DURAVYU: LUGANO and LUCIA primary failures, with the DME test ahead

Angiogenesis and treatment duration

DURAVYU combines the kinase inhibitor vorolanib with a bioerodible intravitreal insert. In wet AMD the objective is sustained suppression of pathological vascular leakage while preserving visual acuity. Receptor inhibition differs from extracellular VEGF binding, but a broader mechanism does not automatically improve vision compared with a well-administered anti-VEGF control. [1] [5]

Clinical interpretation: durability is valuable only if reduced injections do not sacrifice vision. Anatomy, visual acuity and treatment burden are related but distinct outcomes. An eye may remain anatomically controlled yet lose vision from another process; a randomized treatment strategy must account for the total patient outcome rather than redefine success around favorable biological components.

Repeated insert dosing and rescue

LUGANO compared DURAVYU 2.7 mg on a six-month repeat schedule with on-label 2 mg aflibercept in a double-masked randomized trial. Both treatment-naïve and previously treated active wet AMD were eligible. The primary assessment averaged change in BCVA at weeks 52 and 56, so the study tested repeat-treatment performance rather than only the first insert's early effect. [1] [5]

Analytical judgment: supplemental aflibercept is part of the treatment strategy, not evidence to discard a participant. The protocol's rescue threshold, number of visits and missing-data handling determine whether a low injection count represents adequate sustained control. Repeat dosing also requires attention to residual insert material, local exposure and cumulative inflammation.

DAVIO2 supported development but did not predict Phase3

DAVIO2 enrolled 160 patients and used two active dose groups plus aflibercept control. Reported BCVA differences at the blended six-month endpoint were −0.3 and −0.4 letters for the originally labeled 2 mg and 3 mg doses. The detailed presentation lists smaller endpoint-specific denominators, including 50, 52 and 54 for injection analyses; these are not interchangeable with total enrollment. [3] [4] [6]

The presentation reported 63% supplement-free at six months for each active dose, while treatment-burden reductions differed depending on whether comparison used prior treatment history or prospective aflibercept. Mixing those comparators creates an exaggerated durability narrative. A short Phase2 noninferiority result does not ensure a longer, repeatedly dosed Phase3 trial will succeed.

LUGANO failed the full-dataset primary endpoint

On August 17, 2026, EyePoint disclosed that LUGANO did not meet its primary noninferiority endpoint in the full dataset. The company highlighted an ad hoc analysis excluding nine of 211 DURAVYU patients with ≥15-letter loss attributed to causes unrelated to wet AMD; that restricted analysis had nominal p=0.0096. These are separate results. [1] [2]

Analytical judgment: unequal vision-loss causes can explain sensitivity of an estimate without repairing the prespecified test. Post-randomization exclusions may break comparability, particularly when selected after outcomes are known. The primary effect estimate, confidence interval, exact noninferiority margin and full disposition should be supplied before estimating how far the trial missed; the topline prose does not provide all of them.

Noninferiority is a confidence-bound claim, not simply a small difference in average vision. Interpretation requires assurance that the active control retained its expected effect and rescue or missing visits did not make the groups artificially similar. Conversely, calling the control unusually strong relative to other trials does not invalidate the observed randomized comparison. External trials can differ in baseline vision, disease activity and definitions of unrelated vision loss. Transparent adjudication of each major loss, including whether reviewers remained masked, would help explain the imbalance. Removing those eyes answers a conditional question; patients starting treatment cannot know that they belong to the favorable selected subset. If subsequent analyses are clinically persuasive, they may inform regulatory discussion, but their evidentiary role must remain distinct from the original prespecified full-dataset primary endpoint and its failed result.

Secondary durability and selection effects

LUGANO reported 76% supplement-free through week32 and54% through week56, with42% less treatment burden than control. These signals support further investigation, but secondary nominal p-values do not recover confirmatory status after primary failure unless the prespecified testing plan explicitly allows it. The two timepoints also cannot be turned into patient-level relapse rates without longitudinal records. [1] [3]

Selecting only supplement-free eyes conditions on a post-treatment outcome. That can preferentially retain eyes with better response or prognosis and produces a different question from the original randomized comparison. An anatomical difference of only a few microns cannot substitute for the visual-acuity endpoint that was selected to protect patients from an inferior treatment strategy.

