Safety of an Intravitreal Injection of Triamcinolone: Results From a Randomized Clinical Trial

Objective  To determine the safety of a single intravitreal injection of triamcinoloneacetonide (4 mg) in patients with subfoveal choroidal neovascularization causedby age-related macular degeneration.

Methods  A double-masked, placebo-controlled, randomized clinical trial was conductedat a public tertiary referral eye hospital. Patients participating had age-relatedmacular degeneration with evidence of choroidal neovascularization, any partof which was classic; age older than 59 years; and best-corrected visual acuityof 20/200 or better. Eyes were assigned to active study treatment or to placebo.Intraocular pressure and cataract grading were performed every 6 months for3 years. Adverse events, from mild to vision-threatening or life-threatening,were recorded as procedure-related or corticosteroid-related.

Results  Seventy-five eyes were assigned to study treatment and 76 eyes to placebo.There were no moderate or severe adverse events related to the surgical procedurein either group. Triamcinolone-treated eyes had a significantly increasedrisk of developing mild or moderate elevation of the intraocular pressure.Topical glaucoma medication reduced intraocular pressure to acceptable levelsin all patients. There was significant progression of cataract in the triamcinolone-treatedeyes.

Conclusion  Despite a significant adverse event profile, intravitreal triamcinoloneis generally well tolerated by the human eye as long as patients are carefullyfollowed up by their surgeon and treated appropriately, when necessary.

Periocular and orbital injections of long-acting corticosteroids havebeen standard treatments for various inflammatory conditions of the eye formany years.^1,2 Some authoritieshave proposed that peribulbar administration of corticosteroids might notbe ideal for the treatment of chronic ocular disease, suggesting that theiraction is partly due to systemic absorption of the drug and that their effectivedose lasts only a few days to weeks.³ Intraocularadministration of corticosteroids has the potential to give extended dosesof a drug at high local concentrations.

The safety of intraocular corticosteroids has yet to be studied systematicallyin humans. The risks of procedure-related (eg, endophthalmitis and retinaldetachment) or corticosteroid-related (cataract and elevated intraocular pressure[IOP]) adverse events need to be balanced against the potential beneficialeffects of any treatment.

The Intravitreal Triamcinolone Study is a prospective, single-center,double-masked, placebo-controlled, randomized clinical trial to test the hypothesisthat a single intravitreal injection of triamcinolone acetonide (4 mg) willreduce the risk of severe visual loss in eyes with classic neovascular age-relatedmacular degeneration. Despite angiographic evidence that the treatment significantlyinhibited the growth of subretinal neovascular fronds 3 months after injection,it did not reduce the risk of severe visual loss during the first year ofthe study.⁴ The study also presents a uniqueopportunity to examine the adverse events associated with the intraocularadministration of a corticosteroidal agent.

Patients were recruited from the retina clinics of the Sydney Eye Hospital,a major public tertiary referral center in New South Wales.

Eligibility criteria included age older than 59 years, neovascular age-relatedmacular degeneration with any classic component that was unsuitable for argonlaser photocoagulation, and best-corrected visual acuity of 20/200 or bettertested on a logMAR chart. The trial was conducted before the advent of photodynamictherapy; all patients had declined argon laser treatment for subfoveal neovascularization.Exclusion criteria included other serious eye diseases, including uncontrolledglaucoma or a glaucomatous visual field defect, and a history of using systemiccorticosteroids. Patients were enrolled in the study after they provided written,informed consent; standard fundus photographs, including fluorescein angiograms,had been taken; and the treatment allocation had been issued by the independentdesignated officer in the clinic.

Intravitreal triamcinolone was injected into the vitreous within 1 weekof the baseline angiogram and on the day of the baseline visual acuity measurements.The procedure was performed in the outpatient clinic under aseptic conditions.Eyes were prepared with 1 drop of guttae 0.25% apraclonidine hydrochloride,several drops of guttae 1% amethocaine hydrochloride, and 2 flushes of half-strengthpovidone-iodine (Betadine). A small amount of 2% lignocaine hydrochloridewas then administered subconjunctivally with a 30-gauge needle to the siteof the injection, and the IOP was reduced by digital massage. Five minuteslater using a 27-gauge needle, 0.1 mL of triamcinolone acetate (Kenacort 40;Bristol-Myers Squibb Australia, Noble Park, Victoria) was injected into thevitreous. A small amount of chloramphenicol ointment was then instilled. Thecommercially available preparation Kenacort 40 (40 mg/mL of triamcinoloneacetate) was constituted as follows: 40 mg of triamcinolone acetonide, 6.9mg of sodium chloride, 15 mg of benzyl alcohol, 7.5 mg of carmellose sodium,and 0.4 mg of polysorbate 80 per milliliter of drug. At the time of manufacture,the air in the container is replaced by nitrogen. No attempt was made to decantthe solution.

