19 The Role of Intravitreal Steroids in the Management of Diabetic Retinopathy

edema and high-risk proliferative diabetic retinopathy for which immediate pan-retinal photocoagulation is required. Further research is warranted to determine the safest and most efficacious dose of IVTA, and into how ocular steroid therapy can be combined with both retinal laser treatment and the new anti-vascular endothelial growth factor treatments for the safest and most efficacious outcomes for patients.

Key Words: Adverse events; Diabetic macular edema; Intravitreal therapy; Triamcinolone acetonide.

Diabetic macular edema (DME) is the main cause of vision impairment in people with diabetes (1). In the Wisconsin Epidemiologic Study of Diabetic Retinopathy, macular edema developed in 20% of people with type I diabetes over a 10-year period

(2). Laser treatment has been proven effective in reducing the risk of visual loss from DME and is widely employed, but it is also inherently destructive. Progressive loss of vision occurred in up to 26% of patients with DME despite laser treatment in the Early Treatment of Diabetic Retinopathy Study (3). Thus an intervention that could reduce DME in patients who were failing laser treatment would prevent many cases of blindness in people with diabetes.

There has been increased recognition lately that features of chronic inflammation, such as adhesion of leukocytes to the retinal vasculature and migration into the retina, may play a role in the pathogenesis of diabetic retinopathy (4). Glucocorticoids have also been widely used for the treatment of edema in the brain and the lung. Since the blood–brain barrier is similar to the blood–retinal barrier, the long-standing use of corticosteroids in the treatment of brain edema, which is possibly mediated through suppression of vascular endothelial growth factor secretion (5), suggests they should be evaluated for the treatment of macular edema. In asthma, which is also characterized by increased vascular leak, steroids have been found to reduce vascular leak (6) and to suppress the release of endothelial cell activators. However, long-term systemic administration of steroids in people with diabetes would cause problems with glycemic control and elevate the risk of other adverse events to unacceptable levels. Intraocular administration of steroids has the potential to give extended doses of a drug at high local concentrations with minimal risks of systemic complications. Local delivery of high-dose long-acting steroids by periocular and orbital injections has been standard treatment for various inflammatory conditions of the eye for many years (7,8).

Intravitreal therapy with triamcinolone acetonide was first proposed as a treatment for ocular angiogenesis in a series of animal studies mainly performed by Machemer’s group at Duke University in the 1980s (9). Triamcinolone acetonide was chosen because its unique crystalline nature resulted in slow release of the drug over several months, and its vehicle appeared safe when injected into animal eyes (10,11).

The first report of its use in humans was by Penfold et al. from the Save Sight Institute in Sydney, in which intravitreal triamcinolone acetonide (IVTA) was used to treat a group of patients with exudative age-related macular degeneration (AMD) (12). IVTA was a popular treatment for subfoveal neovascularization for a while, particularly since there were no other treatments at that time that might stabilize or improve vision in such eyes, but its use waned with the advent of photodynamic therapy with verteporfin, the safety and efficacy of which had been proven by a phase III randomized clinical

trial; at the same time, a randomized clinical trial of intravitreal triamcinolone detected no benefit for AMD with classic choroidal neovascularization (13). These studies did, however, establish that IVTA had a manageable, albeit significant, adverse event profile, thus providing a foundation for clinical trials of IVTA for other macular diseases.

Subsequent animal studies suggested that IVTA might be efficacious for macular edema. Detecting the leak of a gadolinium-based marker into the vitreous, Wilson et al. found that intravitreal, but not sub-Tenon’s, triamcinolone significantly attenuated blood–retinal barrier breakdown caused by argon-laser panretinal photocoagulation in the rabbit eye (14). Edelman et al. found that a single intravitreal 2 mg dose of IVTA completely blocked VEGF-induced retinal and iris leakage for 45 days after VEGF165 was injected intravitreally in Dutch Belt rabbits, while indomethacin had no effect (15).

CLINICAL EFFICACY

The first report of the clinical use of IVTA for DME came from Jonas et al. in 2001 (16). This was quickly confirmed by a number of short case series which were extremely useful at the time. Martidis et al. presented a series of 16 eyes with DME treated with 4 mg IVTA with 1 month follow-up for 14 eyes and 6 months for 8. Snellen visual acuity improved in association with reduction of central macular thickness (17). Similarly, Jonas et al. reported improvement in Snellen visual acuity and fluorescein leakage in 26 eyes of 20 patients treated with 25 mg of triamcinolone in 0.2 ml with a mean follow-up of 6 months (16).

