8.2 Fluorescein Angiography

8.1 Color Fundus Photography

Color fundus photography is used to document the state of the fundus in RVO. This can be important as the appearance of the fundus changes with time (see Chap. 7).92 Fundus photographs are also used in research. In clinical studies, the photographic variables graded include number of hemorrhages, number of cotton wool spots, presence and degree of disk edema, presence and degree of macular edema, and absolute and relative diameters of retinal arteries and veins. The grading categories are typically expressed by subjective terms such as mild, moderate, and marked and may be guided by reference to standard photographs.39 The reproducibility of grading is not often provided or is provided in vague terms.39 It has been reported that presence of more hemorrhages than one-fourth of the posterior retina has 81Ð84% sensitivity and 72Ð74% speciÞcity for detecting ischemic CRVO.39 There is no evidence that such grading can be reliably generalized to clinical practice.

Absolute values of arterial and venular diameters from fundus photographs are rendered problematic by unknown variation in refraction indices of cornea, lens, and vitreous; unknown axial length of the eye; and unmeasured corneal curvature. There are attempts to correct for this, but they are approximations. By measuring ratios of the diameters of arteries and veins within a given eye, the problems of variable magniÞcation of photographs are circumvented. However, this method has its own problems, such as difÞculty in interpreting the meaning of

ratios when both arterial and venular diameters can change independently.4

8.2 Fluorescein Angiography

Until the introduction of OCT, ßuorescein angiography (FA) was the most commonly used ancillary test used in the evaluation and management of patients with RVO. It is useful for assessing the degree of capillary perfusion and the breakdown of the bloodÐretina barrier, but it has always been fraught with problems.49 The older patients who more often develop RVOs often have cataract which degrades image quality, as does the inability to adequately dilate the pupil for high-quality photographs.10 With conventional FA, it is difÞcult to capture images of the middle and far periphery, and blood can obscure details of capillary perfusion (Fig. 8.2).10 Although FA is used in most cases to photograph the fundus, the same technique can be used to assess iris vessels and look for subtle forms of anterior segment neovascularization.33

Fluorescein angiography does not lend itself well to quantitative analysis because of variability in many test factors and methods of interpretation. The phase of the disease at which the test is administered, the rate and quantity of ßuorescein injected, the vein chosen for injection, patient variability in circulation time, reader variability in choosing end points, camera type (30¡, 45¡, or 60¡ photographic Þelds), patient age, and patient gender all render reproducibility difÞcult.

Fig. 8.2 Frame from the mid-phase ßuorescein angiogram of an eye with a central retinal vein occlusion. The intraretinal hemorrhage in many areas is so dense that the details of capillary perfusion are blocked (the yellow ovals). Such an eye is managed as though it were ischemic, although by ßuorescein angiographic criteria it is indeterminate

Few studies examine the reproducibility of FA.91 Normal perifoveal capillary blood ßow velocity is 3.33 ± 0.09 mm/s. Normal retinal arterial dye velocity is 6.67 ± 1.6 mm/s.37,51 COVs for arteriovenous passage time and capillary transit velocity have been reported to be 15% and 18%, respectively.37 Complicated analysis of ßuorescein dye dilution curves using ßuorescein videoangiography has been performed to assess rate of vascular Þlling, but the most reproducible index (circulation time from beginning of Þlling to 50% of peak Þlling) has a measurement error of 9%.106 A change of 20% after an intervention is required to reliably conclude that a change rises above measurement error.106

Studies often measure some variation of arteriovenous transit time.28 This time is deÞned in different ways. Some studies deÞne it as time from injection until appearance of the dye in retinal arteries. Using this deÞnition, a mean value of 21.2 ± 5.8 s was reported in 72 patients with CRVO and HCRVO.31,38,78 Other studies deÞne early arteriovenous transit time as the interval from Þrst appearance in the central artery to that

in the corresponding vein.27 In normal controls, the early arteriovenous transit time has been reported to range from 5 to 12 s.1 In CRVO and HCRVO, early arteriovenous transit time has varied from 7.7 ± 3.6Ð12.5 ± 2.3 s.31,38,78 No differences in arteriovenous transit times were seen between nonischemic and ischemic CRVOs in one study.38 In a case series of BRVOs, the arteriovenous passage time at Þrst venous Þlling for affected and ipsilateral unaffected branch retinal veins was 4.5 ± SD0.8 and 1.5 ± SD0.2 s, respectively.60 The arteriovenous passage time until maximal venous Þlling for the affected and unaffected branch veins was 11.8 ± SD0.8 and 7.7 ± SD1.0 s, respectively.60

