The fluorescein angiography revolution: a breakthrough with sustained impact

The advent of fundus fluorescein angiography (FA) represents a technological advance that has revolutionized the diagnosis and treatment of chorioretinal disease. This review will report some of the observations made by FA that have formed the basis of our present concept of the nature of many chorioretinal diseases, using diabetic retinopathy, retinal vein occlusion and exudative degeneration of the macula as examples. Fluorescein angiography of the fundus was introduced by Novotny & Alvis (1961). Due to the sodium fluorescein molecule's high degree of fluorescence and the minimal penetrability of the background melanin pigment within the retinal pigment epithelium (RPE), fluorescein dye proved to be ideal for studying retinal circulation and retinal vasculature. It soon became obvious that FA was an important aid in both the diagnosis and study of the pathogenesis of posterior ocular diseases, such as diabetic retinopathy and other retinal vascular diseases and macular diseases, in particular (Dollery et al. 1962; Scott et al. 1963, 1964; Norton et al. 1964, 1965; Norton & Gutman 1965). The most characteristic angiographic features in some of these diseases were discovered as early as the 1960s, although it was some time before they were widely recognized by the ophthalmologic community. An example of this is the pathogenesis of disciform degeneration of the macula. In some diseases, such as retinal venous occlusion, the information obtained by FA was not fully beneficial until larger studies comparing natural history and fluorescein angiographic findings had been carried out. The basics of fluorescein angiography have been described in many handbooks. The key point of the technique is the ability of the sodium fluorescein molecule to emit green light (wavelength of 520–530 nm) when it is stimulated by blue light (wavelength of 465–490 nm). Therefore, two filters are needed to perform FA: a blue exciter filter to produce fluorescence, and a green filter to collect the green fluorescence and block the reflected blue light. After injection, 60% of sodium fluorescein is bound to plasma proteins, particularly to albumin. The free dye passes readily across systemic capillaries and enters the tissues. Retinal capillaries do not leak fluorescein. Loss of integrity of the blood–retinal barrier either in the endothelial cells of the retinal vessels or in the RPE permits leakage of free fluorescein into the extracellular spaces of the surrounding neurosensory retina. New vessels in the retina, choroid and iris leak profusely. Diabetic retinopathy was one of the first entities where FA was used for both research and diagnostic purposes (Scott et al. 1963, 1964; Norton & Gutman 1965; Kohner et al. 1967). The clinical picture of diabetic retinopathy was well known before the advent of FA, and most of the changes that occur in the retinal vasculature in diabetic eye disease had been described in enucleated eyes using in vitro techniques such as flat preparations injected with Indian ink (Ashton 1950), stained trypsin digested preparations (Cogan et al. 1961) and electron microscopy (Bloodworth & Molitor 1965). Fluorescein angiography afforded, however, in vivo information on blood flow and the development and progression of diabetic retinopathy. Fluorescein angiography revealed the full extent of abnormalities of the capillary bed, including microaneurysms, dilated capillaries, capillary closure and abnormal permeability of the retinal vasculature (Scott et al. 1964; Kohner et al. 1967) (Fig. 1A, B). It has become obvious that microaneurysms and other capillary changes are usually more extensive than can be suggested by ophthalmoscopy. Leaking vessels are detected only by angiography, and capillary closure and even small new vessels may be difficult to detect by clinical examination. Using a super-wide montage fluorescein angiographic technique, Shimizu et al. (1981) showed that the midperipheral retina was far more prone to undergo capillary non-perfusion than the posterior retina (Fig. 2A, B). (A, B) Diabetic retinopathy with microaneurysms, capillary dilatation, leakage and capillary closure. (A, B) Diabetic retinopathy with predominantly midperipheral capillary non-perfusion and profuse leakage from early preretinal neovascularization. Before the era of FA, the pathogenesis of hard or lipid exudates was controversial, with debate over whether they were due to neuronal degeneration consequent to capillary damage or to leakage from damaged capillaries. The plasmal origin of the lipid accumulations in the retina had already been