Ranibizumab for Treatment of Neovascular Age-Related Macular Degeneration

Ranibizumab for Treatment of Neovascular Age-Related Macular Degeneration A Phase I/II Multicenter, Controlled, Multidose Study Jeffrey S. Heier, MD, 1 Andrew N. Antoszyk, MD, 2 Peter Reed Pavan, MD, 3 Steven R. Leff, MD, 4 Philip J. Rosenfeld, MD, PhD, 5 Thomas A. Ciulla, MD, 6 Richard F. Dreyer, MD, 7 Ronald C. Gentile, MD, 8 Judy P. Sy, PhD, 9 Gary Hantsbarger, PhD, 9 Naveed Shams, MD, PhD 9 Objective: To assess safety of repeated intravitreal injections of ranibizumab in treating neovascular agerelated macular degeneration (AMD), and to assess changes in visual acuity (VA) and AMD lesion characteristics. Design: Multicenter, controlled, open-label, clinical trial. Participants: Sixty-four patients with subfoveal predominantly or minimally classic AMD-related choroidal neovascularization. Methods: In part 1, subjects were randomized to monthly intravitreal ranibizumab for 3 months (4 injections of 0.3 mg or 1 injection of 0.3 mg followed by 3 injections of 0.5 mg; n 53) or usual care (UC; n 11). In part 2, subjects could continue their regimen for 3 additional months or cross over to the alternative treatment. Main Outcome Measures: Adverse events (AEs), intraocular pressure (IOP), VA, and lesion characteristics assessed by fluorescein angiography and fundus photography. Results: Of the 64 randomized subjects, 62 completed the 6-month study. Twenty of 25 subjects (80%) randomized to 0.3 mg, and 22 of 28 subjects (79%) randomized to 0.5-mg ranibizumab in part 1 continued on that treatment in part 2; 9 of 11 (82%) subjects randomized to UC in part 1 crossed over to ranibizumab treatment in part 2. The most common AEs with ranibizumab were reversible inflammation and minor injection-site hemorrhages. Serious AEs were iridocyclitis, endophthalmitis, and central retinal vein occlusion (1 subject each). Postinjection, IOP increased transiently in 22.6% of ranibizumab-treated eyes in parts 1 and 2. After 4 ranibizumab injections (day 98), mean ( standard deviation) VA had increased 9.4 13.3 and 9.1 17.2 letters in the 0.3- and 0.5-mg groups, respectively, but had decreased 5.1 9.6 letters with UC. In part 2 (day 210), VA increased from baseline 12.8 14.7 and 15.0 14.2 letters in subjects continuing on 0.3 and 0.5 mg, respectively. Visual acuity improved from baseline 15 letters in 26% (day 98) and 45% (day 210) of subjects initially randomized to and continuing on ranibizumab, respectively, and areas of leakage and subretinal fluid decreased. No UC subject had a 15-letter improvement at day 98. Conclusions: Repeated intravitreal injections of ranibizumab had a good safety profile and were associated with improved VA and decreased leakage from choroidal neovascularization in subjects with neovascular AMD. Ophthalmology 2006;113:633 642 2006 by the American Academy of Ophthalmology. Overexpression of vascular endothelial growth factor-a (VEGF-A), a diffusible cytokine that stimulates vascular permeability and angiogenesis, 1 is thought to play a key role in age-related macular degeneration (AMD), 2,3 diabetic retinopathy, and other retinal disorders associated with neovascularization. 4 Inhibitors of VEGF-A activity may therefore be useful in the treatment and management of these disorders. 5,6 Originally received: May 24, 2005. Accepted: October 20, 2005. Manuscript no. 2005-448. 1 Ophthalmic Consultants of Boston, Boston, Massachusetts. 2 Charlotte Eye, Ear, Nose, & Throat Associates, Charlotte, North Carolina. 3 University of South Florida College of Medicine, Tampa, Florida. 4 Retina Vitreous Center, New Brunswick, New Jersey. 5 Bascom Palmer Eye Institute, University of Miami School of Medicine, Miami, Florida. 6 Midwest Eye Institute, Indianapolis, Indiana. 7 Retina Northwest, Portland, Oregon. 8 New York Eye and Ear Infirmary, New York, New York. 9 Genentech, Inc., South San Francisco, California. Presented in part at: American Society of Retinal Specialists Annual Meeting, August, 2003; New York, New York, and American Academy of Ophthalmology Annual Meeting, November, 2003; Anaheim, California. This study was supported by Genentech, Inc., South San Francisco, California. Financial disclosure/conflict of interest: Drs Heier, Antoszyk, Rosenfeld, and Gentile have served on the Lucentis Advisory Board; Dr Pavan has served as a consultant to Genentech for Lucentis; and Drs Sy, Hantsbarger, and Shams are employees of Genentech, Inc. Reprint requests to Jeffrey S. Heier, MD, Ophthalmic Consultants of Boston, 50 Staniford Street, Suite 600, Boston, MA 02114. 2006 by the American Academy of Ophthalmology ISSN 0161-6420/06/$ see front matter Published by Elsevier Inc. doi:10.1016/j.ophtha.2005.10.052 633

Ophthalmology Volume 113, Number 4, April 2006 Ranibizumab (Lucentis, Genentech, Inc., South San Francisco, CA) is a specific affinity-matured, recombinant, humanized anti VEGF-A neutralizing antibody fragment (Fab). 7 9 The fragment is one third the size of a full-length antibody and readily penetrates all layers of the retina after intravitreal injection. 10 Ranibizumab is being developed as a potential treatment for various VEGF-A mediated ocular vascular diseases and is in phase III trials for treatment of neovascular AMD. One-year results from a 2-year phase III trial in treatment of minimally classic and occult-only AMD-related choroidal neovascularization indicate that monthly injection or 0.3- or 0.5-mg ranibizumab halted the growth of the choroidal neovascularization lesion, reduced vascular leakage, and in one quarter to one third of treated patients, respectively, improved visual acuity (VA) [Miller JW, Chung CY, Kim R. Randomized, controlled, phase III study of ranibizumab (Lucentis) for minimally classic or occult neovascular age-related macular degeneration. Presented at: American Society of Retinal Specialists 23rd Annual Meeting, July 18, 2005; Montreal, Canada]. As part of this clinical development program, the phase I/II clinical trial (Genentech study FVF2128g) reported here evaluated the safety of repeated intravitreal injections of ranibizumab (0.3 mg and 0.5 mg) for up to 6 months in patients with active AMD-related choroidal neovascularization. Because animal data suggested that starting with a lower dose and increasing over subsequent injections might lessen inflammatory responses (Genentech, unpublished data), subjects treated with the higher of the 2 dose levels received the lower dose for the first injection only. Secondary objectives of the study were preliminary evaluation of the effects of ranibizumab treatment on VA and lesion characteristics as assessed by fluorescein angiography (FA) and fundus photography. Materials and Methods This was a 30-week, multicenter (8 centers), controlled, open-label study of intravitreally administered doses of ranibizumab (0.3 mg and 0.5 mg) versus usual care (UC). All sites received institutional review board approval before study initiation, and all subjects provided written informed consent before eligibility screening. Eligible subjects were 50 years old with primary or recurrent (i.e., previously treated) active subfoveal AMD-related choroidal neovascularization in the study eye (based on investigator evaluation of a screening fluorescein angiogram obtained 14 days before study day 0) and best-corrected VA in the study eye of 20/40 to 20/400 (Snellen equivalent) using Early Treatment Diabetic Retinopathy Study charts at 2 m. The investigator determined if the AMD lesion (defined as a localized area of neovascular growth, based on FA) met the following additional eligibility criteria: a classic choroidal neovascularization component was observed or the AMD lesion had been previously treated with verteporfin photodynamic therapy; total lesion size 9 optic disc areas (DAs), with choroidal neovascularization component 50% of the lesion (unless a serous pigment epithelial detachment was present, in which case 50% choroidal neovascularization was acceptable); active choroidal neovascularization leakage; and submacular blood 50% and subretinal fibrosis 25% of the total lesion. Key exclusion criteria were any of the following with respect to the study eye: additional eye disease that could compromise VA; choroidal neovascularization unrelated to AMD; ocular inflammation; vitreous hemorrhage; retinal hemorrhage (other than AMDrelated submacular blood) 1 DA; intraocular surgery 1 month before day 0; uncontrolled glaucoma; 3 prior treatments with verteporfin photodynamic therapy 12 months before day 0, or the last treatment was 1 month before day 0; laser photocoagulation or other intervention for AMD 14 days before day 0; previous treatment with external-beam radiation therapy or transpupillary thermotherapy; and history of vitrectomy. Before randomization, subjects study eye lesions were classified into 3 strata based on the investigator s screening assessment: predominantly classic ( 50% classic choroidal neovascularization) and no prior photodynamic therapy, minimally classic ( 50% classic choroidal neovascularization) and no prior photodynamic therapy, and previously treated with photodynamic therapy, independent of the size of the classic choroidal neovascularization component. To maximize the safety information obtained in the relatively small number of subjects and also provide an opportunity for those randomized to UC to later receive ranibizumab, subjects were permitted to cross over to the alternative treatment after 3 months of the randomized treatment. In part 1 (3 months), subjects were enrolled in 2 dose cohorts for reasons of safety. In the first cohort, subjects were randomized to either UC, which was photodynamic therapy in predominantly classic lesions and observation in all other lesions, or 4 injections of 0.3-mg ranibizumab at monthly intervals. If an interim safety review revealed no concerns in this first cohort, a second cohort allowing comparison of UC and a higher-dose ranibizumab regimen (initial injection of 0.3-mg ranibizumab followed by 3 injections of 0.5-mg ranibizumab) was to be enrolled. The target sample size of 24 subjects receiving ranibizumab and 6 subjects receiving UC in each enrollment cohort (i.e., 4:1 ratio) was considered sufficient for safety assessment. In part 2 (3 months), subjects randomized to ranibizumab in part 1 could receive either an additional 4 monthly injections at the same dose or cross over to UC, and subjects randomized to UC for part 1 could cross over to receive 4 monthly injections of ranibizumab (0.3 mg or 0.5 mg). Although ranibizumab treatment in part 2 was intended only for subjects who continued to have eligible lesions, some not meeting this criterion were treated in part 2 because of investigators differing interpretations of the guidance. A schematic of the study design is presented in Figure 1. Ranibizumab was reconstituted from a lyophilized preparation and administered by intravitreal injection (volume, 50 l). Just before injection, the study eye was prepared with an antimicrobial (antibiotic, povidone iodine solution, or both) of the investigator s choice and local administration of anesthetic (2% lidocaine, topical or subconjunctival). Subjects returned to the clinic 1 and 3 days after injection for safety assessments. Standardized VA testing by certified examiners masked to treatment assignment and without knowledge of the subject s prior VA scores was performed every 2 weeks using Early Treatment Diabetic Retinopathy Study charts at an initial testing distance of 2 m. Slit-lamp and dilated fundus examination were also performed every 2 weeks. Certified personnel performed retinal photography and FA monthly, before study treatment. Although individual investigators determined initial eligibility, all photographs and angiograms were evaluated subsequently by masked readers at a central reading facility (University of Wisconsin). Subgroup analyses were based on reading center assessment of the percentage of classic choroidal neovascularization at baseline. The primary outcome of interest was the safety and tolerability of the study treatment. Verbatim descriptions of all adverse events (AEs) were converted to Medical Dictionary for Regulatory Ac- 634