Masked grading of ßuorescein angiograms was used in clinical trials of treatments for BRVO and CRVO.15,93 Variables graded include severity of leakage, degree of capillary nonperfusion (negligible, minimal, moderate, and severe), and disruption of perifoveal capillaries (180¡ or less and more than 180¡).15 Reproducibility of the graders is not often documented, and how this type of grading might generalize to clinical practice is unclear.15,60,81,87 Even in technically excellent studies, the reproducibility of interpretation of capillary nonperfusion across clinicians is poor.97 Part of this comes from vague deÞnitions of ischemia groupings, such as ÒfocalÓ and Òextensive.Ó81

Capillary nonperfusion of the posterior pole is usually accompanied by capillary nonperfusion of zones of retina in the retinal midperiphery. However, often there will be peripheral capillary nonperfusion but intact posterior pole perfusion. For this reason, it is important to attempt to obtain midperipheral sweeping views of the fundus during ßuorescein angiography.39 Typical ßuorescein angiography, even using a wide-angle camera and intentionally sweeping the retinal periphery, fails to capture the degree of nonperfusion often present in the retinal periphery in retinal vascular disease. Ultra-wide- Þeld scanning laser ophthalmoscopy combined with digital image analysis holds promise for overcoming this drawback, but these systems are

not widely available and remain a research interest as yet.54

Ischemia is frequently considered synonymous with capillary nonperfusion, but in fact, they are distinct.39 That is, the retina can have some degree of perfusion and still be ischemic. It takes an estimated 3Ð4 weeks of ischemia before retinal capillaries become nonperfused.39 Methods of quantitating capillary nonperfusion vary widely. In actual clinical practice, clinicians just make an estimate of the ischemic index.83 In automated systems, computerized planimetry is used based on outlining a nonperfused area on a computer display.91 The units may be given as squared millimeters or the area normalized by the area of an idealized circular optic disk with a diameter of 1.5 mm.109

Measurements of arteriovenous transit time and ischemic indices are research techniques. They are not used by clinicians in the care of patients. In clinical care, FA interpretation is qualitative and subjective. Clinicians look at the

FA and decide, based on personal experience, whether the case is ischemic or not. Therefore, it is not possible to know if statements in the literature about prognostic factors that depend on interpretations of FAs in studies have bearing in real-life care.

Although ßuorescein angiography is often used to detect ßuorescein leakage from retinal vessels, it is somewhat insensitive to low-grade leakage. Cases without ßuorescein leakage, yet with the presence of histopathological extracellular edema, have been published, and the same discordance is seen between FA and OCT (Fig. 8.3).96

8.2.1 Branch Retinal Vein Occlusion

In a qualitative sense, delayed venous Þlling is present commonly in acute BRVO. Complete venous obstruction has been reported

Fig. 8.3 Fundus images of a 68-year-old woman with a nonischemic central retinal vein occlusion of the right eye showing the possible presence of macular thickening in the absence of ßuorescein leakage on ßuorescein angiography. (a) Monochromatic fundus photograph of the right eye shows dilated veins and intraretinal hemorrhages in all quadrants. (b) Frame from the late-phase ßuorescein angiogram shows no leakage of ßuorescein in the right macula. (c) OCT false-color map of the right eye. Despite

the absence of ßuorescein leakage, the right macula is thickened compared to the left. The mean central subÞeld macular thickness of the right eye is 250 m (compare d). (d) OCT false-color map of the left eye. The mean central subÞeld macular thickness of the right eye is 225 m. The normal intereye difference in mean central subÞeld macular thickness is 6 ± SD4 m.68 Similar differences in mean subÞeld thickness are present for all of the paracentral subÞelds (compare (c) and (d))

c

d

Fig. 8.3 (continued)

500

277

400

329

335

200

288

100

ILM-RPE thickness ( m)

0 m

500

200

303

100

251

0 m ILM-RPE thickness ( m)

in up to 37% of cases.22 Arteriolar Þlling to the involved region of the retina is not delayed.22 In 15% of cases, a focal hyperßuorescence occurs at the site of the arteriovenous crossing, or occasionally proximal or distal to it, that disappears within 1Ð6 weeks of the acute event (Figs. 8.4 and 8.5).22,36,74 Some have hypothesized that this represents a prethrombotic event indicating turbulent blood ßow and endothelial injury or the thrombus in evolution, but it is uncommon to have an opportunity to witness the hypothesized time course. When eyes having recentonset BRVO are studied, ßuorescein videoan-

giography shows diastolic arterial reßux in the arteries to involved segments.80 Intraretinal hemorrhage may make determination of perfusion difÞcult.70 When blood does not obscure capillary detail, there may be capillary nonperfusion (Fig. 8.5). Late leakage of ßuorescein is commonly seen emanating from the obstructed retinal vein and its tributaries (Figs. 8.4 and 8.5).60,70 Late in the study, the venous walls may stain.74 Collateral vessels do not leak, but new vessels do.74

There is no standard practice with regard to obtaining an FA in the management of BRVO. At