Diabetic Macular Edema (DME): High Unmet Medical Need Significant Global Burden Due to Vision Loss

15th Angiogenesis meeting 2018, Miami

Roche virtual pipeline event Tuesday, 13 February 2018 Agenda

Welcome Karl Mahler, Head of Investor Relations

Roche in ophthalmology - overview Jason S. Ehrlich, M.D. Ph.D., Global Head, Clinical Ophthalmology – Product Development

BOULEVARD: Phase 2 results of the bispecific VEGF/Ang2 antibody in diabetic macula edema Sascha Fauser, M.D., Head pRED Ophthalmology

Q&A Karl Mahler, Head of Investor Relations

Introduction

Karl Mahler Head of Investor Relations 2018: Roche off to a good start Record number of pipeline assets at pivotal stage in multiple disease areas

Hemlibra polatuzumab vedotin Venclexta idasanutlin Alecensa Alecensa ipatasertib Venclexta Tecentriq Tecentriq taselisib Alecensa Hemlibra Hemlibra VEGF-ANG2 biMAb

Tecentriq idasanutlin idasanutlin SMN2 splicer Cotellic taselisib taselisib V1a receptor ant. - autism

lampalizumab lampalizumab lampalizumab anti-myostatin adnectin NMEs gantenerumab gantenerumab satralizumab (IL-6R MAb) gantenerumab Ocrevus Ocrevus gantenerumab crenezumab satralizumab satralizumab crenezumab Ocrevus lebrikizumab lebrikizumab Ocrevus CapEndo etrolizumab etrolizumab etrolizumab etrolizumab

2014 2015 2016 2017

2 5 5 4 1 27 1 1

21 33 31 line extensions line Oncology Ophthalmology Neuroscience Immunology 4 Diabetic macular edema (DME): High unmet medical need Significant global burden due to vision loss

• Rate of T2D continues to grow; 40-45% will develop diabetic retinopathy (DR)1

• DME – is the most common diabetic eye disease and the leading cause of irreversible blindness in working age Americans1

Increased risk of DME with longer duration of diabetes and increase HbA1c levels3 PDR

Proliferative Diabetic Retinopathy

Severe NPDR

DME

1c Occurrence (~10% of DR will

HbA Moderate NPDR 1

Stages of DR of Stages develop DME )

Mild NPDR Non-proliferative diabetic retinopathy Duration of Diabetes

1 National Eye Institute. Facts About Diabetic Eye Disease. Available at: https://nei.nih.gov/health/diabetic/retinopathy. 2 NIH National Eye institute-https://nei.nih.gov/health/macular-edema/fact_sheet. 5 3 Varma et al. Prevalence of and Risk Factors for Diabetic Macular Edema in the United States:JAMA Ophthalmol. 2014 Nov; 132(11): 1334–1340. Roche in ophthalmology - overview

Jason S. Ehrlich, M.D., Ph.D. Global Head, Clinical Ophthalmology - Product Development Roche in ophthalmology

Taking on the leading causes of vision loss through pioneering therapies

Unmet Need Breakthrough Science Patient-Focused Innovation People and Partnerships

Innovative treatment We are pioneers striving for Through our smart We hire the best people to options are needed to transformative therapies innovations in molecular advance science and improve address the leading causes that leverage cutting edge engineering, biomarkers people’s lives. We foster of vision loss affecting 253 science for diseases with and sustained drug delivery, extensive collaborative million people worldwide1 high unmet medical need we strive to design the partnerships with researchers, right therapies for the clinicians and patient groups right patients

1 WHO Fact Sheet, October 2017. Accessed 6 Feb 2018, http://www.who.int/mediacentre/factsheets/fs282/en/ 7

Ophthalmology pipeline

Indication Phase I Phase II Phase III In Submission Approved

ranibizumab Port Delivery System Neovascular AMD Lucentis 0.5 mg PFS anti-VEGF/ANG2 biMAb (RG7716)

anti-VEGF/ANG2 Lucentis 0.3 mg Lucentis 0.3 mg Diabetic Eye Disease biMAb (RG7716) PFS (DME, DR) (DME, DR)

Lucentis 0.5 mg Retinal Vein Occlusion PFS

NME Geographic Atrophy (RG6417)

Lucentis 0.5 mg Myopic CNV PFS

Actemra/RoActemra Giant Cell Arteritis

Multiple Sclerosis Ocrevus

anti-IL6R Neuromyelitis Optica (RG6168/SA237)

Glaucoma NME (RG7945)

Status as of February 1, 2018. Lucentis in collaboration with Forsight and Novartis 8 DME=diabetic macular edema; DR=diabetic retinopathy; PFS=prefilled syringe; biMAb=bispecific monoclonal antibody; NME=new molecular entity Retinal vascular diseases remain leading causes of vision loss

• Leading causes of vision loss in… – working-age people in US, Europe: Diabetic eye disease (DME, PDR) – elderly people in US, Europe: Neovascular AMD

2017 Ophthalmology market Neovascular AMD & DME market outlook

AMD DME 8% nAMD, DME, RVO 5.7 5% USD bn 5.3 Glaucoma 4.2 US 16% Dry Eye 53% 2.8 Ocular inflammation/infections 18% Others EU5 Japan 2016 2026 2015 2025