Heier et al Ranibizumab (Anti-VEGF Antibody) for Neovascular AMD Subjects Randomized to Receive Ranibizumab during Part 1 Crossover Part 1 Part 2 Week -2 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 Dose Group A (n=24) 0.3 mg 0.3 mg 0.3 mg 0.3 mg 0.3 mg 0.3 mg 0.3 mg 0.3 mg Dose Group B (n=24) 0.3 mg 0.5 mg 0.5 mg 0.5 mg 0.5 mg 0.5 mg 0.5 mg 0.5 mg Subjects Randomized to Receive during Part 1 Crossover Part 1 Part 2 Week -2 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 Dose Group A (n=6) Dose Group B (n=6) 0.3 mg 0.3 mg 0.3 mg 0.3 mg 0.3 mg 0.5 mg 0.5 mg 0.5 mg Screening Period Ranibizumab Phase Ranibizumab injection Usual care Ranibizumab injection (only for subjects receiving ranibizumab in part 2) Figure 1. Study schema. In part 1, subjects were randomized to intravitreal ranibizumab every 4 weeks for 12 weeks (4 injections of 0.3 mg or 1 injection of 0.3 mg, followed by 3 injections of 0.5 mg,) or usual care (UC). In part 2, subjects could continue their regimen for 12 additional weeks or cross over to the alternative treatment. During part 2, subjects on UC had their final visual acuity assessment at day 196 (week 28; i.e., 14 weeks after start of part 2), whereas subjects on ranibizumab had their final assessment at day 210 (week 30; i.e., 14 weeks after start of part 2 plus an additional 2 weeks to allow recovery from injection procedure). tivities (version 2.0) terms for statistical analysis. Inflammation was evaluated by grading flare and cells from 0 to 4 using scales modified from those described by Hogan et al 11 (Table 1 [available at http://aaojournal.org]). Secondary outcomes included changes in VA and lesion characteristics (based on FA). The main assessment of study outcomes was made on day 98, with follow-up assessment on day 196 or day 210. All data were collected on case report forms and analyzed using SAS software (SAS Institute, Cary, NC). Results Sixty-nine subjects enrolled from 8 sites were randomized in part 1 to 0.3-mg ranibizumab, 0.5-mg ranibizumab, or UC. Five subjects (2 in the 0.3-mg ranibizumab group, 1 in the 0.5-mg ranibizumab group, and 2 in the UC group) withdrew before starting study treatment. Randomization was stratified (with approximately one third of the subjects in each stratum), based on investigators assessments, into those with predominantly classic choroidal neovascularization (n 21), minimally classic choroidal neovascularization (n 25), and those who had prior photodynamic therapy (n 18). Of the 64 subjects, 30 were enrolled into the first cohort (25 randomized to 0.3 mg of ranibizumab and 5 to UC) and 34 into the second cohort (28 randomized to 0.5 mg of ranibizumab and 6 to UC). Thus, a total of 25 subjects were randomized to 0.3 mg, 28 to 0.5 mg, and 11 to UC. Of the 53 ranibizumab-treated subjects, 52 received all 4 scheduled doses during part 1, and 1 subject received 3 doses. The third of the 4 scheduled doses was held in the latter subject because of severe iridocyclitis, which resolved; this subject crossed over to UC in part 2. Because of scheduling issues, 1 subject received the fifth dose of ranibizumab at the day 98 visit (and thus, technically, 5 doses in part 1 of the study); this subject received a total of 8 doses in parts 1 and 2. In part 2, 9 of 11 subjects (82%) randomized to UC in part 1 crossed over to receive ranibizumab (4 at the 0.3-mg dose and 5 at the 0.5-mg dose). Five of 25 subjects (20%) randomized to 0.3 mg, and 6 of 28 subjects (21%) randomized to 0.5 mg in part 1 crossed over to receive UC in part 2. Of the 64 subjects starting part 2, 62 completed the entire treatment schedule. Two subjects (1 receiving 0.3-mg ranibizumab in both parts and 1 receiving 0.5 mg during both parts) withdrew from the study because of AEs (central retinal vein occlusion [CRVO] and cerebrovascular accident, respectively). At baseline, the treatment groups were similar with respect to demographic characteristics, VA, investigator classification of choroidal neovascularization lesion (predominantly or minimally classic), and photodynamic therapy history (Table 2). Based on central reading facility assessment, the mean total lesion size was approximately 3 DAs (mean standard deviation, 3.55 2.60, 2.99 2.04, and 3.40 2.59 DAs in the UC, 0.3-mg, and 0.5-mg groups, respectively); the total area of choroidal neovascularization was slightly smaller (Table 2). Mean percentages of the lesion that were classic choroidal neovascularization were 21.7%, 41.8%, and 31.3% in the UC, 0.3-mg, and 0.5-mg groups, respectively. Although the study enrolled only subjects whose choroidal neovascularization lesions were assessed by the investigator as having a classic component, in 18 of the 64 subjects (28%) (5 randomized to UC, 4 to 0.3 mg, and 9 to 0.5 mg) the lesions were retrospectively assessed by the reading center as totally occult. Also, the baseline lesion of 1 subject was not gradable because of poor angiogram quality. Among the remaining 45 subjects, the distribution of lesion types was not balanced across treatment arms: predominantly classic choroidal neovascularization in 1 of 11 (9.1%) UC subjects, 10 of 25 (40.0%) subjects in the 0.3-mg group, and 9 of 28 (32.1%) subjects in the 0.5-mg group. The greater representation of minimally classic or occult choroidal neovascularization in the UC group was associated with larger lesion sizes, smaller areas of classic choroidal neovascularization (in DAs and as percentage of total lesion area), and smaller areas of subretinal fluid (SRF) than in the 0.3- and 0.5-mg groups. The angiographic differences between the 0.3- and 0.5-mg groups were generally smaller than those between the UC group and the ranibizumab-treated groups. These comparisons are descriptive only and are not based on any statistical test. 635

Ophthalmology Volume 113, Number 4, April 2006 Characteristic Table 2. Demographic and Baseline Characteristics (n 11) Ranibizumab, 0.3 mg (n 25) Ranibizumab, 0.5 mg (n 28) Age (yrs) [mean (SD)] 74.5 (8.2) 77.8 (7.0) 76.6 (6.5) Female [n (%)] 4 (36.4) 14 (56.0) 15 (53.6) White [n (%)] 11 (100) 23 (92.0) 26 (92.9) Visual acuity in study eye (no. of letters; possible range, 0 85) [mean (SD)] 44.3 (16.2) 47.1 (13.5) 48.8 (15.3) Visual acuity in study eye Median (approximate Snellen equivalent) 20/160 20/125 20/100 20/200 or worse [n (%)] 4 (36.4) 8 (32.0) 4 (14.3) Subject strata at randomization based on investigator evaluation [n (%)] Predominantly classic 4 (36.4) 7 (28.0) 10 (35.7) Minimally classic 3 (27.3) 12 (48.0) 10 (35.7) Prior PDT 4 (36.4) 6 (24.0) 8 (28.6) Prior PDT [n (%)] No 7 (63.6) 17 (68.0) 20 (71.4) Yes 4 (36.4) 8 (32.0) 8 (28.6) Distribution of classic CNV, based on FA central reading center evaluation [n (%)] 0% (purely occult CNV) 5 (45.5) 4 (16.0) 9 (32.1) 0% 50% classic CNV 5 (45.5) 10 (40.0) 10 (35.7) 50% classic CNV 1 (9.1) 10 (40.0) 9 (32.1) Could not grade (poor quality FA) 0 1 (4.0) 0 Mean (SD) angiographic lesion dimensions, based on FA central reading center evaluation N 11 24* 28* Total area of lesion (DAs) 3.55 (2.60) 2.99 (2.04) 3.40 (2.59) Total area of CNV (DAs) 3.25 (2.45) 2.51 (2.00) 3.12 (2.55) Area of classic CNV (DAs) 0.55 (1.12) 1.05 (1.19) 0.85 (1.27) Area of classic CNV (% of total lesion area) 21.7 (28.2) 41.8 (30.9) 31.3 (33.8) Total area of fluorescein leakage from CNV plus intense progressive RPE staining (DAs) 3.30 (2.16) 3.84 (3.03) 3.87 (3.00) Area of serous sensory retinal detachment/subretinal fluid (DAs)* 3.54 (2.69) 4.53 (1.49) 4.63 (2.78) CNV choroidal neovascularization; DA disc area; FA fundus angiogram; PDT photodynamic therapy; RPE retinal pigment epithelium; SD standard deviation. The number of evaluable subjects (n) for each characteristic is the same as the number of randomized subjects as shown in the column headings, except where listed separately. One subject in the usual care group did not have visual acuity data for day 0; this subject s screening assessment was used as baseline. Baseline (screening) angiographic assessments were not available for some subjects either due to missing assessments or because some assessments could not be graded by the central reading center. *For area of serous sensory retinal detachment/subretinal fluid, the numbers of subjects with data were 11, 22, and 26 in the usual care, ranibizumab 0.3 mg, and ranibizumab 0.5 mg groups, respectively. Safety Ranibizumab was generally well tolerated. As previously noted, 52 of 53 ranibizumab-treated subjects in part 1 received all 4 scheduled injections. Table 3 summarizes ocular AEs in the study eye occurring in 5% of the pooled ranibizumab-treated subjects in parts 1 and 2. The most common AE was sterile, painless, reversible inflammation (mild or moderate in most subjects). No instances of hypopyon were reported. Slit-lamp examination during part 1 revealed a 2 to 4 inflammatory response in the aqueous or vitreous in 26% of the pooled ranibizumab-treated subjects and at least a trace inflammatory response in 85% of them. The 2 to 4 inflammatory responses were most severe on the day after injection, and usually resolved without treatment by 14 days after injection. In part 2, 2 to 4 inflammation occurred in 22% of ranibizumabtreated subjects and at least trace inflammation in 88% of them. Inflammation of 2 to 4 occurred after 8% of injections and trace or greater inflammation occurred after 58% of injections in parts 1 and 2 combined. Minor scleral or subconjunctival hemorrhage at the injection site was the second most common AE. Three ranibizumab-treated subjects developed a serious (defined as potentially sight threatening) AE in the study eye. The first was recurrent severe iridocyclitis in a subject in the 0.5-mg group who had previously reacted similarly to intravitreally administered triamcinolone acetonide. Iridocyclitis developed after the second ranibizumab injection, and resolved with topical steroid treatment. The third injection was held, and iridocyclitis recurred after the fourth injection. The subject s VA returned to at least baseline level after each episode. For part 2, this subject crossed over to UC. The second serious AE was infectious endophthalmitis (coagulasenegative staphylococcal infection) with onset 8 days after the fourth injection in a subject in the 0.5-mg group. Ranibizumab was discontinued, and the infection was treated effectively with intravitreal vancomycin, ceftazidime, and oral prednisone. By day 210, VA had recovered to the pre-ae level. As this was the only case of endophthalmitis among 62 ranibizumab-treated eyes and a total of 407 ranibizumab injections, event rates for endophthalmitis were 0.016 per study eye and 0.002 per injection. The third serious AE was ischemic CRVO with onset 2 days after the fifth ranibizumab injection in an 82-year-old diabetic subject in the 0.3-mg group. Ranibizumab was discontinued, and the subject was observed, with VA recovering to the pre-ae level by day 196. Increases in intraocular pressure (IOP) were transient and mostly mild and resolved without treatment. In part 1, 5 of 25 (20.0%) subjects in the 0.3-mg group and 4 of 28 (14.3%) in the 0.5-mg group (overall rate, 17.0%) experienced a clinically significant IOP elevation, defined as an increase from predose IOP of 10 mmhg. In part 2, 3 of 24 (12.5%) subjects in the 0.3-mg group and 4 of 27 (14.8%) subjects in the 0.5-mg group experienced elevated IOP (overall rate, 13.7%). The overall rate of elevated IOP for parts 1 and 2 for both treatments combined 636

Heier et al Ranibizumab (Anti-VEGF Antibody) for Neovascular AMD Table 3. Number (%) of Subjects with Adverse Events That Occurred in 5% of Pooled Ranibizumab-Treated Subjects (i.e., 3 Subjects) during Part 1* and Part 2 Adverse Event Ranibizumab, 0.3 mg Ranibizumab, 0.5 mg Ranibizumab, Pooled Part 1 (as randomized and treated) (n 11) (n 25) (n 28) (n 53) Abnormal sensation in eye 0 (0.0) 5 (20.0) 2 (7.1) 7 (13.2) Conjunctival hemorrhage 0 (0.0) 20 (80.0) 24 (85.7) 44 (83.0) Eye pain 0 (0.0) 6 (24.0) 1 (3.6) 7 (13.2) Increased IOP 0 (0.0) 5 (20.0) 4 (14.3) 9 (17.0) Injection-site hemorrhage 0 (0.0) 4 (16.0) 1 (3.6) 5 (9.4) Injection-site pain 0 (0.0) 2 (8.0) 3 (10.7) 5 (9.4) Iridocyclitis 0 (0.0) 17 (68.0) 22 (78.6) 39 (73.6) Iritis 0 (0.0) 12 (48.0) 17 (60.7) 29 (54.7) Keratoconjunctivitis sicca 0 (0.0) 1 (4.0) 2 (7.1) 3 (5.7) Maculopathy 0 (0.0) 1 (4.0) 2 (7.1) 3 (5.7) Retinal hemorrhage 3 (27.3) 4 (16.0) 9 (32.1) 13 (24.5) Visual acuity reduced 1 (9.1) 3 (12.0) 7 (25.0) 10 (18.9) Visual disturbance 1 (9.1) 3 (12.0) 0 (0.0) 3 (5.7) Vitreous detachment 1 (9.1) 4 (16.0) 2 (7.1) 6 (11.3) Vitreous floaters 0 (0.0) 4 (16.0) 2 (7.1) 6 (11.3) Vitreous hemorrhage 0 (0.0) 2 (8.0) 1 (3.6) 3 (5.7) Vitritis 0 (0.0) 4 (16.0) 9 (32.1) 13 (24.5) Part 2 (by treatment received in part 2) (n 13) (n 24) (n 27) (n 51) Conjunctival hemorrhage 1 (7.7) 21 (87.5) 23 (85.2) 44 (86.3) Corneal edema 0 (0.0) 1 (4.2) 2 (7.4) 3 (5.9) Increased IOP 0 (0.0) 3 (12.5) 4 (14.8) 7 (13.7) Injection-site hemorrhage 0 (0.0) 2 (8.3) 2 (7.4) 4 (7.8) Injection-site erythema 0 (0.0) 1 (4.2) 2 (7.4) 3 (5.9) Injection site pain 0 (0.0) 1 (4.2) 2 (7.4) 3 (5.9) Iridocyclitis 0 (0.0) 15 (62.5) 12 (44.4) 27 (52.9) Iritis 0 (0.0) 14 (58.3) 16 (59.3) 