suggested by Toussaint et al. (1962), but FA confirmed the concept. Using fluorescein angiographic data, Maumenee (1968) showed that, in most instances, hard exudates are a result of leakage of plasma contents, which are selectively absorbed by the surrounding healthy vessels in such a way that the lipid content remains in the extravascular retinal tissue in an area where blood vessels are normal outside the areas in which FA reveals leaking vessels (Fig. 3A, B). Another convincing item of evidence for this conception was the effect of closure of the leaking capillaries by photocoagulation (Meyer-Schwickerath & Schott 1968; Wessing & Meyer-Schwickerath 1968; Spalter 1971). Despite this evidence, the plasmal origin of lipid deposits was still considered questionable in the early 1970s (Ballantine & Michaelson 1973). (A, B) Diabetic retinopathy with localized macular oedema and circinate lipid exudates in the area of leaking capillaries. In the 1960s, the term ‘diabetic maculopathy’ was not yet used, nor was it mentioned in the handbooks of the early 1970s (Wise et al. 1971; Ballantine & Michaelson 1973). Dobree (1970) defined diabetic maculopathy as a condition of visual loss due to macular oedema, in the presence of microaneurysms, haemorrhages and, commonly, hard exudates. From the treatment point of view it became important to differentiate between ‘perfused macular oedema’ with dilated perifoveal capillaries and microaneurysms with leakage, and ‘ischaemic oedema’, related to macular capillary closure with little or no fluorescein leakage (Fig. 4A, B). The extent of fluorescein leakage divided diabetic maculopathy into focal or diffuse, and some of the diffuse cases were cystic (Bresnick et al. 1983). Perfused (A) versus ischaemic (B) diabetic maculopathy. The currently widely used term ‘clinically significant macular oedema’ (CSME) was introduced in the 1980s by the Early Treatment Diabetic Retinopathy Study (ETDRS) Research Group (1985). This diagnosis does not necessitate FA; it can be recognized by ophthalmoscopy or by stereo colour photographs. In the ETDRS, CSME was defined as an area of definite retinal thickening, with or without hard exudates, within 500 µm of the foveola, or an area of thickening greater than one disc area in extent, involving the region within one disc's diameter of the foveola. The ETDRS showed that CSME may lead to moderate visual loss if not treated with focal photocoagulation. The pathogenesis of cotton wool spots as retinal microinfarcts was also confirmed by FA (Hodge & Dollery 1964; Kohner et al. 1967, 1969). The appearance of multiple cotton wool spots was first considered to be a preproliferative sign indicating a rapidly advancing retinopathy and high risk of neovascularization (Kohner & Dollery 1975; Bresnick et al. 1983). Subsequent studies did not, however, agree on their preproliferative status and poor prognostic value, although cotton wool spots are often seen in severe retinopathy. For example, during pregnancy the development of multiple cotton wool spots is a fairly common occurrence, but after delivery they may resolve without leading to the proliferative stage (Laatikainen et al. 1980) (Fig. 5A, B). (A, B) Diabetic retinopathy with multiple cotton wool spots (microinfarcts) during pregnancy. Venous abnormalities in diabetic retinopathy had already been observed by Nettleship in 1888 and have since been noted by many others (Kohner & Dollery 1975). Generalized venous dilatation is an early sign of abnormality. It may even occur in children with diabetes before any other signs of retinopathy develop (Falck & Laatikainen 1995). Focal venous changes, like venous beading and loops, have been found to occur characteristically in diabetes. Before FA, venous beading was explained as a feature of venous obstructive disease (Wise et al. 1971). Fluorescein angiography showed that venous beading is related to retinal capillary non-perfusion (Fig. 6A, B) and is thus an important preproliferative sign in diabetic retinopathy (Bresnick et al. 1983). (A, B) Diabetic retinopathy with venous beading and cotton wool spots in relation to extensive capillary non-perfusion. Another preproliferative set of symptoms beautifully demonstrated by FA concern intraretinal microvascular abnormalities (IRMA). These include dilated capillaries and/or intraretinal new vessels that develop in the non-perfused areas. Since the Airlie House classification of diabetic retinopathy was defined (Davis et al. 1969), IRMA has been regarded as one of the poor prognostic indicators of diabetic retinopathy. Preretinal and disc neovascularization, known as ‘rete mirabile’ or ‘retinitis proliferans’, as