Market size: CHF 18.6bn Wet AMD & DME: High growth disease areas

Source: Evaluate Pharma (January 2018) Source: Decision Resources (January 2018)

Note: Figures for 2017 involve forecast values for Q4 2017 as investor reports of all companies are not yet released 9 DME=diabetic macular edema; PDR=proliferative diabetic retinopathy; AMD=age-related macular degeneration Vision can be recovered in neovascular AMD and DME Anti-VEGF therapy can improve outcomes and has become SOC

DME phase III studies: Aflibercept vs laser; DME phase III studies: Lucentis vs sham;

mean change from baseline in VA during 52-wks1 mean change from baseline in VA during 24-months2

* VISTA VIVID

*with censoring of values after additional treatment was given (LOCF)

1 Ophthalmology 2014; 121: 2247-2254; 2 RIDE/RISE: Ophthalmology 2011;118:1594-1602 10 AMD=age-related macular degeneration; DME=diabetic macular edema; VA=visual acuity Still, many patients remain visually impaired Improved efficacy is a major unmet need in retinal vascular disease

RIDE and RISE phase III trials – DME1

Sham injection 0.3 mg ranibizumab 0.5 mg ranibizumab

(n=130; 127) (n=125; 125) (n=127; 125)

or better (%) better or

Snellen equivalent of 20/40 of equivalent Snellen

24 months 24 months

Ranibizumab versus sham-injection: Change from baseline in percentage of patients with a Snellen equivalent of 20/40 or better at 24 months

Despite significant improvements, many people do not achieve a visual acuity score of 20/40 or better

1 RIDE/RISE: Ophthalmology 2011;118:1594-1602; DME=diabetic macular edema 11 Snellen (eye chart to measure visual acuity) equivalent of 20/40: a person needs to approach to a distance of 6 metres (20 ft) to read letters that a person with normal acuity could read at 12m (40 feet). In an even more approximate manner, this person could be said to have "half" the normal acuity of 6/6 or 20/20.

DME is underdiagnosed and undertreated today

• Rate of type 2 diabetes continues to grow Treatment received during the first year following diagnosis • 40-45% of people with diabetes will develop diabetic retinopathy1,2

• DME is a type of DR that is a leading cause of vision loss for people 10000 75% of all patients receive no treatment in first 28 days

with diabetes3

8000

5.4 mn DME prevalent cases (major markets) in 2015* Ranibizumab Aflibercept 6000 number Bevacizumab

Patient 4000 39% 52% Drug- Diagnosed

Treated 2000 Number ofPatients

0 28 365 28 365 28 365 28 365 28 365 Days 4 ~Only 50% of people diagnosed today Observation anti-VEGF Laser Corticosteroid Combination

J.R. Willis et al., AAO 2017 *Source: Decision Resources (January 2018) 1 National Eye Institute. Facts About Diabetic Eye Disease. Available at: https://nei.nih.gov/health/diabetic/retinopathy. 2 NIH National Eye institute-https://nei.nih.gov/health/macular- edema/fact_sheet. 3 American Academy of Ophthalmology. What is Diabetic Retinopathy? Available at: http://www.geteyesmart.org/eyesmart/diseases/diabetic-retinopathy/index.cfm. Accessed October 2, 2015. 4 Bressler NM, Varma R, Doan Q, et al. Prevalence of Visual Impairment from Diabetic Macular Edema and Relationship to Eye Care from the 2005 − 2008 12 National Health and Nutrition Examination Survey (NHANES)[abstract]. The Retina Society 45th Annual Scientific Meetings, Washington, DC; October 4−7, 2012 (accepted for presentation).

DME is underdiagnosed and undertreated today

• Rate of type 2 diabetes continues to grow Treatment received during the first year following diagnosis • 40-45% of people with diabetes will develop diabetic retinopathy1,2

• DME is a type of DR that is a leading cause of vision loss for people 10000 with diabetes3 60% of all patients receive

no treatment in 1st year

8000

5.4 mn DME prevalent cases (major markets) in 2015* Ranibizumab Aflibercept 6000 number Bevacizumab

Patient 4000 39% 52% Drug- Diagnosed

Treated 2000 Number ofPatients

0 28 365 28 365 28 365 28 365 28 365 Days 4 ~Only 50% of people diagnosed today Observation anti-VEGF Laser Corticosteroid Combination

J.R. Willis et al., AAO 2017 *Source: Decision Resources (January 2018) 1 National Eye Institute. Facts About Diabetic Eye Disease. Available at: https://nei.nih.gov/health/diabetic/retinopathy. 2 NIH National Eye institute-https://nei.nih.gov/health/macular- edema/fact_sheet. 3 American Academy of Ophthalmology. What is Diabetic Retinopathy? Available at: http://www.geteyesmart.org/eyesmart/diseases/diabetic-retinopathy/index.cfm. Accessed October 2, 2015. 4 Bressler NM, Varma R, Doan Q, et al. Prevalence of Visual Impairment from Diabetic Macular Edema and Relationship to Eye Care from the 2005 − 2008 13 National Health and Nutrition Examination Survey (NHANES)[abstract]. The Retina Society 45th Annual Scientific Meetings, Washington, DC; October 4−7, 2012 (accepted for presentation).