Patients were seen 1, 6, 12, and 26 weeks after treatment and then every6 months for up to 3 years. At each visit, the IOP was determined and corticalcataracts, nuclear sclerosis, and posterior subcapsular cataracts were gradedindividually (range, 0 to ≥4) using a semiquantitative scale with the aidof photographic standards from the Age-Related Eye Disease Study. Data wereentered onto standard data collection forms and transferred to an electronicdatabase by dual data entry. Source data verification was performed by 2 independentstudy monitors on all patients for eligibility, demographics, and outcomeand safety data.

Adverse events were graded as follows: grade 1, mild; grade 2, moderate;and grade 3, severe (vision-threatening or life-threatening). Elevation ofIOP was graded according to the level before or after the first line of glaucomamedication had been started, when appropriate (Table 1).

Safety data were reviewed by a safety monitoring committee after 50,80, 110, 140, and 151 eyes had been entered into the study. This committeewas instructed to break the randomization code and consider stopping the trialif a highly significant (P<.01) difference betweenthe 2 groups in the development of severe adverse events was found. The studywas approved by the human ethics committees of the University of Sydney andthe South Eastern Sydney Area Health Service (Eastern Section).

Data were analyzed on the basis of intention to treat. The numbers ofadverse events in the treated and control groups were compared using Fisherexact test or the exact test for trend in proportions in SAS version 8 (SASInstitute Inc, Cary, NC). The mean IOP was plotted against time for the 2groups and compared using a 2-sided t test when itpeaked in the triamcinolone group. No allowance was made in the analysis forpossible correlation between paired eyes of the same individual, as only 12patients had both eyes treated in the trial.

One hundred fifty-one eyes of 139 patients were entered into the studyfrom November 4, 1996, to June 29, 1999. All patients were white. Seventy-fiveeyes were treated with triamcinolone and 76 with placebo. Data in this reportinclude observations made until 44 months after the entry of the first patient.One hundred forty-two eyes (94.0%) were available for analysis at 12 months,100 (66.2%) at 2 years, and 35 (23.2%) at 3 years. Baseline patient demographicdata are shown in Table 2. The2 groups of patients were similar, with a slight preponderance of patientsin the placebo-treated group being female and having a history of systemichypertension. Figure 1 shows theflow of patients through the study.

Procedure-related adverse events are shown in Table 3. There were no severe or moderate adverse events in treatedor control eyes related to the intraocular or periocular injection. A fewpatients in both groups reported low-grade symptoms associated with the injection,such as transient discomfort or blurring.

Corticosteroid-related adverse events are shown in Table 3. Treated eyes had a significantly increased risk of developingmild or moderate elevation of the IOP.

Elevated IOP was adequately treated with topical medication in all eyes,without recourse to laser or surgical treatment. In particular, all eyes withIOPs greater than 25 to 40 mm Hg during treatment were receiving only 1 medication.The pressure was quickly normalized (<25 mm Hg in eyes without cupped discsand <18 mm Hg in eyes with preexisting glaucoma) in all eyes by the additionof further medications, as appropriate. The decision to treat elevated IOPwas based on conventional considerations in each patient, including the degreeof elevation, the extent of cupping of the optic nerve head, and whether therewas a history or family history of glaucoma. Twenty-one (28.0%) of 75 eyesreceiving triamcinolone required treatment with topical glaucoma therapy tocontrol elevated IOP, compared with 1 (1.3%) of 76 placebo-treated eyes (P<.001). A single medication was sufficient in 18 (85.7%)of these eyes, while the other 3 (14.3%) required 2 medications. This treatmentwas discontinued in 15 (71.4%) of 21 eyes receiving triamcinolone after amean of 8 months (range, 1.5-32 months). Treatment was discontinued in 11(52.4%) of 21 eyes at the 6-month posttreatment visit. Six (8.0%) of 75 eyesthat were not receiving glaucoma medication at entry into the study continuedreceiving glaucoma medication at the last study visit.

