|Home | About | Journals | Submit | Contact Us | Français|
NVG is a severely blinding, intractable disease. The objective of this review is to provide detailed information on its basic and clinical aspects, to enable us to manage it logically. Therefore, its causes, pathogenesis and pathology, methods of early diagnosis and management are discussed. To prevent or reduce the extent of visual loss caused by NVG, the first essential is to have a high index of suspicion of its development. The most common diseases responsible for development of NVG are ischemic CRVO, diabetic retinopathy and ocular ischemic syndrome. In the management strategy, the first priority should be to try to prevent its development by appropriate management of the causative diseases. If NVG develops, early diagnosis is crucial to reduce the extent of visual loss. Management of NVG primarily consists of controlling the high IOP by medical and/or surgical means to minimize the visual loss. Currently we still do not have a satisfactory means of treating NVG and preventing visual loss in the majority, in spite of multiple modes of medical and surgical options advocated over the years and claims made. This review discusses pros and cons for the various advocated treatments.
Neovascular glaucoma (NVG) is a blinding, intractable disease, difficult to manage and often resulting in disastrous visual loss. For a logical understanding and scientific rationale for management of any disease, one first has to know the basic issues involved and the scientifically valid information available on the disease. To prevent or reduce the visual loss caused by NVG, the first essential is to have a high index of suspicion of its development, i.e. to be aware of the various ocular diseases in which it can develop. Once it develops, early diagnosis and rational management are important to minimize the visual loss. Therefore, the objective of this review on NVG is to discuss its causes, pathogenesis and pathology, methods of early diagnosis and finally management.
These can be divided into two categories: (a) the most common causes, and (b) uncommon causes.
Diabetic retinopathy, ischemic central retinal vein occlusion (CRVO) and ocular ischemic syndrome are by far the most common causes of NVG. There is little controversy about NVG in diabetic retinopathy. However, many of the concepts about NVG in various types of retinal vascular occlusive diseases and ocular ischemic syndrome are controversial. In the Ocular Vascular Clinic of the University of Iowa Hospitals and Clinics we have systematically investigated the various retinal vascular occlusive diseases as well as ocular ischemic syndrome in detail in longitudinal prospective studies since 1973; these studies have provided new information, contradicting several of the conventional wisdoms on these disorders. Those require detailed discussion in the interest of clarifying the confusion and controversies on these important disorders (see below);
Association of NVG with diabetic retinopathy is a well-established clinical entity, and a huge volume of literature has accumulated on this subject. NVG is an advanced manifestation of diabetic retinopathy. NVG may occur without retinal or optic disc neovascularization (NV), however, it is more commonly seen in association with proliferative diabetic retinopathy.
NVG may occur in the setting of ischemic CRVO or more rarely following ischemic hemi-CRVO, simultaneous multiple branch retinal vein occlusion involving large areas of the retina, or when venous occlusions are superimposed upon a background of non-proliferative diabetic retinopathy. There are many unfounded theories regarding NVG and vein occlusions. These have spawned controversy and some confusion on the etiology, natural course and management of NVG in such patients. I have discussed the various misconceptions elsewhere [Hayreh 2005]. I have included here an abbreviated discussion of some of the most common misunderstandings relevant to NVG in retinal vascular occlusive diseases.
There is a common notion among ophthalmologists that every eye with CRVO is at risk of developing NVG. It is well established now that CRVO is of two distinct types – non-ischemic and ischemia CRVO, with very different clinical findings, complications, course, prognoses and managements (Hayreh 1965,1976,1983,1994, 2003, 2005; Hayreh et al. 1983, 1990a). NVG is a complication only of ischemic CRVO (Hayreh et al. 1983). Eyes with nonischemic CRVO do not develop ocular NV or NVG (Hayreh et al. 1983), unless there is associated diabetic retinopathy or ocular ischemic syndrome - the latter two associated conditions being the sole cause of ocular NV in those eyes, which may wrongly be attributed to nonischemic CRVO. Therefore, the first essential step in the management of CRVO is to determine whether the CRVO is ischemic or non-ischemic. Hayreh et al. (1990a) in their study investigated the various clinical tests that can help to differentiate the two types of CRVO. In almost all the published series, a criterion of “10 disc area of retinal capillary obliteration” has been used to classify CRVO as ischemic. However, studies by Hayreh et al. (1990a) have shown that this is not at all a valid criterion. The presence of isolated, small, focal retinal capillary obliteration is compatible with nonischemic CRVO. The results of a large multicenter CRVO study (The Central Retinal Vein Occlusion Group, 1995) supported their conclusions. This multicenter study clearly showed that eyes with less than “30” disc diameters of retinal capillary nonperfusion and no other risk factor are at low risk for developing iris/angle neovascularization (i.e. ischemic CRVO), “whereas eyes with 75 disc diameters or more are at highest risk”. Thus, “10 disc area of retinal capillary obliteration” on fluorescein angiography is totally unreliable parameter in differentiating ischemic from non-ischemic CRVO. It can result in incorrect diagnosis, prognosis and management.
Hayreh et al. (1990a) in their study, to differentiate ischemic from non-ischemic CRVO during the early acute phase, found a number of much more sensitive and specific diagnostic clinical tests, which make such a differentiation accurately. They divided these tests into two categories:
|Peripheral visual fields*:|
|Relative afferent pupillary defect:|
|≥ 0.9 log units||80%||97%|
Thus, their study showed that combined information provided by the six functional tests helps to differentiate ischemic CRVO from nonischemic CRVO most reliably during the early acute stage.