Real world outcomes in nAMD, DME limited by treatment burden Potentially addressed by extending durability of anti-VEGF, targeting additional MOAs, or both Anti-VEGF injections* in 1st year of treatment in patients provided Mean change in visual acuity from baseline over time for patients with AMD1 anti-VEGF within 28 days of initial DME diagnosis2

Germany (n=420) Italy (n=365) 20% 19.3% 8 France (n=398) The Netherlands (n=350)

United Kingdom (n=410) Total (n=2227)

16.4% a

6 15% 14.1%

MORE VISITS & INJECTIONS 4 10.2% 9.6% 10% 7.7% 2 7.3% 5.9% Patients (%) Patients 5.4%

5% 4.0% 0 FEWER

Mean VA MeanVA difference 0 90 180 270 360 450 540 630 720

-2 0% from baseline frombaseline country by (LOCF) 1 2 3 4 5 6 7 8 9 10+ -4 Days Number of Anti-VEGF Injections • Patients received a mean of 5.0 and 2.2 injections in the first and second year, • Almost 50% of patients receiving anti-VEGF treatment within 28 days of initial respectively. diagnosis received ≤3 injections in the first year of treatment. • There were substantial differences in visual outcomes and injection frequency between countries. More frequent visits and injections were associated with greater improvements in visual acuity

1 F.G. Holz et al., Br J Ophthalmol 2015; 2 J.R. Willis et al., AAO 2017 14 AMD=age-related macular degeneration; DME=diabetic macular edema; MOA=mechanism of action; VA=visual acuity

Approaches to potentially achieve better outcomes 1. Sustained delivery of intravitreal therapies for months at a time

Solid-state reservoir implant Implanted in OR Refillable in the office using slowly elutes ranibizumab by vitreoretinal surgeon proprietary needle assembly

LADDER: Phase 2 multicenter, randomized, interventional, active treatment-controlled study (NCT02510794)

RPDS Implant RBZ formulation 1 N=60

RPDS Implant RBZ formulation 2 Wet AMD population N=60 (N = 220) R RPDS Implant RBZ formulation 3 N=60

ITV SOC RBZ 0.5 mg/mL Monthly N=40 Primary endpoint: Time to first refill Study fully enrolled, data expected 2H2018 ForSIGHT VISION4 acquired by Roche, January 2017 15 AMD=age-related macular degeneration; RPDS=ranibizumab port delivery system; RBZ=ranibizumab; ITV=intravitreal; SOC=standard of care Approaches to potentially achieve better outcomes 2. Address vascular homeostasis by targeting Ang-2 along with VEGF

Dual Ang-2/VEGF inhibition in DME may improve both efficacy and durability

Anti-VEGF/Ang2 biMAb

anti-VEGF anti-Ang2

Fc region

Human vitreous levels

• First bispecific antibody in ophthalmology • Levels of Ang-2 elevated in retinal vascular diseases1 • Ang2 and VEGF are key drivers of angiogenesis binding to VEGF and Angiopoetin-2 (Ang2) • Angiopoietin/Tie2 axis modulates endothelial • Optimized Fc for faster systemic clearance cell stabilization2 (FcRn), no effector function (FcγR)

1 EMBO Mol Med 2016;8:1265; 2 Am J Pathol 2002;160:1663 16 DME=diabetic macular edema; biMAb=bispecific monoclonal antibody VA2 RG7716 phase 2 program in both DME and nAMD

Ranibizumab 0.3 mg Diabetic macular edema N=150 anti-VEGF naïve, 60 anti- Observational period up to BOULEVARD RG7716 1.5 mg VEGF previously-treated R week 36 NCT02699450 RG7716 6 mg Primary outcome: Mean change from Data at Angiogenesis 2018 baseline in BCVA letter score at week 24 Every fourth week up to week 20, for a total of 6 administrations

Ranibizumab 0.5 mg Q4 weeks (9 months)

RG7716 D1 Q4 weeks (9 months) Dose response nAMD, Treatment naïve H2H N=273 R RG7716 D2 Q4 weeks (9 months) AVENUE NCT02484690 RG7716 D2 Q4 X 4 mo RG7716 D2 Q8 x 5 months Q8 weekly durability Primary outcome: Mean change from RBZ Q4 X 3 mo VA2 D2 Q4 x 6 months Sub group analysis baseline in BCVA letter score at week 36 incomplete responders 9 month primary efficacy readout Data expected in 2018

Ranibizumab 0.5 mg q4w nAMD, Treatment naïve N=75 R RG7716 short interval duration STAIRWAY NCT03038880 RG7716 long interval duration Primary outcome: Mean change from Baseline in BCVA letter score at week 40 Data expected in 2018

AMD=age-related macular degeneration; DME=diabetic macular edema; BCVA=best corrected visual acuity 17

BOULEVARD: Phase 2 results of the bispecific VEGF/Ang2 antibody in diabetic macula edema Sascha Fauser, M.D., Head pRED Ophthalmology Anti-VEGF/Anti–Angiopoietin-2 Bispecific Antibody RG7716 in Diabetic Macular Edema Results From the Phase 2 BOULEVARD Clinical Trial

Pravin U. Dugel, MD1 Jayashree Sahni, MBBS, FRCOphth, MD2; Shamil Sadikhov, MSc2; Meike Pauly-Evers, PhD2; Piotr Szczesny, MD, PhD2; Robert Weikert, MSc2 on behalf of the BOULEVARD Study Investigators