30 (58.8) Keratoconjunctivitis sicca 0 (0.0) 1 (4.2) 2 (7.4) 3 (5.9) Visual acuity reduced 0 (0.0) 4 (16.7) 3 (11.1) 7 (13.7) Vitritis 0 (0.0) 5 (20.8) 5 (18.5) 10 (19.6) IOP intraocular pressure. *Study eye ocular adverse events reported in 2 of 53 subjects (3.8%) in the pooled ranibizumab groups in part 1 were eye irritation, eye pruritus, asteroid hyalosis, injection site irritation, and keratitis. Study eye ocular adverse events reported in 1 of 53 subjects (1.9%) were arcus lipoides, blepharitis, conjunctival hyperemia, corneal edema, endophthalmitis, episcleritis, eye discharge, glare, infective conjunctivitis injection-site reaction, keratitis, lacrimation disorder, macular degeneration, mydriasis, ocular discomfort, pruritus, pupillary disorder, retinal tear, retinoschisis, subretinal fibrosis, vision blurred, and xanthelasma. Ocular adverse events reported in 2 of 51 subjects (3.9%) in the pooled ranibizumab groups in part 2 were blurred vision, eye pain, retinal hemorrhage, vitreous detachment, and vitreous hemorrhage. Ocular adverse events reported in 1 of 51 subjects (2.0%) were blepharitis, nuclear cataract, corneal abrasion, corneal erosion, eye redness, eyelid edema, glaucoma, injection-site reaction, lacrimation disorder, pruritus, iris atrophy, retinal artery embolism, retinal vein occlusion, and visual disturbance. was 22.6%. The greatest increase from baseline IOP in any subject during the study was 25 mmhg. In each case, the elevated IOP was discovered at either 30 or 60 minutes after injection, and returned to normal by the next measurement (i.e., at 60 minutes or 1 3 days later). No patient required anterior chamber paracentesis to treat elevated IOP. Among the 53 ranibizumab-treated eyes in part 1, vitreous hemorrhage occurred in 3 (5.7%) and retinal tear in 1 (1.9%). Among the 51 ranibizumab-treated eyes in part 2, vitreous hemorrhage occurred in 2 (3.9%) and retinal tear in none. In 1 subject, the vitreous hemorrhage was severe, but it occurred in association with CRVO (case described above) and resolved after 208 days. The only nonocular AE judged by an investigator to be a possible effect of ranibizumab was a case of metallic taste in the mouth (dysgeusia, or taste perversion) in a single patient in part 1. Serum antibody assays performed under masked conditions indicated that no subject developed new anti-ranibizumab antibodies. Visual Acuity at 3 Months Visual acuity was assessed at all visits. At day 98, only 2 of 53 ranibizumab-treated subjects (3.8%) had lost 15 letters from baseline, compared with 2 of 11 UC subjects (18.2%) (Table 4, Fig 2 [available at http://aaojournal.org]). Fourteen of 53 ranibizumab-treated subjects (26.4%) gained 15 letters, compared with 0 of 11 UC subjects. Ranibizumab-treated subjects had a mean gain of 9.2 15.3 letters, whereas the UC subjects on average lost 5.1 9.6 letters (P 0.0048 for the 0.3-mg group and P 0.0024 for the 0.5-mg group, vs. UC). Also, whereas an approximate Snellen equivalent of 20/200 or worse was reported for 32.0%, 14.3%, and 36.4% in the 0.3-mg, 0.5-mg, and UC groups, respectively, at baseline, by day 98 the proportions had decreased to 6 of 25 (24.0%) in the 0.3-mg group and 3 of 28 (10.7%) in the 0.5-mg group, but had increased to 5 of 11 (45.5%) in the UC group. This difference was statistically significant (P 0.0276) for the 0.5-mg group versus the UC group, but this may have been partly due to the difference in the proportions at baseline. 637

Table 4. Visual Acuity Changes from Baseline at Day 98, after 3 Months of Study Treatment Outcome Measure Ophthalmology Volume 113, Number 4, April 2006 (n 11) Ranibizumab, 0.3 mg (n 25) Ranibizumab, 0.5 mg (n 28) 0.3 mg vs. P Value 0.5 mg vs. No. of letters, change from baseline N 11 25 28 Mean (SD) 5.1 (9.6) 9.4 (13.3) 9.1 (17.2) 0.0048 0.0024 95% CI 6.4 5.5 6.7 Frequency distribution [n (%)] 15-letter increase 0 8 (32.0) 6 (21.4) 15-letter change (stable) 9 (81.8) 16 (64.0) 21 (75.0) 15-letter decrease 2 (18.2) 1 (4.0) 1 (3.6) Cumulative distribution [n (%)] 10-letter increase 1 (9.1) 11 (44.0) 12 (42.9) 5-letter increase 3 (27.3) 17 (68.0) 18 (64.3) 1-letter increase 3 (27.3) 20 (80.0) 24 (85.7) No. with 15-letter increase from baseline (%) 0 (0) 8 (32.0) 6 (21.4) 0.0757 0.1578 95% CI 0% 28.5% 14.9% 53.5% 8.3% 41.0% No. with 15-letter decrease from baseline (%) 2 (18.2) 1 (4.0) 1 (3.6) 0.2157 0.1866 95% CI 2.3% 51.8% 0.1% 20.4% 0.1% 18.3% Approximate Snellen equivalent of 20/200 or worse n (%) 5 (45.5) 6 (24.0) 3 (10.7) 0.2519 0.0276 95% CI 16.7% 76.6% 9.4% 45.1% 2.3% 28.2% CI confidence interval; SD standard deviation. P values were based on the Wilcoxon rank-sum test for continuous variables and Fisher exact test for binary variables. The 95% CIs for percentages were based on the exact method for binomial proportions. One subject in the usual care group did not have a day 0 assessment; this subject s screening assessment was used as baseline. Seven of 11 subjects in the usual care group received verteporfin in the study eye on day 0. Of the 7, 3 received additional verteporfin therapies after day 0 during part 1 of the study. Visual Acuity at 6 Months There were 7 possible subgroups in part 2 (Table 5). Of 11 subjects randomized to UC in part 1, 9 elected to cross over to ranibizumab treatment for part 2: the 4 who were in the first enrollment cohort received 0.3 mg, and the 5 who were in the second enrollment cohort received 0.5 mg. Of the 25 subjects randomized to 0.3-mg ranibizumab in part 1, 20 elected to continue on this treatment (including the subject who was withdrawn because of CRVO 2 days after the fifth injection), and 5 elected to cross over to UC (no further ranibizumab injections). Of 28 subjects receiving 0.5-mg ranibizumab in part 1, 22 elected to continue this treatment, and 6 elected to cross over to UC (which was observed only up to day 196). Figure 3 summarizes change from baseline in the number of Table 5. Summary of 7 Subgroups in Part 2 of Study Part 1 Part 2 Label No. of Subjects No. Evaluable for VA at 6 Months UC UC UC/UC 2 2 UC 0.3 mg UC/0.3 4 4 UC 0.5 mg UC/0.5 5 5 0.3 UC 0.3/UC 5 3* 0.3 0.3 mg 0.3/0.3 20 19 0.5 UC 0.5/UC 6 5* 0.5 0.5 mg 0.5/0.5 22 21 0.3, 0.5 ranibizumab doses (mg); UC usual care; VA visual acuity. *Two subjects in the 0.3/UC group and 1 subject in the 0.5/UC group completed part 2 but did not