Key Takeaways
- An experimental oral drug called gildeuretinol acetate slowed the expansion of retinal atrophic lesions in a phase 2 study of people with Stargardt disease.
- In the primary analysis, lesion growth was 21.6% slower with gildeuretinol than in an untreated comparison group that included placebo-treated periods and natural-history cases.
- A prespecified analysis restricted to randomized participants also favored treatment, with retinal atrophy growing 14.9% more slowly during gildeuretinol treatment than during placebo treatment.
- The study did not show that gildeuretinol preserved visual acuity. Vision measured by best-corrected visual acuity changed little in either group over 24 months and did not differ significantly between treatment and placebo.
- The trial was small, its primary comparison incorporated nonrandomized natural-history data, and the clinical importance of the slower lesion growth remains uncertain.
Introduction
An ophthalmologist performs a slit-lamp eye examination. A phase 2 trial found that experimental gildeuretinol acetate slowed the expansion of retinal atrophy in people with Stargardt disease, but it has not yet been shown to preserve vision.
An experimental modified form of vitamin A slowed the expansion of retinal atrophy in people with Stargardt disease, according to a phase 2 clinical trial published September 3, 2026, in JAMA Ophthalmology.
The treatment, called gildeuretinol acetate, reduced the rate at which areas of damaged retina enlarged compared with an untreated comparison group (Kay et al., 2026).
The result is notable because Stargardt disease is a progressive inherited retinal disorder for which no approved treatment currently exists. But the study did not establish that gildeuretinol slows the loss of usable vision.
Best-corrected visual acuity remained generally stable during the two-year trial and was not significantly different between the treatment and placebo groups.
The most accurate interpretation, therefore, is that gildeuretinol affected an anatomical measure of disease progression. Whether that translates into meaningful preservation of vision over longer periods is still unknown.
What the Study Examined
Stargardt disease is the most common inherited macular dystrophy and is usually linked to disease-causing variants in the ABCA4 gene. The condition damages photoreceptors and the retinal pigment epithelium, gradually producing areas of retinal atrophy and central vision problems (Tanna et al., 2017).
Gildeuretinol acetate, also known as C20D3-vitamin A acetate or ALK-001, is chemically similar to vitamin A. Three hydrogen atoms at a specific position in the molecule are replaced with deuterium.
The purpose of that modification is to slow vitamin A dimerization, a chemical process that contributes to the formation of potentially toxic bisretinoids in the retina. Unlike simply lowering vitamin A exposure, the approach is intended to preserve vitamin A’s normal biological functions while reducing its tendency to form these byproducts.
Preclinical studies provided the rationale for testing the treatment in people. In an Abca4 knockout mouse model, deuterated vitamin A reduced lipofuscin accumulation and other abnormalities associated with Stargardt disease without impairing retinal function (Charbel Issa et al., 2015).
The TEASE-1 study was a 24-month, multicenter, double-masked, placebo-controlled randomized clinical trial conducted at seven US sites. Fifty participants were randomized, and 46 completed the study. Their mean age was 44.3 years, with a range of 18 to 60 years (Kay et al., 2026).
Participants were eligible from age 12, although no randomized participant in the reported study was younger than 18.
The 50 participants were assigned in a 1.5:1.5:1:1 ratio to one of four regimens: 14 mg of gildeuretinol daily, 24 mg daily, placebo for 24 months, or placebo for the first year followed by gildeuretinol during the second year.
Fifteen participants began at 14 mg, 15 began at 24 mg, 10 received placebo, and 10 entered the crossover arm. Participants initially assigned to 24 mg were switched to 14 mg after one year after pharmacokinetic analyses showed that the lower dose produced comparable replacement of native vitamin A.
For efficacy analyses, treatment exposure was handled according to the prespecified statistical plan. Crossover participants contributed to the untreated group while receiving placebo and to the gildeuretinol group after switching to active treatment.
The primary outcome was the rate of growth of retinal atrophic lesions measured using fundus autofluorescence imaging. Two masked graders assessed the retinal images, and optical coherence tomography was used to confirm atrophic lesions.
What Researchers Found
For the primary analysis, researchers compared gildeuretinol-treated observations with an untreated group that combined placebo-treated observations and natural-history data.
The model-estimated growth rate of square-root-transformed retinal atrophic lesion area was 0.182 millimeters per year with gildeuretinol and 0.232 millimeters per year in the untreated group.
The difference was 0.050 millimeters per year, with a 95% confidence interval from 0.027 to 0.072 millimeters per year in favor of treatment. That represented a 21.6% relative reduction in lesion growth, with P<.001 (Kay et al., 2026).
Because the primary comparison included nonrandomized natural-history cases, the trial’s prespecified randomized-only analysis is particularly important.