1 Retina Consultants of Arizona, Phoenix, AZ, USC Eye Institute, Keck School of Medicine, University of Southern California, Los Angeles, CA RG7716 2 Roche Pharma Research and Early Development, Roche Innovation Center, Basel, Switzerland

Presented at Angiogenesis, Exudation, and Degeneration 2018, February 10, 2018

19 Disclosures

• Relevant financial disclosures – Dr. Dugel is a consultant for Genentech/Roche and Novartis – Dr. Dugel sits on the Aerpio board of directors

• Study disclosures – This study includes research conducted on human subjects – Institutional Review Board approval was obtained prior to study initiation – Funding was provided by F. Hoffmann-La Roche Ltd. for the study and third-party writing assistance, which was provided by Betsy C. Taylor, PhD, CMPP, of Envision Pharma Group

20 Take-Home Points RG7716 met its primary endpoint in the BOULEVARD phase 2 DME trial

• RG7716 is the first bispecific antibody designed for intravitreal use that simultaneously binds and neutralizes both Ang-2 and VEGF-A

• RG7716 demonstrated statistically significant BCVA gains over ranibizumab at week 24 in anti-VEGF treatment-naïve patients with DME

• Secondary functional and anatomical endpoints (OCT, DRSS) support BCVA primary outcome

• RG7716 was well tolerated and showed no new or unexpected safety signals

BOULEVARD clinical trial (NCT02699450). 21 Ang-2, angiopoietin-2; BCVA, best-corrected visual acuity; DME, diabetic macular edema; DRSS, Diabetic Retinopathy Severity Score; OCT, optical coherence tomography; VEGF-A, vascular endothelial growth factor A. Significant Unmet Need Remains for Patients With DME There is a need for therapies that improve efficacy and/or reduce treatment burden

DME is an increasing global burden

• Global rate of diabetes is growing and is expected to increase to affect 592 million by 20351 • DR, including associated DME, is a leading cause of vision loss in working-age adults2

Anti–VEGF-A monotherapies have hit a ceiling for efficacy

• In RCT, only ~ 30%–45% of anti–VEGF-treated patients with DME gained ≥ 15 letters*,3-8 • Real-world data show that patients are not receiving the optimal number of injections, highlighting the need for therapies that can reduce treatment burden3-12 • DR/DME is a multifactorial disease involving multiple pathways beyond VEGF13,14 • Current anti-VEGF treatments only target 1 pathway, primarily addressing vessel leakage and proliferation15 • Targeting additional pathways beyond VEGF could improve BCVA outcomes and decrease treatment burden

* Results for years 1 and 2 of treatment. 1. Guariguata L et al. Diabetes Res Clin Pract. 2014;103(2):137-149. 2. Bourne RR et al. Lancet Glob Health. 2013;1(6):e339-349. 3. Massin P et al. Diabetes Care. 2010;33(11):2399-2405. 4. Korobelnik JF et al. Ophthalmology. 2014;121(11):2247-2254. 5. DRCR.net. N Engl Med. 2015;372(13):1193-1203. 6. Nguyen QD et al. Ophthalmology. 2012;119(4):789-801. 7. Brown DM et al. Opthalmology. 2015;122(10):2044-2052. 8. DRCR.net. Ophthalmology. 2016;123(6):1351-1359. 9. Granstrom T et al. Diabetes Res Clin Pract. 2016;121:157-165. 10. Wecker T et al. Br J Ophthalmol. 2017;101(3):353-359. 11. Willis JR et al. Presented at AAO 2017. 12. Ciulla TA et al. Presented at ARVO 2017. 13. Cohen SR et al. Dev Ophthalmol. 2016;55:137-146. 14. Romero-Aroca P et al. J Diabetes Res. 2016;2016:2156273. 15. Boyer DS et al. Ther Adv Endocrinol Metab. 2013;4(6):151-169. 22 BCVA, best-corrected visual acuity; DME, diabetic macular edema; DR, diabetic retinopathy; RCT, randomized clinical trial; VEGF, vascular endothelial growth factor. Ang-2 Signaling Implicated in the Pathology of DR and DME Evidence supports rationale for Ang-2 inhibition

Ocular Characteristics of Associated With DR and DME Ang-2 Signaling

2,5 Elevated vitreous levels of Ang-21-4

2,7,8 Retinal microvascular inflammation5,6

Blood-retinal-barrier breakdown5 2,5,7,8 2,5 • Pericyte dropout Vascular destabilization 5 • Endothelial cell destabilization Microaneurysms • Retinal vessel leukostasis5,6 Fluid leakage Ischemia 9,10 Capillary sprouting and remodeling9