well as iris neovascularization were well known before the advent of angiography, but the pathogenesis of neovascularization remained unsolved. The existence of a vasoproliferative factor as a cause of new vessel formation was proposed by Ashton (1953), and Wise (1956) suggested that the stimulus for neovascularization arose from ischaemic retinal tissue. Angiographic pictures confirmed the coexistence of new vessels and capillary non-perfusion (Fig. 7). It was observed that all eyes with new disc vessels had large areas of peripheral non-perfusion (Kohner & Dollery 1975); neovascularization of the angle (Fig. 8) and neovascular glaucoma (NVG) were almost never seen in diabetes patients unless they had large areas of peripheral non-perfusion (Kohner 1978). Proliferative diabetic retinopathy with preretinal neovascularization and extensive capillary non-perfusion. Diabetic iris neovascularization with profuse leakage. Although FA has contributed greatly to our understanding of the development of diabetic retinopathy, it has been considered too demanding and cumbersome a technique to be used in the grading of retinopathy severity in large clinical treatment trials and in many other clinical studies carried out on diabetic retinopathy during the last 30 years. For example, in the two clinical trials on photocoagulation therapy, the Diabetic Retinopathy Study (DRS) Research Group (1981) and the Early Treatment Diabetic Retinopathy Study (ETDRS) Research Group (1985, 1991), the grading of retinopathy and maculopathy was based on colour stereo photography. The latest proposal for an international clinical classification of the severity of diabetic retinopathy and diabetic macular oedema is based on fundus photographs (Wilkinson et al. 2003). At present, FA is not regarded as essential before every laser treatment, particularly before panretinal photocoagulation for proliferative retinopathy. However, in the management of macular oedema FA is most helpful; it can demonstrate macular ischaemia and define the area of leakage. It may also assist in the investigation of unexplained visual loss and may be used to monitor the effect of any treatment of macular oedema. For the demonstration of macular oedema, angiography may nowadays be partly replaced or supplemented by optical coherence tomography (OCT). The clinical picture of retinal vein occlusion (RVO), characterized by dilated tortuous veins and varying amounts of retinal oedema and haemorrhage, was also well known before the era of FA, but it was not known which factors determined the course of the disease and the development of complications. In a retrospective clinical study on retinal vein occlusion, Raitta (1965) found that the only local factor having prognostic value was the development of papillary anastomoses. These had a favourable influence. However, early visual acuity (VA) after occlusion, with the exception of very poor vision (counting fingers at less than 2 metres), was not of prognostic value. Raitta (1965) noticed that 20% of eyes with central retinal vein occlusion (CRVO) but no eyes with branch retinal vein occlusion (BRVO) developed haemorrhagic glaucoma, but no local reasons for this difference were identified. Older age and the presence of glaucoma influenced late VA unfavourably and these factors were also found to be important for the development of haemorrhagic glaucoma. Raitta (1965) supposed that with regard to retinal venous occlusion ‘fluorescence photography will in the future clear many questions’. The first fluorescein angiographic findings in RVO were reported in the early 1960s (Dollery et al. 1962; Paton et al. 1964; Hill 1968). The first studies dealt mainly with the alterations of blood flow in the main retinal vessels, but other details – except leakage – were not very well documented. A more detailed and complete description of the recent and late fluorescein angiographic features of RVO was published by Gass (1968a). In fact, Gass reported all the characteristic features of RVO, such as delayed capillary and venous perfusion, capillary dilatation and tortuosity, occasional microaneurysms, focal areas of absent or deficient capillary filling, leakage of veins and capillaries, and pooling of dye within the cystoid spaces in the macula, as well as development of collateral capillaries, neovascular proliferations and cellophane maculopathy. Based on this study, Gass (1968a) suggested that FA was helpful in the differential diagnosis of macular oedema and in determining the degree of capillary and retinal damage secondary to venous occlusion. In the 1970s, the prognostic value of fluorescein angiographic findings in