new peripheral in · borderclassification borderclassification classification i iia iib i iia iib...

Bnrtish Journal ofOphthalmology 1994; 78: 681-689

New classification of peripheral retinal vascularchanges in sickle cell disease

Alan D Penman, John F Talbot, Elaine L Chuang, Peter Thomas, Graham R Serjeant,Alan C Bird

Medical ResearchCouncil Laboratories(Jamaica) University ofthe West Indies,Kingston, JamaicaA D PenmanP ThomasG R Serjeant

The Royal HallamshireHospital, SheffieldJ F Talbot

University ofWashington, Seattle,Washington, USAE L Chuang

Institute ofOphthalmology,Moorfields Eye Hospital,LondonA C BirdCorrespondence to:Professor A C Bird, Instituteof Ophthalmology, MoorfieldsEye Hospital, City Road,London EC1V 2PD.

Accepted for publication5 May 1994

AbstractThe systemic complications of homozygoussickle cell disease (SS) are more severe than insickle cell haemoglobin C (SC) disease, and yetvisual loss due to proliferative retinopathy ismore common in the latter. This anomaly isunexplained. It is believed that proliferativedisease occurs in response to closure of theperipheral retinal vasculature, yet a systematiclongitudinal survey of the peripheral retinalvascular bed has not been undertaken. In theJamaica Sickle Cohort study all subjects are

scheduled to receive annual ocular examina-tion and fluorescein angiography. The resultshave now been analysed in subjects with SSand SC disease using a new classificationsystem based on a comparison of the periph-eral retinal vascular bed with that recorded inthe cohort with normal haemoglobin (AA)genotype. The vascular patterns could be clas-sified as qualitatively normal (type I) or abnor-mal (type II). An abnormal vascular patternwas identified more commonly with age, in a

significantly larger proportion of subjects withSC than SS disease, and was associated withthe development of proliferative disease. Inorder to establish the dynamics of change, theangiograms were analysed in the 18 subjects(24 eyes) who developed proliferative disease.It is shown that a qualitatively normal vascularpattern may be retained despite loss of capill-ary bed and posterior displacement of thevascular border. A border which is quali-tatively abnormal does not revert to normal,and once abnormal, continuous evolution mayoccur before development of proliferativelesions. The peripheral border of the retinalvasculature was too peripheral to photo-graphed in 13 of the 24 eyes before it becomingqualitatively abnormal. It is concluded that anormal border, if posterior, results from grad-ual modification of the capillary bed and indi-cates low risk of proliferative disease. Aqualitatively abnormal vascular border occurs

as a radical alteration of retinal perfusion insubjects in whom little modification of thevascular bed occurred before the event, andsignals risk ofproliferative disease. This classi-fication system is useful in identifying thelikelihood of threat to vision in young Jamai-cans with sickle cell disease, and the higherfrequency of proliferative retinopathy in SCcan be explained by the higher prevalence of a

qualitatively abnormal peripheral retinal vas-

culature.(BrJ Ophthalmol 1994; 78: 681-689)

The major threat to vision in patients with sickle

cell diseases is vascular proliferation in theperipheral retina (proliferative sickle retino-pathy, PSR), which may lead to vitreous haemor-rhage or retinal detachment.' The developmentof PSR is believed to occur in response to pro-gressive peripheral retinal vascular closure, andcentripetal recession of the peripheral margin ofperfusion. This process has been described,'and extensive remodelling of the peripheralcapillary bed has been reported in adults.5 Asequence of events was proposed by Goldberg in1972,6 and later modified by Condon andSerjeant.7 However, the relation between periph-eral vascular closure and PSR is not well under-stood. Although vascular obstruction would bepredicted to be more common in homozygous-sickle cell (SS) disease than in sickle cell haemo-globin C (SC) disease on the basis of systemicinvolvement,8 peripheral retinal non-perfusionand PSR are more common in adults with SCdisease. 9 10

In an attempt to explain this dilemma we havedocumented the evolution of sickle retinopathyby undertaking serial fluorescein angiograms in a

cohort of children with SS and SC genotypes.The patterns ofthe peripheral retinal vasculaturewere characterised and compared with those seenin subjects with a normal (AA) haemoglobingenotype." It appears that an abnormal patternof the peripheral retinal vasculature is more

common in SC than SS disease, and in additioncorrelates with the subsequent development ofPSR.