1. Regula JT et al. EMBO Mol Med. 2016;8(11):1265-1288. 2. Saharinen P et al. Nat Rev Drug Discov. 2017;16(9):635-661. 3. Patel JI et al. Br J Ophthalmol. 2005;89(4):480-483. 4. Tuuminen R et al. Acta Ophthalmol. 2015;93(6):e515-e516. 5. Klaassen I et al. Prog Retin Eye Res. 2013;34:19-48. 6. Cohen SR et al. Dev Ophthalmol. 2016;55:137-146. 7. Scholz A et al. Ann N Y Acad Sci. 2015;1347:45-51. 8. Huang H et al. Nat Rev Cancer. 2010;10(8):575-585. 9. Campochiaro PA. Prog Retin Eye Res. 2015;49:67-81. 10. Flecht M et al. J Clin Invest. 2012;122(6):1991-2005. 23 Ang-2, angiopoietin-2; DME, diabetic macular edema; DR, diabetic retinopathy. RG7716: the First Bispecific Antibody Designed for Intravitreal Use Engineered for efficacy, duration within the eye, and fast systemic clearance

1 molecule – 2 targets

Anti–Ang-2 Fab Anti–VEGF-A Fab • Enhanced activity • Proven efficacy through through Ang-2 inhibition VEGF-A inhibition

Optimized Fc • Faster systemic clearance (FcRn) • No effector function (FcγR)

Regula JT et al. EMBO Mol Med. 2016;8(11):1265-1288. 24 Ang-2, angiopoietin-2; Fab, fragment antigen binding; Fc, fragment crystallizable; FcRn, neonatal Fc receptor; VEGF-A, vascular endothelial growth factor A. RG7716 Simultaneously Binds Both Ang-2 and VEGF-A Dual inhibition independent of binding order

45 40 2nd VEGF-A 35 binding VEGF-A

Ang-2 30 2nd Ang-2 25 binding 20 1st Ang-2 15 binding Response, RU Response, 10 5 1st VEGF-A binding 0 RG7716 0 100 200 300 400 500 Time, s

Regula JT et al. EMBO Mol Med. 2016;8(11):1265-1288. 25 Ang-2, angiopoietin-2; RU, response units; VEGF-A, vascular endothelial growth factor A. The BOULEVARD Study: RG7716 in DME

26 Study Design Randomized, active comparator-controlled, double-masked, phase 2 clinical trial

Study Off-Treatment Treatment Observation

229 patients with 6.0 mg RG7716 center-involving DME* 1.5 mg RG7716 0.3 mg ranibizumab • 168 anti-VEGF treatment- R naïve patients† Primary endpoint • 61 previously anti–VEGF- 0.3 mg ranibizumab treated patients‡ if required§ 0 4 8 12 16 20 24 28 32 36 Time, Weeks

* Safety data set consists of 224 patients; 2 patients removed due to GCP non-compliance at a single site; 3 patients were randomized but did not receive treatment. † Patients randomized 1:1:1 into the 3 treatment arms; 2 patients removed from analysis due to GCP non-compliance at a single site. ‡ Patients randomized 1:1 into 0.3 mg ranibizumab and 6.0 mg RG7717 treatment arms. § Ranibizumab treatment given and patient exited study if both CST increased by ≥ 50 μm from week 24 and BCVA decreased by ≥ 5 ETDRS letters from week 24 due to DME. BOULEVARD clinical trial (NCT02699450). 27 BCVA, best-corrected visual acuity; CST, central subfield thickness; DME, diabetic macular edema; GCP, Good Clinical Practice; R, randomized; VEGF-A, vascular endothelial growth factor A. Key Inclusion and Exclusion Criteria

Key inclusion criteria Key exclusion criteria

• Age ≥ 18 years All patients • Any signs of high-risk PDR • Center-involving DME • Any prior PRP • CST ≥ 325 μm* • Any macular laser photocoagulation within 3 months prior to study start • BCVA 73–24 ETDRS letters (20/40–20/320 Snellen equivalent) Previously anti–VEGF-treated patients only • Anti-VEGF treatment within 3 months prior to study start

* CST as measured by Spectralis (Heidelberg) at screening; ≥ 315 μm for Cirrus and Topcon and ≥ 295 μm for Optovue. BOULEVARD clinical trial (NCT02699450). BCVA, best-corrected visual acuity; CST, central subfield thickness; DME, diabetic macular edema; ETDRS, Early Treatment Diabetic Retinopathy Study; PDR, proliferative diabetic retinopathy; 28 PRP, panretinal photocoagulation; VEGF, vascular endothelial growth factor. Key Study Objectives

• Primary objective – Efficacy of RG7716 compared with ranibizumab in anti-VEGF treatment-naïve patients at week 24 • Mean BCVA change from baseline using a linear model adjusting for baseline BCVA and randomization stratification factors*

• Key secondary and exploratory objectives – Ocular and systemic safety – Anatomical outcomes – DR severity outcomes – Duration of effect – Outcomes in previously anti–VEGF-treated patients – Systemic and ocular pharmacokinetics – Analysis of plasma, aqueous humor, and vitreous biomarkers

* Linear model adjusted for baseline BCVA, previous macular laser treatment status at randomization, and BCVA category (≥ 64 letters vs ≤ 63 letters) at baseline. BOULEVARD clinical trial (NCT02699450). 29 BCVA, best-corrected visual acuity; DR, diabetic retinopathy; VEGF, vascular endothelial growth factor. Study Results: Week 24 Data for Anti-VEGF Treatment-Naïve Patients