Figure 2 shows the mean IOPduring the study for treated and control eyes. A significant elevation intreated eyes was found at 6 weeks (21.8 vs 17.1 mm Hg, P = .03). The curves converged during the rest of the study as affectedeyes were prescribed topical medication.

Treated eyes had a significantly increased risk of moderate progressionof cataract. Table 4 shows theproportion of eyes in the treated and control groups that developed progressionby 2 or more grades on the Age-Related Eye Disease Study semiquantitativescale for nuclear, cortical, and posterior subcapsular lens opacities duringthe study. There was significant progression of cataract in the triamcinolone-treatedeyes. By the 24-month visit, 8 (24.2%) of 33 triamcinolone-treated eyes hadprogression by 2 or more Age-Related Eye Disease Study grades (5 progressedby 2 grades, and 3 by 3 grades) compared with 0 of 22 placebo-treated eyes(P = .02). Cataract surgery was performed in 16 (28.6%)of 56 treated eyes of patients who completed at least the 12-month study visitvs 2 (5.0%) of 40 eyes receiving placebo (P = .003).Cataract surgery was performed at a mean of 25 months (range, 12-34 months)after treatment. The decision to recommend cataract surgery was based on conventionalconsiderations, including the density of the opacification, the level of visualacuity in the fellow eye, and the patient's willingness to undergo surgery.

One eye treated with triamcinolone developed a hemiretinal vein occlusion2 months after entry into the study. The IOP at the preceding clinical visitwas 16 mm Hg. Visual acuity (20/200) was not affected. This adverse eventwas judged to be possibly related to treatment with triamcinolone.

Severe adverse events thought to be unrelated to treatment occurredin 1 eye receiving triamcinolone and in 3 control eyes (P = .62). The eye receiving triamcinolone developed presumed arteriticanterior ischemic optic neuropathy 6 months after treatment. Elevation ofthe erythrocyte sedimentation rate and C-reactive protein was detected, butresults of a temporal artery biopsy were negative. Three control eyes developedmassive subretinal or breakthrough vitreous hemorrhage 1, 3, and 26 weeksfollowing treatment.

The results of this study indicate that a single 4-mg injection of triamcinoloneacetonide appears to have an acceptable adverse effect profile in older humaneyes. Some mild adverse events were associated with the surgical procedureof intravitreal injection in the 75 eyes treated. A significant incidenceof corticosteroid-related adverse events was found, including moderate elevationof the IOP and the development of cataract requiring surgery. All eyes withelevated IOP were adequately controlled with topical therapy alone. Surgeryfor corticosteroid-induced cataract was undertaken, when appropriate. Thesafety monitoring committee was satisfied that all adverse events were adequatelymanaged.

Few data exist on the complications of intravitreal injection of corticosteroidsin human eyes. In a series of 28 eyes receiving triamcinolone, severely elevatedIOP in 4 eyes was reported, but these had received a second injection within4 months of the first.⁵ A brief report of 113eyes receiving a single 4-mg injection of triamcinolone acetonide found elevationof the IOP by 5 mm Hg or higher in 32% of eyes 3 months after treatment.⁶ In 6 eyes that received intravitreal triamcinolonefor cystoid macular edema in uveitis, 5 developed elevated IOP and 2 developedcataract.⁷

We tried to define a grading system for the severity of the IOP risethat would be relevant to clinical practice. A system that only consideredthe degree of IOP rise without treatment might be unnecessarily severe, becausea very high pressure rise is not necessarily a severe adverse event if itresponds satisfactorily to treatment. On the other hand, a system that onlyincluded the degree of IOP elevation after full treatment might have the reverseeffect. In the present study, all IOP rises were adequately treated with topicaltherapy. We chose to examine the IOP level before treatment and after thefirst-line glaucoma medication had been introduced. It might be alleged thatan IOP in the high 30s after institution of first-line therapy is a severeadverse event rather than moderate, but this should be viewed in the contextof a regimen of careful postinjection follow-up, in which patients are checkedregularly for adverse events, such as elevated IOP, which was treated adequatelyin all cases with topical medication.

