Efficacy and Safety of Retinal Gene Therapy Using Adeno-Associated Virus Vector for Patients With Choroideremia: A Randomized Clinical Trial

Key Points

Question  What are the efficacy and safety outcomes 24 months after gene therapy using an adeno-associated virus vector designed to deliver a functional version of the CHM gene for the treatment of choroideremia?

Findings  In this phase 2, open-label randomized clinical trial of 6 patients, no statistically significant differences between treated and control eyes were identified for mean change in best-corrected visual acuity or retinal sensitivity. Safety outcomes were consistent with vitrectomy.

Meaning  Gene therapy with adeno-associated virus vector designed to deliver a functional version of the CHM gene was associated with maintenance or improvement of visual acuity.

Importance  Choroideremia (CHM) is a rare, degenerative, genetic retinal disorder resulting from mutation of the CHM gene, leading to an absence of functional ras-associated binding escort protein 1 (REP1). There is currently no approved treatment for CHM.

Objective  To assess the safety and efficacy of retinal gene therapy with an adeno-associated virus vector (AAV2) designed to deliver a functional version of the CHM gene (AAV2-REP1) for treatment of patients with choroideremia.

Design, Setting, and Participants  Tübingen Choroideremia Gene Therapy (THOR) was a single-center, phase 2, open-label randomized clinical trial. Data were collected from January 11, 2016, to February 26, 2018. Twenty-four–month data are reported for 6 men with a molecularly confirmed diagnosis of CHM. Intention-to-treat analysis was used.

Interventions  Patients received AAV2-REP1 by a single, 0.1-mL subretinal injection of 10¹¹ genome particles during vitrectomy into 1 eye randomly assigned to receive treatment.

Main Outcomes and Measures  Primary end point was change in best-corrected visual acuity (BCVA) on the Early Treatment Diabetic Retinopathy Study chart from baseline to month 24 in the treated eye vs the control eye. Secondary end points included microperimetry variables, change in fundus autofluorescence, and spectral-domain optical coherence tomographic evaluations from baseline to month 24 in the treated eye vs the control eye.

Results  On enrollment, the mean (SD) age of the 6 men included in the study was 54.9 (4.1) years. The mean (SD) BCVA score was 60.3 (13.4) (approximately 20/63 Snellen equivalent) in the study eyes and 69.3 (20.6) (approximately 20/40 Snellen equivalent) in the control eyes. At 24 months, the BCVA change was 3.7 (7.5) in the treated eyes and 0.0 (5.1) in the control eyes (difference, 3.7; 95% CI, −7.2 to 14.5; P = .43). Mean change in retinal sensitivity was 10.3 (5.5) dB in the treated eyes and 9.7 (4.9) dB in the control eyes (difference, 0.6; 95% CI, −10.2 to 11.4; P = .74). A total of 28 adverse events were reported; all were consistent with the surgical procedure (eg, conjunctival hyperemia, foreign body sensation), and none were regarded as severe.

Conclusions and Relevance  Among 6 participants, gene therapy with AAV2-REP1 was associated with maintenance or improvement of visual acuity, although no significant difference was found from control eyes. All safety issues were associated with the surgical procedure and none were judged severe. Continued investigations could more precisely define the efficacy and safety of gene therapy with AAV2-REP1 in CHM.

Trial Registration  ClinicalTrials.gov identifier: NCT02671539

Choroideremia (CHM) is an X-linked disorder of the retina and choroid, affecting approximately 1 in 50 000 individuals.¹ Choroideremia presents in childhood as nyctalopia, followed by progressive constriction of visual fields, leading to vision loss in early adulthood and ultimately total blindness.^2,3 The disorder is caused by deletion or mutation of the CHM gene (OMIM 300390) that encodes ras-associated binding (Rab) escort protein 1 (REP1), triggering retinal epithelium cell death and photoreceptor degeneration, resulting in progressive loss of visual acuity (VA).^1,3-5 Choroideremia is characterized by photoreceptor dysfunction, retinal pigment epithelium depigmentation, neuronal cell death, and retinal remodeling.^3,4