Subjects and methods

THE JAMAICAN SICKLE COHORTBetween 1973 and 1981, 315 children with SSdisease, 173 with SC disease, and 250 age/sex-matched controls with normal haemoglobin (AA)genotype, were recruited to the study.2 13 Theprotocol of the ocular study called for annualophthalmoscopic examinations of the sickle cellchildren aged 5 years and over, and fluoresceinangioscopy and/or angiography over the age of 6years.The ophthalmic component of the cohort

study is now in its 12th year, and a very large database has accumulated, with clinical and ophthal-mic information in 392 subjects, and over 2000fluorescein angiograms. Serial angiography ofcorresponding retinal areas is available on a largenumber of subjects.

STUDY DESIGNThe ophthalmic records and fluorescein angio-grams performed on all eligible subjects with SS

681

on June 12, 2020 by guest. Protected by copyright.

http://bjo.bmj.com

/B

r J Ophthalm

ol: first published as 10.1136/bjo.78.9.681 on 1 Septem

ber 1994. Dow

nloaded from

http://bjo.bmj.com/

Penman, Talbot, Chuang, Thomas, Serjeant, Bird

Figure 1 Left eye of15-year-old male with Hb SS disease.Type I border: continuous arteriovenous loops withprogressive thinning ofcapillary bed towards periphery.

or SC disease in the Jamaican Sickle Cell CohortStudy from 1981 to 1992 were reviewed. Theangiograms were classified, and the resultsassessed in relation to genotype, age, and'thesubsequent evolution of disease.

Figure 4 Left eye of11-year-old male with SC diseaseshowing a type hIa pattern with hyperfluorescence ofcapillaries extending into the avascular retina.

METHODSAt each annual ocular examination the visualacuity was measured, and retinal drawings indi-cated the number and position of haemorrhages,chorioretinal scars, retinal schisis cavities, andPSR lesions. Fluorescein angioscopy andangiography were undertaken unless there was aspecific contraindication. A bolus dose of 2-4 mlof 20% sodium fluorescein was injected intra-venously, and the circumferential extent of thevisible peripheral margin of retinal perfusionrecorded by angioscopy. The retinal vasculaturewas photographed with a Zeiss fundus camera(Zeiss, Oberkochen, Germany) using Ilford FP4black and white film. Every effort was made tophotograph the peripheral margin ofthe vascularbed over 900 of the temporal periphery, and theposterior pole of each eye. In those in whom lessthan 3600 of the peripheral margin had been seenpreviously, fluoroscopy was undertaken initiallyfollowed by photography. If the whole of themargin had been seen previously in both eyes,photography was performed first.

Figure 2 Left eye of 14-year-old male with Hb SS disease.Type I border: continuous arteriovenous loops with irregularlarge capillary free areas and small vascular anomaly.

Figure 3 Left eye of 10-year-old male with Hb SC diseaseshowing a type IIa border: irregular margin withoutarteniovenous loops or attenuation ofcapillary density.Capillary stumps extend into non-perfused retina.

Figure 5 Left eye of 16-year-oldfemale with Hb SCdisease. Type IIb border: irregular border without continuousarteriovenous loops or attenuation ofcapillary density. Nocapillary stumps extend into non-perfused retina.

682


http://bjo.bmj.com

/B

r J Ophthalm


ber 1994. Dow

nloaded from

http://bjo.bmj.com/

New classification ofperipheral retinal vascular changes in sickle cell disease

Table 3 The number ofeyes with readable angiograms on 2successive years which convertedfrom a type I to a type IIborder in which a type I border assessed according to age

SC disease SS disease

Type I Type IAge previous Change to previous Change to(years) year type II year Type II

7891011121314151617

S614324340383131248

23378610535

2672125161152158141115899158

000311312-633

Figure 6 Left eye ofa 10-year-old male with Hb SCdisease showing obstruction ofa blood vessel. The presence ofa hypofluorescence column continuous with the cessation offluorescence within the vessel implies recent occlusion. Thecapillary bed appears to conform with a type II, andyetreperfusion may occur such that this border was deemedunclassifiable.

Figure 7 Left eye of10-year-old male with Hb SC disease.Recently obstructed vessels appear as hypofluorescent lines,but the appearance (IIa) of the margin has not been alteredby this event.