have visual acuity assessed at their visit at 6 months (day 196). One subject in the 0.3/0.3 group and 1 subject in the 0.5/0.5 group discontinued the study early because of an adverse event. discernible letters at each assessment through 6 months for the 42 subjects treated with ranibizumab in both part 1 and part 2. The change from baseline up to 3 months for the 11 subjects who received UC in part 1 is presented as reference. Table 6 (available at http://aaojournal.org) presents 6-month VA data both for subjects who received either 0.3 mg or 0.5 mg of ranibizumab in part 1 but crossed over to UC (i.e., received no further ranibizumab injections) in part 2 and for subjects who continued on their ranibizumab dose in part 2. In subjects treated with ranibizumab in both part 1 and part 2 who were evaluable for VA at 6 months (day 210), a 15-letter loss occurred in 5.3% (1/19) in the 0.3/0.3 group, 0% (0/21) in the 0.5/0.5 group, and 2.5% (1/40) in the pooled groups. (One subject in each group did not have a VA evaluation at day 210.) No subject in the UC/0.3 or UC/0.5 group (n 9 overall) had a 15-letter loss at 6 months. Mean changes from baseline VA at 6 months were 12.8 14.7 letters in the 0.3/0.3 group and 15.0 14.2 in the 0.5/0.5 group. Among these same 40 subjects, mean changes from baseline to month 3 were 9.3 13.3 (n 19) in the 0.3/0.3 group and 13.1 14.8 (n 21) in the 0.5/0.5 group. Thus, the VA benefit of ranibizumab seen at 3 months appeared to persist through 6 months. Also, subjects who received UC in part 1 and ranibizumab in part 2 showed improved VA at 6 months: mean gains from baseline of 7.3 13.1 letters (n 4) in the UC/0.3 group and 3.2 9.0 letters (n 5) in the UC/0.5 group. Eleven subjects randomized to ranibizumab in part 1 received UC in part 2, but only 2 of these subjects both originally in the 0.5-mg group had not tolerated ranibizumab during part 1. Both subjects developed severe iridocyclitis and a loss of VA near the end of part 1, and both recovered from the VA loss (one by the day 98 visit, the other during part 2). Five subjects (excluded from Table 6) had missing VA assessment at 6 months. Table 7 (available at http://aaojournal.org) provides the VA data for these 5 subjects at 3 months and at their final VA assessment for part 2. 638

Heier et al Ranibizumab (Anti-VEGF Antibody) for Neovascular AMD Figure 3. Visual acuity at each assessment up to 6 months in subjects who received ranibizumab in both part 1 and part 2, and up to 3 months in subjects who received usual care in part 1: mean ( standard error of the mean) change from baseline in number of letters (study eye). Because treatments in part 2 were elected rather than randomized, it could be suspected that VA differences between treatment groups at 6 months reflect a greater likelihood of subjects who most benefited from ranibizumab in part 1 electing to continue on the study drug and remaining in the study, whereas subjects who had less favorable outcomes on ranibizumab might be more likely to cross over to UC or have missing final assessments. To address this possibility, we performed an analysis on the last available nonmissing VA assessment for all subjects originally randomized to ranibizumab (regardless of their treatment in part 2) or who stayed on UC for the entire study, and the last VA assessment before the first injection for subjects who crossed over to ranibizumab (Table 8). Subjects randomized to ranibizumab in part 1 had mean gains of 11.6 14.0 letters (0.3-mg group) and 11.8 14.1 letters (0.5-mg group) at 6 months, compared with a mean loss of Table 8. Visual Acuity in Study Eye at 6 Months (Day 196 or Day 210) or Last Available Observation, by Treatment Received in Part 1 (i.e., Treatment Groups as Originally Randomized) Outcome Measure (n 11) 0.3 mg (n 25) Ranibizumab 0.5 mg (n 28) Pooled (n 53) No. of letters, change from baseline N 11* 25 28 53 Mean (SD) 5.3 (8.7) 11.6 (14.0) 11.8 (14.1) 11.7 (13.9) Frequency distribution [n (%)] 15-letter increase 0 11 (44.0) 9 (32.1) 20 (37.7) 15-letter change (stable) 8 (72.7) 12 (48.0) 19 (67.9) 31 (58.5) 15-letter decrease 3 (27.3) 2 (8.0) 0 2 (3.8) Cumulative distribution [n (%)] 10-letter increase 0 15 (60.0) 11 (39.3) 26 (49.1) 5-letter increase 2 (18.2) 19 (76.0) 19 (67.9) 38 (71.7) 1-letter increase (any increase) 2 (18.2) 19 (76.0) 23 (82.1) 42 (79.2) SD standard deviation. Treatment group represents treatment received in part 1, the original randomized treatment. *Includes day 112 measurements for 9 subjects who switched to ranibizumab during part 2 and day 196 measurements for subjects who continued on usual care during part 2. One subject discontinued from the study early due to a serious adverse event of central retinal vein occlusion that occurred after the fifth injection; the subject did not have a scheduled visit at day 210, but had an unscheduled visit close to the 6-mo time point. Two subjects who were on usual care in part 2 had a visit at day 196 but did not have visual acuity assessment; the day 98 assessment is used here for those subjects. One subject discontinued from the study early due to a serious adverse event of cerebrovascular accident, which occurred 14 days after the last (eighth) injection; the subject could not come back for the day 210 visit but had the last visit at day 199, 3 days after the last injection. One subject who received usual care in part 2 had a visit at day 196 but did not have visual acuity assessment; the day 98 assessment is used here for this subject. 639