The subsequent explosion of short-term, uncontrolled studies of IVTA for practically every conceivable acquired macular condition has, however, added little more information, particularly for DME. Once efficacy has been suggested in phase I/II studies, interventions need to be tested in randomized clinical trials. As experience with AMD demonstrated, the apparent initial efficacy of interventions for macular disease frequently does not stand up to the scrutiny of a formal, double-masked clinical trial in which primary end points are prospectively identified and measured by trained observers.

The Triamcinolone for Diabetic Macular Edema study (TDMO), also conducted at the Save Sight Institute, was the first randomized clinical trial that was adequately powered to test the hypothesis that an intravitreal injection of triamcinolone acetonide would be beneficial for DME that had failed laser treatment. It was also the first study published to follow patients for long enough (2 years) to provide a realistic estimation of the risks of treatment.

The TDMO study was a prospective, double-masked, placebo-controlled randomized clinical trial that enrolled 69 eyes of 43 patients, with 34 eyes receiving active treatment and 35 placebo. The procedure was designed to be performed in the office under sterile conditions with topical and subconjunctival anesthesia. Triamcinolone acetonide (0.1 ml of 40 mg ml−1) was injected through the pars plana using a 27G needle. Patients with persistent DME involving the central fovea persisting for 3 months or more after adequate laser treatment and best-corrected visual acuity in the affected eye(s) of 6/9 or worse were included. Eyes randomized to placebo received a subconjunctival injection of saline. The main outcome measures were improvement of best-corrected Log MAR visual acuity by five or more letters and incidence of moderate or severe adverse events. Retreatment was considered at each visit as long as treatments

438 Gillies

Table 1

Effect of Triamcinolone on Change in Eye Outcomes 3 Months from Baselinea

(Reprinted from (18) with permission) a Plus-minus values are means ± SE

* P-value using generalized estimating equations (GEE) to allow for correlations between paired eyes

were at least 6 months apart. Eyes with a reduction of visual acuity of at least five letters from previous peak value and persistent central macular thickness greater than 250 m received retreatment with study medication. If visual acuity had not improved significantly when measured 4 weeks later and macular thickening persisted, then fluorescein angiography was performed and further laser treatment was applied if the investigator thought it would be beneficial.

The data were analyzed at 3 months to ensure that a single treatment was in fact efficacious in the short term (18). Vision improved by five or more letters in 18/33 (55%) eyes treated with IVTA that completed this visit compared with 5/32 (16%) eyes treated with placebo (P = 0.002) (Table 1). Central macular thickness decreased by a mean of 152 m in the treated eyes compared with 36 m in placebo-treated eyes. One IVTA-treated eye developed infectious endophthalmitis, which was promptly treated with persistent improvement of visual acuity compared with baseline throughout the study.

After 2 years, the beneficial effect of IVTA continued to hold up (19). Vision had improved by five or more letters in 19/34 (56%) IVTA-treated eyes compared with 9/35 (26%) eyes treated with placebo (P = 0.006). Only 6/34 (18%) treated eyes lost five or more letters compared with 13/35 (37%) untreated eyes. The mean improvement in visual acuity was 5.7 (95% CI: 1.4–9.9) letters more in the IVTA-treated eyes than in those treated with placebo (Table 2). The mean number of injections received in the IVTA-treated group was 2.6 with a maximum of 5 possible (Table 3). Although some of the placebo-treated eyes had done well, emphasizing the importance of randomized

Table 2

Effect of Triamcinolone on Change in Eye Outcomes 24 Months from Baseline

(Reprinted from (19) with permission) aN = 21 for each group

*P-value from exact Mantel–Haenszel trend test

**P-value using generalized estimating equations (GEE) to allow for correlations between paired eyes

Table 3

Distribution of Number of Treatments Given by Treatment

Group for Patients Completing 2-year Follow-up

Mantel–Haenszel trend test c12 = 7.57, P = 0.006 (reprinted from (19) with permission)

clinical trials for treatments for DME, IVTA treatment had roughly doubled the chance of improving vision and halved the risk of visual loss. The mean improvement in visual acuity was not especially large. However, many eyes in the study had 20/30 vision on entry. If only eyes with 20/40 visual acuity had been accepted, then improvements would likely have been greater. The study continues, with 5-year visits to be completed by March 2008.

It has been reported that repeated intravitreal injections may not be as effective as the initial treatment (20). However, we have yet to find any evidence of this. In the TDMO study, the mean number of injections was 2.4 over 2 years with a total potential of 5. We found that there was no difference in the reduction in central macular thickness and