prevalence of 3.8%.84 This percentage is probably still higher than the true value as it derived from a tertiary referral clinic and was not population based.84 In the DRS, 2.3% of eyes with PDR had NVI by clinical examination.85 A higher percentage of cases of NVI are bilateral in association with diabetic retinopathy than other causes of NVI, reflecting the generally symmetric nature of diabetic retinopathy.79 Bilateral involvement of NVI in association with diabetic retinopathy has been reported in 5–76%.74 NVI occasionally spon-

taneously resolves clinically, although histological evidence of the vessels may be detected.74,86,87

11.4 Clinical Detection

Iris neovascularization usually follows the development of proliferative diabetic retinopathy, but cases

have been described in which NVI was seen in the absence of PDR.83,88,89 In these cases, however,

severe nonproliferative retinopathy was present or intracapsular cataract extraction had been performed.83,89 Studies have reported widespread

areas of peripheral retinal capillary nonperfusion in eyes with PDR and in eyes with NVI.62,63 There

is a controversy regarding the sequence of anterior segment neovascularization.

Most authors report that in diabetic patients pupillary margin iris neovascularization consistently precedes angle neovascularization, pathological evidence exists to support this sequence, and

some classification schemes are based on this pre-

sumption.59,75,79,83,87,90,74,94,80,95 In five clinical ser-

ies comprising 245 eyes, no case of NVA was seen without concomitant NVI.59,83,86,96,80 Based on

Hanley’s rule of three, we can be 95% confident

that the probability of NVA without NVI is at most 3/245 (1.2%).96,97 However, rare cases have

been reported in which angle new vessels were seen in the absence of pupillary margin new vessels.63,98

The authors consider the general clinical rule to be true, conceding that rare counterexamples may exist, and we have illustrated our understanding of the clinical sequence of anterior segment neovascularization in association with diabetic retinopathy based on this understanding (Fig. 11.4).

Unlike the situation with diabetic retinopathy, NVA is often found before pupillary margin NVI in central retinal vein occlusion, occurring in 12% of cases developing anterior segment neovascularization.91 If it can occur in diabetic retinopathy, the frequency of its occurrence is much less, and there seems to be little reason to do undilated

Fig. 11.4 Diagram illustrating the natural history of uninhibited anterior segment neovascularization associated with diabetic retinopathy. A – new vessels begin at the pupillary margin. B – next new vessels can appear in the angle. C – next the iris stroma is involved with new vessels and angle

synechiae develop. D – pupillary margin posterior synechia to the lens capsule develop and iris bombe may occur. E – hyphema may occur when iris neovascularization

bleeds. Adapted from and reproduced with permission from Roth and Brown172

gonioscopy in search of NVA in diabetic eyes with no pupillary margin NVI.99 NVI in association with diabetic retinopathy behaves differently from NVI in association with central retinal vein occlusion. In the latter, the course of NVI is compressed and neovascular glaucoma can occur

rapidly, whereas on an average the course is slower in the case of diabetic retinopathy.88,100 This gen-

eral observation may reflect the generally lower levels of intraocular VEGF seen in diabetic retinopathy compared with ischemic central retinal vein occlusion.

Clinical detection of NVI is clouded by the difficulty of distinguishing dilated normal iris vessels from NVI. Eyes with inflammation can have dilated iris vessels that may masquerade as NVI. Iris neovascularization must also be distinguished from iris vascular tufts, which are located along the pupillary

margin and do not occur on the surface of the iris stroma or in the angle.90,102

Iris color influences the ease of detection of NVI.79 Clinical detection at the slit lamp is easier in a lightcolored iris.74 Subtle NVI may be difficult to discern in a dark-colored iris. Red-free light may make detection of the red NVI easier.103 In difficult cases, iris fluor-

escein angiography and gonioangiography can make detection of NVI and NVA easier.62,80 Iris fluorescein

angiography is more sensitive than slit lamp biomicro-

scopy in the detection of NVI and NVA, but is also less specific.80,104,105 Dilated iris vessels in eyes with

iritis, normal iris vessels in older patients, and eyes of diabetics with retinopathy but no NVI can leak fluorescein on angiography just as does NVI, thus iris

fluorescein angiography is not always a reliable method for distinguishing the two conditions.59,106–108

Iris fluorangiography is an often forgotten method of assessing the presence of diabetic retinopathy in eyes with dense cataracts before proceeding to surgery. Presence of NVI implies that severe NPDR or worse is present and allows for preoperative planning such

as use of intravitreal bevacizumab or early postoperative PRP.83 Indocyanine green iris angiography has also been investigated in patients with diabetic retinopathy and is not as useful as iris fluorescein angiograms.109

Figures 11.7, 11.8, and 11.9 show standards useful in interpreting iris fluorescein angiograms.