When researchers restricted the analysis to randomized participants, the estimated growth rate was 0.206 millimeters per year during gildeuretinol treatment and 0.242 millimeters per year during placebo treatment.
That difference of 0.036 millimeters per year represented a 14.9% relative reduction. The 95% confidence interval for the difference ranged from 0.009 to 0.062 millimeters per year, and the result was statistically significant at P=.008 (Kay et al., 2026).
This analysis included 36 randomized participants with evaluable data during gildeuretinol treatment and 17 with evaluable data during placebo treatment. Those figures reflect treatment periods rather than a conventional parallel comparison between two separate groups, because crossover participants could contribute observations to both conditions at different points in the study.
A second prespecified sensitivity analysis used the untransformed lesion area. In that analysis, retinal atrophy expanded by 0.867 square millimeters per year with gildeuretinol and 1.230 square millimeters per year in the untreated group, a difference of 0.363 square millimeters per year and a 29.5% relative reduction.
The anatomical results therefore pointed consistently in the same direction.
Visual acuity told a different story.
After 24 months, best-corrected visual acuity had changed by an average of minus 1.6 letters with gildeuretinol and minus 1.1 letters with placebo. Researchers reported no significant difference between the groups at month 24 or at any other study visit (Kay et al., 2026).
This means the trial showed slower expansion of retinal atrophy, not demonstrated preservation of visual acuity.
How It Compares With Earlier Research
The treatment strategy is based on the biology of ABCA4-related disease.
When ABCA4 function is impaired, vitamin A-related compounds can accumulate and contribute to formation of bisretinoids and lipofuscin in the retinal pigment epithelium. Progressive damage to photoreceptors and retinal pigment epithelium is central to Stargardt disease (Tanna et al., 2017).
Animal research previously showed that replacing conventional vitamin A with deuterated vitamin A could slow vitamin A dimerization and reduce retinal abnormalities in Abca4 knockout mice (Charbel Issa et al., 2015).
TEASE-1 is important because it moves that mechanism from preclinical models into controlled human evidence and shows an effect on the expansion of retinal lesions.
The drug has also been tested in a different retinal disorder. In the 198-participant SAGA trial involving people aged 60 or older with geographic atrophy from age-related macular degeneration, gildeuretinol did not meet its prespecified primary endpoint. Geographic atrophy grew by 1.62 square millimeters per year with treatment and 1.87 square millimeters per year with placebo, corresponding to a 13.4% relative reduction and P=.075 (Boyer et al., 2026).
Some prespecified secondary and sensitivity analyses in that trial favored gildeuretinol, but the investigators concluded that further adequately powered trials were needed.
The SAGA findings do not negate the Stargardt result because the diseases and populations differ. They do reinforce the need to establish clinical benefit independently in each condition.
What the Results May Mean
Retinal atrophic lesions represent irreversible structural damage, so slowing their enlargement could be valuable if the effect persists and ultimately preserves useful vision.
That second step has not yet been demonstrated.
Visual acuity can be an imperfect and slow-changing measure in Stargardt disease. Some patients use areas of retina outside the damaged fovea for fixation, and visual acuity may remain relatively stable even while atrophy expands.
This is one reason the investigators used retinal lesion growth rather than visual acuity as the primary endpoint.
Still, patients ultimately care about how well they can see and function, not only what retinal imaging shows. Longer and larger studies will need to determine whether slower atrophy translates into preservation of central vision, reading ability, visual fields, mobility, or other meaningful aspects of daily life.
The accompanying JAMA Ophthalmology commentary described the study as an important step in efforts to turn understanding of Stargardt disease biology into therapy while noting that no approved treatments currently exist (Huckfeldt & Comander, 2026).
The results should also not be interpreted as evidence that ordinary vitamin A supplements can treat Stargardt disease. Gildeuretinol is a specifically engineered drug designed to alter the chemistry of vitamin A. It is not conventional supplemental vitamin A.
Safety Findings
Most treatment-emergent adverse events were classified as mild or moderate, and the investigators reported that adverse events were similarly distributed across study cohorts.
Three serious adverse events occurred, but investigators judged all three unrelated to the study drug. One participant died in a road traffic crash.
There were no treatment-related serious adverse events, no clinically significant liver abnormalities attributed to treatment, and no participant-reported night blindness or difficulty adapting to darkness (Kay et al., 2026).
One safety event deserves particular attention. A participant receiving the 24-mg dose developed transient visual obscuration, blurred optic-disc margins and increased intracranial pressure at the 12-month visit. The participant was diagnosed with grade 1 papilledema.
The investigators considered the symptoms potentially related to excess vitamin A activity, although measured vitamin A concentrations remained within the normal range. The event resolved after temporary interruption of gildeuretinol and treatment with acetazolamide. The participant later resumed the study drug and completed the trial without recurrence.