Based upon this and additional criteria, Hayreh et al. (1983) have demonstrated that every eye with ischemic CRVO does not develop ocular NV and/or NVG. The cumulative risk in their study of ocular NV in ischemic CRVO is illustrated in figure 1. It shows that the risk of developing NVG in eyes with ischemic CRVO reaches a maximum of about 45% in aggregate over several years – the maximum risk being during the first 7–8 months only [Hayreh et al. 1983]. They also found that only 20% of all eyes with CRVO are of the ischemic type [Hayreh et al. 1983; Hayreh 1994]. Therefore, the risk of an eye with CRVO developing NVG is not 100% but about 9–10% [Hayreh 2003]. This puts the threat of NVG in CRVO into proper perspective. However, if an ischemic CRVO is identified, one should have a high index of suspicion for development of NVG.
Similar to CRVO, hemi-central retinal vein occlusion (HCRVO) may be divided into ischemic and non-ischemic types, based upon clinical findings [Hayreh and Hayreh 1980]. Like the CRVO, NVG is only a complication of the ischemic hemi-CRVO and not in non-ischemic type. In one series, NVG developed in 3% (one of 31) ischemic HCRVO eyes (Hayreh et al. 1983).
In general, the development of NVG in retinal vein occlusion depends upon the severity and extent (area) of retinal ischemia (Hayreh et al. 1983; The Central Vein Occlusion Study 1997). Hemi-CRVO typically involved one hemisphere of the eye. Thus, the risk of developing NVG due to sufficient stimulus in ischemic hemi-CRVO is very low.
There is a prevalent mistaken belief that branch retinal vein occlusion can cause NVG. The principal factors in the development of ocular NV following retinal vein occlusion are the severity and spatial extent (area) of retinal ischemia [Hayreh et al. 1983; The Central Vein Occlusion Study 1997]. In the study by Hayreh et al. (1983) on ocular NV associated with retinal vein occlusion, NONE of the 264 eyes with branch retinal vein occlusion developed NVG. Their studies have indicated that it usually requires at least half or more of the retina to be involved by ischemia to provide adequate neovascular stimulus. The vast majority of major branch retinal vein occlusions involve smaller segments of the retinal vasculature, typically a quarter or less of the retinal surface area. Concerns of development of NVG in branch retinal vein occlusion may stem from equating ischemic CRVO with ischemic changes seen following branch retinal vein occlusion, which represent different clinical entities and scales of stimuli. Thus, there is little risk of NVG following typical or isolated branch retinal vein occlusion.
Some authors have proposed that CRAO, similar to ischemic CRVO, may be followed by ocular NV and NVG. A review of all the publications associating NVG with CRAO from 1874 to 1982 revealed a number of problems with this contention (see details in Hayreh and Podhajsky 1982). Studies by Hayreh et al. [Hayreh and Podhajsky 1982; Hayreh and Zimmerman 2007], which evaluated 248 eyes with CRAO, have not shown development of NVG in CRAO. Thus, their studies have shown no basis for development of NVG in CRAO (Hayreh and Podhajsky 1982). The reasons are:
In my study of more than 150 eyes with branch retinal artery occlusion, no eye developed any ocular NV, including NVG. The reason for not developing NVG in these eyes is the same as discussed above in CRAO.
Ocular ischemic syndrome is a serious but uncommon blinding condition. It is a frustrating condition for the ophthalmologist. The visual prognosis and treatment outcomes are poor. There is no well-established treatment. Moreover, it may be overlooked or misdiagnosed, primarily because of its diverse and sometimes subtle presentation. Ocular ischemic syndrome can also masquerade several other ocular conditions. Mizener et al. (1997) in their study, on ocular ischemic syndrome in 39 eyes, found a variable range in the symptoms, signs, and course of the disease. Hayreh and Podhajsky (1982) have discussed its pathogenesis in detail. Ocular ischemic syndrome is caused by reduction of global blood flow to the eyeball, which can produce anterior and/or posterior segment ischemia. Anterior segment ischemia results in development of iris and angle NV and NVG. Most patients with ocular ischemic syndrome have severe carotid artery occlusive disease, but not all; it can be associated with vascular occlusive disease of the aortic arch, or of the ophthalmic, central retinal or ciliary arteries. It is not uncommon in clinical practice to find that carotid artery disease is ruled out as the cause of ocular ischemia based on findings of absence of occlusion or severe stenosis of the internal carotid artery on carotid Doppler. However, carotid Doppler evaluates only the artery in the neck and not above or below that, where it may be markedly stenosed. Hayreh and Dass (1962), in their anatomical studies on the ophthalmic artery, found on rare occasions that the ophthalmic artery was markedly stenosed at its origin, while the internal carotid artery was unobstructed and fully patent. Cerebral angiography or magnetic resonance angiography may be necessary to provide information which carotid Doppler cannot demonstrate, such as a significant stenosis in the carotid siphon or ophthalmic artery.
