Subretinal timrepigene emparvovec in adult men with choroideremia: a randomized phase 3 trial

Choroideremia is a rare, X-linked retinal degeneration resulting in progressive vision loss. A randomized, masked, phase 3 clinical trial evaluated the safety and efficacy over 12 months of follow-up in adult males with choroideremia randomized to receive a high-dose (1.0 × 1011 vector genomes (vg); n = 69) or low-dose (1.0 × 1010 vg; n = 34) subretinal injection of the AAV2-vector-based gene therapy timrepigene emparvovec versus non-treated control (n = 66). Most treatment-emergent adverse events were mild or moderate. The trial did not meet its primary endpoint of best-corrected visual acuity (BCVA) improvement. In the primary endpoint analysis, three of 65 participants (5%) in the high-dose group, one of 34 (3%) participants in the low-dose group and zero of 62 (0%) participants in the control group had ≥15-letter Early Treatment Diabetic Retinopathy Study (ETDRS) improvement from baseline BCVA at 12 months (high dose, P = 0.245 versus control; low dose, P = 0.354 versus control). As the primary endpoint was not met, key secondary endpoints were not tested for significance. In a key secondary endpoint, nine of 65 (14%), six of 35 (18%) and one of 62 (2%) participants in the high-dose, low-dose and control groups, respectively, experienced ≥10-letter ETDRS improvement from baseline BCVA at 12 months. Potential opportunities to enhance future gene therapy studies for choroideremia include optimization of entry criteria (more preserved retinal area), surgical techniques and clinical endpoints. EudraCT registration: 2015-003958-41.

Ocular conditions are optimal for gene therapies.Retinal cells are post-mitotic, enabling sustained gene expression without the need for genomic integration of transgenic material; the blood-ocular barrier facilitates immune privilege, limiting immunological response to gene therapy products; and gene transduction may be achieved at a low dose, potentially reducing manufacturing burden 18,19 .Viral vector-based gene therapy is being widely studied in both preclinical and clinical settings for the treatment of choroideremia and other inherited retinal dystrophies 6,20 .Adeno-associated viral vectors, such as adeno-associated virus serotype 2 (AAV2), have been shown to achieve efficient transduction of photoreceptors and RPE after subretinal injection and have an acceptable safety profile 19,[21][22][23][24] .The first ocular gene therapy approved by the US Food and Drug Administration (FDA) uses an AAV2 vector delivered via subretinal injection 19 .Clinical data suggest that AAV2 vectors have no long-term retinal toxicity at the subretinal dose range of 1.0 × 10 10 to 1.0 × 10 11 vector genomes (vg) and, in addition to high specificity for RPE transduction, may be able to target rod photoreceptors more effectively than some other AAV serotypes.Inclusion of the inactivated woodchuck hepatitis virus post-transcriptional regulatory element (WPRE) may further boost retinal gene expression of the REP1 protein by up to a log unit in some cases 25 .These observations of both safety and tropism efficacy provide the optimal rationale for the AAV2 choroideremia gene therapeutic strategy 19,23 .
Timrepigene emparvovec (BIIB111/AAV2-REP1) is an AAV2 vector-based gene therapy encoding the wild-type CHM cDNA sequence driven by the strong ubiquitous CAG promoter and augmented by an inactivated WPRE sequence 26 .By restoring absent REP1 expression, timrepigene emparvovec aims to address the underlying genetic cause of choroideremia 23 .Data from phase 1/2 studies have demonstrated that timrepigene emparvovec-based gene therapy improved when there are very low levels of correctly spliced mRNA; as little as 2-5% of the wild-type transcript levels may significantly attenuate disease progression 7 .The condition is likely underdiagnosed because of its similarities in early stages to other inherited retinal diseases, such as retinitis pigmentosa 2,8 .Choroideremia initially presents in childhood and early adolescence as night blindness [9][10][11] .Slow and progressive vision loss associated with choroideremia starts from the periphery of the visual field, and best-corrected visual acuity (BCVA) decreases as the disease advances with age [12][13][14] .In a retrospective study of 71 males with choroideremia, the average age of onset of night blindness symptoms was reported to be 12.6 years (±1.0 year), with loss of peripheral vision at 19.7 years (±1.3 years) (ref.9).Individuals with choroideremia generally retain good central vision until approximately 40 years of age, followed by a rapid reduction of visual acuity in the advanced stage as degeneration starts impacting the fovea (for representative retinal images from a patient with advanced choroideremia, see Fig. 1) 15 .This impairment phase may last 5-10 years until vision is no longer recordable and provides the only potential period during which visual acuity changes might be assessed against a potential treatment 16 .
Choroideremia is caused by mutations in the CHM gene, which encodes Rab escort protein 1 (REP1) (refs.3,17).CHM mutations decrease REP1 expression, leading to degeneration of the retinal pigment epithelium (RPE), photoreceptors and choroid 3,17 .REP1 serves as a mediator of intracellular trafficking of prenylated Rab proteins in the retina and RPE 3 .Most CHM gene mutations responsible for clinical phenotypes cause loss of function either via deletion or nonsense mutations 3 .Missense mutations in the CHM gene have also been reported on rare occasions and may result in decreased levels of REP1 expression and protein structure destabilization 3 .