Ophthalmology Volume 113, Number 4, April 2006 5.3 8.7 letters at the end of UC treatment. Thus, it did not seem that selection bias contributed importantly to the maintenance of improvement in VA at month 6 in subjects treated with ranibizumab in both parts 1 and 2. Visual Acuity Outcomes by Subgroup Because in many cases the retrospective reading center assessments of baseline lesion morphology were discordant with the investigators screening assessments, post hoc analyses of VA were performed based on the reading center s assessments. The percentage of the baseline choroidal neovascularization lesion deemed classic was used to group subjects into reading center predominantly classic and reading center minimally classic/occult subgroups; Tables 9 and 10 (available at http://aaojournal.org) show changes in VA at 3 months and 6 months, respectively, in these subgroups. Although the numbers of subjects were small, making interpretation difficult, the data suggest that at 3 months VA was improved in both the minimally classic/occult and predominantly classic subgroups. At 6 months, the improvement in the subgroups was still present in subjects who received ranibizumab in both parts 1 and 2. Angiographic Results Fluorescein angiograms and color photographs obtained on day 0 (baseline) and days 14, 28, 56, and 84 were evaluated by 2 masked readers at the central reading center. At day 84 (Table 11), growth of choroidal neovascularization (by at least 0.3 disc diameters from baseline) was seen in 2 of 25 (8.0%) and 8 of 27 (29.6%) subjects in the 0.3-mg and 0.5-mg groups, respectively, compared with 4 of 10 (40.0%) in the UC group (P 0.0428 for the 0.3-mg group, vs. UC). Mean changes from baseline in total area of choroidal neovascularization were 0.14 1.48 DAs and 0.16 2.26 DAs in the 0.3- and 0.5-mg groups, respectively, and an increase of 0.20 1.29 DAs in the UC group (differences vs. UC group not statistically significant). The mean total lesion area decreased in both ranibizumab-treated groups (changes, 0.07 1.31 and 0.23 2.15 DAs in the 0.3- and 0.5-mg groups, respectively) but increased by 0.28 1.22 DAs in the UC group (differences vs. UC group not statistically significant). Large mean decreases from baseline in total area of fluorescein leakage from choroidal neovascularization (plus intense progressive staining of the retinal pigment epithelium) occurred in the 0.3- and 0.5-mg ranibizumab groups (mean change, 2.39 2.48 and 1.57 2.53 DAs, respectively), compared with a small increase in the UC group (0.03 1.72 DAs). The difference between the UC and 0.3-mg groups was statistically significant (P 0.0028), as was the overall difference between the UC and pooled ranibizumab groups (P 0.0098). The mean area of SRF decreased in the 0.3-mg group ( 2.57 2.15 DAs) and 0.5-mg group ( 0.83 2.37 DAs) but increased in the UC group (0.41 2.47 DAs) (P 0.0022 for the 0.3-mg group, vs. UC). (Note: in this open-label study, no adjustment was made for multiple comparisons.) Discussion This phase I/II clinical trial evaluated the safety profile of repeated intravitreal injections of ranibizumab a Fab antibody fragment that binds and inactivates all active forms of VEGF-A in subjects with neovascular AMD. In general, repeated monthly injections for up to 6 months were very well tolerated. The most common AE was a transient, painless, reversible inflammatory response that was most severe on the day after injection, did not increase in severity Outcome Measure Table 11. Fluorescein Angiography and Fundus Photography Results at Day 84 (n 11) Ranibizumab, 0.3 mg (n 25) Ranibizumab, 0.5 mg (n 28) 0.3 mg vs. P Value 0.5 mg vs. Growth of choroidal neovascularization ( 0.3-DD increase from baseline) N 10 25 27 n (%) 4 (40.0) 2 (8.0) 8 (29.6) 0.0428 0.6964 95% CI 12.2% 73.8% 1.0% 26.0% 13.8% 50.2% Total area of choroidal neovascularization, change from baseline (DAs) N 10 24 27 Mean (SD) 0.20 (1.29) 0.14 (1.48) 0.16 (2.26) 0.6230 0.3557 Total area of lesion, change from baseline (DAs) N 10 24 27 Mean (SD) 0.28 (1.22) 0.07 (1.31) 0.23 (2.15) 0.5326 0.3468 Total area of fluorescein leakage from choroidal neovascularization plus intense progressive RPE staining, change from baseline (DAs) N 10 24 27 Mean (SD) 0.03 (1.72) 2.39 (2.48) 1.57 (2.53) 0.0028 0.0699 Area of serous sensory retinal detachment/subretinal fluid, change from baseline (DAs) N 10 21 24 Mean (SD) 0.41 (2.47) 2.57 (2.15) 0.83 (2.37) 0.0022 0.1976 CI confidence interval; DA disc area; DD disc diameter (approximated by square root of DA); RPE retinal pigment epithelium; SD standard deviation. P values were based on the Wilcoxon rank-sum test for continuous variables and Fisher exact test for binary variables. Ninety-five percent CIs for percentages were based on exact method for binomial proportions. 640

Heier et al Ranibizumab (Anti-VEGF Antibody) for Neovascular AMD with repeated injections, and usually resolved without treatment within 14 days. Similar inflammatory responses occurred with intravitreal ranibizumab administration in animals (Genentech, unpublished data), but the etiology remains uncertain. Some component of the intravitreal injection procedure alone may account for the response. 12 The vehicle may be another contributory factor, but vehicle injections in animal models did not result in a cellular response (Genentech, unpublished data). The possibility that some of these inflammations are immunoreactions to ranibizumab has not yet been ruled out. Subconjunctival hemorrhage after subconjunctival injection of anesthetic and intravitreal injection of ranibizumab was another common AE, and was likely the result of these procedures rather than ranibizumab. It was not possible to distinguish between ranibizumab-related events and those related to the intravitreal injection procedure in this study, because performing multiple intravitreal injections of a placebo is ethically unacceptable. However, the rates of AEs in this 6-month open-label trial suggest an acceptable safety profile for ranibizumab injections. Previous experience with multiple intravitreal injections of protein therapeutic agents has shown that repeated injections can be safely tolerated and are justifiable based on risk/benefit considerations. 6,13 Endophthalmitis is a risk inherent to administration of any agent by intravitreal injection. Although an aseptic technique was used throughout this study, 1 study eye of 62 treated (2%) did develop endophthalmitis with coagulasenegative staphylococcal infection after the fourth ranibizumab injection. Previous controlled clinical trials, cited by Jager et al, 14 are consistent with an endophthalmitis rate of 1 to 2 cases per 1000 injections. This does not differ substantially from the rate of 1 case in 407 injections (0.0025 per injection) in this study, or the rate of 12 cases in 7545 injections (0.0016 per injection) in the pivotal studies of pegaptanib. 