Pivotal Studies and Protocol Selected Studies
Description of Studies
Study 301 (N = 31) was a phase III, open-label, randomized controlled trial designed to evaluate the efficacy and safety of sequential subretinal injection of voretigene neparvovec into each eye in patients diagnosed with LCA due to RPE65 mutations. Randomization, which followed screening and confirmation of study eligibility, occurred in a 2:1 ratio of intervention (voretigene neparvovec) to control and used a block design stratified by age (≥ 10 years versus < 10 years) and mobility testing passing level (pass at ≥ 125 lux versus < 125 lux), as determined at screening. A total of 31 patients in 2 study sites in the US (the study enrolled international patients, including 1 patient from Canada) were randomized to either the treatment group (n = 21) or the control group (n = 10). Patients randomized to the voretigene neparvovec group received a dose of 1.5 × 1011 vg of voretigene neparvovec in each eye; the non-simultaneous subretinal injections were to occur within an 18-day period (12 days ± 6 days). Patients randomized to the control group did not receive voretigene neparvovec, sham injection, or corticosteroids for a period of at least 1 year from baseline evaluations. Following repeated retinal and visual function analysis, including mobility testing at 1 month, 3 months, 6 months, and 1 year, patients in the control group were crossed over to receive non-simultaneous injections of 1.5 × 1011 vg of voretigene neparvovec in each eye (within 18 days) after 1 year of randomization, provided they still met all study eligibility criteria. The primary end point was change in bilateral multi-luminance mobility testing (MLMT) performance at year 1 relative to baseline. Secondary end points were change in full-field sensitivity threshold (FST) averaged over both eyes at year 1 relative to baseline, change in assigned first eye MLMT performance at year 1 relative to baseline, and change in VA averaged over both eyes at year 1 relative to baseline.
In Study 301, there were more females (n = 18, 58%) than males (n = 13, 42%) and patients were primarily White (n = 21, 68%). At the time of randomization, the mean patient age was 15.1 (standard deviation [SD] = 10.9) years, with a range of 4 years to 44 years. Patients randomized to the voretigene neparvovec group were slightly younger (mean = 14.7 years; SD = 11.8 years) than those randomized to the control group (mean = 15.9 years; SD = 9.5 years). More patients with better baseline MLMT performance (< 125 lux) were enrolled in the voretigene neparvovec group (57%) than in the control group (40%).
Efficacy Results
Treatment with voretigene neparvovec resulted in statistically significant improvements in navigational ability in low-to-moderate light conditions (measured by MLMT score). The mean bilateral MLMT change score after 1 year was 1.8 (SD = 1.1) for the voretigene neparvovec group and 0.2 (SD = 1.0) for the control group. The difference in change from baseline in bilateral MLMT between the voretigene neparvovec and control treatment groups at 1 year was 1.6 (95% confidence interval [CI], 0.72 to 2.41; P = 0.001) which was statistically significant in favour of voretigene neparvovec; however, the difference between the treatment groups did not exceed 2 points. Eleven patients (52%) in the voretigene neparvovec group had an MLMT score change of 2 or more (the difference considered meaningful by the FDA and the European Medicines Agency [EMA]). In contrast, only 1 patient (10%) in the control group had a score change of 2 and none of the patients in the control group had a score change greater than 2. Although 62% of patients in the voretigene neparvovec group achieved a score of 6 on the MLMT (the maximum possible score) following administration of voretigene neparvovec, compared with no patients in the control group achieving a score of 6 on the MLMT, the observed mean increase in MLMT score of 1.8 observed in the voretigene neparvovec group could be an underestimate of the within-groups magnitude of the change due to the potential ceiling effect. Improvements in the MLMT score observed at 1 year seemed to be maintained until the 4-year follow-up.