Classification ofangiographic patternsThe proposed system of classification charac-terises the changes in the peripheral retinal

Table I Abnormal vascular border in different genotypes(no (%)) according to most abnormal pattern recorded

Pattern type SC disease SS disease

I 48 (32) 210 (86)Ila 57 (38) 4 (2)Ilb 37 (25) 17 (7)Other 6 (4) 13 (5)Total 148 244

Table 2 Correlation ofabnormal vascular borders with age and with genotype (no (%))

SC disease SS disease

Age Readable Readable(years) angiograms IIa IIb angiograms IIa IIb8 52 9 (17) 15 (29) 171 0 3 (2)9 92 14 (14) 27 (29) 237 0 5 (2)10 117 18 (15) 32 (27) 235 0 8 (3)11 145 29 (20) 42 (29) 263 0 10 (4)12 150 35 (23) 48 (32) 221 0 12 (5)13 125 33 (26) 37 (30) 212 0 13 (6)14 121 32 (26) 40 (33) 163 0 12 (7)15 113 33 (29) 34 (30) 141 1 10 (7)16 87 34(39) 19(22) 118 2 13(11)17 54 22 (42) 12 (22) 82 1 11 (13)

Table 4 Symmetry between eyes in the pattern oftheperipheral vasculature in SC disease (kappa=0-6; SE0 072)

Left eye border typeRight eyeborder type I II III

I 38 3 8II 9 29 7III 6 6 23

vasculature, and distinguishes normal vascularpatterns from abnormal on the fluorescein angio-gram. The normal pattern had been establishedin a study of the subjects with haemoglobin AAwithin the cohort." Each eye was designated asconforming to a pattern which was indistinguish-able qualitatively from normal (type I), abnormal(type II), or to a category implying that such adesignation could not be made. If the vascularborder in an eye showed normal and abnormalpatterns in different regions the eye was classifiedaccording to the region which was most abnor-mal.Type I. This pattern was qualitatively similar tonormal." It was required that the capillariesbecame gradually less dense in distribution andlonger as the border was approached, and theborder was smooth and formed of arteriovenousloops of varying length (Fig 1). Additional fea-tures were allowed within this type, includingavascular lacunae within the peripheral capillarybed, hairpin loops, or tortuous vessels (Fig 2).Type II. The peripheral retinal vasculature dif-fered from the normal pattern because a densecapillary bed existed up to the margin of perfu-sion with abrupt terminations of small ormedium calibre vessels creating an irregular or'moth eaten' appearance to the border.Type II presented two morphological patterns.

In some cases there were capillary 'buds' or'stumps' extending from the border of the vas-culature into the non-perfused retina which weresometimes hyperfluorescent or even leaked fluo-rescein slightly (Figs 3, 4). In some cases, thecapillary 'buds' or 'stumps' showed bifurcationswhich simulated the capillary pattern seen whenthere is active reperfusion of recently infarctedretina. 1 A border with this feature was classifiedas IIa, and otherwise as IIb (Fig 5).Type III. The pattern was indeterminate becauserecent acute arteriolar occlusion involving thevascular border had given rise to a type IIpattern, but which may have reverted to normalfollowing subsequent reperfusion of the vascular

683


http://bjo.bmj.com

/B

r J Ophthalm


ber 1994. Dow

nloaded from

http://bjo.bmj.com/

Penman, Talbot, Chuang, Thomas, Seyjeant, Bird

Table S The border type compared posterior position ofvascular margin shows that asignificantly higher percentage oftype IIa and type IIb borders had a posterior location thantype I in SC disease (no (%)) (RE X2=4l16, df=2, p<0 001; LE x2=30-4 df=2,p<0 001). The numbers in SS disease were too small to reach significant levels

SC disease SS diseaseBorder classification Border classification

Classification I IIa IIb I IIa IIb

Posterior 30 (30) 78 (87) 55 (76) 157 (40) 3 17 (71)Not posterior 70 (73) 11 (13) 17 (24) 235 (60) 1 7 (29)Total 100 85 72 392 4 24

Table 6 Classification ofborder type compared with theextent ofthe vascular margin seen byfluoroscopy at the timeof the last angiogram. A significantly higher proportion ofatype II border had >2700 ofthe periphery seen than type I inSC disease (RE x: d=17-5, df=2, p<0 001; LE x2=26l1,df=2, p<0 001) andfor SS disease (p=0-053 for the REand 0-060 for the LE, Fisher's exact test)