30 Baseline Demographics Well balanced across treatment arms

0.3 mg Ranibizumab 1.5 mg RG7716 6.0 mg RG7716 All Patients Characteristic n = 59 n = 54 n = 53 N = 166 Mean age, years (SD) 61.6 (9.5) 61.4 (7.7) 60.5 (9.1) 61.2 (8.8) Male, n (%) 37 (62.7%) 19 (35.2%) 33 (62.3%) 89 (53.6%) Race, n (%) American Indian or Alaska native 0 (0%) 0 (0%) 2 (3.8%) 2 (1.2%) Asian 0 (0%) 0 (0%) 1 (1.9%) 1 (0.6%) Black or African American 9 (15.3%) 11 (20.4%) 10 (18.9%) 30 (18.1%) White 49 (83.1%) 42 (77.8%) 39 (73.6%) 130 (78.3%) Unknown 1 (1.7%) 1 (1.9%) 1 (1.9%) 3 (1.8%) Ethnicity, n (%) Hispanic or Latino 11 (18.6%) 8 (14.8%) 9 (17.0%) 28 (16.9%) Mean duration of diabetes at 14.0 (10.5) 15.6 (10.0) 14.5 (9.3) 14.7 (10.0) randomization, years (SD) Mean HbA1c, % (SD) 7.8% (1.6) 8.2% (1.6) 7.7% (1.8) 7.9% (1.7)

Observed data; anti-VEGF treatment-naïve patients only. BOULEVARD clinical trial (NCT02699450). 31 HbA1c, glycated hemoglobin. Baseline Ocular Characteristics Generally well balanced across treatment arms

0.3 mg Ranibizumab 1.5 mg RG7716 6.0 mg RG7716 All Patients Characteristic n = 58 n = 54 n = 51 N = 163 Vision Mean BCVA, ETDRS letters (SD) 61.2 (9.9) 60.9 (11.1) 60.0 (11.0) 60.8 (10.6) 20/40 or better vision, n (%) 13 (22.4%) 15 (27.8%) 11 (21.6%) 39 (23.9%) Worse than 20/40 vision, n (%) 45 (77.6%) 39 (72.2%) 40 (78.4%) 124 (76.1%) Anatomic Mean CST, μm (SD) 490.9 (139.0) 535.4 (163.1) 496.5 (135.0) 507.4 (146.7) Diabetic retinopathy status, n (%) n = 59 n = 54 n = 53 N = 166 DR questionable 0 (0.0%) 1 (1.9%) 0 (0.0%) 1 (0.6%) Mild NPDR 5 (8.5%) 5 (9.3%) 5 (9.4%) 15 (9.0%) Moderate NPDR 14 (23.7%) 8 (14.8%) 10 (18.9%) 32 (19.3%) Moderately severe NPDR 23 (39.0%) 21 (38.9%) 25 (47.2%) 69 (41.6%) Severe NPDR 15 (25.4%) 16 (29.6%) 9 (17.0%) 40 (24.1%) Prior PRP 1 (1.7%) 1 (1.9%) 1 (1.9%) 3 (1.8%) Moderate PDR 0 (0.0%) 0 (0.0%) 1 (1.9%) 1 (0.6%) Missing/cannot grade 1 (1.7%) 2 (3.7%) 2 (3.8%) 5 (3.0%) Observed data; anti-VEGF treatment-naïve patients only. BOULEVARD clinical trial (NCT02699450). 32 BCVA, best-corrected visual acuity; CST, central subfield thickness; DR, diabetic retinopathy; ETDRS, Early Treatment Diabetic Retinopathy Study; NPDR, non-proliferative DR; PDR, proliferative DR; PRP, panretinal photocoagulation. Mean BCVA Gains From Baseline (Observed Values) RG7716 treatment provided dose-dependent improvements in vision

16 + 14.8 14

12 + 12.3

10 + 10.1

4 Baseline, Baseline, Letters ETDRS Mean BCVA Change From Mean BCVA 2

0 0 4 8 12 16 20 24 Time, Weeks 0.3 mg ranibizumab (n = 59) 1.5 mg RG7716 (n = 54) 6.0 mg RG7716 (n = 53)

Observed data; anti-VEGF treatment-naïve patients only. Error bars represent 80% CI. 33 BOULEVARD clinical trial (NCT02699450). BCVA, best-corrected visual acuity; ETDRS, Early Treatment Diabetic Retinopathy Study. Adjusted Mean BCVA Gains From Baseline RG7716 met its prespecified primary endpoint of efficacy

Linear model adjusted for baseline BCVA and randomization stratification factors*

14 + 13.9 12 + 11.7 + 3.6 10 + 10.3

8 P = 0.03† 80% CI 1.53, 5.61 6

Adjusted Mean BCVA Change Adjusted Mean BCVA From From Baseline, Letters ETDRS 0 0 4 8 12 16 20 24 Time, Weeks 0.3 mg ranibizumab (n = 59) 1.5 mg RG7716 (n = 54) 6.0 mg RG7716 (n = 53) Observed data; anti-VEGF treatment-naïve patients only. Error bars represent 80% CI. * Linear model adjusted for baseline BCVA, previous macular laser treatment status at randomization, and BCVA category (≥ 64 letters vs ≤ 63 letters) at baseline. † Protocol prespecified significance level, 34 P < 0.2. BOULEVARD clinical trial (NCT02699450). BCVA, best-corrected visual acuity; ETDRS, Early Treatment Diabetic Retinopathy Study. Mean CST Change From Baseline (Observed Values) CST reduction directionally supports BCVA primary outcome