Patients treated with voretigene neparvovec experienced a mean improvement in FST greater than 2 log units, whereas mean FST did not change in the control group; the mean change from baseline to year 1 was −2.08 (standard error [SE] = 0.29) log10(candela second per square metre [cd.s/m2]) for the voretigene neparvovec group and 0.04 (SE = 0.44) log10(cd.s/m2) for the control group. There were statistically significant improvements in full-field light sensitivity with voretigene neparvovec (mean difference versus control = −2.11 log units; 95% CI, −3.19 to −1.04; P < 0.001) at 1 year. This between-group difference exceeded the suggested clinical significance threshold of 10 decibels (dB) or 1 log unit for clinical significance.14 The improvements were sustained for 4 years after the second-eye injection, where for all patients who received treatment with voretigene neparvovec, the mean change from injection baseline at 4 years after the second-eye injection was −2.00 (SE = 1.35) log10(cd.s/m2). The clinical experts explained that the changes seen would be clinically meaningful in terms of improving visual function.
A longitudinal repeated measures analysis of VA using the Holladay scale averaged over both eyes showed a mean change from baseline to 1 year of −0.16 (SE = 0.07) logarithm of the minimum angle of resolution (LogMAR) for the voretigene neparvovec group and 0.01 (SE = 0.10) LogMAR for the control group, resulting in a mean treatment-effect difference of −0.16 LogMAR (95% CI, −0.41 to 0.08; P = 0.17), an 8-letter improvement. This difference was neither statistically significant nor clinically meaningful. Using the Lange scale for off-chart VA results, the mean between-group treatment difference was −0.16 LogMAR (95% CI, −0.31 to −0.01; nominal P = 0.035) corresponding to a 7.5-letter improvement on the eye chart for people who had voretigene neparvovec. All changes were smaller than the accepted clinically meaningful change (≥ 0.30 LogMAR). By year 3, little further change was seen in VA for either arm after treatment. The clinical experts consulted by CADTH for this review explained that even a small change would be important for patients. The clinical experts also noted that even if there were no improvement, preventing vision deterioration would be important for the patient’s quality of life.
For the patient-completed Visual Function Questionnaire (VFQ), the mean change from baseline to year 1 was 2.6 (SD = 1.8) for the voretigene neparvovec group and 0.1 (SD = 1.4) for the control group, for a mean between-group treatment difference of 2.4 (95% CI, 1.0 to 3.8; nominal P = 0.001). For the parent-completed surveys, the mean change from baseline to year 1 was 3.9 (SD = 1.9) for the voretigene neparvovec group and −0.2 (SD = 1.3) for the control group, for a mean between-group treatment difference of 4.0 (95% CI, 2.1 to 6.0; nominal P = 0.002). Although the VFQ assessed the patients’ ability to perform ADL for those patients who received voretigene neparvovec, the questionnaire did not contain any items to specifically assess HRQoL for patients. In addition, the VFQ used in Study 301 was not assessed psychometrically and, given the modifications made to the original National Eye Institute VFQ, the minimal important differences (MIDs) identified in the literature for that measure were not considered directly generalizable to the version used in Study 301.
Long-term data for patients in the voretigene neparvovec group suggest durable improvements in visual performance across multiple end points for at least 4 years following voretigene neparvovec administration. Similarly, after crossing over to voretigene neparvovec, patients in the control/voretigene neparvovec group exhibited visual performance improvements comparable to those observed in the voretigene neparvovec group, effects that were maintained through at least 3 years following bilateral voretigene neparvovec administration. Study 301 aims to follow patients for up to 15 years after treatment. The current duration of follow-up is limited to 5 years and longer-term efficacy and safety data for voretigene neparvovec are awaited. The duration of treatment effect is unclear. Also, given the small sample size (30 patients), there is uncertainty around the generalizability of the results observed for the long-term treatment effect. The clinical experts indicated that voretigene neparvovec would likely provide long-term benefits, although this was associated with substantial uncertainty.