Article
https://doi.org/10.1038/s41591-023-02520-3visual acuity in a subset of patients with choroideremia who received treatment [26][27][28][29] .In most patients after treatment, BCVA in the study eye improved or remained stable, and, for patients with moderate vision loss at baseline (that is, 34-73 Early Treatment Diabetic Retinopathy Study (ETDRS) letters), higher mean gains in vision at 24 months were observed (5.6 letters) compared to all patients (3.1 letters) 30 .Three patients (9%) achieved and maintained a clinically significant gain of ≥15 ETDRS letters at 24 months 30 .Here we report the results from the randomized, parallel-controlled, phase 3 STAR clinical trial that evaluated the efficacy and safety of timrepigene emparvovec versus a non-surgical control in adult males with genetically confirmed diagnosis of choroideremia.

Participant disposition and baseline characteristics
A total of 169 participants were randomized (high dose, n = 69; low dose, n = 34; control, n = 66), and 164 participants completed their surgery or attended a post-baseline visit in the study (high dose, n = 65; low dose, n = 34; control, n = 65) (Fig. 2).Demographics were generally balanced across the three study groups (Table 1).Most participants were White, and most were 40-60 years of age.Participants across study groups were well distributed in the study sites located in Finland, Germany, the United Kingdom and the United States.

Safety
The overall incidence of treatment-emergent adverse events (TEAEs) was generally higher in the timrepigene emparvovec groups (59/65 participants in the high-dose group (91%); 32/34 participants in the low-dose group (94%)) than in the control group (33/65 participants (51%)) (Table 2).No participants died or discontinued from the study because of TEAEs.Most TEAEs were mild or moderate in severity.Ocular TEAEs and severe ocular TEAEs occurred more frequently in the treated participants compared to the control group and more commonly in the study eye than in the fellow eye for treated groups.Ocular inflammation-related TEAEs and visual acuity-related TEAEs were common and occurred more frequently in the treated groups.The incidence (n (%)) of ocular inflammation-related TEAEs in the high-dose, low-dose and control groups was 33/65 (51%), 16/34 (47%) and 1/65 (2%), respectively.One participant (2%) in the high-dose group experienced a serious ocular inflammation-related TEAE (non-infective retinitis).Visual acuity reduction events were the most reported serious ocular TEAEs.Three of these visual acuity reduction events were related to the study drug, whereas seven of these events were related to the study procedure.Cataracts as TEAEs were observed more frequently in the high-dose group (9/65 (14%)) and low-dose group (4/34 (12%)) than in the control group (3/65 (5%)).Most participants in each treatment group had no shift in lens opacity grade from baseline to month 12, and changes in the lens opacity in the fellow eye for the same follow-up period for all study groups were similar to those in the study eye of the control group.No clinically meaningful changes in vital sign measurements were observed.