In the study of Mizener et al. (1997) on ocular ischemic syndrome, the prevalence of diabetes mellitus in these patients was much higher than in the comparable general population. Conventionally, retinal ischemia in the form of extensive retinal capillary non-perfusion (as seen for example in ischemic CRVO and diabetic retinopathy) is considered the exciting factor for the development of anterior segment NV and NVG. However, in that study, no patient with ocular ischemic syndrome and ocular NV had retinal capillary non-perfusion on fluorescein fundus angiography, not even those with diabetes mellitus. In contrast, in their studies in eyes with ocular ischemic syndrome, they have invariably found evidence of uveal vascular insufficiency [Hayreh and Podhajsky 1982; Mizener et al. 1997]. This was further confirmed by an experimental study in rhesus monkeys, where anterior segment NV (similar to that seen with ocular ischemic syndrome) developed with uveal ischemia alone, without any retinal ischemia (Hayreh and Baines 1973). These data suggest that uveal ischemia may be an important contributing factor for the NV seen in ocular ischemic syndrome.
The following is a list of the diseases in which NVG has been reported sometimes. In many of them, the associated retinal ischemia seems to be the most likely factor for ocular NV.
There are large number of reports of development of NVG following radiation for a variety of ocular and orbital lesions, e.g., iris melanoma (Shields et al. 2003), posterior uveal melanomas (Shields et al. 2002), choroidal metastatic tumors (Tsina et al. 2005), retinoblastoma (Kingston et al. 1996), orbital lymphoma (Bhatia et al. 2002), and nasal and paranasal malignancies (Takeda et al. 1999). In ocular radiation, in most eyes the main factor responsible for the development of NVG most likely is the development of secondary radiation retinopathy, in which there is development of retinal capillary non-perfusion and retinal ischemia.
NVG has been reported in association with prior to treatment in ring melanoma of the anterior uvea (Allaire et al. 1997), adenocarcinoma of the nonpigmented ciliary epithelium (Terasaki et al. 2001), medulloepithelioma (Singh et al. 2001), circumscribed choroidal hemangioma (Shields et al. 2001a), metastatic cutaneous melanoma to the vitreous (Gunduz et al. 1998), retinoblastoma (De Potter 2002), and metastatic malignant lymphoma (Matsui et al. 2005).
Development of NVG has been observed following anterior as well as posterior uveitis. It is not known whether the stimulus for NV is from the inflammation or related secreted cellular products or may be a complication of underlying systemic diseases associated with uveitis, e.g., Whipple’s disease (Nishimura et al. 1998), Crohn’s disease (Salmon et al. 2000) and Behçet’s disease (Elgin et al. 2004).
These include retinal vasculitis per se or when it is associated with systemic diseases, such as Crohn’s disease (Salmon et al. 2000) and Behçet’s disease (Elgin et al. 2004). Other miscellaneous retinal conditions which may be associated with development of NVG include Coat’s disease (Shields et al. 2001b,c), Eales’ disease (Atmaca et al. 2002; Perentes et al. 2002), frosted branch angiitis (Seo et al. 1998), giant cell astrocytoma of the retina (Gunduz et al. 1999), peripheral retinal detachment (Barile et al. 1998), and X-linked retinoschisis (Rosenfeld et al. 1998). Similarly other system diseases, for example, cryoglobulinemia and Churg-Strauss syndrome, by causing retinal vascular occlusion can be associated with NVG
There is a large volume of literature on the subject of angiogenesis and NV and is beyond the scope of this discussion. However, a brief summary of the most pertinent information concerning ocular NV and NVG, especially to its clinical aspects, is warranted. Chen et al. (Chen et al. 1999) reported that increase in the inflammatory cytokine interleukin (IL) - 6 in aqueous humor correlated spatially and temporally with the grade of iris NV in patients of NVG secondary to CRVO. They postulated that the increased level of IL-6 might have a putative role, along with other angiogenic factors in angiogenesis of NVG.
Since 1996, several studies have implicated vascular endothelial growth factor (VEGF) as an important and likely the predominant factor in the pathogenesis of intraocular NV and NVG (Pe’er et al. 1996, 1998; Sone et al. 1996; Tolentino et al. 1996; Kozawa et al. 1998; Tripathi et al. 1998; Atmaca et al. 2002; Hu et al. 2002). Boyd et al. (2002) found a close temporal correlation between aqueous VEGF levels and the course of iris NV and permeability in ischemic CRVO, indicating that increased aqueous VEGF level may predict the need for treatment. Itakura et al. (2004) reported that in proliferative diabetic retinopathy a high VEGF level was maintained in the vitreous cavity after vitrectomy. They stated that their results suggest that there is persistent secretion of VEGF into the vitreous cavity even after vitrectomy in these eyes. This observation is supported by the experimental findings of ocular NV in rhesus monkeys, where Virdi and Hayreh (1982) found a correlation between retinal vascular leakage and the development of ocular NV.
The main reason for visual loss with high IOP in NVG is ischemia of the optic nerve head and/or retina. Blood flow in the various intraocular vascular beds can be calculated by using the following formula:
According to this formula and assuming no change in vascular resistance, a rise of IOP reduces the perfusion pressure and thus decreases the blood flow in the retina, choroid and optic nerve head. Therefore, the higher the IOP and the lower the blood pressure, the greater is the reduction of blood flow, and the worse the ischemic damage to the optic nerve head and retina, particularly the former. Thus, in the management of NVG, although lowering the IOP is crucial, one also has to make sure that the treatment does not lower the systemic arterial blood pressure.