-50

m μ -100

-150 – 177.9 μm

-200 – 200.3 μm

Mean Mean Change CST From From Baseline, – 237.5 μm -250

-300 0 4 8 12 16 20 24 Time, Weeks 0.3 mg ranibizumab (n = 59) 1.5 mg RG7716 (n = 54) 6.0 mg RG7716 (n = 53)

Observed data; treatment-naïve patients only. Error bars represent 80% CI. 35 BOULEVARD clinical trial (NCT02699450). BCVA, best-corrected visual acuity; CST, central subfield thickness. Adjusted Mean CST Change From Baseline CST reduction directionally supports BCVA primary outcome in a dose-dependent manner

Linear model adjusted for baseline CST and randomization stratification factors*

-50

m μ -100

-150

– 204.7 μm -200

– 217.1 μm From From Baseline,

-250 – 225.8 μm Adjusted Adjusted Mean CST Change

-300 0 4 8 12 16 20 24 Time, Weeks 0.3 mg ranibizumab (n = 59) 1.5 mg RG7716 (n = 54) 6.0 mg RG7716 (n = 53) Observed data; anti-VEGF treatment-naïve patients only. Error bars represent 80% CI. * Linear model adjusted for baseline CST, previous macular laser treatment status at randomization, and BCVA category (≥ 64 letters vs ≤ 63 letters) at baseline; protocol prespecified significance level, 36 P < 0.2. BOULEVARD clinical trial (NCT02699450). BCVA, best-corrected visual acuity; CST, central subfield thickness. RG7716 Resulted in Improved Proportions of 2- and 3-Line Gainers Supporting BCVA primary outcome in a dose-dependent manner

≥ 2-Line Gainers* ≥ 3-Line Gainers† at Week 24 at Week 24

80 80 70.5 61.2

57.1

60 60

43.2 36.7

40 40 32.7

Patients, Patients, % Patients, % 20 20

0 0

0.3 mg ranibizumab (n = 49) 1.5 mg RG7716 (n = 49) 6.0 mg RG7716 (n = 44)

* ≥ 10 ETDRS letters from baseline. † ≥ 15 ETDRS letters from baseline. Observed data, week 24; anti-VEGF treatment-naïve patients only. BOULEVARD clinical trial (NCT02699450). 37 BCVA, best-corrected visual acuity; ETDRS, Early Treatment Diabetic Retinopathy Study. RG7716 Resulted in Improved DR Severity Scores ≥ 2-step DRSS improvement data support BCVA primary outcome in a dose-dependent manner

≥ 2-Step DR Improvement at Week 24

38.6 40

27.7

Percent of Patients 20 12.2

0.3 mg ranibizumab (n = 49) 1.5 mg RG7716 (n = 47) 6.0 mg RG7716 (n = 44)

Observed data, week 24; anti-VEGF treatment-naïve patients only. BOULEVARD clinical trial (NCT02699450). 38 BCVA, best corrected visual acuity; DR, diabetic retinopathy; DRSS, Diabetic Retinopathy Severity Score. Safety: All Patients

39 Safety – All Patients (Anti-VEGF Treatment-Naïve and Previously Treated) No new safety signals or unexpected findings

Selected Adverse Events, Study Eye 0.3 mg Ranibizumab 1.5 mg RG7716 6.0 mg RG7716 All Patients or Systemic, n (%) n = 89 n = 55 n = 80 N = 224 Intraocular inflammation 0 0 0 0 Endophthalmitis 0 0 0 0 Retinal detachment 0 0 0 0 Vitreous hemorrhage 3 (3.4%) 0 1 (1.3%) 4 (1.8%) Hypertension 7 (7.9%) 5 (9.1%) 6 (7.5%) 18 (8.0%) Non-fatal myocardial infarction 1 (1.1%) 0 0 1 (0.4%) Non-fatal stroke 0 0 0 0 Vascular death 1 (1.1%) 0 1 (1.3%) 2 (0.9%) All deaths attributed to cardiac, cerebral, hemorrhagic, 1 (1.1%) 0 2 (2.5%) 3 (1.3%) embolic, other vascular, or unknown causes Any other death 1 (1.1%)a 1 (1.8%)b 0 2 (0.9%) IOP, mmHg Baseline; week 24 (pre-dose); (change from baseline) 15.1; 15.4 (0.1) 16.0; 16.6 (0.6) 15.3; 15.3 (-0.1) NA

a Kidney failure. b Gangrene. Observed data; safety population includes both anti-VEGF treatment-naïve and previously anti–VEGF-treated patients. 40 BOULEVARD clinical trial (NCT02699450). IOP, intraocular pressure; NA, not applicable. Safety – All Patients (Anti-VEGF Treatment-Naïve and Previously Treated) No new safety signals or unexpected findings

• Ocular and systemic safety profile of RG7716 was consistent with the safety profile observed in patients with DME treated with intravitreal anti-VEGF drugs, with no new safety signals observed

• No adverse events of intraocular inflammation or endophthalmitis

• No increase in mean intraocular pressure

• No increase in mean blood pressure

Observed data; safety population includes both anti-VEGF treatment-naïve and previously anti–VEGF-treated patients. BOULEVARD clinical trial (NCT02699450). 41 DME, diabetic macular edema; VEGF, vascular endothelial growth factor. Outlook Presentation of additional data and analyses at future meetings