SC disease SS diseaseBorder classification Border classification

Marginextent I IIa Ilb I IIa IIb

<1800 11 1 1 31 0 0<2700 58 17 17 186 1 5>2700 43 68 55 207 3 19Total 112 86 73 424 4 24

Table 7 Vascular obstruction in eyes ofSC subjects inwhich change was recordedfrom type I to type II (index eye)compared with eyes ofSC subjects and SS subjects in whichthe border retained type I characteristics as pairs matched asto age ofthe subject, number ofvisits, and age at the time ofthe last visit. Using McNemar's test (paired X2), indexpatients were different from the SC comparison group(x2= 11 03, df= 1, p<0 001), and the SS comparison group(X2= 138, df= 1, p<0 001). The SC and SS comparisongroups were not different (x2=0 00, df=1, p=l 0)

Index SC SSeye comparison comparison

No obstruction 22 44 43Old obstruction 10 6 4Recent obstruction 29 11 14Total 61 61 61

bed. The recent nature of the obstruction wasidentified by the presence of an abrupt cessationof the intravascular dye column of an arteriolewith a hypofluorescent column extendingbeyond it caused by absorption of light byintravascular blood or altered blood products(Fig 6). An eye was not classified as type III ifother regions of the border could be classified astype II (Fig 7).

Figure 8 The left eye ofa 9-year-old male with Hb SCdisease. Proliferative sickle retinopathy in association withtype IIa border.

The vascular border was designated as unread-able if the edge was too peripheral to be photo-graphed, the photographs were blurred owing toastigmatic aberration, or the fluorescence was toofaint.

Other angiographic featuresEvidence ofprevious acute arteriolar occlusion inany frame was recorded, whether or not thevascular border was involved. A distinction wasmade between vessels which did or did notcontinue as a hypofluorescent linear column. Thepresence of this feature was believed to signify arecent occlusion and, therefore, the possibility ofsubsequent reperfusion.The extent ofnon-perfused retina was assessed

according to its posterior displacement and thecircumferential extent of the vascular marginthat could be seen by fluoroscopy.2 The presenceof a posterior location of the border was desig-nated on fluoroscopy if the vascular border waspost-equatorial.

Scoring procedurePositive and negative transparencies were exam-ined under magnification, and the classificationofthe vascular border was limited to the temporalretina. Up to 1991, angiograms of each eye for

Fig 9A

Fig 91

Figure 9 Right eye ofa 10-year-old male with SC diseasewith a type I border (A). Within 1 year there is a change inthe vasculature with loss ofa large peripheral loop (arrow)but without change in the designation (B).

684


http://bjo.bmj.com

/B

r J Ophthalm


ber 1994. Dow

nloaded from

http://bjo.bmj.com/


Fig IOA Fig JOB

Figure 10 Left eye ofa 16-year-old male with SC disease designatea type 1 (A). Oneyear later there was a major change ofthe vascular bed with loss ofcapillaries and vascular obstruction (arrow) (B). However, the edge retains type I characteristics.

Fwig IIA

every year were scored according to the proposedclassification system. The four observers were

masked from one another and from the genotypeof the subject. There was no masking of theangiogram scores of that subject in previousyears. Those angiograms for which there was notuniversal agreement were reviewed by the groupas whole, and a final coding agreed upon. Ameasure of interobserver agreement was calcu-lated using kappa statistic. Good reliability was

established, and the proposed coding system wasshown to be practicable. For 1992, two observersclassified the marging independently, beingmasked as to the previous classification of theangiogram, and a third arbitrated in the case ofdisagreement.

Correlates ofclassificationThe data were analysed to ascertain whether or

not correlations existed between the type ofvascular pattern and age, sex, and genotype ofthe subject. Comparison was also made withcertain retinal features:

(a) The circumferential extent of visibleperipheral retinal vascular non-perfusion as

Fig JPIB

Figure 11 Left eye ofmale with SS disease at 14years hada type I border (A). Two years later there was a majorrecession of the border which now had type II characteristics(B). Corresponding artenral divisions have been marked withan arrow.

judged by fluoroscopy.(b) The posterior location of the vascular

margin.(c) The number of retinal haemorrhages,

chorioretinal scars, and schisis cavities.(d) The presence of vessel occlusions. Two

matched pair studies were performed. In thefirst, subjects with SC disease in whom a type Iborder changed to a type II were matched bygenotype, age in years, and number of visits (± 1)with an individual with persistent type I border.In the second, each SC subject with a type IIborder was also paired with one with SS diseasewho retained a type I border using the same

criteria. The pairs were compared with respect tothe number with a recorded vascular obstruc-tion, the number of obstructive events detected,and a distinction was made between acute and oldobstruction.