6 More extensive use of prophylactic antibiotics and povidone iodine, as well as greater emphasis on aseptic technique (including use of a lid speculum), was designed into the protocols of subsequent trials of ranibizumab in an effort to decrease the chance of further cases of endophthalmitis. Although the case of endophthalmitis was considered to be most likely related to intravitreal injection rather than to the study drug, 2 of 62 (3%) treated eyes developed sightthreatening AEs that may have been related to ranibizumab. These events were severe recurrent iridocyclitis and, in an 82-year-old diabetic woman, CRVO. The iridocyclitis resolved, but the CRVO event was still ongoing at the end of follow-up. In both subjects, VA returned to the premorbid level. The possibility that ranibizumab may stimulate an immune antibody response is a safety concern that continues to be evaluated. Because ranibizumab is a humanized antibody fragment, it is expected to be less antigenic than full-length monoclonal antibodies, chimeric antibodies, or other foreign large molecules. None of the 53 subjects in this study developed new antibodies to ranibizumab during the initial 98-day study period or in the subsequent variable follow-up (up to 210 days), and none of the 9 subjects who crossed over from UC to ranibizumab developed new antibodies to ranibizumab during treatment. This lack of antibody generation may be due to humanization of the molecule, low systemic exposure levels, or both. Improvement in VA occurred rapidly after the first ranibizumab injection in this study. By day 14, most of the change from baseline seen at day 98 was already evident in the 0.5-mg group, and half was already evident in the 0.3-mg group. Visual acuity continued to improve gradually between days 14 and 98, and this benefit persisted between days 98 and 210. The more gradual improvement after day 14 may reflect continued inhibition of VEGF-associated vascular permeability, blockade of VEGF angiogenesis effects, or both. The change in mean VA occurred in both ranibizumab dose groups, with no clear dose response relationship. It is also encouraging that, although subjects randomized to UC demonstrated a decline in overall VA during part 1 of the study, those who crossed over to ranibizumab treatment in part 2 gained an average of 5 letters from baseline during the second 3-month period. Further studies are needed to determine if disease duration has any impact on the effectiveness of ranibizumab treatment. In addition to improving VA, ranibizumab treatment resulted in decreases in the areas of leakage from choroidal neovascularization and serous sensory retinal detachment (SRF), which were statistically significant in the 0.3-mg ranibizumab group. Larger studies with longer durations of treatment are addressing the potential effects of ranibizumab on this and other lesion characteristics, such as the total area of choroidal neovascularization or the total area of the lesion, and also investigating whether any adverse morphologic events occur with repeated intravitreal administration of ranibizumab. The small number of subjects and relatively short follow-up period in this phase I/II study limit extrapolation of the results. Furthermore, although a randomized UC group was included, the absence of a masked sham injection group for comparison is a major limitation. These comparisons are being addressed in the ongoing phase III trials. Other limitations include the open-label design, absence of a control group during part 2, and the larger lesions in the subjects randomized to UC in part 1. Also, eligibility for the study and stratification of the treatment groups were based on investigators screening assessments of lesion characteristics, which were in some cases discordant with subsequent findings by the reading center. Nonetheless, the results of this study provide encouraging preliminary support for the use of anti- VEGF therapy for treatment of neovascular AMD. References 1. Ferrara N, Davis-Smyth T. The biology of vascular endothelial growth factor. Endocr Rev 1997;18:4 25. 2. Kvanta A, Algvere PV, Berglin L, Seregard S. Subfoveal fibrovascular membranes in age-related macular degeneration express vascular endothelial growth factor. Invest Ophthalmol Vis Sci 1996;37:1929 34. 3. Lopez PF, Sippy BD, Lambert HM, et al. Transdifferentiated retinal pigment epithelial cells are immunoreactive for vascular endothelial growth factor in surgically excised age-related mac- 641

Ophthalmology Volume 113, Number 4, April 2006 ular degeneration-related choroidal neovascular membranes. Invest Ophthalmol Vis Sci 1996;37:855 68. 4. Aiello LP, Avery RL, Arrigg PG, et al. Vascular endothelial growth factor in ocular fluid of patients with diabetic retinopathy and other retinal disorders. N Engl J Med 1994;331:1480 7. 5. Aiello LP, Pierce EA, Foley ED, et al. Suppression of retinal neovascularization in vivo by inhibition of vascular endothelial growth factor (VEGF) using soluble VEGF-receptor chimeric proteins. Proc Natl Acad Sci U S A 1995;92: 10457 61. 6. Gragoudas ES, Adamis AP, Cunningham ET Jr, et al, VEGF Inhibition Study in Ocular Neovascularization Clinical Trial Group. Pegaptanib for neovascular age-related macular degeneration. N Engl J Med 2004;351:2805 16. 7. Presta LG, Chen H, O Connor SJ, et al. Humanization of an anti-vascular endothelial growth factor monoclonal antibody for the therapy of solid tumors and other disorders. Cancer Res 1997;57:4593 9. 8. Muller YA, Chen Y, Christinger HW, et al. VEGF and the Fab fragment of a humanized neutralizing antibody: crystal structure of the complex at 2.4 A resolution and mutational analysis of the interface. Structure 1998;6:1153 67. 9. Chen Y, Wiesmann C, Fuh G, et al. Selection and analysis of an optimized anti-vegf antibody: crystal structure of an affinity-matured Fab in complex with antigen. J Mol Biol 1999;293:865 81. 10. Mordenti J, Cuthbertson RA, Ferrara N, et al. Comparisons of the intraocular tissue distribution, pharmacokinetics, and safety of 125I-labeled full-length and Fab antibodies in rhesus monkeys following intravitreal administration. Toxicol Pathol 1999;27:536 44. 11. Hogan MJ, Kimura SJ, Thygeson P. Signs and symptoms of uveitis. I. Anterior uveitis. Am J Ophthalmol 1959;47:155 70. 12. Perry CM, Balfour JA. Fomivirsen. Drugs 1999;57:375 80. 13. Au Eong KG, Beatty S, Charles SJ. Cytomegalovirus retinitis in patients with acquired immune deficiency syndrome. Postgrad Med J 1999;75:585 90. 14. Jager RD, Aiello LP, Patel SC, Cunningham ET Jr. Risks of intravitreous injection: a comprehensive review. Retina 2004; 24:676 98. 642