Iris and angle NV almost invariably develops before the pressure rises. This is associated with the development of a fibrovascular membrane on the anterior surface of the iris and iridocorneal angle of anterior chamber. Membrane development is followed by development of progressive anterior synechiae, and angle closure, and precipitous rise of IOP, which may be of fairly acute onset. In some of the eyes, anterior segment NV may be associated with development of hyphema, which may contribute or precipitate an acute rise of IOP. It is worth remembering that the iridocorneal membrane may be difficult to visualize and that the angle may appear to be open and the IOP elevated before synechiae develop.
Advanced NVG is straightforward to diagnose. However, early in the course, NVG may present subtle findings and one must have a high index of suspicion of its development in the settings of various diseases discussed above (particularly ischemic CRVO, diabetic retinopathy and ocular ischemic syndrome). In addition, a careful examination of the iris and angle of the anterior chamber is essential, before the pupil is dilated and any drops put in the eye. Once the pupil is dilated, it may not be easy to find the NV. During the early stages, iris NV is essentially at the pupil margin and is very fine and delicate in character. Similarly, careful gonioscopy is essential to detect early angle NV and early anterior synechiae. The anterior chamber in these eyes often shows the presence of flare. Some time a few cells may be seen in the anterior chamber, which may erroneously be diagnosed as a sign of uveitis. Fluorescein iris angiography can be helpful in some doubtful cases because it shows leakage of fluorescein which is normally not seen. However, in a fully developed NVG with very high IOP, the clinical picture may be dramatically different; the eye may be painful and, when the IOP goes very high fairly fast, there is usually corneal epithelial edema which can make examination for iris and angle NV difficult.
From time to time, NVG has been confused with other ocular conditions. For example, eyes with severe non-granulomatous uveitis with dilated iris vessels and proteinous aqueous and high IOP can be misdiagnosed to have NVG. There are some eyes where normal iris vessels are seen easily, particularly in blue eyes, which may be mistaken for iris NV or even angle NV when the vessels are seen near the root of the iris. Eyes with carotid-cavernous fistula erroneously may be diagnosed to have NVG because of the blood in Schlemm’s canal, and elevated IOP.
This is highly challenging, unpredictable, difficult and controversial. It involves several considerations, including the following:
Sivak-Callcott et al. (2001) conducted a review of the literature to evaluate evidence-based recommendations for treatment of NVG. They stated: “The current standard of care includes retinal ablation and control of increased intraocular pressure with medical and surgical therapy.” They concluded: “The current literature on neovascular glaucoma has few articles that provide strong evidence in support of therapy recommendations” based on the “data that provided strong evidence in support of clinical recommendation”.
We can divide the management of NVG into two categories: (1) management of the underlying disease and (2) management of high IOP when it develops.
As discussed above, the main diseases, which are responsible for inducing NVG, are diabetic retinopathy, ischemic CRVO and ocular ischemic syndrome. Management of these 3 conditions is discussed.
There is strong evidence that panretinal photocoagulation (PRP) is the treatment of choice for prevention of development of NVG in diabetic retinopathy (The Diabetic Retinopathy Study Research Group 1976). Recently, Bandello et al. (2006), in a series of nine patients, reported that intravitreal triamcinolone before PRP may be useful in improving the effect of PRP in eyes with proliferative diabetic retinopathy by reducing NV and retinal thickening.
NVG is the most dreaded and blinding complication of ischemic CRVO. Unlike diabetic retinopathy, management of ischemic CRVO to prevent NVG is highly controversial. One reason for this longstanding controversy is that application of PRP for ischemic CRVO has been advocated by some based primarily upon the limited similarities in retinal vascular changes in diabetic retinopathy and ischemic CRVO. The controversy is increasingly confused since the role of growth factors in ischemic retinal disease has been elucidated. Clinically and in therapeutic response to PRP, ischemic CRVO and proliferative diabetic retinopathy behave very differently in nature and course (Hayreh 2003; Hayreh et al. 1990b).
The theoretical justification advocated for PRP in ischemic CRVO is, as in diabetic retinopathy, to prevent development of ocular NV and NVG. Some time ago, Hayreh et al. (1990b) reviewed the literature on the subject, and found flaws in most of the studies claiming beneficial effects of PRP in ischemic CRVO. For example, Magargal and co-workers (1981, 1982) claimed that not one of the 100 eyes treated by them with PRP developed NVG attributable to ischemic CRVO and “iris neovascularization subsequently regressed in each case” (Magargal et al. 1982). They concluded that “Prophylactic PRP in high-risk ischemic CRVO eyes appears to eliminate virtually the devastating complications of NVG” (Magargal et al. 1982). Yet no subsequent study has been able to confirm their enthusiastic claim.