Primary efficacy measure
Proportion of participants with ≥15-letter ETDRS increase from baseline in BCVA at month 12.The primary endpoint (proportion of participants with a ≥15-letter ETDRS improvement from baseline in study eye BCVA at 12 months) was not statistically different between high-dose (n = 3/65 (5%)) and control (n = 0/62 (0%)) groups (P = 0.245; Fig. 3a).Although all comparisons for the key secondary endpoints would not be tested (per the hierarchical procedure) if efficacy were not claimed for the primary endpoint, those comparisons were conducted in an exploratory nature.The difference between the proportion of participants in the low-dose group (n = 1/34 (3%)) and the control group (n = 0/62 (0%)) experiencing ≥15-letter ETDRS improvement from baseline in study eye BCVA at 12 months was also not statistically significant (P = 0.354).
The volume of the blebs raised and the degree of reflux of vector into the vitreous would have been variable in these surgically challenging participants, which complicates the simple binary assumption of a low or high dose.Because the low dose is also known to be therapeutic 26 , it is not unexpected to find responders in this group.In a post hoc analysis pooling data from both treatment groups, n = 4/99 (4%) treated eyes gained ≥15-letter ETDRS improvement compared to n = 0/62 (0%) in the control group.

Key secondary efficacy measures
Change from baseline in BCVA score at month 12.The least squares (LS) mean difference (95% confidence interval (CI)) in the change from baseline in study eye BCVA at month 12 between the high-dose group and the control group was a gain of 2.1 (−2.0, 6.2) ETDRS letters, favoring the high dose (Table 3 and Fig. 3b).The LS mean difference in the change from baseline in study eye BCVA score at 12 months between the low-dose group and the control group was a gain of 0.9 ETDRS letters, favoring the low dose.https://doi.org/10.1038/s41591-023-02520-3

Proportion of participants with
improvement from baseline in study eye BCVA at month 12 compared to the control group (n = 1/62 (1.6%)) (Fig. 3c).

Proportion of participants with no decrease or <5-letter ETDRS letter decrease from baseline BCVA at month 12.
There was a greater proportion of participants with no decrease from baseline or a decrease of <5 ETDRS letters from baseline in study eye BCVA at month 12 in the high-dose group (n = 54/65 (83.1%)) and the low-dose group (n = 24/34 (70.6%)) compared to the control group (n = 42/62 (67.7%)) (Fig. 3d).Surgical detachment of the fovea is normally associated with a reduction in visual acuity, as is vector-related inflammation.Hence, the finding that the proportions of participants maintaining at least one line (that is, five letters) of ETDRS acuity in both high-dose (83%) and low-dose (71%) groups were numerically greater than in the unoperated control group (68%) does at least support the safety of the gene therapy treatment.

Other prespecified secondary efficacy measures
The LS mean differences in the change from baseline to month 12 in the study eye mean retinal sensitivity, bivariate contour ellipse area 63% and bivariate contour ellipse area 95% between the high-dose group and the control group were −0.1573 dB, 0.6606 deg 2 and −1.4877 deg 2 , respectively.The LS mean differences in the change from baseline to month 12 in the study eye mean retinal sensitivity, bivariate contour ellipse area 63% and bivariate contour ellipse area 95% between the low-dose group and the control group were 0.0478 dB, −0.1610 deg 2 and −4.4940 deg 2 , respectively.The microperimetry data, however, were generally found to be inconsistent, with most participants unable to perform the test accurately or scoring zero because of the advanced nature of their disease when the degeneration has undermined the fovea 31 .
At month 12, despite a difference in the change from baseline (95% CI) in the study eye total area of preserved autofluorescence (AF) between the high-dose and control groups (−0.1264 mm 2 (−0.2308, −0.0220)), as well as between the low-dose and control groups (−0.1541 mm 2 (0.2781, −0.0302)), the treatment groups showed a greater decrease, indicating a worsening condition relative Skin and subcutaneous tissue disorders LLVA, low luminance visual acuity; SOC, system organ class.a TEAEs were defined as AEs starting on or after the day of the surgery (or, for control group participants, visit 2, day 0).If a participant had multiple events of severity and outcome, then this participant was counted only once in the worst hierarchy in each category.However, participants could have been counted more than once in action taken.

Alternative data ascertainments due to coronavirus disease 2019
The impact of coronavirus disease 2019 (COVID-19) on usage of alternative data ascertainment and study completion was examined (Extended Data Tables 1 and 2).Overall, 32 participants (19%) had alternative data ascertainment (that is, assessment at local non-study sites or out-of-window visits) at month 12 due to COVID-19 travel restrictions.Most alternative data ascertainments were via extension of out-of-window visits.There were 10 COVID-19-related major protocol deviations, with six involving out-of-window visits and four pertaining to BCVA performed by unmasked assessors.Two participants were seen locally in Brazil because of travel restrictions to the United States.The sensitivity analyses assessing the impact of COVID-19 were conducted for the primary endpoint and the key secondary endpoints.The results were consistent with the primary analyses.

