Prompt PRP vs. Ranibizumab + Deferred PRP for PDR Study Protocol S Prompt PRP vs. Ranibizumab + Deferred PRP for PDR Study DRCR.net

Background DRCR.net Protocol S (PRP vs. Ranibizumab for PDR) Primary outcome: mean change in VA from baseline to 2-years with ranibizumab no worse than (non-inferior to) PRP Secondary outcomes: Ranibizumab: superior mean VA over course of 2 years (AUC) superior mean visual field outcomes at 2 years (although not necessarily perceived by study participant per NEI VFQ-25 questionnaire) 3-fold decrease in occurrence of vitrectomies decreased development of central-involved DME with vision impairment over 2 years among eyes without this at baseline PRP rarely given for failure or futility of ranibizumab to control PDR more visits and injections only cost effective (within range frequently cited as cost-effective in the United States) when DME present and causing vision impairment

Randomized, multi-center clinical trial (55 Sites) Study Design Randomized, multi-center clinical trial (55 Sites) Participants meeting all of the following criteria: Age ≥18 years with Type 1 or type 2 diabetes Study eye(s) meeting all of the following criteria (a participant can have 2 study eyes): PDR No history of PRP Best corrected visual acuity letter score ≥24 (~Snellen equivalent 20/320 or better) Eyes with or without central-involved DME were eligible Primary Objective: Compare the efficacy and safety of PRP with that of intravitreous ranibizumab (0.5-mg in 0.05 mL) for proliferative diabetic retinopathy (PDR)

Follow-up Schedule PRP group: Visits every 16 weeks* Ranibizumab group: Visits every 4 weeks to assess for PDR treatment Both groups simultaneously evaluated for DME treatment Baseline to 1 Year PRP group: Visits every 16 weeks* Ranibizumab group: Visits every 4wk to 16wk to assess for PDR treatment Interval is extended if injections for PDR and DME deferred (“Defer and Extend”) 1 Year to 5 Years *Eyes with DME could be seen more frequently for DME treatment as needed.

Visit Completion 394 Eyes Randomized (305 Participants) Baseline Ranibizumab Group N = 191 PRP Group N = 203 Baseline 2-Years Excluding Deaths 88% 86% 5-Years Excluding Deaths 69% 65% 5-Years Overall 61% 61%

Ocular Baseline Characteristics at Enrollment Ranibizumab Group (N = 191) PRP Group (N = 203) Randomization VA letter score, mean 75 ~Snellen Equivalent, mean (20/32) Randomization OCT CST*, mean 262 249 < 250 µm, % 66% 67% 250 to 349 µm, % 19% 26% ≥ 350 µm, % 15% 7% Presence of CI-DME with VA loss**, % 22% 23% Required ranibizumab at baseline *Stratus equivalent; 2 eyes in each group were excluded because OCT CST measurements were missing. **Eyes with visual acuity letter score ≤ 78 (20/32 or worse) AND OCT CST ≥ machine and gender specific thresholds

Ocular Baseline Characteristics at Enrollment All Eyes Had PDR Per Investigator Determination Prior to Randomization Ranibizumab Group (N = 191) PRP (N = 203) DR Severity Confirmation by Reading Center* NPDR 10% 13% Mild to moderate PDR 52% 49% High risk PDR to advanced PDR 38% 37% *Diabetic retinopathy level data were missing for 2 in the ranibizumab group and 4 in the PRP group.

Ocular Baseline Characteristics by 5-Year Completion Status Did Not Complete Completed Ranibizumab Group (N = 74) PRP Group (N = 80) (N = 117) PRP Group (N = 123) VA Letter Score, Median (25th, 75th percentile) 75 (67, 81) 74 (65, 81) 79 (75, 85) 80 (72, 85) ~ Snellen Equivalent, Median 20/32 20/25 OCT CSF*, 227 (197, 295) 234 (197, 268) 220 (196, 260) 229 (203, 263) *Stratus equivalent; 1 eye in each group was excluded because OCT CST measurements were missing.