Hayreh et al. (1990b) conducted the first long-term (10-year) prospective, planned study of argon laser PRP in a large group of ischemic CRVO eyes. They started this study in 1977, soon after the publication of the Diabetic Retinopathy Study (The Diabetic Retinopathy Study Research Group 1976) in 1976, which showed the beneficial effect of PRP in proliferative diabetic retinopathy. Their hypothesis at the start of the study was that they would find a similar beneficial response in ischemic CRVO, since both retinopathies share common features of retinal capillary non-perfusion and ocular NV. However, on completion of their study, on comparing the lasered eyes with the non-lasered eyes, they found to their complete surprise that there was NO statistically significant difference between the two groups in the incidence of development of angle NV, NVG, retinal and/or optic disc NV, or vitreous hemorrhage, or in visual acuity (in Fig. 3 both eyes developed iris NV in spite of PRP) (Hayreh et al. 1990b). This study did show, however, a statistically significant (p=0.04) difference in the incidence of iris NV between the two groups, with iris NV less prevalent in the lasered group than in the non-lasered group, but only when the PRP was performed within 3 months after the onset of CRVO; however, iris NV per se is of little clinical importance for the following reasons: (a) in one third of the eyes it does not progress to NVG (Fig. 1), and (b) in that study there was no significant difference in development of NVG between the lasered and control groups in spite of that difference in iris NV (Hayreh et al. 1983). The most important finding of this PRP study in ischemic CRVO was the statistically significant difference in the loss of peripheral visual fields between the lasered and non-lasered eyes - the lasered group suffered a significantly (p≤0.03) greater loss than the non-lasered group (Fig. 3). This showed that ischemic CRVO cannot be equated to diabetic retinopathy; Hayreh et al. (1990b) discussed the reasons for this disparity in response to PRP in the two diseases.
A multicenter prospective clinical trial by “the Central Vein Occlusion Study” (CVOS) group investigated the role of PRP in ischemic CRVO (The Central Retinal Vein Occlusion Group 1995) to find out whether PRP prevents progression of iris/angle NV to NVG. The authors recommended “careful observation with frequent follow-up examinations in the early months (including undilated slit-lamp examination of the iris and gonioscopy) and prompt PRP of eyes in which 2’clock iris/angle NV develops.” Such a multicenter, multimillion-dollar study conducted under the aegis of the National Institutes of Health carries tremendous prestige, and the ophthalmic community considers its recommendations the “gold standard” in the management of ischemic CRVO. It is extremely important to place the results of this study in its true perspective; to do that, I pointed out the various flaws in the study and its conclusions (Hayreh 1996); briefly, they are as follows.
It could be argued that this is simply my opinion (Hayreh 1996) about the CVOS; but the Editor of Ophthalmology (where this study was published) sent my comments to the authors of the CVOS Group and in their response (Central Vein Occlusion Study Group 1996) they agreed with all the concerns I had raised. Nevertheless, the ophthalmic community, unfortunately, still considers that study as the “gold standard”.
Management of ocular ischemic syndrome remains difficult and controversial. Reduction of blood flow to the eyeball can produce anterior and/or posterior segment ischemia (see Fig. 2). Anterior segment ischemia manifests as NV and NVG. As discussed above, in the ocular ischemic syndrome studies of Hayreh and Podhajsky (1982) and Mizener et al. (1997), no patient with ocular ischemic syndrome and ocular NV had retinal capillary non-perfusion on fluorescein fundus angiography, even those with diabetes mellitus; however, those studies invariably found evidence of uveal vascular insufficiency. Therefore, the conventional use of PRP for anterior segment NV and NVG in diabetic retinopathy cannot be extrapolated to ocular ischemic syndrome. Moreover, rise of IOP following PRP may further compromise the already highly precarious ocular and optic nerve head circulation and result in severe visual loss due to development of anterior ischemic optic neuropathy (Brown 1986) or retinal ischemia.
Since internal carotid artery occlusive disease is the most common cause of ocular ischemic syndrome, carotid endarterectomy seems a logical management. However, the benefit of carotid endarterectomy in patients with ocular ischemic syndrome is unknown and controversial. I reviewed the multiple reports in the literature dealing with this topic. Those described that, following carotid endarterectomy or internal-external carotid anastomosing procedure, there was stabilization of vision if initial vision was good, regression of iris NV, and, rarely, improvement in already poor vision (Mizener et al. 1997). Some studies showed that selected patients with asymptomatic severe carotid artery disease might benefit from carotid endarterectomy (Thompson 1993). In the study by Mizener et al. (1997), most patients with ocular ischemic syndrome who underwent carotid endarterectomy had poor vision and visual outcome was unchanged by the surgery. However, clinical decisions for carotid endarterectomy are usually driven by the patient’s entire clinical picture, as determined by neurologists and vascular surgeons.
Since, in ocular ischemic syndrome, perfusion pressure in the various ocular vascular beds is low, lowering IOP to as low a level as possible is crucial to improve the blood flow (Fig. 2), and thereby to prevent visual loss from various types of ocular vascular occlusion (including CRAO and anterior ischemic optic neuropathy) and/or glaucoma.
Studies dealing with 24-hour ambulatory blood pressure monitoring in ocular and optic nerve head ischemic disorders have shown that nocturnal arterial hypotension plays an important role in their pathogeneses, by lowering the perfusion pressure during sleep (Hayreh et al. 1994, 1999b; Hayreh 1999). These studies also showed that aggressive antihypertensive therapy, particularly administration of blood pressure lowering drugs in the evening or at bedtime, can result in marked nocturnal arterial hypotension, which may precipitate ischemic visual loss. For example, 73% of patients with non-arteritic anterior ischemic optic neuropathy gave a history of discovering the visual loss on waking up from sleep (Hayreh et al. 1997). Similarly, in the study by Mizener et al. (1997) on ocular ischemic syndrome, some patients reported discovery of visual loss on waking up from sleep in the morning. Therefore, in patients with ocular ischemic syndrome, it is important to avoid nocturnal arterial hypotension. A study by Hayreh et al. (1999a) showed that topical beta-blockers eye drops produced significant nocturnal arterial hypotension. Therefore, it would be advisable to avoid beta-blocker eye drops for lowering the IOP in these patients. In patients who develop NVG from any cause, it is also important to avoid development of nocturnal arterial hypotension, to prevent visual loss. Unfortunately, the important role played by nocturnal arterial hypotension has not been stressed in the prevention of ocular ischemic conditions and NVG.