Subgroup analyses of interest to this review were based on age, clinical phenotype, and number of viable retinal cells. No preplanned subgroup analyses were conducted in Study 301. The sponsor conducted a post hoc subgroup analysis by age (younger than 18 years at first injection versus 18 years or older at first injection); however, this analysis should be considered hypothesis-generating. Also, it is not clear if this subgroup analysis was conducted at the request of a regulator, and no rationale was provided for the age cut-off. The clinical experts consulted by CADTH indicated that greater clinical benefit from treatment is expected earlier in the condition when there are more viable retinal cells for the gene replacement therapy to restore. While patient age can be used as a proxy to estimate how advanced the condition is, due to the heterogeneous nature of RPE65 mutation–associated retinal dystrophy and the differences in the age of onset and disease progression, retinal cell viability should be assessed on an individual patient basis, regardless of patient’s age (as long as the patient is at least 4 years of age), to determine if the specific individual is likely to respond to voretigene neparvovec.
Harms Results
All patients in Study 301 experienced at least 1 treatment-emergent adverse event (TEAE). Most adverse events were mild in severity and no patient had any adverse events that led to study discontinuation or death.
The most frequently reported TEAEs in the voretigene neparvovec group were leukocytosis (in 45% of patients); vomiting (in 40% of patients); nasopharyngitis, headache, and pyrexia (in 35% of patients for each); oropharyngeal pain, cough, and nausea (in 30% of patients for each); increased intraocular pressure (in 20% of patients); and cataract and hematuria in 15% of patients for each.
Overall, 13 (65%) patients in the voretigene neparvovec group had at least 1 TEAE considered to be related to the study drug administration procedure. The TEAEs most often considered to be probably related to the administration procedure were cataract and increased intraocular pressure (n = 3 [15%] patients for each).
During the control period, 2 (10%) patients in the voretigene neparvovec group experienced 3 serious adverse events (SAEs) at time points distant from vector administration. One patient experienced a possible seizure requiring hospitalization and 1 patient experienced an adverse reaction to medications administered during oral surgery, which required hospitalization.
At the time of the data cut-off for the Clinical Study Report (CSR), which provided updated results of safety data through July 2, 2018, including follow-up for up to 5 years after the second injection for some patients, 6 SAEs occurred in 5 patients, including convulsion (1 event), adverse drug reactions (2 events) and retinal disorder (1 event of foveal thinning and loss of vision), retinal detachment (1 event), pneumonia (1 event), and menorrhagia (1 event).
One ocular SAE occurred in Study 301, where a patient with pre-existing atrophy of the retina who received voretigene neparvovec experienced a retinal disorder (foveal thinning and a loss of central vision) related to the subretinal injection of the treatment.
In the voretigene neparvovec group during the first year after the randomization period, 3 patients experienced a cataract, 2 patients experienced a retinal tear, and 1 patient developed an asymptomatic full-thickness macular hole. During the follow-up, 1 patient who was originally in the control group and crossed over to voretigene neparvovec experienced a retinal disorder (foveal thinning and a loss of central vision). Another patient experienced 1 SAE of retinal detachment.
Summary of Key Results From Study 301.
Critical Appraisal
Study 301 used accepted methods to conceal allocation where a randomization list was generated by an independent biostatistician. However, because it was an open-label trial, patients were aware of the treatment allocation following randomization. Therefore, the evaluation of patient-reported outcomes (such as those measured by VFQ) and adverse events may be biased by treatment knowledge. The treatment effect on these subjective outcomes can potentially be overestimated as a consequence of the patient’s expectation of the efficacy of a new drug. Certain steps have been taken to ensure appropriate blinding of the assessment of the primary outcome measure (MLMT) and it seems there is potential for an unbiased outcome assessment with the MLMT, despite the open-label design. On the other hand, while the trial states that orientation and mobility assessors were masked, there is insufficient detail provided in the CSR to judge if adequate blinding of the outcome assessment was performed for all secondary outcome measures (e.g., VF and VA).