Discussion
The primary endpoint of a three-line gain was not met in this phase 3 trial.Although a spontaneous gain of three lines of vision was not observed in any of the control group participants, there were not enough participants gaining three lines of vision in the treatment groups to meet statistical significance.However, there were notable observations related to BCVA changes in participants undergoing retinal gene therapy for choroideremia in all visual acuity endpoints tested, including the proportion of participants reporting a two-line gain and mean gain in vision and preservation of at least one line of vision.
A ≥10-letter ETDRS improvement from baseline in BCVA has been considered clinically relevant from the patient's perspective according to scientific advice given at a workshop by the European Medicines Agency (EMA) 32 .A ≥10-letter ETDRS improvement would be more appropriate for younger patients with better baseline vision and more intact retinal structure who would likely be the most optimal patient for a future gene therapy.Although a three-line gain is typical for FDA 100 0/62, 0%   https://doi.org/10.1038/s41591-023-02520-3 approval, it might be challenging to achieve this in many inherited retinal degenerations with significant anatomical damage.This is not the case in age-related macular degeneration or diabetic retinopathy, for instance, as BCVA in these retinal degenerations drops because of fluid leakage in the retina, which can quickly be reversed with drug treatments.Considering that there is no approved treatment for choroideremia, stabilization or any improvement of vision in affected individuals could be considered beneficial.The proportion of individuals gaining three lines of vision in a meta-analysis of the phase 1/2 choroideremia trials was more than twice the rate observed in this pivotal STAR study 30 .The participants selected for the STAR study, however, had far more advanced disease than those in the earlier trials because of the low BCVA entry requirement.This would limit the BCVA gain potential, and the much thinner retina is likely to be more susceptible to surgically induced damage.This highlights the difficulty in obtaining homogeneous patient groups for clinical trials on rare single-gene retinal degenerations.For future trials, a two-line gain in vision may be a better endpoint given the more advanced stages of disease in participants able to be recruited to a trial because it would be more achievable and yet still likely be noted as a clinically significant change by most clinicians.Furthermore, the potential for a spontaneous two-line gain in the control group may have been less because of their more advanced disease (as was observed).
Although it was not considered for significance per the hierarchical procedure, it is interesting that there was a greater numerical difference in BCVA score from the high-dose group versus control than the low-dose group versus control (Table 3), despite there being a slightly worse mean baseline BCVA for the high-dose group compared to the control group.It is possible that the number of available vector genomes associated with the high dose offset the variability in surgery and dose administration, thus leading to the highest numerical mean gain of ETDRS letters among the three groups.
Although no efficacy claims can be made on the basis of this study, it is interesting to note that the treated eyes underwent iatrogenic retinal detachment in thin and degenerate tissues, which would ordinarily be associated with reduced visual acuity.Instead, there was a trend for improvements in visual acuity to occur more frequently in the high-dose group compared to the control group, albeit not enough to meet statistical significance in this cohort.Because of the low percentage of participants who achieved three lines of visual acuity gain across the study, including none in the control group, a larger study size would be needed to confirm the differences between the treatment groups and control group.Alternatively, the results from this cohort suggest that another trial with a two-line gain as the primary endpoint (if acceptable to the regulators) could be more appropriate for smaller enrollment populations, based on the power calculation derived from the current STAR study data.
The safety profile of timrepigene emparvovec was determined to be acceptable.Most TEAEs in the timrepigene emparvovec treatment groups were related to study procedure rather than to study drug.Most ocular inflammation-related TEAEs occurred within 30 d of study drug administration or surgery and were likely related to the surgical procedure.The occurrence of ocular inflammation-related and visual acuity reduced-related events was not dose dependent.Visual acuity reduction events were the most reported serious ocular TEAEs.
Natural history studies have reported a transition age from slow to rapid BCVA decline of ~39 years 15 .In a cohort of patients with choroideremia who were grouped by age <50 years or ≥50 years, visual acuity was observed not to change significantly in either age group over the 1-year follow-up period 33 .It can, therefore, be difficult to assess improvements over a 12-month period in patients if there is no measurable decline in vision in the control group over this period 34 .
The efficacy of AAV2-based gene therapy in younger patients is unknown.For instance, although improvement in BCVA was not demonstrated, it is possible that preservation of vision with timrepigene emparvovec is more likely to be seen in younger patients who are able to tolerate the risks associated with subretinal gene therapy administration.
To get a large gain in vision, three factors need to be aligned.First, the retina needs to be healthy enough to be able to improve visual acuity by two or more lines; second, the surgery needs to be completed without damaging the retinal architecture; and third, the vector needs to be sequestered at a high enough dose subretinally, without being refluxed back into the vitreous.With the high dose, this is more likely to be achieved, because the third factor is countered by having a log unit higher dose of vector particles.Nevertheless, the other factors remain because there will be some retinas that are simply too advanced to be able to achieve large gains in vision.Similarly, in others, there will be surgical complications that might offset the potential gains.It should be expected, though, that large gains in vision might be seen in some low-dose patients when all three factors go well.