There are currently no approved treatment options for CHM. There is evidence that loss of VA may have a reversible component, which could enable some patients to experience improvements in VA with treatment; however, the aim for gene therapy is to stop or slow this degeneration to maintain VA.¹

Choroideremia is a good candidate for gene therapy owing to the small size of the CHM gene, enabling packaging within an adeno-associated virus 2 (AAV2) capsid¹ and the easy access and immune-privileged state of the eye.^6,7 The first human clinical trial, in which an AAV2 vector encoding REP1 (NSR-REP1; Nightstar Therapeutics) was administered in 6 patients with CHM, was reported in 2014.¹ Two patients with advanced CHM and low baseline best-corrected VA (BCVA) on the Early Treatment Diabetic Retinopathy Study chart gained 21 and 11 letters in their treated eyes, which was sustained at 3.5 years of follow-up.⁸ Of the other 4 patients with good VA at baseline, VA was maintained at 6 months¹; in 3 of the patients, this VA was sustained at 3.5 years.⁸ The fourth patient, who experienced a decline in VA in both eyes, had received a lower dose of vector than expected and the decline was assumed to be a result of foveal degeneration.^1,8

The present study was designed to assess the safety and efficacy of AAV2-REP1 in 6 patients with CHM. This study is part of a larger, investigator-initiated program of phase 2 studies being conducted in the United States (NCT02553135), Canada (NCT02077361),⁹ and the United Kingdom (NCT02407678). Herein, we report the final 24-month data from all 6 patients treated in Tübingen, Germany.

THOR (Tübingen Choroideremia Gene Therapy) was a 24-month, phase 2, open-label, single-center randomized clinical trial using AAV2-REP1; it has been described previously.¹⁰ A single, 0.1-mL injection of 10¹¹ genome particles of the gene therapy medicinal product AAV2-REP1 (Nightstar Therapeutics) was administered subretinally into the study eye of patients with CHM during vitrectomy using a 2-step technique as previously described.^1,11 A prior study had shown this dose to be effective and well tolerated.⁸ The injection was administered on day 0 and patients were followed up for 24 months after surgery. Oral prednisone, 1.0 mg/kg, was administered to patients from day −1 until day 19 to reduce the risk of an immune response to the vector. From day −1, moxifloxacin, 0.5%, and dexamethasone, 0.5%, eyedrops were administered 4 times per day for 21 days. The untreated fellow eye served as the study control.

The study was designed and conducted in accordance with Good Clinical Practice and all other relevant regulatory requirements. Ethical approval was provided by the ethics committee of the University of Tübingen Faculty of Medicine, Tübingen, Germany, and approval was provided by the Regulatory Authority at the Paul-Ehrlich-Institute, Langen, Germany. Written informed consent was obtained from all patients prior to participation in the study; participants received travel reimbursement. The study protocol is available in Supplement 1.

Preparation of the clinical grade AAV2-REP1 vector has been described previously.¹ Briefly, the expression cassette consists of a ubiquitous promoter driving the CHM complementary DNA.

The primary end point was change in BCVA from baseline to month 24 in the treated eye vs the control eye. Secondary end points included microperimetry variables, change in the area of remaining fundus autofluorescence (FAF), and spectral-domain optical coherence tomographic (SD-OCT) evaluations (ellipsoid zone [EZ] and subfoveal choroidal thickness [SFCT]) from baseline to month 24 in the treated eye vs the control eye. Safety assessments were performed throughout the study.

Men who were 18 years or older with a clinical and confirmed genetic diagnosis of CHM were eligible for the study. Eligible patients had clinically visible active disease within the macular region, BCVA letter score of 78 to 34 (Snellen equivalent 20/32 to 20/300; logMAR, 0.14 to 1.02) in the study eye. Exclusion criteria included a history of amblyopia in the treated eye, any other genetic mutation leading to a pathologic retinal condition, intraocular surgery within 6 months or use of oral corticosteroids within 14 days before study entry, any ocular morbidity that confounded use of the fellow eye as a long-term control, unwillingness to use barrier contraceptive methods, high fever or high-fever disease, a history of autoimmune conditions and/or other systemic diseases that may have ocular manifestations, and neurodegenerative conditions. Patients with any other ocular or nonocular diseases or disorders or retinal surgery that put the patient at risk because of study participation or may have influenced study results or the patient’s ability to participate were also excluded.