(e) The presence of PSR.

Dynamics ofchangeThe serial angiograms of the 24 eyes of 18subjects in which PSR developed during theperiod of the cohort study were examined for

685


http://bjo.bmj.com

/B

r J Ophthalm


ber 1994. Dow

nloaded from

http://bjo.bmj.com/

Penman, Talbot, Chuang, Thomas, Serleant, Bird

change between annual visits. AM but one of thesubjects had SC disease.

ResultsOf the 308 SS subjects and the 173 SC subjects

Fig 12A

Fig 12B

Fig 12C

Figure 12 Left eye of 10-year-oldfemale with SC diseaseat which time the edge was not photographed (A). Within 1

year the edge had type II characteristics (correspondingvessels marked with a curved arrow) (B). A furtheryear laterthere was major recession ofthe edge (corresponding vesselsare marked with a straight arrow) (C).

recruited to the study at birth, 260 and 154respectively were available for examination at theage of 5 years. Of those not seen 68 were eitherlost to review soon after birth, died, oremigrated. Four were unsuitable for studybecause ofother eye disease and 17 did not attendan annual review. In each year, between 93% and96% ascertainment was achieved of those knownto be available for study. Photographs of thevascular border could be classified at least once asnormal or abnormal in 373 of the subjects. In theremainder the border was either too peripheralfor it to be photographed, or no designation asnormal or abnormal could be made (type III orunclassifiable).Of the 390 subjects categorised up to 1991,

there was universal agreement on the assignmentofboth eyes for all years ofstudy between all fourreaders in 336. The agreement between pairs ofobservers for the distinction between normal(type I) and abnormal (type II) (six comparisons)gave kappa values between 0 79 and 0 71. Therewas more disagreement as to the distinctionbetween IIa and IIb although the lowest kappavalue for'this was 0-64.When compared using the unpaired t test, the

SC and SS groups were similar in terms of themean number of ophthalmic examinations (5-9(SD 2 5) and 5 8 (2 5) respectively; p=0 87), thenumber of years between the first and last visits(5 8 (2 8) and 5 7 (2 8) respectively; p=0 76) andthe number of angiograms in which the bordercould be classified as normal or abnormal (4 6(2 4) and 4 9 (2-6) respectively; p=031). In SCsubjects the prevalence of pattern IIa borderincreased with age (X2 for trend=29-3, p<001),and a type II border was considerably morecommon in SC than in SS disease (Tables 1, 2).This was not influenced by sex in either genotype(x2 test overall=0-73, p=0 69). In the majorityof cases a change from normal to abnormal wasrecorded although in some an abnormal borderwas seen on the first angiogram in which theborder was photographed. The age ofconversionwas lower in SC disease than SS (Table 3). Theborder classification tended to be similar for botheyes of a single individual, although asymmetrywas not uncommon (Table 4).The borders classified as II were more post-

erior and had a greater circumference of theborder seen than those classified as type I,although a posterior position allowing full visu-alisation of 3600 of the border was not exclusiveto fundi with a type II designation (Table 5, 6).The number with documented vascular

obstruction, both acute and long term, and thenumber of recorded occluded vessels for eachyear recorded was greater in those who developedan abnormal border, although such events werealso seen in those in whom the border remainednormal (Table 7).

Proliferative retinopathy was been seen in only24 eyes of 18 subjects (17 with Hb SC and onewith Hb SS), and was always associated with atype II border (23 hIa and one IIb) (Fig 8). Serialphotographs of the same area were not availablein one of these subjects. In this case the borderwas too peripheral to be photographed before theonset of a type II border, and at the three annualvisits thereafter either the same area was not

.686


http://bjo.bmj.com

/B

r J Ophthalm


ber 1994. Dow

nloaded from

http://bjo.bmj.com/

687New classification ofperipheral retinal vascular changes in sickle cell disease

Figure 13 Right eye ofmale aged 12 years with SCdisease shozving a type IIaborder (A). Within I yearthere was loss ofa majorperipheral loop (B).

