in the left eye. Subsequently, a vitreous traction band developed from the hemorrhage to the macula. This band presumably tugged on the macula, leading to cystic changes therein, and reduced central visual acuity from 20/15 to 20/200.

Comment

The small number of eyes (25) and the relatively short follow-up period (6 to 77 months) make additional study mandatory before precise conclusions can be reached concerning the rate and mode of progression of untreated PSR. Nonetheless, certain tentative conclusions seem possible. From the limited data available, untreated PSR appears to have progressed slowly during the follow-up period by having passed through the stages and substages of the classification used herein (Fig 2 to 4,7,10, and color Fig 1 to 5). In this sense the severity and extensiveness of PSR may be considered somewhat age dependent. In addition, data from

the original population suggest that severity of PSR possibly may be related to an increasing average patient age, although the age ranges of the various stages of PSR overlap (Table 4).

Of perhaps more practical importance is the observation that the stage of neovascular and fibrous proliferations (stage III) had been reached by 11 patients whose mean age was only 25 (Table 4). Moreover, the subsequent two stages, during which clinical symptomatology is first manifested (stage IV, vitreous hemorrhage; stage V, retinal detachment [color Fig 1 to 5]), had been reached by ten patients whose mean ages were also 30 or less (Table 4). Consequently, although inspection of Fig 2 to 4, 7, and 10 indicates that progression of PSR seems to be relatively slow and may be somewhat age-dependent, stages of clinical significance (stages III, IV, and V1,2) occur at a relatively early age.

Once stage III has been reached, further increase in its severity (ie, an increase in its

*If two eyes of one patient differed in stages, the more advanced stage of the two was used for purposes of this tabulation. One patient was not included because his age was unknown.

Because most patients' retinas proceeded sequentially through earlier stages prior to showing signs of later stages, the categories involving later stages are listed, for example, I-IV, rather than simply IV.

Three patients in the original population had retinal detachments. Because they required surgery, these eyes were not included in the follow-up study and thus do not appear in Table 2 and 3.

case 3 and 5, Table 1, Fig 8 and 9), thereby threatening the eye with potential vitreous hemorrhage or retinal detachment or both (color Fig 1 to 5).1 Those factors causing or potentiating such a progression are currently obscure. Since pregnancy was associated with florid progression of stage III lesions in one patient, it is possible that the "pregnant state" may be one of those factors in certain circunstances. An analogous situation has

been postulated for pregnant diabetics with PDR, although the data are not entirely clear cut.8

In SC disease almost all eyes with vitreous hemorrhage appear to have undergone neovascular proliferations prior to any episode of bleeding. The reverse sequence of events (ie, spontaneous vitreous hemorrhage followed by previously nonexistent neovascularization) only rarely, if ever, appears to occur, but may possibly apply to the right eye of case 13 (Table 1). Alternatively, the responsible patch of neovascularization may simply have been small enough to escape detection in this eye.

Initially, vitreous hemorrhages may be totally asymptomatic because the extravasated blood may remain localized in the peripheral vitreous, which overlies the bleeding neovascular tissue (color Fig 1 to 3). Such blood slowly turns from red to white and may become organized. The underlying neovascular tissue may then proliferate further or may re-bleed in an erratic, unpredictable fashion with varying severity. Occasionally,

437

a vitreous hemorrhage breaks into the center of the vitreous space, thereby producing visual symptoms. Marked reduction of visual acuity (of variable duration) may occur. Blunt trauma may uncommonly precipitate a fresh vitreous hemorrhage, 10 but most bleeding episodes are apparently spontaneous. Although vitreous detachment is intimately related to the pathogenesis of vitreous hemorrhages in diabetes mellitus, its possible role in PSR is not yet known.

This investigation was supported in part by the Bailey Fund of the Wilmer Institute and by National Institutes of Health special fellowship 2F11 NB01675-02VSN. Hemoglobin analyses were carried out in the Hematology Division, Department of Medicine, Johns Hopkins Hospital (supported by grant HE-02799 from the National Heart Institute). Photographic assistance was provided by Terry George and Norbert Jednock.

Key Words. Sickle cell disease; retinopathy, proliferative; natural history; classification; arteriolar occlusions; arteriolar-venular anastomoses; sea-fans; neovascularization; vitreous hemorrhage; retinal detachment; ischemia.

1. Goldberg MF: Proliferative sickle retinopathy: Classification and pathogenesis. Amer J Ophthal, to be published.

2. Ryan SJ, Goldberg MF: Anterior segment ischemia following scleral buckling in sickle cell hemoglobinopathy. Amer J Ophthal, to be published.

3. Romayananda N, Green WR, Goldberg MF: Histopathologic study of sickle cell retinopathy including trypsin digestion study. Read in part before the Association for Research in Ophthalmology, Sarasota, Fla, 1970.

