resolve the current confusionregarding the relation between OAG andthe different forms of retinal veinocclusion and may also provide newinsights into the pathogenesis of glau-coma.

ReferencesBertelsen TI (1961). The relationship between

thrombosis in the retinal veins and primaryglaucoma. Acta Ophthalmol (Copenh) 39:603–613.

Dobree JH (1957). Venous obstruction andneovascularization at the disc in chronicglaucoma. Trans Ophthal Soc UK 77: 229–237.

Dryden RM (1965). Central retinal vein oc-clusion and chronic simple glaucoma. ArchOphthalmol 73: 659–663.

Hayreh SS, Hayreh MS (1980). Hemi-centralretinal vein occlusion. Pathogenesis, clinicalfeatures and natural history. Arch Ophthal-mol 98: 1600–1609.

Hayreh SS (1994). Progress in the understand-ing of the vascular etiology of glaucoma.Curr Opinion Ophthalmol 5: 26–35.

Kohner E, Shilling J (1976). Retinal vein oc-clusion. In: Clifford Rose, ed. Medical Oph-thalmology. London: Chapman & Hall,chap. 28.

Rader J, Feuer WJ, Anderson DR (1994). Peri-papillary vasoconstriction in the glaucomasand the anterior ischemic optic neuro-pathies. Am J Ophthalmol 117: 72–80.

Rubinstein K, Jones EB (1976). Retinal veinocclusion: long-term prospects. 10 yearsøfollow-up of 143 patients. Br J Ophthalmol60: 148–150.

Sanborn GE, Magargal LE (1984). Character-istics of the hemispheric retinal vein oc-clusion. Ophthalmology 91: 1616–1626.

Sonnsjo B, Krakau CET (1993). Argumentsfor a vascular glaucoma etiology. ActaOphthalmol (Copenh) 71: 433–444.

Zhao J, Sastry SM, Sperduto RD, Chew EY,Remaley NA, The Eye Disease Case-ControlStudy Group (1993). Arteriovenous crossingpatterns in branch retinal vein occlusion.Ophthalmology 100: 423–428.

Received on April 14th, 1997.Accepted on August 18th,1997.

Corresponding author:

Bertil Lindblom, MDDepartment of OphthalmologySahlgrenska University HospitalS-413 45 GoteborgSwedenTel: π46-31 601000, Fax: π46-31 412904E-mail: bertil.lindblom/oftalmologi.gu.se