Eyes with uveitis and retinal vasculitis are at risk of developing NVG (see above). Thus, their appropriate management is important to prevent development of NVG. The treatment depends upon cause of these inflammatory diseases. In most of the eyes with uveitis, use of topical steroids and mydriatics are primary management; however, some of patients with uveitis require systemic corticosteroids or other immunosuppressive therapies. For retinal vasculitis, usually systemic corticosteroids are required, although some advocate using subtenon or intravitreal steroids. Since most of these patients have an associated systemic disease, management of that is also essential.
This can be done by medical or surgical methods.
These are primarily meant to lower the IOP in eyes with NVG. However, more recently, other medical therapies aimed at treating intraocular NV have been reported.
Different medical strategies have been tried to control NVG and the high IOP. This is invariably the first step to prevent visual loss and relieve pain or discomfort associated with NVG. IOP is lowered invariably by means of various aqueous suppressants (beta-blockers, alpha adrenergics, and carbonic anhydrase inhibitors). There is no role for cholinergic eye drops. Prostaglandins may not be of much help because they work by increasing the uveal outflow, which may be covered by a membrane. At the same time, it is helpful to give topical steroids to reduce any inflammatory component that may be present.
As discussed above, there is evidence now that VEGF is an important factor in the pathogenesis of ocular NV and NVG. There are a large numbers of reports about the use of anti-VEGF therapy for choroidal neovascularization associated with age-related macular degeneration; however, only a few recent reports deal with its use in ocular NV and NVG, essentially with Bevacizumab (Avastin).
Most of the published reports deal with anti-VEGF therapy in ocular NV associated with proliferative diabetic retinopathy (Avery et al. 2006; Davidorf et al. 2006; Mason et al. 2006; Oshima et al. 2006; Spaide et al. 2006). Regression of retinal and iris NV (Avery et al. 2006 in 45 eyes), retinal and optic disc NV (Mason et al. 2006 in 3 patients), retinal NV (Spaide et al. 2006 in 2 patients), and iris NV (Davidorf et al. 2006 in one eye; Oshima et al. 2006 in 7 eyes) have been reported. Spaide et al. (2006) had to re-inject after 3 months because of reactivation of retinal NV. Grisanti et al. (2006) gave intracameral injection of 1.0 mg bevacizumab in 6 eyes with iris NV and claimed a decrease in leakage from the iris vessels on angiography. All the reports published so far show only a short-term beneficial effect of this treatment.
There is much less information about the role of anti-VEGF therapy in ocular NV and NVG associated with CRVO. Genaidy et al. (2002), in an experimental study on retinal vein occlusion in monkeys, found that intravitreal anti-VEGF therapy did not affect the development of iris NV. Iliev et al. (2006) gave intravitreal Avastin in 6 eyes with NVG associated with CRVO and reported marked regression of anterior segment NV within 48 hours in all and a “substantial” fall of IOP in 3 eyes, while the other 3 eyes required cyclophotocoagulation to control IOP on a follow-up of 4–16 weeks. Kahook et al. (2006) gave intravitreal bevacizumab (1.0 mg) to a patient with NVG following CRVO and reported “IOP improved within 2 days”.
Thus, so far the published experience is in only in a small number of eyes, except that of Avery et al. (2006), with extremely short follow-up and limited information on IOP control in NVG, other than that iris NV responded. As yet, there is no study providing definite information for long-term control of ocular NV and NVG and complications of the anti-VEGF therapy. As usual, there is invariably an initial marked enthusiasm with a new therapy in a disease with poor prognosis; it is only a long-term experience, which provides realistic information. Moreover, the success rate of this therapy in NVG may also depend upon the cause of NVG. For example, ocular NV and NVG associated with diabetic retinopathy are not as aggressive and sudden in onset as in ischemic CRVO, and that may influence the success and failure rates of this mode of treatment. Unfortunately, no such information is available in the reported series so far. However, a decrease in iris and angle NV can permit safer surgical intervention.
Jonas et al. (2001) in 4 eyes with NVG gave intravitreal injection of 20 mg of crystalline triamcinolone acetonide and found the mean IOP decreased from 26.5± 12.1 mm Hg to 21.75±11.3 mmHg; however, in one of the eyes with IOP of 40 mm Hg, there was no change although iris NV decreased.
In conclusion, so far, information on anti-VEGF and corticosteroid therapy in NVG is available from only a few anecdotal reports, and no worthwhile conclusions can be drawn about their beneficial effect in controlling IOP and preventing visual loss, and their complications. The primary medical management still remains the use of drugs that lower the IOP.
If medical methods do not control IOP in NVG, then one has to resort to surgery. A variety of such methods have been advocated with differing claims, as is evident from the following. In these eyes, there is no role for laser trabeculoplasty because there is no visible trabecular meshwork.