RVO, both BRVO and CRVO, received increasing attention. It was noticed that the main responses in the retina were: capillary dilatation and hyperpermeability, and non-perfusion of the capillaries (Laatikainen 1976; Laatikainen & Kohner 1976; Shilling 1976; Shilling & Kohner 1976) (Fig. 9A−C). These were the key factors that determined the visual prognosis and the occurrence of complications. In BRVO, chronic capillary leakage resulted in persistent macular oedema and lipid exudates, and capillary non-perfusion was related to new vessel formation, either from the retina or the optic disc (Archer et al. 1974; Kohner & Shilling 1976) (Fig. 9A−C). The efficiency of primary arterial perfusion to the territory of the obstructed vein was also thought to be important (Archer et al. 1974), but in later studies arteriolar changes proved to be progressive and mainly secondary to venous occlusion (Shilling 1976). Retinal branch vein occlusion. (A, B) Early and late picture of non-ischaemic macular branch vein occlusion. (C) Ischaemic upper temporal branch vein occlusion. Before FA, CRVO had been classified according to the completeness of the occlusion or the clinical picture as ‘impending’ versus ‘complete’, or ‘mild’ versus ‘severe’, or ‘venous stasis retinopathy’ versus ‘haemorrhagic retinopathy’ (Hayreh 1983). In a prospective series in classified retinal venous occlusion based on angiographic findings as either by dilated and leaking veins and or by vascular (Laatikainen 1976; Laatikainen & Kohner 1976) (Fig. This classification is In some more recent studies the of capillary non-perfusion has been more The Study Group the extent of capillary non-perfusion in photographs using a disc area eyes with less than disc areas of retinal capillary non-perfusion in the photographs were defined as and eyes with at disc areas of non-perfusion as intraretinal of perfusion the eye was defined as of disc areas of non-perfusion may be considered to fairly (A, B) Ischaemic peripheral branch vein occlusion with preretinal neovascularization. (C) neovascularization on the disc branch vein occlusion. of are non-ischaemic and are ischaemic (Laatikainen & Kohner 1976; et al. & from non-ischaemic to ischaemic however, (Fig. In our study to ischaemic within (Laatikainen & Kohner 1976). et al. reported in to the Study Group at study of eyes were and were non-perfused or However, during the course of of the eyes to ischaemia Study Group according to Study Group of the eyes to the ischaemic The only sign of ischaemia to be a in the of the veins on of non-ischaemic (A, B) to ischaemic central retinal vein occlusion within 2 In non-ischaemic CRVO full may occur and the visual is In our study of the non-ischaemic cases VA of or (Laatikainen & Kohner 1976). of the non-ischaemic eyes developed chronic macular oedema or pigment changes, with VA of or found that if the perifoveal capillary was was a to but in eyes with a persistent oedema was (Fig. B). This with the observed in eyes with BRVO et al. 1973). central retinal vein occlusion with perifoveal capillary (A) and persistent perifoveal leakage after In in ischaemic CRVO poor visual is the and neovascular are common (Fig. B). eyes with extensive central and peripheral vascular closure or during the have poor visual and iris new vessels and/or angle new vessels by neovascular glaucoma develop in of (Laatikainen & Kohner 1976; et al. et al. In the of the eyes defined as non-perfused or developed or Study Group New vessels on the optic disc or retina with or without are less Ischaemic central retinal vein occlusion with preretinal (A) and iris (B) neovascularization. The of the of CRVO and the or prognosis may be difficult if the fundus details be seen due to or extensive In these cases et al. or & et al. may be used to eyes with ischaemic fluorescein angiography may also be and leakage of the iris vessels with or without iris neovascularization to the ischaemic of CRVO (Laatikainen & The early of prognosis may also be by the development of extensive vascular closure in some eyes with an capillary perfusion as mentioned At after the symptoms the visual however, be in most cases by The present of the development and natural history of neovascular or glaucoma (NVG) is based on fluorescein angiographic The that retinal capillary non-perfusion was the main factor leading to iris neovascularization and both in CRVO (Laatikainen & Kohner 1976) and in diabetic eye disease (Kohner to treatment trials for the of using panretinal photocoagulation. Based on the first clinical (Laatikainen et al. and some series et al. 1976; Laatikainen panretinal photocoagulation of ischaemic CRVO was as soon as