Ocular Baseline Characteristics by 5-Year Completion Status All Eyes Had PDR Per Investigator Determination Prior to Randomization Did Not Complete Completed Ranibizumab Group (N = 74) PRP Group (N = 80) (N = 117) PRP Group (N = 123) DR Severity (ETDRS) Confirmation by Reading Center, %  Severe NPDR or less (≤ level 53) 4% 11% 14% Mild to moderate PDR (level 61 or 65) 49% 51% 54% 48% High risk PDR (level 71 or 75) 46% 38% 31% 36% Advanced PDR (level 81 or 85) 1% 2% <1% * Diabetic retinopathy level data were missing for 2 non-completers in the Ranibizumab group, and 4 completers in the PRP group.

Results

Treatment

Mean Number of Injections 5-Year Completers Only Ranibizumab Group (N = 117) PRP Group (N = 123) Year 1  7.1 2.3  Year 2 3.3 1.1 Year 3 3.0 0.9 Year 4 2.9 0.6 Year 5 0.4 Cumulative Through 5 Years 19.2 5.4

Number of Intravitreous Injections by Year Ranibizumab Group, 5-Year Completers Only

Panretinal Photocoagulation at Follow-up Ranibizumab Group (N = 191) PRP Group (N = 203) Received PRP During Follow-up, N (%) 26 (14%) 103 (51%) Received PRP prior to 104 Weeks, N 13 92 Received PRP after 104 Weeks, N 15 31 Received PRP during Vitrectomy, N 18 29 Received PRP outside of Vitrectomy, N 10 89

Visual Acuity

Visual Acuity at 5-Years Ranibizumab (N = 117) PRP (N = 123) VA Letter Score and Approximate Snellen Equivalent Mean 80 81 ~Snellen Equivalent, Mean 20/25 Median (25th, 75th percentile) 84 (89, 78) (89, 77) ~Snellen Equivalent, Median 20/20 (20/16, 20/32)

Mean Changes in VA From Baseline Over Time - Overall Cohort 5-Year Adjusted Mean Difference: +0.6 letters 95% Confidence Interval: (-2.3, +3.5), P = .68 +3.1 +3.0 N = 191 N = 117 N = 203 N = 123

Mean Changes in VA From Baseline Over Time for 5-Year Completers Only 5-Year AUC Difference: +1.6 letters 95% Confidence Interval: (0, 3.2), P = .05 +3.3 +1.5 N = 117 of 191 N = 123 of 203 Outlying values were truncated to 3 SD from the mean

Change in VA Letter Score at 5-Years Ranibizumab (N = 117) PRP (N = 123) Adjusted Difference (95% CI) ≥ 15 letters improvement, % 26% 23% 1% (-12%, 15%) ≥ 10 letters improvement, % 52% 41% 6% (-10%, 21%) ≥ 10 letters worsening, % 9% -3% (-11%, 5%) ≥ 15 letters worsening, % -1% (-7%, 5%)

VA at 5-Years Stratified by Baseline DME with Decreased VA Ranibizumab PRP Adjusted Difference (95% CI) P -Value With Vision-Impairing DME at Baseline Eyes, N 20 24 Mean Change from Baseline 2.5 4.6 -4.1 (-20.6, 12.3) .48 Without Vision-Impairing DME at Baseline 96 98 3.2 2.4 0.03 (-2.7, 2.7) .98

Mean Change in Cumulative Visual Field Total Point Score (30-2 + 60-4) - Overall Cohort -330 -527 5-Year Adjusted Mean Difference: 208 dB 95% Confidence Interval (-9, 408), P = .04 N = 81 N = 41 N = 86 N = 38 Outlying values were truncated to 3 SD from the mean

Median Change in Cumulative Visual Field Total Point Score (30-2 + 60-4) - Overall Cohort -244 -480 N = 81 N = 41 N = 86 N = 38 Outlying values were truncated to 3 SD from the mean

Mean Change in Cumulative Visual Field Point Score (30-2) - Overall Cohort -157 -190 5-Year Adjusted Mean Difference: 29 dB 95% Confidence Interval (-91, 148), P = .64 N = 84 N = 44 N = 88 N = 41 Outlying values were truncated to 3 SD from the mean