This has been oldest of the surgical procedures used to try to control high IOP in NVG. Various such methods have been used for this purpose. The objective of this mode of treatment is to try to reduce the formation of aqueous by partially destroying the ciliary body. Initially cyclodiathermy was the most popular mode of doing that but this frequently had post-operative complications. That was later replaced by cyclocryotherapy, which has much less post-operative complications. Another similar procedure advocated in the past was cyclo-electrolysis. However, since 1997 (Bloom et al. 1997), for cycloablation various cyclophotocoagulation methods have been advocated.
A report on cyclophotocoagulation by the American Academy of Ophthalmology showed that transscleral cyclophotocoagulation with noncontact Nd:YAG and semiconductor diode laser is useful in acute-onset NVG (Pastor et al. 2001). Cyclophotocoagulation with a diode laser compared to cyclocryoablation has the advantage that it causes less pain and is better tolerated by the patients; however, in my experience it takes longer to lower the IOP. Oguri et al. (1998) reported that diode laser transscleral cyclophotocoagulation appears to be as effective as free-running mode Nd:YAG laser transscleral cyclophotocoagulation and better than continuous-wave mode Nd:YAG laser transscleral cyclophotocoagulation for treating NVG. However, there are reports of complications of laser cyclophotocoagulation, the most common being hypotony. Yap-Veloso et al. (1998) reported that 26% of eyes had severe long-term complications, including loss of vision (22%), corneal decompensation (2%), and phthisis bulbi (2%). Development of necrotizing scleritis 10 months after diode laser cyclophotocoagulation for NVG has been reported (Shen et al. 2004). Delgado et al. (2003) reported that use of noncontact transscleral neodymium:yttrium-aluminum-garnet cyclophotocoagulation for NVG in 115 eyes, while providing long-term IOP reduction, was associated with complications that included inflammation, visual loss, and hypotony, and that repeat treatments may be necessary to main good control of IOP.
The most common filtering procedure tried has been trabeculectomy combined with or without mitomycin C or 5-fluororacil. Elgin et al. (2006) claimed that trabeculectomy with mitomycin C combined with direct cauterization of peripheral iris in NVG decreased the incidence of both intraoperative bleeding and early postoperative hyphema, and provided reduction of IOP and the number of antiglaucomatous medications in 96% at one week, 86% at one month, 83% at 3 months and 66% at 6 months, in cases with a 6-month follow-up period. Kiuchi et al. (2006) found that pars plana vitrectomy, followed by PRP and trabeculectomy with mitomycin C in eyes with NVG associated with diabetic retinopathy effectively reduced the elevated IOP. Kono et al. (2005) performed pars plana vitrectomy combined with filtering surgery and claimed success in control of IOP for more than 12 months. By contrast, Mietz et al. (1999) found that trabeculectomy without antimetabolites failed to control IOP in 80% of eyes with NVG. Thus, there are conflicting claims, and it seems the effectiveness of these procedures decreases with time. There are several limitations in the reported studies to provide definite information for long-term control of IOP. For example, in all of them the follow-up was only 12 months or less. Moreover, the success rate of these procedures in NVG may also depend upon the cause of NVG. For example, NVG associated with diabetic retinopathy is not as aggressive and sudden in onset as in ischemic CRVO, and that may influence the success and failure rates of these procedures. Unfortunately, no such information is available in the reported series, where NVG from various causes were usually lumped together.
Glaucoma drainage devices have been considered as an option in the management of NVG where there is a high risk of failure from conventional filtering surgery. Various drainage devices used have been Molteno implant, Baerveldt implant, Ahmed glaucoma valve and Krupin valve. Different reports make conflicting claims about the success of different devices to control IOP in NVG. For example, Every et al. (2006) in a prospective study of 145 eyes followed up for a mean of 3.3 years found Molteno implant to control the IOP at ≤21 mm Hg up to 5 years, but the success rate progressively decreased from one year onwards. Failure to control IOP was significantly correlated with persistent iris NV. Broadway et al. (2001) found that, in the long term, Molteno implant tended to fare poorly in NVG. Krishna et al. (2001) claimed that the 350-mm(2) Baerveldt glaucoma implants are a safe and effective treatment for intermediate-term IOP control in patients with NVG. According to Yalvac et al. (2006), the Ahmed glaucoma valve and Molteno single-plate implant were successful for early and intermediate-term of IOP control but in the long term both implants failed to achieve control of IOP. Hong et al. (2005) recently reviewed the literature on various glaucoma drainage devices and compared their success rate in controlling IOP. They concluded that the Moltino implant, Baerveldt implant, Ahmed glaucoma valve and Krupin valve showed no statistically significant difference in either the percentage change in IOP or the overall surgical success rate at the last follow up. One has to bear in mind that each glaucoma drainage device has its own complications and limitations. Placing a glaucoma drainage device in an eye with 360 degrees of synechiae is technically challenging and the fibrovascular membrane can encase the tube.
In 1984, in an experimental study in rhesus monkeys with iris NV, Packer et al. (1984) reported marked reduction of leakage from iris NV on fluorescein angiography after photodynamic therapy, but the effect was temporary and required repeated application to control iris NV. Parodi and Iacono (2005) recently reported that photodynamic therapy might be a promising approach for NVG. In their series of 16 eyes with NVG, there was a 39% decrease in IOP overall and treatment did not control IOP satisfactorily in 31%.