significant non-perfusion was detected and glaucoma may dilatation of the and the treatment The Study Group treatment only in eyes in which iris or angle neovascularization had already The to early neovascularization may be difficult to Before the advent of FA, the pathogenesis of the disciform or exudative degeneration of the macula was In the 1960s, described disciform degeneration as localized and of tissue the retina in the macular the was found to be a thickening of by a with or without between and the RPE and between the RPE and the Subsequent and of this resulted in a and of the retinal the was that the neovascular from the choroid seen in is a late that a in the into the after the of FA, in the early 1970s, the presence and the of neovascularization were not recognized (Wise et al. 1971; Ballantine & Michaelson 1973). The present that new vessel is an important factor in the formation of the disciform of the macula or exudative macular degeneration was demonstrated by Gass that the in the pathogenesis of disciform degeneration was the degeneration of the and but found that in some eyes FA demonstrated neovascular already in the stage on to that FA may helpful in the of the capillary of the and may thus as an aid in the treatment of the to the development of disciform At present it is that the primary in macular degeneration is of the It diffuse and localized deposits in and these are by pigment and/or these neovascularization is the at which are Gass the well defined new vessels can be localized with FA (Fig. and treatment of new vessels with laser proved Study Group in many cases the new vessels are or they are by blood or exudates, or are not well defined on FA (Fig. In a recent study, more than 60% of all eyes with neovascular were found to have defined et al. (A, B) well defined neovascularization with profuse leakage. neovascularization partly by and with large pigment In the well defined was classified as and defined as be if blood or lipid is with a that is not well on the fluorescein et al. The Study Group defined as either pigment with areas of early and late or leakage, or leakage of The and the extent of may not be determined by In the the fluorescein angiographic of were classified and with Study Group The classification of was developed during of clinical trials of of of with Study Group It is based on the of the in the predominantly where at of the area of the where less than of the and with no The of FA to the and the extent of was found to be This was for the of laser treatment of new vessels in et al. This was partly by the development of green angiography, which the of circulation not in exudative but also in other with green dye from fluorescein in it and light in the at and The light the of the fundus and than the used to the fluorescein of vessels, and of to with 60% of the of the & et al. (1970) and all reported However, also reported that had used fluorescein and green angiography in From the first it almost for the technique to be developed to a stage where clinical become & & et al. et al. Due to in including the and filters and of that were defined with FA & and proved helpful for to chorioretinal persistent or new vessels after the are with but the neovascular with fluorescein. Based on a et al. the of in the of and neovascularization with pigment as well as in of and et al. that may also be helpful in vessels in and it may a diagnosis and for in diseases the vasculature. green angiography however, The of small retinal vessels and of well defined is greater with FA than with In to other features of such as the presence of of and were demonstrated by FA as early as during the 1960s of the pigment either or laser treatment, was by FA later et al. Laatikainen The presence of new vessels in other exudative such as in high and focal was also discovered by FA during the 1960s new vessels were also found in relation to chorioretinal (Laatikainen & and without any obvious cause in (Laatikainen & 1983). The advent of fundus fluorescein angiography represents a technological advance that has revolutionized the diagnosis and treatment of chorioretinal Fluorescein angiography has revealed that the main abnormalities in diabetic retinopathy and in other retinal vascular diseases are abnormal vascular permeability and retinal capillary closure Despite in some FA has the of neovascularization in and other exudative of the macula. In to the diseases described as in this FA has to features and the pathogenesis of many other well known and discovered diseases of the ocular In fact, it be that photocoagulation and fluorescein angiography key factors the of on the over the last years.