For examining the dose effect more accurately, however, it is more logical to look at the effects on BCVA across the whole cohort because then data from every participant can be included, rather than from only the few who have large BCVA gains.If there were no beneficial effect at all from the vector, then one would predict that the mean BCVA loss from the retinal detachment would be similarly worse in both the high-dose and low-dose groups compared to the control group, because of the negative effects of iatrogenic retinal detachment.The observations of the mean BCVA change from this study, however, are the reverse.Although these values were not formally tested for significance per the hierarchical procedure, the direction and pattern of BCVA changes across the entire participant group with these subanalyses is consistent with a treatment effect that is better with the high dose, especially given that these participants underwent an invasive surgery.
To summarize some previous points in this discussion, this study had several limitations that would be important to address in future trials of gene therapy in choroideremia.COVID-19 resulted in some major protocol deviations in this trial as noted previously, including some out-of-window visits and performance of BCVA measurements by unmasked assessors.However, on the basis of consistent results of analyses of the per-protocol and intent-to-treat populations for the primary endpoint, the protocol deviations did not appear to have a substantial effect on the study outcomes.Additionally, the primary endpoint requiring a three-line gain in BCVA may not have been realistic in this cohort, and a two-line gain should be considered, especially in cohorts with more advanced disease.Participants in the treatment arm also underwent vitrectomy and pre-injection subretinal formation before subretinal injections, inducing a transient and localized detachment of the central retina.Variable success with the surgery could, therefore, potentially be an additional confounder for the treatment effect in participants with advanced choroideremia, as there may have been only a small preserved retinal area for vector delivery, which might not have been accurately targeted in every case 35 .Furthermore, the small preserved retinal area may not be normal given that BCVA is reduced and the area may not tolerate subretinal injection well compared to intervention in milder disease with better-preserved retinas 35,36 .Lastly, given the advanced disease of the participants in the STAR study, the microperimetry data were largely unreliable because patients need reasonably good BCVA and stable fixation to be able to perform the test accurately.There has been a recent uptake in the use of microperimetry in interventional retinal disease trials, including in choroideremia, and it has potential to succeed in a future trial as a clinical endpoint for earlier-stage patients 34 .
Although the primary endpoint was not met in this pivotal trial, gene therapy for choroideremia remains a promising therapeutic approach.Furthermore, visual acuity gains lower than the prespecified threshold for primary endpoint can still be clinically meaningful.However, for advanced disease with a rapid decline in visual acuity, perhaps a more realistic expectation may be to slow down Article https://doi.org/10.1038/s41591-023-02520-3 Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material.If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/.© The Author(s) 2023 https://doi.org/10.1038/s41591-023-02520-3excluded both the participants for whom visual acuity readings were so poor that they became unreliable and participants for whom visual acuity readings were so good that there would be no potential for a three-line gain without hitting a ceiling of 6/5 or 20/16 Snellen acuity.
Exclusion criteria.Participants were not eligible for study participation if they had a history of amblyopia in the eligible eye or were unwilling to use barrier contraception methods or abstain from sexual intercourse for 3 months if treated with timrepigene emparvovec, as is standard for gene therapy trials.Furthermore, participants should not have had a previous intraocular surgery in the study eye within 3 months of the first visit or any significant ocular or non-ocular disorder that, in the opinion of the investigator, might put the participant at risk, influence the results of the study or affect the ability of the individual to participate in the study.This includes, but was not limited to, individuals with a contraindication to an oral corticosteroid (for example, prednisolone/prednisone), with a clinically significant cataract and who, in the clinical opinion of the investigator, was not an appropriate candidate for subretinal surgery.To be eligible for participation, individuals should also not have taken part in another research study involving an investigational product in the past 12 weeks or received a gene or cell therapy at any time in the past.
Individuals with advanced choroideremia can have variable BCVA readings when the disease causes partial collapse of the fovea-a concept referred to as 'foveal splitting' by the investigators.For this reason, participants were recruited from the natural history of choroideremia (NIGHT) study group (ClinicalTrials.gov:NCT03359551), which allowed identification and exclusion of individuals with variable BCVA.Individuals from the NIGHT study whose baseline value at visit 1 was ≥10 letters different in the study eye compared to the previous NIGHT study visit, as well as all individuals who were not recruited from the NIGHT study, underwent three baseline BCVA readings, with the highest reading selected to determine eligibility for the STAR study.At least two of the three values were required to meet eligibility requirements, and the difference between the three assessments could not be ≥10 letters.A BCVA reading was not repeated for those recruited from the NIGHT study whose BCVA on day 1 was <10 letters different from the previous NIGHT study visit.Several individuals who had stable BCVA in one eye but variable readings in the fellow eye could still be recruited into the trial but only with the stable BCVA eye.Others who had a fellow eye outside the trial inclusion criteria range of BCVA were also recruited.Hence, although these individuals were entered into the STAR study for the stable eye, the asymmetric nature of the end-stage choroideremia in a large proportion of participants meant that the fellow eye was not a suitable control.For this reason, the randomization after recruitment included a non-operated control group.