To make the eye feel comfortable, it is advisable to try first topical corticosteroids, cycloplegics, cyclodestruction, and even alcohol injection. If all else fails, as a last resort one may consider enucleation. Our policy is to try to avoid doing enucleation whenever possible, because even a blind eye is less bothersome in the long run (if cosmetically acceptable) than to maintain an artificial eye and socket. Some ophthalmologists advocate doing evisceration of such eyes.
While control of IOP helps to prevent visual loss in NVG, one has to keep in mind that visual outcome also depends upon the severity of underlying ocular disease and postoperative complications. Thus, successful control of IOP does not always correspond with visual outcome in NVG.
Finally, based on our studies on NVG in CRVO and ocular ischemic syndrome, in the Ocular Vascular Clinic at the University of Iowa Hospitals and Clinics, since 1973, I will discuss our management regimen of NVG.
As discussed above, PRP study by Hayreh et al. (1990b) in ischemic CRVO showed no significant difference in the development of NVG between the treated and the control group. I discussed above the serious flaws with the conclusions of the multicenter CVOS about the role of PRP (Hayreh 1996). Naturally, the question arises: if none of the advocated treatments is beneficial in ischemic CRVO, how should patients with ischemic CRVO be managed? I have discussed this topic in detail elsewhere (Hayreh 2003).
For a logical management of any disease, one has first to understand the basic issue involved and the available information, which should act as guidelines. In ischemic CRVO, to reiterate briefly what has been said above, we currently have the following definite information:
The most important fact to bear in mind in the management of NVG in ischemic CRVO is that the primary factor, which is going to completely blind these eyes, is the high IOP, producing optic nerve head damage, and NOT the ischemic CRVO per se. If one can keep the IOP under control by any means available, one can keep their peripheral visual fields intact. In the light of these facts, I follow the following regimen with these patients:
As in NVG due to any cause, our first priority is to lower the IOP to prevent further visual loss, particularly since in these eyes perfusion pressure in the ocular vascular bed is already low (Fig. 2). To do that, we use IOP lowering eye drops, along with topical steroids, and if need be add systemic carbonic anhydrase inhibitors. We avoid using beta-blocker eye drops because of their arterial hypotensive effect (Hayreh et al. 1999a). Cycloplegics may be useful to decrease ciliary pain and prevent posterior synechiae. If the IOP is uncontrolled with these measures, as with extensive peripheral anterior synechiae formation, or the medications are not tolerated, we do graduated cycloablation (see above) in addition to the above measures. Out of 30 eyes in our study (Mizener et al. 1997), 18 developed NVG but only four patients required cycloablation. In our series, only one eye developed phthisis bulbi, and this blind, painful eye required a retrobulbar alcohol injection after cycloablation failed to give relief.
NVG is a severely blinding disease. To prevent or reduce the extent of visual loss caused by NVG, the first essential is to have a high index of suspicion of its development; if NVG develops, early diagnosis and aggressive control of high IOP is crucial to minimize the visual loss. The most common diseases responsible for development of NVG are ischemic CRVO, diabetic retinopathy and ocular ischemic syndrome. In the management strategy, the first priority should be to try to prevent its development by appropriate management of the causative diseases.
Currently there is no satisfactory means of treating NVG and preventing visual loss in the majority, in spite of multiple modes of medical and surgical options advocated over the years and claims made. Currently our ability to prevent NVG and treat it is not satisfactory. For that, the primary aim of our future research should be prevention of the development of anterior segment NV and NVG. Primarily advances in basic sciences, which enable us to understand the disease process, make advances in medicine. As a clinical scientist, I have found that once one understands the basic scientific facts about a disease and its pathogenesis, one can almost mathematically calculate whether a particular treatment is going to work or not. Unfortunately, the major problem in clinical medicine has been lack of adequate knowledge of those basic scientific facts about the disease process among clinicians who invariably manage them and advocate various therapies. Not infrequently, with the best of intentions, treatment methods without any scientific rationale are advocated which in the long run not only do not work but even can be harmful. Thus, in the management of NVG, we first need scientifically valid means of management of the main diseases, which cause NVG, i.e. ischemic CRVO, diabetic retinopathy and ocular ischemic syndrome. While conventionally PRP has been advocated for the prevention and management of NVG in these diseases, as discussed above, apart from its effectiveness in diabetic retinopathy, it has been shown that not only it may not be effective in preventing NVG in ischemic CRVO and ocular ischemic syndrome but also it can cause further visual loss. After all, PRP is basically a destructive procedure. Currently there is considerable interest in the anti-VEGF drugs for management of ocular NV in age-related macular degeneration, but, so far, we have little long-term worthwhile information in a large series of patients about its effectiveness in prevention or control of NVG or ocular NV elsewhere. Various medical means to control high IOP are usually only temporary measures. Claims about the success of various surgical means to control IOP in NVG have not yet been substantiated in the long run. Thus, we are still far from having a satisfactory management method to prevent and treat NVG, to prevent visual loss.
I am grateful to my colleagues Drs. Wallace L.M. Alward, Young H. Kwon and Stephen R. Russell for valuable suggestions.
For comprehensive bibliography on he various topics, please consult the full bibliography given in the references listed above.