With only 50 randomized participants, the study cannot reliably rule out uncommon or delayed adverse effects.
Limitations
The trial’s small size is a major limitation. Fifty people were randomized, and only 36 participants with evaluable data contributed observations during gildeuretinol treatment to the primary efficacy analysis.
The primary comparison also incorporated 54 eligible natural-history cases to enlarge the untreated control group. Natural-history controls can be useful in rare-disease research, where recruiting large placebo groups is difficult, but they do not provide the same protection against bias as concurrent randomization.
The investigators attempted to address this issue with a prespecified analysis restricted to randomized participants. That analysis still favored gildeuretinol, but there were only 17 randomized participants with evaluable data during placebo treatment.
Another methodological point is that the trial was designed as an early phase 2 study and used a prespecified two-sided significance level of 0.20 for its primary efficacy endpoint. That threshold is considerably less stringent than the conventional 0.05 level.
In this case, however, the primary result had P<.001 and the randomized-only sensitivity analysis had P=.008, so both results also crossed the conventional 0.05 threshold.
The most clinically important limitation remains the absence of demonstrated visual benefit. Best-corrected visual acuity did not differ significantly between treatment and placebo during the two-year study.
The participants also already had retinal atrophic lesions. The findings therefore cannot establish whether treating patients earlier in the disease process would have a different effect.
There is also substantial industry involvement to consider. Alkeus Pharmaceuticals and the US Food and Drug Administration’s Office of Orphan Products Development provided financial support. Alkeus designed the study and participated in interpretation of the results, manuscript preparation, approval of the manuscript, and the decision to submit it for publication.
Several authors reported financial relationships with Alkeus. Corresponding author Leonide Saad reported a pending patent involving gildeuretinol for Stargardt disease, and other investigators reported company grants, personal fees or clinical trial funding (Kay et al., 2026).
These disclosures do not determine whether the findings are valid, but they make independent confirmation especially valuable.
Final Thoughts
TEASE-1 provides promising controlled human evidence that gildeuretinol acetate can slow the expansion of retinal atrophic lesions in Stargardt disease.
The primary analysis found a 21.6% relative reduction in lesion growth compared with an augmented untreated group. A prespecified analysis restricted to randomized participants found retinal atrophy grew 14.9% more slowly during gildeuretinol treatment than during placebo treatment.
For an inherited retinal disease with no approved therapy, that is an encouraging biological signal.
It is not yet evidence that the drug preserves sight.
Visual acuity did not differ significantly between treatment and placebo over 24 months, and the trial was too small to establish uncommon safety risks or the long-term clinical importance of slower lesion growth.
The next question is therefore more consequential than whether gildeuretinol changes retinal images. Larger and longer studies need to determine whether slowing retinal atrophy allows people with Stargardt disease to retain useful vision for longer.
References
Boyer, D. S., Saad, L., Washington, I., Melamud, A., Graham, K. B., O’Malley, A. E., DeBartolomeo, G. M., Jiang, H., Harris, L. D., Kay, C. N., Ferrone, P. J., Khan, S., & SAGA Study Team. (2026). Gildeuretinol acetate versus placebo for geographic atrophy secondary to age-related macular degeneration: The Study of ALK-001 in Geographic Atrophy randomized clinical trial. Ophthalmology Retina. Advance online publication. https://doi.org/10.1016/j.oret.2026.05.018
Charbel Issa, P., Barnard, A. R., Herrmann, P., Washington, I., & MacLaren, R. E. (2015). Rescue of the Stargardt phenotype in Abca4 knockout mice through inhibition of vitamin A dimerization. Proceedings of the National Academy of Sciences of the United States of America, 112(27), 8415–8420. https://doi.org/10.1073/pnas.1506960112
Huckfeldt, R. M., & Comander, J. I. (2026). Deuterated vitamin A for Stargardt disease in TEASE-1 trial. JAMA Ophthalmology. Advance online publication. https://doi.org/10.1001/jamaophthalmol.2026.3926
Kay, C. N., Saad, L., DeBartolomeo, G., Bressler, N., Tsang, S. H., Stepien, K., Bernstein, P., Lam, B. L., Washington, I., & Gorin, M. B. (2026). Safety and effects of gildeuretinol acetate on retinal atrophic lesions in Stargardt disease: The TEASE-1 randomized clinical trial. JAMA Ophthalmology. Advance online publication. https://doi.org/10.1001/jamaophthalmol.2026.3662
Tanna, P., Strauss, R. W., Fujinami, K., & Michaelides, M. (2017). Stargardt disease: Clinical features, molecular genetics, animal models and therapeutic options. British Journal of Ophthalmology, 101(1), 25–30. https://doi.org/10.1136/bjophthalmol-2016-308823