To minimize selection bias, patients had 1 eye randomly assigned to receive treatment, and the fellow eye served as the study control (Figure 1). Randomization was performed by the principal investigator (M.D.F.) after informed consent was completed and was based on randomly permuted blocks using the random allocation sequence generator.¹² Fundus autofluorescence, microperimetry, and SD-OCT were conducted by an appropriately qualified masked examiner to minimize bias in evaluation of treated and control eyes. In addition, BCVA was assessed by a masked assessor who was not part of the core study team and did not have any other involvement in the trial.

Clinical assessments were conducted at day 7 and months 1, 3, 6, 9, 12, 18, and 24 following surgery. Early Treatment Diabetic Retinopathy Study (ETDRS) vision charts were used to determine BCVA, and a basic ophthalmic evaluation (slitlamp) was performed. Fundus-controlled microperimetry (MAIA; Centervue SpA) was used to evaluate visual fields, using a 10-2 (68 stimuli) grid after 30 minutes of dark adaptation as described previously.^11,13 Fundus autofluorescence assessments and SD-OCT were conducted (Spectralis HRA + OCT system; Heidelberg Engineering) as described previously.^11,13 Change in FAF was measured by quantifying the area of remaining FAF at each visit and calculating the change in the area over time. To determine the area of preserved EZ, the extent of intact EZ line was measured in 97 consecutive B-scans of a 20° × 20° volume scan by masked investigators at the Doheny Image Reading Center, Los Angeles, California.

The safety population comprised all patients who received at least 1 dose of AAV2-REP1 or corticosteroids and for whom at least 1 posttherapy safety assessment was available. The intention-to-treat population included all patients who were randomized and received AAV2-REP1 and had at least 1 postbaseline assessment. No formal sample size calculation was done. The aim of our study was to contribute to a number of investigator-initiated trials.^9,11,14 The number and proportion of patients in each category was calculated with 95% CIs for categorical/binary data. For continuous data, the mean, 95% CI, and SD were calculated. For intergroup analysis of microperimetry data, a repeated-measures analysis of variance with Bonferroni correction was used. With 1-sided testing, findings were considered significant at P = .05.

Six men with genetically confirmed CHM were enrolled (Figure 1). The mean (SD) age was 54.9 (4.1) years, and all patients were of white race/ethnicity. Baseline characteristics of the groups are summarized in the Table. The CHM genotype and baseline BCVA varied between patients. Age, lens status, and individual BCVA letter scores are summarized in eTable 1 and eTable 2 in Supplement 2.

Throughout the study period, a total of 28 adverse events were reported; none of them were regarded as severe. Fifteen adverse events were ocular, mostly common symptoms following vitreoretinal surgery with sutured sclerotomies (eg, conjunctival hyperemia, foreign body sensation). Five adverse events were unresolved at the last visit: patient 401 reported worsening diplopia over the 2 years (likely due to progressive loss of an adequate visual field to fuse)¹⁵; a preexisting cataract in patient 405 worsened following vitrectomy in the treated eye; 3 patients (401, 403, and 405) developed localized idiopathic thickening of the inner retina (eFigure 1 in Supplement 2). No plausible relationship to the study procedure and/or study drug was documented for any of the nonocular adverse events.