Interventions and cohorts
Participants were randomized in a 2:1:2 ratio at baseline to receive a volume of up to 100 μl subretinally of a high dose of timrepigene emparvovec (1.0 × 10 11 vg), a low dose of timrepigene emparvovec (1.0 × 10 10 vg) or no treatment.The dose range of vector employed was based on previous clinical trials using the AAV2 vector with a chicken β-actin promoter 26,37 and investigator-driven clinical studies in which AAV2-REP1 was administered to patients with choroideremia [26][27][28]38 . Forindividuals randomized to receive treatment, timrepigene emparvovec was administered at the planned dose as a subretinal injection targeting the preserved retinal region of the macula via vitrectomy and after formation of a subretinal bleb using balanced salt solution on the surgical date (day 0) (ref.39).Participants who received timrepigene emparvovec were given a 21-d course of oral corticosteroid to prevent potential inflammation resulting from surgery and immune responses, beginning 2 d before the study dose.

Endpoints
Primary and key secondary endpoints.The primary endpoint of the study was the proportion of participants with a ≥15-letter improvement (amended from ≥10-letter improvement, in accordance with US regulatory requirements) from baseline in BCVA at 12 months as measured by the ETDRS chart.The key secondary endpoints were the mean change from baseline in BCVA at 12 months measured by the ETDRS chart, the proportion of participants with a ≥10-letter ETDRS improvement from baseline in BCVA and the proportion of participants with no decrease in BCVA from baseline or a decrease of <5 ETDRS letters from baseline at 12 months.Safety endpoints.The safety-related assessments included overall AEs, SAEs and AEs or SAEs leading to discontinuations from the clinical trial.
Additional secondary endpoints.Other secondary endpoints included the change from baseline to month 12 in the following measures: BCVA, total area of preserved AF, area of preserved ellipsoid zone, microperimetry, contrast sensitivity score, Color Vision Total Error score, reading speed and VFQ-25 score.Change of BCVA from baseline at months 4 and 8 were also secondary endpoints, but the results were not included in this report.Fundus AF was performed to evaluate the changes in the area of viable retinal tissue.Contrast sensitivity was measured before pupil dilation using a Pelli-Robson chart.Color vision was tested separately before pupil dilation.The International Reading Speed Texts (IReST) was used to evaluate reading speed.Self-reported vision-targeted health status responses (individual, subscale and overall composite scores) were obtained using the VFQ-25 questionnaire.