Safety and the remaining clinical program

The release reported no observed insert migration, retinal vasculitis or severe intraocular inflammation and no difference in several ocular safety categories. Complete arm-level incidence and exposure data were not supplied in that summary. Absence of a named severe event is not evidence that all eye-related adverse outcomes were absent or that rare repeat-dose risks are excluded. [1] [5] [6]

LUCIA is the separate remaining pivotal wet-AMD trial, with Q4 2026 guidance. A positive LUCIA could add independent evidence but would not retroactively change LUGANO's prespecified outcome. Regulatory interpretation will require the combined efficacy and safety package, consistency of endpoints and sensitivity analyses; a planned 2027 filing is a management intention, not an accepted application.

October 25 update: LUCIA is a second failed primary test

EyePoint disclosed LUCIA on October 19. The separate trial is now a historical result, not the earlier Q4 forecast and not a reclassification of LUGANO as a success. LUCIA and LUGANO both missed their prespecified full-dataset vision endpoint. The earlier LUGANO analysis excluding nine patients remains post hoc; a second failure weakens the claim that the first failure was only an unusual cohort imbalance. The original evidence is retained above so the evolution of the argument remains visible. [7] [8] [10]

LUCIA (NCT06683742) was a randomized, double-masked active-control trial enrolling 475 treatment-naïve or previously treated people with active wet AMD. After aflibercept loading, DURAVYU 2.7 mg was given every 24 weeks versus aflibercept 2 mg every eight weeks, with protocol-directed supplemental injections. The primary comparison was the blended change in best-corrected visual acuity at weeks 52 and 56; the non-inferiority margin was −4.5 letters. Reduced visits cannot compensate for failing this prespecified vision comparison.

The slides report baseline tables for 232 versus 238 and a safety table for 232 versus 243. These denominators are reproduced separately; the five-person discrepancy is not labeled dropout, crossover or an ITT exclusion because the disposition explanation was not extracted. The baseline groups were about 76.8/77.2 years old and approximately 76%/75% treatment-naïve. Baseline BCVA was 65.0/64.5 letters. The dataset is expressly preliminary.

LUCIA safety and treatment-burden limits

The safety table uses 232 DURAVYU and 243 aflibercept recipients. The percentages below are not converted to integer counts. Selected ocular events were more frequent with DURAVYU; the favorable-safety phrase is management characterization, not proof of equivalence. Severe intraocular inflammation and endophthalmitis were each below 1% in both groups; the source reports no insert migration into the anterior chamber, anterior chamber opacities, free-floating particles or retinal vasculitis. Total TEAE, serious-event and death counts by arm remain unextracted. [8]

About 8% discontinued through week 56 in each arm; no study or treatment discontinuation was attributed to DURAVYU. The burden analysis excludes early discontinuers without any week-8-or-later study-drug injection and excludes loading and the week56 study injection. Thus the 38% reduction is a defined analysis, not an unconditional real-world annual saving. Rescue use is part of the strategy being evaluated, not a nuisance to remove after viewing the outcome.

DME becomes the next test, with a smaller capital base

On October 19 the board approved approximately 53% workforce reductions and early conclusion of both wet-AMD pivotal trials without completing their second year. Estimated annual savings are $30 million, while separation charges are approximately $9 million. Preliminary September30 cash, equivalents and short-term investments were about $118 million. Savings are a future annual run-rate, not immediate cash received; the preliminary cash figure is not a new audited quarter. [7] [8] [9]

COMO and CAPRI in diabetic macular edema enrolled 551 in total and remain guided to Q4 2027. DME differs biologically from wet AMD, but a plausible inflammatory rationale does not erase the failed late-stage visual comparisons. VERONA was a small phase2 trial; its 2.7mg arm and control had only11 and6 participants, and its control strategy differed from repeated aflibercept in phase3. The October deck also displays an analysis excluding one outlier. That selected figure cannot replace the full-analysis result or establish replication.

Analytical judgment: the decisive future evidence is a prespecified, controlled vision result together with the full safety and missing-data analysis, not another favorable fluid-control subgroup. Longer-term tolerability after repeated implants and financing through the DME readout require continuing review. This update does not convert a forecast quarter into an exact date, assume an NDA will be filed, or infer that a low share price makes the remaining asset undervalued.

Sources / References

1. EyePoint LUGANO topline release, August 17, 2026
2. EyePoint SEC 8-K, August 17, 2026
3. EyePoint original DAVIO2 results presentation
4. EyePoint DAVIO2 topline release
5. LUGANO registry NCT06668064
6. EyePoint 2024 second-quarter update: 12-month DAVIO2 follow-up
7. EyePoint · LUCIA actual result · October 19, 2026
8. EyePoint · SEC-filed LUCIA slides · October 19, 2026
9. EyePoint · strategic prioritization · October 19, 2026
10. LUCIA registry NCT06683742 · current record requires history verification