Fig 13A

Fsig 131B

photographed or the extent of remodelling of thevasculature prevented recognition of the same

area. In the remaining 23 eyes, the evolution ofchange in the peripheral vasculature variedgreatly from one subject to another. In 11 eyes ofnine subjects a type I border was recorded at twoor more (range 2-7) annual visits before a type IIbeing identified. In three eyes changes in thevascular bed were identified without there beingan alteration of category (Figs 9, 10). In eighteyes the border was stable for as long as 6 years

before conversion to a type II border. In theremaining 12 eyes good angiograms were

obtained but the border was too peripheral to bephotographed before demonstration of a type IIborder, and after conversion it was possible toidentify that the new border corresponded withan area photographed that had been within thevascular bed (Figs 11, 12).

In 18 eyes of 14 subjects serial photographswere available of the same area at two or more

annual visits (range 2-7) after a type II borderhad been recognised before development ofPSR.In all but four eyes there was progressive changeof the vascular bed (Figs 12, 13). In the remain-ing five eyes, PSR was recorded at the same

examination or within 1 year of demonstration ofa type II border (Fig 14).

DiscussionThe peripheral retinal vasculature in sickle celldisease presents different patterns which corre-late with genotype, and indicates the risk of thedeveloping visually threatening disease.

Interobserver agreement, as measured bykappa statistic, was good such that the classifica-tion system appears to be reliable and practic-able. The main area of disagreement was in thesubclassification of abnormal patterns ratherthan in the distinction between normality andabnormality.There are clear limitations inherent in this

cohort study when seeking incidence data, in thatthe ophthalmic findings are recorded only once ayear. For example, multiple events may occurduring any year which would be recorded only asa single event, and vascular obstruction followedby full restoration ofthe circulation would not bedetected; such an event has been well described.15In addition, since only 900 of the peripheralretina was photographed the survey is far fromcomplete. Alterations in the retinal vasculatureoutside this area were not recorded routinely andwere ignored in this study when available.Nevertheless, from these observations certaingeneral conclusions can be drawn concerning theimplications of different vascular patterns andthe dynamics of change in the peripheral retinalcirculation.Although the type I border is qualitatively

similar to that seen in normal subjects this doesnot imply that the type I border is unmodified bydisease. The posterior location and the extent ofthe visible border were apparently different fromnormal," and it was a strong impression that thetype I border in sickle cell disease was mucheasier to photograph than in subjects with normalhaemoglobin (AA) genotype. The findings mightbe influenced by greater cooperation of subjectsas they became used to the examination pro-cedure; those with SS and SC disease werephotographed on several occasions whereas thosewith normal haemoglobin genotype were seenonly once. Better cooperation could also accountfor the increasing proportion in whom the bordercould be photographed as the study progressed,although the proportion in whom the border wasphotographed did not change in the sickle sub-jects after the age of9 years. It is also evident thatthe degree of pupillary dilatation may influenceobservation. For this reason the examination wasnot undertaken until full dilatation was achieved.The procedure limited classification ofthe vascu-lar border to the temporal retina to reducevariability in the data.That alteration of the vascular border may

occur without the vascular pattern changing indesignation from type I illustrates the type Iborder is not always stable in sickle cell disease.This is illustrated further by a postmortemmorphological study of an eye from a 20 monthold infant with SS disease.'6 Using adenosinetriphosphate labelling of flat retinal vascularpreparations, there was good evidence of arterio-lar obstruction and consequent loss peripheralretinal capillary bed. In addition, vascular occlu-sive events are well recorded in early life outsidethe retina. 7 These observations collectively indi-cate that retinal vascular remodelling, which is a


http://bjo.bmj.com

/B

r J Ophthalm


ber 1994. Dow

nloaded from

http://bjo.bmj.com/

Penman, Talbot, Chuang, Thomas, Serjeant, Bird

Figure 14 Left eye in a9-year-old male with SCdisease with a type IIaborder (A). One year laterthere was further loss ofperipheral vessels, occlusior.ofa major vessel (arrow),and proliferative disease(B).

Fig 14A

Fig 14B

feature of normal early postnatal develop-ment,18.2' continues well after the age of 6 monthsin sickle cell disease resulting in posterior dis-placement of the vascular border. The changewould be slow with occasional loss of capillariesat the border of vascular bed or within it. Weconclude that the increase of the circumferenceseen with age and the posterior position of theborder in sickle cell disease reflects alteration ofthe vascular bed rather than an artefact inherentto the study design. Such a gradual evolutionappears not to evoke neovascularisation, since atype I border is not associated with proliferativedisease however posterior, implying that a post-erior border with qualitatively normal perfusioncharacteristics protects against proliferativedisease. The gradual progression of ischaemiamay cause loss of viable retina which mightotherwise have been the source of diffusibleagents causing vasoproliferation.