Statistical methods
Sample size.Sample size estimation was performed using Fisher's exact test.Considering that choroideremia is a degenerative disease, it was assumed that a ≥15-letter BCVA gain would not be observed in participants without treatment.Assuming that 16.7% of the treated participants would gain ≥15 letters in BCVA at 12 months, 56 participants in the high-dose group and the control group would provide ≥90% power at a 0.05 level of significance with a two-sided test.To be conservative, 64 participants in the high-dose group and 64 participants in the control group were needed to ensure 85% power in case one participant in the untreated control group had ≥15-letter BCVA gain by chance, which corresponded to a total of 160 participants completing the study (64 participants in the high-dose group, 32 in the low-dose group and 64 in the control group).

Analysis of outcomes.
All analyses and summaries were produced using SAS version 9.4 or higher.Primary and key secondary efficacy endpoints were tested under a hierarchical procedure to maintain the type I error for the comparison between the high-dose group and the control group.Nominal P values were calculated for comparisons of the high-dose group or low-dose group versus the control group, with a prespecified threshold of significance set at 0.05.Statistical tests and 95% CIs were two-sided.The primary efficacy endpoint was tested first, and, if the P value was less than 0.05, then the key secondary endpoints would be tested in the following prespecified order: change from baseline in BCVA at month 12, proportion of participants with a ≥10-letter improvement from baseline in BCVA at month 12 and proportion of participants with either no decrease or a <5-letter decrease from baseline in BCVA at month 12.
Analysis of the primary endpoint was based on the intent-to-treat population, defined as all participants who were randomized, completed visit 2 (that is, received the study treatment or received a phone call (if in control group)) and had at least one post-treatment BCVA measurement.Change from baseline BCVA score was compared between the study groups (that is, high dose versus control and low dose versus control) using Fisher's exact test supported by a Fisher's exact Boschloo test with a Berger-Boos correction of beta = 0.001, in which the reported P value was two times the one-sided P value to maintain the test at 0.05 two-sided level.Results were further described over time using summary statistics for categorical data, including counts, percentages and 95% CI.Missing data were imputed as failures.

Fig. 1 |
Fig. 1 | Retinal images and microperimetry plot from a patient with advanced choroideremia.a, AF imaging reveals the fluorescent shapes that represent the surviving retinal pigment epithelium centrally (green arrows).b, The microperimetry plot shows that the central triangular area of AF is broadly correlated with the surviving visual field (red dots).c, The green box shows the region of optical coherence tomography scan, with the scan along the green arrow shown in d. d, The surviving outer nuclear layer is the area above the green arrows, with disruption of the outer segments indicating early degeneration.

Fig. 3 |
Fig. 3 | Visual acuity changes in trial participants.a-d, In all cases, the eyes treated with gene therapy had numerically better outcomes than unoperated control eyes, both in terms of the proportion gaining lines of vision and in the mean changes in visual acuity.Proportion of participants with ≥15-letter ETDRS improvement from baseline (a), LS mean change from baseline (b), proportion of participants with ≥10-letter ETDRS improvement from baseline (c) and https://doi.org/10.1038/s41591-023-02520-3Extended Data Fig. 1 | Study design for the STAR phase 3 trial.IOP, intraocular pressure; SLE, slit lamp examination.Extended Data Table 1 | Impact of COVID-19 on patient disposition among all randomized participantsExtended Data Table2| Sensitivity analysis of COVID-19 impact on change from baseline in BCVA in study eye at month 12 for the intent-to-treat population a

Table 1 (continued) | Participant demographics and baseline characteristics: safety population Table 1 | Participant demographics and baseline characteristics: safety population Timrepigene emparvovec
Age was calculated as the number of years between the date of birth and the informed consent date.b Self-reported.The LS mean difference (95% CI) in the change from baseline in the study eye reading speed at month 12 as compared to the control group was 22.1 (−7.9, 52.1) words per minute for the high-dose group and 27.4 (−7.8, 62.7) words per minute for the low-dose group, and neither difference was noteworthy versus the control group.At month 12, the LS mean difference (95% CI) in the change from baseline in the Visual Function Questionnaire 25 (VFQ-25) composite score was 3.4934 (0.0304, 6.9565) for the high-dose group versus the control group and 4.4207 (0.3095, 8.5319) for the low-dose group versus the control group. a