Mean Change in Cumulative Visual Field Point Score (60-4) - Overall Cohort -199 -348 5-Year Adjusted Mean Difference: 140 dB 95% Confidence Interval (16, 265), P = .03 N = 82 N = 43 N = 87 N = 41 Outlying values were truncated to 3 SD from the mean

Mean Change in Cumulative Visual Field Point Score for 5-Year Completers in Ranibizumab Group without PRP Treatment -96 -119 Outlying values were truncated to 3 SD from the mean

Median Change in Cumulative Visual Field Point Score for 5-Year Completers in Ranibizumab Group without PRP Treatment -87 -87 Outlying values were truncated to 3 SD from the mean

Development of Vision- Impairing DME: Central Subfield Thickening with Visual Acuity ~20/32 or Worse

Kaplan-Meier Estimates Development of Vision-Impairing DME Among Eyes Without Vision-Impairing DME at Baseline Hazard Ratio = 0.4 95% CI: 0.3 to 0.7 P < .001 Kaplan-Meier Estimates Years 2 3 4 5 Ranibizumab (N = 147) 11% 14% 17% 22% Prompt PRP (N = 155) 29% 35% 38% 38% 22% 147 125 109 98 85 37 155 108 95 80 72 41

Diabetic Retinopathy

DR on Fundus Photographs at 5 Years* Ranibizumab (N = 90) PRP (N = 93) Without PDR (≤ level 60), % 43% 37% With Regressed NV (level 61A), % 28% 33% With Active NV (≥ level 61B), % 29% 30% Improved from PDR (≥ level 61) to NPDR (≤level 53)**, % N/A Without DR (≤ level 20)*, % 10% Improved ≥2 steps in DR severity on fundus photos at 5 years**, % 46% *Observed data only, only include eyes with active NV at baseline as graded by reading center. **Not applicable or cannot determine for PRP group.

DR Adverse Events: Over 5 Years Ranibizumab (N = 117) PRP (N = 123) Adjusted Difference (95% CI) Any Retinal detachment, % 6% 15% -9% (-14%, -4%) Retinal Detachment involving Center of the Macula, % 1% 4% -3% (-7%, 0%) Neovascular Glaucoma, % 3% -2% (-6%, 2%) Neovascularization of the Iris, % (-1%, 3%) Vitreous Hemorrhage, % 48% 46% 2% (-6%, 11%) Vitrectomy, % 11% 19% -7% (-14%, -1%)

Development of Retinal Detachment Hazard Ratio = 0.4 95% CI: 0.2 to 0.8 P = .004 18% 7% 191 167 148 134 121 53 203 175 152 132 17 59

Development of First Vitreous Hemorrhage after Baseline Difference = 4% 95% CI: -7% to 16% P = .47 58% 54% 191 150 112 89 69 29 203 147 110 91 73 37

Safety

Ocular Adverse Events Ranibizumab (N = 191) PRP (N = 203) P-Value Number of Injections 3132 981 Endophthalmitis, % <1% Inflammation, % 2% 5% .05 Retinal Tear, % Cataract Surgery, % 16% 19% .62 Elevated IOP, % 18% .58 Event defined as occurring at least once through 5 years.

Systemic Adverse Events 2 Study Eyes (N = 89) Ranibizumab (N = 102) PRP (N = 114) P-Value Nonfatal Myocardial Infarction 3% 6% 4% .64 Nonfatal Stroke .65 Death due to potential vascular cause or unknown cause 7% 2% .13 Any APTC Event* 13% 18% 11% .31 Event defined as occurring at least once through 5 years. *Anti-platelet Trialists’ Collaboration Arterial Thromboembolic Events.