The mean (SD) final BCVA ETDRS score of the treated eyes was 64.0 (0.4) (20/50 Snellen equivalent), and the control eyes scored a mean of 69.3 (0.5) letters (20/40 Snellen equivalent). The treated eyes gained a mean of 4.7 (10.9) letters at month 3 and maintained a gain of 3.7 (7.5) letters at month 24 (Figure 2A). The control eyes showed no mean change from baseline at month 24 (mean [SD], 0.0 [5.1] letters). The difference between BCVA change in the groups was 3.7 letters (95% CI, −7.2 to 14.5 letters; P = .43). At month 24, 2 patients (33%) gained 10 or more letters and 1 patient (17%) gained 15 or more letters in the treated eyes, but no such change was observed in the control eyes (difference, 33%; 95% CI, −21% to 71%; P = .50). Patients with moderate VA loss (letter score of 73-34 [approximate Snellen equivalent 20/32 to 20/200]); n = 4) in the treated eye at baseline appeared to experience larger gains in VA (mean, 5.5 letters; range, −1 to 15) than patients with better VA letter score at baseline (>73 [approximate Snellen equivalent 20/32]; loss of 2 and 3 letters). No patient in either group lost 10 or more letters when baseline values were compared with month 24 values. Individual letter scores are reported in eTables 2 to 5 in Supplement 2.

Retinal sensitivity increased by 10.3 (5.5) dB in the treated eyes and 9.7 (4.9) dB in the control eyes (difference, 0.6; 95% CI, −10.2 to 11.4; P = .74). Changes in retinal sensitivity and peak retinal sensitivity are shown in Figure 2B and C. Although 5 of the 6 treated eyes achieved improvements in all or some measures (mean retinal sensitivity, peak retinal sensitivity, and/or gaze fixation area), patient 401 experienced no improvement. He developed a macular hole due to the surgery, which closed spontaneously by the end of the study. Quantitative analysis was performed on the complete cohort (the intent-to-treat analysis is reported in eTable 9 in Supplement 2) after excluding patient 401 to highlight the potential efficacy in a scenario without surgical complications (Table).

Analysis of the qualitative changes in fixation between baseline and month 24 shown in eFigure 2 and eFigure 3 in Supplement 2 indicates that the preferred retinal locus was maintained in the area treated in 2 of 3 patients with good, central fixation at baseline (patients 401 and 405). The preferred retinal locus shifted toward the treated area in 2 of 3 patients with less-defined and/or eccentric fixation at baseline (patients 404 and 406).

At month 24, treated eyes showed mean declines from baseline of 21% in preserved FAF area, 23% in preserved EZ, and 5% in SFCT. The control eyes showed similar mean declines in preserved FAF area (17%) and EZ (23%) and a larger mean decline of 24% in SFCT from baseline over 24 months (Figure 3 and eTables 6-8 in Supplement 2). The Table reports the absolute values of anatomic end points at baseline, the change from baseline to month 24, and differences between study and control eyes, the 95% CIs of those differences, and the P values.

In this randomized clinical trial among 6 participants, no differences between treated and control eyes were identified for mean change in BCVA or retinal sensitivity. However, treated eyes gained a mean of 3.7 letters on the BCVA ETDRS score, while the control eyes showed no change from baseline to 24 months after vector administration (P = .43).

Although 2 patients gained 10 or more letters in their treated eyes, no patient had lost more than 3 letters at 24 months in the treated eye. The best-performing patient (403; a 15-letter gain) already had significant cataract at enrollment; his lens status did not change much after surgery, and no cataract extraction was performed within the 24 months of follow-up. In contrast, patient 404 (an 11-letter gain) had the highest Lens Opacities Classification System score in the cohort, and his cataract progressed substantially on both eyes following surgery. Although cataract extraction was performed in both eyes, only the treated eye gained 11 letters; the untreated eye lost 8 letters compared with baseline.