• Efficacy in previously anti–VEGF-treated patients • Additional anatomical outcomes • Duration of effect

BOULEVARD clinical trial (NCT02699450). 42 VEGF, vascular endothelial growth factor. Conclusions RG7716 met its primary endpoint in the BOULEVARD phase 2 DME trial

RG7716 • The first bispecific antibody specifically designed for intravitreal use that simultaneously binds and neutralizes both Ang-2 and VEGF-A

BOULEVARD phase 2 study in DME* ••RG7716X demonstrated robust visual acuity gains in patients with DME at 6 months, with a mean of + 13.9 letters gained from baseline ••RG7716X demonstrated statistically significant BCVA gains over ranibizumab at 6 months (mean gain of + 3.6 letters over ranibizumab, P = 0.03) • X • ≥ 2- and ≥ 3-line gainers, CST reduction, and DRSS improvement data support BCVA primary outcome • Both primary and secondary outcomes showed a dose-dependent response • RG7716 was well tolerated and showed no new or unexpected safety signals

43 * Results for anti-VEGF treatment-naïve patients. Thank You to the More Than 70 Participating Study Sites

44 Thank You to All Involved in the BOULEVARD Trial

• The authors would like to thank all those involved in the BOULEVARD trial, in particular all of the patients and their families, and all of the investigators • Principal investigators (all US): Adrean S, Retina Consultants of Orange County; Alfaro V, Charleston Neuroscience Inst; Antoszyk A, Char Eye Ear & Throat Assoc; Awh C, Tennessee Retina PC.; Baker C, Paducah Retinal Center; Barbazetto I, Vitreous-Retina-Macula; Brooks-Jr H, Southern Vitreoretinal Assoc; Brown D, Retina Consultants of Houston; Busquets M, Associates in Ophthalmology; Callanan D, Texas Retina Associates; Calzada J, Charles Retina Institution; Campochiaro P, Wilmer Eye Institute; Castellarin A, California Retina Consultants; Chan C, Southern CA Desert Retina Cons; Chittum M, Retina Consultants of Southern; Clark L, Palmetto Retina Center; Danzig C, Rand Eye; Das A, University of New Mexico; Dreyer R, Retina Northwest; Dugel P, Retinal Research Institute, LLC; Eichenbaum D, Retina Vitreous Assoc of FL; Eifrig C, Retina Associates of Orange County; Elman M, Elman Retina Group; Ghuman T, National Ophthalmic Research Institute; Goldberg R, Bay Area Retina Associates; Googe J, Southeastern Retina Associates; Gordon A, Associated Retina Consultants; Gupta S, Retina Specialty Institute; Hampton R, Retina Vit Surgeons/Central NY; Hershberger V, Florida Eye Associates; Higgins P, Retina Center of New Jersey; Hunter, III A, Oregon Retina, LLP; Jain A, Ophthalmic Clinical Trials San Diego; Javid C, Retina Associates Southwest PC; Jhaveri C, Retina Research Center; Keyser B, New Jersey Retina Research Foundation; Khanani A, Sierra Eye Associates; Khurana R, Northern California Retina Vitreous Associates; Kiss S, Weill Cornell Eye Associates, New York-Presbyterian Hospital/Weill Cornell Medical Center; Kwun R, Retina Associates of Utah; Ladd B, Eye Surgeons of Richmond Inc. Virginia Eye Institute; Lee P, Retina Consultants of Western New York; Leys M, West Virginia University Eye Institute; Lim J, Univ of Illinois at Chicago; London N, Retina Consultants, San Diego; MacCumber M, Illinois Retina Associates SC; Marcus D, Southeast Retina Center; Maturi R, Midwest Eye Institute; McCabe F, Vitreo-Retinal Associates; Miller D, Cincinnati Eye Institute; Nielsen J, Wolfe Eye Clinic; Ober M, Retina Consultants of Michigan; Ohr M, OSU Eye Physicians & Surgeons; Parke D. Wilken, Vitreoretinal Surgery; Patel S, W Texas Retina Consultants PA; Payne J, Palmetto Retina Center; Peace J, United Med Res Inst; Pearlman J, Retinal Consultants Med Group; Prensky J, Pennsylvania Retinal Spec.; Rathod R, Orange County Retina Med Group; Schadlu R, Arizona Retina and Vitreous Consultants; Schwartz S, Jules Stein Eye Institute/ UCLA; Sheth V, University Retina and Macula Associates, PC; Shirkey B, Retina Associates of Kentucky; Singer M, Med Center Ophthalmology Assoc; Singerman L, Retina Assoc of Cleveland Inc; Singh R, Cleveland Clinic Foundation, Cole Eye Institute; Stern J, Capital Region Retina; Stoller G, Ophthalmic Consultants of LI; Stoltz R, Georgia Retina PC; Stone C, West Carolina Retinal Assoc PA; Torti R, Retina Specialists; Wirthlin R, Spokane Eye Clinical Research; Wong R, Austin Retina Associates; Wykoff C, Retina Consultants of Houston; Yates P, Univ of Virginia Ophthalmology; Zilis J, Colorado Retina Associates, PC.