Pharma Update

Roche Roche application of deep learning in ophthalmology Vabysmo reduces HRF volumes more vs. aflibercept, indicating better disease control ARVO. The application of deep learning to segment hyperreflective foci Representative predictions on YOSEMITE and RHINE (3 separate study eyes, all baseline) Vabysmo shown to reduced HRF volumes significantly more 300- compared to aflibercept Aflibercept 2.0 mg Q8W Faricimab 6.0 mg Q8W Faricimab 6.0 mg T&E Original B-scan B-scan with predictions En face view Adjusted Mean (95% CI) HRF Volume 250- 200 150 100- P value for comparison vs Aflibercept Q8W averaged across all visits: Faricimab Q8W: P = 0.0006 Faricimab T&E: P < 0.0001 Week 1 Week 4 Week 8 Week 12 Week 16 Week 20 Week 24 Week 28 Week 32 Week 36 Week 40 Week 44 Week 48 Visit HRF are small bright objects in the retina in patients with DME and may be imaging biomarkers for inflammation and disease progression DL was used to segment HRF, which were compared with HRF drawn by expert graders ⚫ HRF model predictions were consistent with expert assessments • • A post hoc analysis of YOSEMITE and RHINE demonstrated greater retinal HRF volume reductions in Vabysmo- vs. aflibercept-treated eyes at week 48 Volume reductions indicate suppression of pathways that mediate inflammation • Better resolution of the pathological manifestations of DME with Vabysmo support the therapeutic potential of dual Ang-2/VEGF-A inhibition in retinal diseases Maunz et al. ARVO 2023, Poster # 2697 -B0519; HRF-Hyperreflective foci; DME=Diabetic macular edema; nAMD=Neovascular age-related macular degeneration; DL-Deep learning; Ang-2-Angiopoietin-2; VEGF-A-Vascular endothelial growth factor A; Q8W-Every 8 weeks; T&E-Treat and Extend; CI=Confidence interval 130

View entire presentation