Fig. 11.7 Example of a normal iris fluorescein angiogram. No dye leakage is present, although in patients over the age of 50 years, a small amount of pupillary margin leakage of fluorescein can be considered normal. Reprinted with permission from Bandello et al.83

Fig. 11.8 Example of nonproliferative diabetic iridopathy. Dye leakage is prominent in the late phase but neovascularization is absent. Reprinted with permission from Bandello et al.83

provide a prognosis for visual outcome and to allow correlations to be made with stages of diabetic retinopathy.88,80 They differ not only in conception, but in the method used for detecting NVI and NVA. Although standardized classifications have attraction in common conditions such as diabetic retinopathy, the relative uncommonness of NVI makes the ability to recall details of any system difficult for the clinician, and we favor a method of straightforward description of pupillary margin and angle involvement by neovascularization as being most practical (Tables 11.1 and 11.2).

11.6Risk Factors for Iris Neovascularization

Fig. 11.9 Example of iris neovascularization present around the entire pupillary margin and at several places on the iris stroma. Reprinted with permission from Bandello et al.83

11.5 Classification

Several classification schemes for NVI and NVA

have been proposed and none is accepted by the majority of clinicians.62,83,88,110 There are also dif-

ferent classification schemes for the clinical and histopathological stages of NVA (Fig. 11.10).82,111,112

The purpose of these schemes is to attempt to

Several risk factors for NVI relate to characteristics of diabetic retinopathy severity and fluorescein angiography. Hamanaka and colleagues performed nearly simultaneous goniofluorescein angiography and panoramic fundus fluorescein angiography in a series of eyes with proliferative diabetic retinopathy.62 They graded eyes as having any capillary nonperfusion in the temporal raphe and radial peripapillary capillary network regions and graded whether midperipheral retinal capillary nonperfusion exceeded 50% or not. Figure 11.11 illustrates these regions. Presence of disk neovascularization (NVD) was also noted. They then correlated the presence of NVA with the findings on fundus fluorescein angiography and the status of NVD. The strongest risk factor for the presence of NVA was midperipheral capillary nonperfusion greater than 50% by area with a relative risk of 16.7 compared to eyes without this characteristic (Table 11.3). The relative risk is much higher for this factor because of the rarity of finding an eye with NVA when capillary nonperfusion of the midperiphery falls below 50%. Only 1 out of 28 such eyes had NVA.

Risk factors for development of NVI in eyes with diabetic retinopathy include previous vitrectomy and previous cataract extraction, both of which

remove a diffusion barrier for VEGF access to the aqueous humor.89,115 Attempts to predict post-

operative NVI after vitrectomy by using preoperative iris fluorescein angiography have not been successful. Aphakia has been reported to be a risk factor for NVI in eyes with diabetic retinopathy in

the past, but is rarely encountered in recent years because almost all eyes receive intraocular lens implants.89,115 Posterior pseudophakia is probably not a risk factor for NVI, although anterior pseudophakia may be.116 Silicone oil present in the vitreous cavity can reduce the risk of NVI in such cases, but not eliminate it.117 Rhegmatogenous or peripheral tractional retinal detachment in an eye with diabetic

retinopathy increases the risk of NVI, and repair of such detachments can lead to NVI regression.68,118

Neodymium-YAG capsulotomy after extracapsular cataract extraction is a potential risk factor for anterior segment neovascularization.119

In eyes with diabetic retinopathy undergoing vitrectomy, the risk factors for postoperative development of rubeosis are intraoperative lensectomy, increased preoperative retinopathy severity, and absence of preoperative PRP associated with relative risks of 3.3 (95% CI 2.0, 5.4), 2.1 (95% CI 1.3, 3.4), and 1.7 (95% CI 1.1, 2.6), respectively.68 If an eye undergoing vitrectomy has preoperative NVI, the odds are increased that it will have postoperative NVI.95 In these cases, presence of a postoperative retinal

detachment negatively affects regression of NVI.95