Sustained maintenance of VA can be considered a clinically meaningful long-term treatment goal for patients with CHM, which is a progressive, degenerative disease with limited loss of VA until later stages.⁸ We observed that patients with moderate VA loss appeared to experience larger gains in VA than patients with better VA letter scores at baseline. Although the small sample size does not allow any firm conclusion, this observation supports the hypothesis that gene therapy is likely to be most effective in improving VA in patients with CHM after the point at which foveal function begins to decline but before photoreceptors are irreversibly lost. Although the small sample size does not allow any firm conclusion, it supports the hypothesis that gene therapy is likely to be most effective in improving VA in patients with CHM after the point at which foveal function begins to decline but before photoreceptors are irreversibly lost. It could be expected that patients with highly functional fovea and very good VA would be unlikely to experience a further gain in BCVA, similar to previous reports.^1,8

Microperimetry, which corrects for eye movements, is another potentially useful assessment of visual function. Although reproducibility of this test in healthy participants has been published, no data are available for patients with CHM, to our knowledge. Generally, microperimetry is subject to concentration and learning effects because it relies on stimulus recognition; when tests are conducted at 6-month intervals, it could also be subject to seasonal effects.^1,16 Using the contralateral eye of the same patient as a control accounted for these potentially confounding effects and provided a measure of the natural course of degeneration in each patient in this study. In contrast to the 12-month results, the good central fixation in patient 402 was not maintained at 24 months despite treatment and beneficial effect in patient 403 (preferred retinal locus shift toward treatment area seen at 12 months was not maintained).

Anatomic end points can also be used to evaluate therapeutic efficacy and are valuable objective variables and surrogate markers of treatment response. Natural disease progression in CHM is characterized by progressive, irregular loss of metabolically active retinal pigment epithelium,³ which can be recorded as the area of preserved autofluoresence.^17,18 The EZ signal in SD-OCT has been established as a surrogate marker for vision in several studies because it is linked to photoreceptor function.^19-23 Across 24 months, no difference was detected in the rate of decline of preserved FAF area in treated and control eyes. This lack of response may be due to loss of fluorescence in CHM at the edge of retinal degeneration, whereas the gene therapy was targeted to the center of the fovea.¹⁸ To test this hypothesis, further follow-up of FAF area decline in treated eyes compared with untreated control eyes is warranted. Because the FAF area is directly correlated with overlying photoreceptors,¹⁸ the similar decline observed in EZ is not surprising. Greater SFCT has been associated with better VA,^3,21,24,25 possibly owing to improvements in photoreceptor metabolism and enhanced oxygenation of the outer retina, with subsequent improvements in function.²⁴ Although the interim 12-month analysis suggested that the progressive loss of tissue was not slowed or halted within the first year following gene therapy, this longer-term follow-up at 24 months shows a slowdown in mean tissue loss in treated vs control eyes from baseline, as assessed by SFCT (−5% vs −24%, respectively). This observed delay in response could be because the surgical trauma offset any short-term therapeutic effect and is in line with reports from an independent study.⁹

There were no vector-related adverse reactions and no severe adverse events throughout the study period (January 11, 2016, to February 26, 2018). However, a conservative assessment of the safety profile according to the rule of 3 would mean that any adverse event that was not observed during the study could still have an actual rate of 50%.²⁶ Although larger cohorts are needed to more precisely describe the safety profile, independent studies reported similar safety outcomes and as such support the notion that gene therapy with AAV2-REP1 has an acceptable safety profile.^1,8,9,11,14

Three other clinical gene therapy trials have so far been conducted using the identical vector construct. The Canadian study (NCT02077361) (n = 6) showed somewhat mixed results regarding visual acuity (1 patient gaining ≥15 letters, 1 patient losing 8 letters) and reported the same rate of retinal pigment epithelium loss in the treated and untreated eye over a 2-year period.⁹ In contrast, Xue et al¹¹ recently reported improved VA in 14 of 14 treated eyes, with 6 treated eyes gaining more than 1 line of vision (>5 letters) (NCT01461213). Another study (NCT02553135) on 6 patients reported 1 treated eye with a 10-letter gain, 1 treated eye with a 5-letter gain, and 1 untreated eye with 4-letter gain; however, BCVA letter score remained unchanged (±2 letters) in all other eyes.¹⁴

The results of this study should be considered within the context of its limitations, which include small patient numbers and a relatively short follow-up period. In addition, the unmasked study design with a subjective outcome measure, such as BCVA, makes it difficult to rule out placebo effects. Patients with retinal dysfunction can demonstrate greater test-retest variability, and the baseline characteristics demonstrate some asymmetry (9-letter score difference) between the groups. Although this study did not perform multiple baseline VA and microperimetry tests to establish test-retest variability, learning effects can effectively be ruled out because all participants performed these tests more than 3 times during routine clinical care before enrollment in the study.

