Three days after EyePoint Pharmaceuticals released LUGANO Phase 3 topline data, Jay S. Duker, MD, President & CEO, joins John Kitchens, MD, and Scott Krzywonos to explain what actually happened. Duravyu (vorolanib, Durasert E sustained-release insert) missed its primary non-inferiority endpoint in the full dataset — driven by nine patients in the treatment arm who lost 15+ letters for reasons unrelated to wet AMD control. Dr. Duker walks through the mechanism, the outlier analysis, and why he remains optimistic heading into LUCIA.

The Vitreoretinal View features Miles Harper, independent ophthalmology consultant, who uses two case studies — the collapse of Bionics' VR surgical headset and the slow adoption of heads-up 3D visualization — to examine why promising OR technologies fail to reach commercial success. The conversation closes with a grounded look at where AI and robotics are actually headed in the surgical suite.

Hosts: John Kitchens, MD, Scott Krzywonos

Topics Covered

Jay S. Duker, MD, on the LUGANO Data

Duravyu mechanism: small molecule TKI with pan-VEGF receptor activity (intracellular), plus PDGF (antifibrotic) and JAK1 (anti-inflammatory via IL-6) activity; delivered as a 2.7 mg bioerodible intravitreal insert every six months; systemic absorption at picogram levels only.

Study design: randomized, double-masked non-inferiority trial; aflibercept 2mg loading phase for all patients; 75% treatment-naive, 25% previously treated — the first wet AMD Phase 3 to enroll previously treated patients; primary endpoint averaged BCVA change at weeks 52 and 56.

Primary endpoint miss: control arm lost only 0.5% of patients to 15+ letter vision loss vs. historical ~4% for aflibercept 2mg — an unexpectedly strong comparator performance. Nine Duravyu patients lost 15+ letters for non-AMD reasons: six due to GA, two due to glaucoma, one due to repaired retinal detachment with resulting cataract.

Re-randomization analysis: CMO Ramiro Ribeiro, MD, PhD, re-randomized all 15+ letter losers across both arms ten times; every simulation produced non-inferiority with a mean letter difference of -1.3 to -2.0 — well within the margin.

Secondary data: 42% reduction in treatment burden vs. aflibercept (~2 fewer injections/year); 76% supplement-free at week 32; 54% supplement-free at week 56; four-micron OCT difference vs. control; supplement-free subgroup alone was non-inferior with a three-micron OCT difference.

GA data: Duravyu arm showed lower GA rates at end of trial (24%) vs. aflibercept arm (29%); no evidence drug causes or accelerates GA.

Regulatory path: LUCIA topline data expected Q4 2026; NDA submission planned 1H 2027 if positive. Precedents for one missed/one positive Phase 3 in retina: pegcetacoplan (DERBY/OAKS) and Lytenava.

Vitreoretinal View: OR Technology, Adoption & AI (Miles Harper)

Bionics case study: Israeli company developed a VR surgical headset with gesture-based controls — too disruptive to workflow, unclear value proposition vs. the analog microscope, and priced beyond what the marginal benefit could justify. First-mover disadvantage: arrived before the use case was fully defined.

Heads-up 3D visualization: similar adoption ceiling — compelling technology, real ergonomic benefits, but incremental rather than transformative improvement over the standard microscope; declining surgical reimbursements raise the break-even bar for surgical centers. Multiple entrants (Alcon, Zeiss, Heidelberg, Bausch, others) will drive prices down; generational familiarity with screens will drive adoption up.

AI and robotics: autonomous robotic surgery is coming — already deployed outside ophthalmology — but surgeon oversight will remain essential. Near-term AI value in the OR: image reading, fluid monitoring, cell counting, vitreous sensing — marginal, non-intrusive efficiencies that integrate without disrupting workflow. Contrast with AI as Skynet: the most useful applications will feel like spell check, not science fiction.

Key Takeaways

LUGANO missed its primary endpoint, but the secondary data and re-randomization analysis give EyePoint a credible argument that nine atypical patients — not drug failure — drove the results.

A positive LUCIA trial may be sufficient for NDA approval; precedent in retina for one missed and one positive Phase 3 exists.

OR technology adoption fails when it disrupts workflow without sufficient value exchange — the lesson from Bionics applies equally to heads-up 3D and will apply to AI and robotics.

The near-term AI opportunity in the OR is incremental efficiency, not autonomy.

Sponsors

This episode is made possible by Genentech, maker of Vabysmo (faricimab-svoa). Hear peer perspectives and case studies — including Dr. David A. Eichenbaum, MD, unpacking what drying looked like in a real-world DME patient — in the Purple Chair series at Vabysmo-HCP.com.

This episode is brought to you by Alcon, a global leader in eye care with an industry-leading breadth of surgical products and solutions, including Unity VCS. Learn more at unityvcs.com.

Credits
Production & Marketing: Laura Brown | Business Operations: Liz Hogan

🎧 Listen at RetinUp.com | Apple Podcasts | Spotify | YouTube