Time to First APTC Event* Kaplan-Meier Estimates Year 2 3 4 5 2-Study Eyes (N = 89) 9% 10% 13% 17% Ranibizumab (N = 102) 18% 21% Prompt PRP (N = 114) 8% *Anti-platelet Trialists’ Collaboration Arterial Thromboembolic Events. 89 78 68 61 55 27 102 91 80 71 60 24 114 99 84 74 37

Systemic Adverse Events 2 Study Eyes (N = 89) Ranibizumab (N = 102) PRP (N = 114) P-Value Death from Any Cause 9% 13% 6% .24 Hospitalization 61% 65% 54% SAE 63% 67% 55% .21 Hypertension 31% 37% 25% .13 Event defined as occurring at least once through 5 years.

Summary: 5-Year Results Visits, Injections, PRP Sessions 66% of participants (excluding death) completing 5-year visit (117 eyes in ranibizumab and 123 eyes in PRP group) Median number of visits* Ranibizumab = 43 PRP = 21 Mean number of injections Ranibizumab = 19 PRP = 5 Mean number of PRP sessions Ranibizumab = 0 PRP = 2 * Counting participants with one study eye only

Summary: 5-Year Results Central and Peripheral Vision Mean change in VA letter score Ranibizumab = 3.1±14.3 letters PRP = 3.0±10.5 letters Difference = 0.6 (95% CI: -2.3 to 3.5) Mean change in cumulative visual field total point score (HFA 30-2 + 60-4) Ranibizumab = -330±645 dB (N = 41) PRP = -527±635 dB (N = 38) Difference = 208 (95% CI: 9 to 408)

Summary: 5-Year Results Complications Development of Vision-Impairing DME (~Snellen equivalent 20/32 or worse) Ranibizumab = 22% PRP = 38% Hazard Ratio = 0.4 (95% CI: 0.3 to 0.7) Underwent Vitrectomy Ranibizumab = 15% PRP = 22% Hazard Ratio = 0.5 (95% CI: 0.3 to 0.8) No statistically significant differences between groups in major SAE rates were identified.

Discussion VA in both treatment groups was good (mean ~20/25) and patient reported vision-related quality of life scores were similar between the groups over 5 years, even though more eyes in the PRP group developed vision-impairing DME. Overall, for participants completing the 5-year visit, few eyes in either group developed neovascular glaucoma or iris neovascularization and the rates were similar between groups. The overall rate of retinal detachment in the PRP group was higher (15% vs 6%). Most of these did not progress to involve the center of the macula and did not undergo vitrectomy, suggesting the detachments were not vision-threatening.

Discussion Few eyes in either group (6%) had substantial visual loss (≥15 letters from baseline). On average the PRP group on average had more peripheral visual field loss through 2 years than ranibizumab group; however, visual field loss, on average, progressed in both the PRP and ranibizumab groups from years 2 through 5 and the difference between groups diminished in this small sub-study. Further analysis of this finding seems warranted. Almost half of all eyes developed vitreous hemorrhage with 22% vs 41% needing vitrectomy in the ranibizumab and PRP groups respectively. The decision to perform vitrectomy was made by an unmasked investigator.

Discussion Despite the reduction in the mean number of injections for participants in the ranibizumab group, mean VA change from baseline remained fairly constant between 2 and 5 years. While there have been inconsistent reports of increased APTC events for patients receiving anti-VEGF, this full 5- year dataset does not show any definitive differences. No ocular safety concerns were identified.

Discussion Major Study Limitations Retention for this study was less than 65% for each group. No differences were identified between baseline factors and treatment groups for the completers vs non-completers. Eyes in the ranibizumab group that did not complete follow-up had more improvement in VA and OCT CST at their last measured follow-up visit. The outcomes for participants who did not complete the study are unknown.

Conclusion Loss to follow-up was relatively high. Visual acuity in most study eyes that completed follow-up was very good. Severe vision loss or serious PDR complications were uncommon in either group. Ranibizumab group had lower rates of developing vision-impairing DME and less visual field loss. There were no meaningful treatment group differences in patient-centered outcomes. These findings support either ranibizumab or PRP as viable treatments for PDR. Patient-specific factors, including anticipated visit compliance, number of visits, cost, and systemic health should be considered when choosing a treatment for PDR.