The results and data from this study show the potential of AAV-mediated gene therapy for CHM. Our study further adds to the evidence that gene therapy with AAV2-REP1 may be well tolerated, although no differences in VA from control eyes were identified in the 6 participants. The phase 3 STAR trial (NCT03496012) may help to assess more confidently the safety and efficacy of treatment with AAV2-REP1 in patients with CHM.

Accepted for Publication: June 16, 2019.

Corresponding Author: M. Dominik Fischer MD, DPhil, University Eye Hospital, University of Tübingen, Elfriede-Aulhorn-Strasse 7, 72076 Tübingen, Germany (dominik.fischer@uni-tuebingen.de).

Published Online: August 29, 2019. doi:10.1001/jamaophthalmol.2019.3278

Author Contributions: Drs Fischer and Wilhelm had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Fischer, Reichel, Kahle, Peters, Ueffing, MacLaren, Wilhelm.

Acquisition, analysis, or interpretation of data: Fischer, Ochakovski, Beier, Seitz, Vaheb, Kortuem, Kuehlewein, Girach, Zrenner, MacLaren, Bartz-Schmidt, Wilhelm.

Drafting of the manuscript: Fischer, Ochakovski.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Fischer, Ochakovski, Girach.

Obtained funding: Fischer, Zrenner, Ueffing.

Administrative, technical, or material support: Fischer, Ochakovski, Reichel, Kahle, Peters, MacLaren, Bartz-Schmidt, Wilhelm.

Supervision: Fischer, Ueffing, MacLaren, Bartz-Schmidt, Wilhelm.

Conflict of Interest Disclosures: Dr Fischer reported receiving nonfinancial support from Nightstar Therapeutics during the conduct of the study, grants and personal fees from Casebia Therapeutics, and personal fees from Nightstar Therapeutics, Novartis, Sanofi, Adelphi Values, EyeServ, and Retina Implant outside the submitted work. In addition, Dr Fischer reported having a patent to WO2017/042584 pending, licensed, and with royalties paid. Mr Beier reported receiving grants from Tistou and Charlotte Kerstan Foundation during the conduct of the study. Dr Vaheb reported receiving grants from Kerstan Foundation during the conduct of the study. Dr Kortuem reported receiving grants from Nightstar Therapeutics during the conduct of the study. Dr Kahle reported receiving grants from Tistou and Charlotte Kerstan Foundation during the conduct of the study. This work was supported by the Tistou and Charlotte Kerstan Foundation by a grant to the University of Tübingen. Dr MacLaren reported receiving grants and personal fees from Nightstar Therapeutics outside the submitted work. In addition, Dr MacLaren reported having a patent for choroideremia gene therapy issued and licensed. Dr Bartz-Schmidt reported receiving personal fees and nonfinancial support from Retina Implant outside the submitted work. Dr Wilhelm reported receiving grants from Tistou and Charlotte Kerstan Foundation during the conduct of the study. No other disclosures were reported.

Funding/Support: This work was supported by the Tistou and Charlotte Kerstan Foundation by a grant to the University of Tübingen (Drs Fischer and Wilhelm as principal investigators and Drs Kahle and Vaheb and Mr Beier as members of those research groups).

Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Data Sharing Statement: See Supplement 3.

Additional Contributions: We thank the patients involved in this study. Editorial support was provided by Rebecca Franklin, PhD (Fishawack Communications Ltd), and funded by Nightstar Therapeutics. Ahmad Zhour, Lea Kontostanou, Andrea Rindtorff, Regine Grund, Michael Breuninger, Susanne Schweyer, Giuseppa Conte, Gina Fischer, Paul Richter, Alexandra Feidt, and Oksana Faul (University Eye Hospital, University of Tübingen, Tübingen, Germany) recorded clinical data and/or processed clinical samples. There was no financial compensation.