This slow process can be contrasted with themore radical modification of the vascular patterngiving rise to a type II border. The proportion ofeyes that developed PSR in which the anteriorborder of perfusion was too peripheral to bephotographed before the documentation of a

type II border is not different from that seen insubjects with normal (AA) haemoglobin geno-

type," and is less than that in the whole cohortwith sickle cell disease. This indicates that theremay have been little modification of the vascular

bed before the type II border was recognised, andthat conversion occurs in a vascular bed which isrelatively unaltered by the sickle cell disease. Itfollows that the qualitative attributes of theperipheral retinal vasculature in sickle celldisease may reflect the temporal profile ofchangeas much as the presence or absence of change.The higher prevalence of circulatory stasis insubjects who developed a type II border impliesthat vascular obstruction is important in causingthe conversion. The detection of obstructedvessels in 64% of eyes on at least one occasion,and the short period of vascular stasis in suchepisodes suggest that this is a very commonphenomenon in young individuals with sicklecell disease. It also indicates that the majority ofepisodes do not result in a major alteration of thecapillary bed given the stability of the vascularsystem in most subjects. Complete reperfusion ofthe vascular bed has been recorded even follow-ing obstruction of large retinal arterioles.22No ready explanation exists for the different

temporal profile of retinal vascular changesbetween Hb SS disease when compared with HbSC, since circulatory stasis was identified in both.The relatively lower figure of detected vascularobstruction in subjects with SS genotype impliesthat either vascular obstruction is less common orthe duration of circulatory disturbance is muchshorter in this group. Either alternative could bedue to the lower haematocrit which characterisesHb SS when compared with Hb SC. Thisargument has some support in that systemiccomplications of vascular obstruction arerecorded to be more common in subjects with ahigh haemoglobin and packed cell volume.8 Ahigher haemoglobin has been found in subjectswith PSR than in those without, although there issome doubt as to whether this was an agedependent relation rather than being a risk factorfor PSR.' It is also possible that remodelling ofthe vascular bed in early life in subjects with HbSS giving rise to larger peripheral capillaries at amore posterior locus protects the vascular systemfrom the consequences of vascular obstruction.Although a type II border was not invariably

recorded before development of PSR, the possi-bility of a change in the vascular bed before thedevelopment of PSR cannot be excluded giventhe frequency of vascular events and the longinterval between study visits. In no instance wasa type I pattern seen to coexist with PSR in thesame region, though rarely other areas of thesame eye retained a normal pattern. The demon-stration that PSR follows a change of the periph-eral vascular organisation, and that this change isassociated with SC rather than SS diseaseaccounts for the more serious visual prognosiswith this genotype despite the relative mildnessof their systemic disease. This has been adilemma for many years, and the various expla-nations put forward in the past have not beenconvincing. It has been suggested that althoughperipheral non-perfusion, which is the putativestimulus to proliferation, is more common in SSthan SC, the frequency of autoinfarction ofproliferative complexes in SS disease preventsPSR becoming extensive.22 Although this cannotbe excluded as a possible protective factor, thedata in this study suggest that this is unlikely to

688


http://bjo.bmj.com

/B

r J Ophthalm


ber 1994. Dow

nloaded from

http://bjo.bmj.com/


be the more important determinant of the differ-ential threat to vision in the two genotypes.We believe that this classification of the

peripheral retinal vasculature has potential clini-cal application in sickle cell retinopathy and isimportant for screening. Identification of thepattern of the peripheral vasculature allowspatients at risk of developing PSR to be identi-fied. A subject with a type I border at a posteriorlocation is most unlikely to develop PSR,whereas the reverse is the case if there are type IIcharacteristics. Perhaps of most importance isthe implication that the time course of change inthe vascular bed is the major determinant of theobserved angioarchitecture, and, therefore, themagnitude of risk of proliferative disease.We acknowledge the assistance of Mr Rob Acheson, Mr BrendanMoriarty, and Mr Peter Fox. Statistical calculations were done byDr J Morris and Dr P Thomas. Film processing and contactprinting were done by D Ballantyne.

