EyePoint Pharmaceuticals announced on October 19 that Duravyu (vorolanib intravitreal insert, EYP-1901), its sustained delivery tyrosine kinase inhibitor implant, failed the primary endpoint of the Phase 3 LUCIA trial in neovascular (wet) age related macular degeneration. The 475 patient trial compared Duravyu 2.7 mg dosed every 24 weeks against aflibercept 2 mg (Regeneron’s Eylea) every 8 weeks, with supplemental aflibercept as needed. Duravyu improved vision but did not achieve statistical noninferiority to aflibercept on average change in best corrected visual acuity at weeks 52 and 56.
It is the second failure of the same design. The companion Phase 3 LUGANO trial, 432 patients, also missed noninferiority earlier this year. In both trials the secondary story was real: LUCIA cut treatment burden by 38 percent, 52 percent of Duravyu eyes needed no supplemental injection through week 56 and 71 percent needed no more than one, with anatomic disease control and a favourable safety profile on redosing. Fewer injections was the product’s whole commercial premise, and the premise was not enough to carry a noninferiority claim against the standard of care.
The corporate consequence landed the same day. EyePoint will reduce its workforce by approximately 53 percent, terminate the wet AMD trial program early rather than complete the second year of both pivotal studies, and prioritise Duravyu in diabetic macular edema, where Phase 3 topline data are now expected in the fourth quarter of 2027. The board approved the reset on October 17 after reviewing the LUCIA results. Shares lost roughly 80 percent of their value when the market opened on the news, falling from 2.99 dollars to about 0.58 dollars, and a shareholder investigation notice followed within hours.
The case is a clean illustration of the noninferiority trap in retinal disease. A sustained delivery product can win on burden, durability and patient preference and still have no registration path if it cannot match the vision outcomes of an entrenched comparator dosed on label. EyePoint’s remaining value now rests on whether the DME program, a smaller and less crowded indication, can clear the bar that wet AMD set twice.