A peculiar condition known as noninfectious endophthalmitis has beendescribed in eyes receiving triamcinolone.⁸ Therewere no such events in eyes treated in the present study.

A particular strength of this study is that the data were collectedin a masked, standardized fashion from eyes treated in a randomized clinicaltrial, with an equally sized group of untreated eyes, which provides a controlagainst the development of events, such as cataract or breakthrough vitreoushemorrhage, that might occur even without corticosteroid treatment. We wereconscious that masking would be an issue in a study such as this. Consequently,the observations of changes in the anterior segment of the eye that constitutemost of the data in the present study were made and recorded before examinationof the posterior segment, when it might have been possible for the observerto detect whether the patient had been treated with triamcinolone or not.It should be noted that the study is not large, having only a limited powerto exclude any serious adverse event that occurred in a small but significantproportion of treated eyes.

The duration of the IOP rise, when it occurred, provides some indicationof how long a single 4-mg dose of triamcinolone acetonide persists in theeye at a significant concentration. Studies^9,10 inrabbit eyes have found that triamcinolone persists in the vitreous for 3 to6 weeks after it has been injected. Pharmacologically active triamcinolonewas, however, identified up to 13 months after subconjunctival administrationin humans, when it was excised for persistent elevation of the IOP that couldnot be controlled with topical medication.¹¹ Thepressure returned to normal within 1 week in 6 of 7 patients. Mills et al¹² reported 2 similar patients in whom ocular hypertensionpromptly resolved after the subconjunctival triamcinolone was excised 6½and 10 months after it had been injected. In the present study, the mean durationof treatment of the corticosteroid-induced elevation of IOP was 8 months,suggesting that the elevated IOP persisted for at least 6 months in most eyes.These data suggest that a single intravitreal injection of 4 mg of triamcinoloneacetonide may persist in the human eye in significant concentrations for around3 to 4 months.

The periorbital administration of long-acting corticosteroids is alsoassociated with a significant incidence of adverse effects. Adverse eventsassociated with the treatment of intermediate uveitis with triamcinolone injectedinto the posterior sub-Tenon space, which is thought to be the safest routeof delivery,^13,14 were reportedby Helm and Holland.¹⁵ They reported developmentof significant cataract in 4 (36.4%) of 11 phakic eyes 10 months to 4 yearsafter treatment. They also found significant elevation of the IOP (median,32 mm Hg; range, 25-40 mm Hg) in 6 (30.0%) of all 20 eyes studied. In thepresent study, 16 (28.6%) of 56 patients receiving triamcinolone underwentcataract surgery, and 28% required glaucoma medication. It would appear, therefore,that the risk of adverse events after an intravitreal injection of triamcinoloneis similar to that associated with an orbital injection. Some adverse eventsare particularly related to periocular injection and would not be expectedto occur after intraocular injection. These include embolic retinal arteryocclusion and accidental penetration or perforation of the globe.^16-19

The findings from our study neither advocate nor support the use ofintravitreal corticosteroids for the treatment of retinal neovascularizationor any other condition. Intravitreal triamcinolone has been studied in animalsfor retinal neovascularization and proliferative vitreoretinopathy.^19-23 Severalcase reports or small case series have recently appeared concerning the clinicaluse of intravitreal triamcinolone for macular edema in uveitis,^7,24,25 diabeticmacular edema,²⁶ and proliferative vitreoretinopathy.²⁷ The use of intraocular corticosteroids in these andother conditions should be guided by the results of appropriate clinical trialsas they are reported and discussed. Intraocular delivery of corticosteroidsappears to be safe and has the potential to be more efficacious and sustainedthan other routes of administration.

Corresponding author: Mark C. Gillies, MD, PhD, Save Sight and EyeHealth Institute, Department of Clinical Ophthalmology, Sydney Eye Hospital,University of Sidney, GPO Box 4337, Sydney, New South Wales, Australia 2001(e-mail: mark@eye.usyd.edu.au).

Submitted for publication August 15, 2002; final revision received July30, 2003; accepted August 19, 2003.

This study was funded by grant 974052 from the National Health and MedicalResearch Council, Canberra, Australia, and the Sydney Eye Hospital Foundation.

Members of the safety monitoring committee were Dr Simpson (chair),Jeremy Smith, FRANZCO, Justin Playfair, FRANZCO, and Paul Power, MS.