1 Goldberg MF. Retinal neovascularization in sickle cell retin-opathy. Trans Am Acad Ophthalmol Otolaryngol 1977; 83:409-31.

2 Talbot JF, Bird AC, Serjeant GR, Hayes RJ. Sickle cellretinopathy in young children in Jamaica. BrJ3 Ophthalmol1982; 66: 149-54.

3 Talbot JF, Bird AC, Rabb LM, Maude GH, Serjeant GR.Sickle cell retinopathy in Jamaican children - a search forprognostic factors. BrJ Ophthalmol 1983; 67: 782-5.

4 Talbot JF, Bird AC, Maude GH, Acheson RW, Moriarty BJ,Serjeant GR. Sickle cell retinopathy in Jamaican children:further observations in a cohort study. Br Ophthalmol1988; 72: 727-32.

S Galinos SO, Asdourian GK, Woolf MB, Stevens TS, Lee CB,Goldberg MF, et al. Spontaneous remodelling of the periph-eral retinal vasculature in sickling disorders. AmJ3 Ophthal-mol 1975; 79: 853-70.

6 Goldberg MF. Classification and pathogenesis of proliferativesickle retinopathy. AmJ Ophthalmol 1971; 71: 649-65.

7 Condon PI, Serjeant GR. Ocular findings in homozygous sicklecell anaemia in Jamaica. AmJ Ophthalmol 1972; 73: 533-43.

8 Serjeant GR. Sickle cell disease. Oxford: Oxford UniversityPress, 1992: Chapter 21, 332.

9 van Meurs JC. Relationship between peripheral vascularclosure and proliferative retinopathy in sickle cell disease.Graefes Arch Clin Exp Ophthalmol 1991; 229: 543-8.

10 Fox PD, Dunn DT, Morris JS, Serjeant GR. Risk factors forproliferative sickle retinopathy. Brj Ophthalmol 1990; 74:172-6.

11 Penman A, Talbot JF, Chuang EL, Bird AC, Serjeant GR.Peripheral retinal vasculature in normal Jamaican children.Brj Ophthalmol 1994; 78: 615-7.

12 Serjeant GR, Grandison Y, Lowrie Y, Mason K, Phillips J,Serjeant BE, et al. The development of haematologicalchanges in homozygous sickle cell disease; a cohort studyfrom birth to 6 years. Brj Haematol 1981; 48: 533-43.

13 Serjeant GR, Serjeant BE, Forbes M, Hayes RJ, Higgs DR,Lehman H. Haemoglobin gene frequencies in the Jamaicanpopulation: a study in 100,000 newborns. Br Haematol1986; 64: 253-62.

14 Hamilton AM, Marshall J, Kohner EM, Bowbyes JA. Retinalnew vessel formation following experimental vein occlusion.Exp Eye Res 1975; 20: 493-7.

15 Condon PI, Whitelock RAF, Bird AC, Talbot JF, SerjeantGR. Recurrent visual loss in homozygous sickle cell disease.BrJ Ophthalmol 1985; 69: 700-6.

16 McLoed GS, Goldberg MF, Lutty GA. Dual-perspectiveanalysis of vascular formations in sickle cell retinopathy.Arch Ophthalmol 1993; 111: 1234-45.

17 Rogers DW, ClarkeJM, Ramlal AM, Sparke BR, Serjeant GR.Early deaths in Jamaican children with sickle cell disease.BMJ 1978; 281: 1515-6.

18 Ashton N. Studies or the retinal capillaries in relation todiabetic and other retinopathies. BrJ Ophthalmol 1963; 47:22-38.

19 Cogan DG. Development and senescence of the human retinalvasculature. Trans Ophthalmol Soc UK 1965; 83: 465-89.

20 Wise GN, Dollery CT, Henkind P. The retinal circulation. NewYork: Harper, 1971: 1-18.

21 Foos Y, Kopelow SM. Development of the retinal vasculaturein paranatal infants. Clin Path Rev 1973; 18: 117-27.

22 Condon PI, Serjeant GR. Behaviour of untreated proliferativesickle retinopathy. BrJ Ophthalmol 1980; 64: 404-11.

689


http://bjo.bmj.com

/B

r J Ophthalm


ber 1994. Dow

nloaded from

http://bjo.bmj.com/

new peripheral in · borderclassification borderclassification classification i iia iib i iia iib...

Documents