4. Welch RB, Goldberg MF: Sickle-cell hemoglobin and its relation to fundus abnormality. Arch Ophthal 75:353-362, 1966.

5. Newcomb N, Goldberg MF, et al: Intravenous fluorescein photography of the ocular fundus in

Proliferative retinopathy is characteristic of certain abnormal hemoglobin states, particularly sickle cell hemoglobin C disease (SC disease).1 Spontaneous vitreous hemorrhage, which can be either localized or diffuse, and retinal detachment, which is exceedingly resistant to successful surgical repair,3 are the ultimate results of a predictable sequence of events which occur in the retinae of such patients.

This paper will present a classification of proliferative sickle retinopathy (PSR), which reflects both severity of the disorder as well as its postulated pathogenesis. Repeated ophthalmoscopic and fluorescein angiographic observations of 24 untreated SC patients were utilized in an attempt to reconstruct the ostensible series of retinal and preretinal events that may eventuate in vitreous hemorrhage and retinal detachment.

MATERIALS AND METHODS

Twenty-four Negro patients (47 eyes) in various untreated stages of PSR were personally evaluated after referral from the outand in-patient services of the Johns Hopkins Hospital and the Wilmer Institute. Nine of the patients were male and 15 were female. Observations made in 13 of these patients provided the clinical bases for a description of the characteristic ophthalmoscopic finding of SC retinopathy, the sea-fan sign.1 The seafan is an arteriolar-venular fan-shaped vasoproliferative lesion which projects from the retina into the vitreous. The name was cho

From the Departments of Ophthalmology and the Divison of Medical Genetics, The Johns Hopkins Hospital, Baltimore, Maryland. Supported in part by the Bailey Fund of the Wilmer Institute and by USPHS Special Fellowship 2F11 NBO167502VSN.

Reprint requests to Morton F. Goldberg, M.D., Professor and Head, Department of Ophthalmology, University of Illinois Eye and Ear Infirmary, 1855 W. Taylor Street, Chicago, Illinois 60612.

sen because superficially the typical lesion resembles a common marine invertebrate called the sea-fan (Gorgonia flabellum).

Evaluation of all fundi was accomplished by indirect ophthalmoscopy through dilated pupils, scleral depression, fluorescein angiography, and, when indicated, Goldmann 3mirror lens studies, although 3-mirror evaluation of the peripheral vitreoretinal interface was somewhat difficult in most of these patients, making it impossible to evaluate precisely any peripheral vitreoretinal adhesions or vitreous detachments in the SC cases.

Evaluation of most patients was repeated on several occasions and findings were recorded on 8 X 12-inch fundus drawings. In addition, color and black and white fundus photographs, and fluorescein angiographic photographs were made according to techniques published previously.*,5

Diagnosis of SC hemoglobin was confirmed by Samuel Charache, M.D., of the Special Hematology Laboratory, Johns Hopkins Hospital, via hemoglobin electrophoresis.

CLASSIFICATION

As the result of observations from indi

rect ophthalmoscopy and fluorescein angiography, the sequence of fundus events in these SC patients could be divided into five naturally occurring, progressive stages, one leading into the next (Table 1):

Stage I: Peripheral arteriolar occlusions.

Stage II: Peripheral arteriolar-venular

anastomoses.

Stage III: Neovascular and fibrous proliferations.

Stage IV: Vitreous hemorrhage.
Stage V: Retinal detachment.

Each of these five stages was further sub

650

MARCH, 1971

ment of 91 or more degrees of the fundus.

This classification reflects largely the equatorial and pre-equatorial pattern that PSR naturally assumes. It places major emphasis on the initial 90 degrees of involvement (substages 1, 2, and 3), because most lesions of PSR are confined to an area in the fundus less than one quadrant in circumferential extent.

In the case of Stages I, II, and III, further modification of the substages was employed, based on the number of characteristic fundus abnormalities visualized with the binocular indirect ophthalmoscope of or by fluorescein angiography. For example, modifications of Stage I (peripheral arteriolar occlusions) included:

Substage 1: 30° circumferential involvement and <3 occluded arterioles. Substage 2: 31-60° circumferential involvement or 4-6 occluded arterioles. Substage 3: 61-90° circumferential involvement or 7-9 occluded arterioles. Substage 4: 91° circumferential involvement or ≥ 10 occluded arterioles. The number of peripheral arteriolar-venular anastomoses, as visualized with the in direct ophthalmoscope or by fluorescein angiography, was also used to quantitate Stage I and the number of neovascular and fibrous proliferations was also used to quantitate Stage III (Table 1). For purposes of classification, no attempt was made to distinguish between neovascular proliferations and fibrous proliferations, because both forms of proliferation frequently co-existed in the same lesion.

If the circumferential involvement and the number of abnormal arterioles, anastomoses, or proliferations did not coincide within one substage, the substage used for purposes of classification was the more advanced of the two parameters: e.g., if a patient had four small neovascular proliferations totalling 20 degrees of angular extent, the classification for neovascularization was Stage III-2 (and not III-1). Also, if there was one large con