Imbalances in the baseline patient characteristics between the voretigene neparvovec and control groups included age and visual performance. It is unclear whether differences in baseline age between the voretigene neparvovec and control groups may introduce a risk of bias; on the one hand, the clinical experts indicated that age on its own should not be a criterion for treating patients but rather the presence of sufficient viable retinal cells. The imbalance at baseline in MLMT performance following assignment to the voretigene neparvovec or control group may bias the observed treatment effect estimate; however, the direction of the bias is unclear. As differences in MLMT, VA, and VF were noted at baseline in Study 301 — and as these outcomes represent the key clinical data — there is uncertainty associated with the effect estimate for these outcomes given that patients in the voretigene neparvovec treatment group had better MLMT scores at baseline (a larger percentage of patients with a baseline MLMT performance < 125 lux were enrolled in the voretigene neparvovec group [57%] than in the control group [40%]). Also, at baseline, patients randomized to the voretigene neparvovec group had better acuity, on average (less VA loss), and better VF than patients randomized to the control group. However, there is no evidence to indicate how visual performance at baseline could affect the treatment effect and, if bias exists, it is not possible to judge the direction of the bias.
Scores on the MLMT may underestimate the treatment effect of voretigene neparvovec due to the potential ceiling effect, where patients who passed the test at the second-lowest light level at baseline were able to achieve a maximum increase of only 1 unit. Although 62% of patients in the voretigene neparvovec group achieved the maximum possible increase following administration, compared with none in the control group, the observed mean increase in MLMT score of 1.8 may be an underestimate of the treatment effect because of this ceiling effect. The MLMT also potentially has a floor effect, where 1 patient in the voretigene neparvovec group did not pass the MLMT at the highest light level at baseline and also failed at year 1. It is worth noting that of the 5 patients who did not pass screening, 2 (40%) were not eligible based on mobility test performance, 1 was excluded due to ceiling effect, and 1 was excluded due to floor effect. In addition, it is not clear whether the 12 unique MLMT course configurations were of equivalent difficulty.
While patients in Study 301 were required to have a diagnosis of LCA due to RPE65 mutations to be enrolled in the pivotal trial of voretigene neparvovec, the clinical experts indicated the results reported for these patients should be generalizable to patients with RP as long as they have confirmed biallelic RPE65 mutations. Voretigene neparvovec is designed to deliver a normal copy of the RPE65 gene to cells of the retina in patients with a reduced level of or no biologically active RPE65; it is therefore intended to treat the underlying mechanism of the disease, which is the same in patients with confirmed RPE65 mutations regardless of clinical phenotype.13
The clinical experts indicated that measuring viable retinal cells is not a straightforward procedure, and that the methods used in the pivotal trial to determine whether a patient has sufficient viable retinal cells do not give a complete picture of the health of the retina or the number of viable photoreceptors. They also indicated that the technology to assess the structure and function of the retina has evolved since the trial. The clinical experts indicated that OCT can measure the thickness of the retina; however, it may not inform the treating physician whether there are viable retinal cells.
Other Relevant Evidence
Description of Studies
The sponsor submitted 2 supportive trials, Study 101 and Study 102. Study 101 (N = 12) was a phase I, open-label, single-arm, dose-escalation study that assessed the safety and tolerability of 3 different doses of voretigene neparvovec administered via subretinal administration to 1 eye (first-treated eye) of patients with LCA due to RPE65 mutations. Study 101 also evaluated the clinical efficacy; however, no formal hypothesis testing was conducted. Study 102 was a follow-on to Study 101 in which patients received voretigene neparvovec treatment in the previously uninjected eye (second-treated eye). In Study 101, 12 patients, who were 8 years of age or older at the time of administration, received unilateral subretinal injection in the eye with the worse function (first-treated eye). Three doses of voretigene neparvovec were tested sequentially: 1.5 × 1010 vg, 4.8 × 1010 vg, and 1.5 × 1011 vg. The dose of voretigene neparvovec under review is 1.5 × 1011 vg. Study 102 was a follow-on study to Study 101. Eleven of the 12 treated patients in Study 101 received a subretinal injection in the contralateral eye (second-treated eye) consisting of 1 dose of 1.5 × 1011 vg of voretigene neparvovec in a total volume of 300 μL.
Efficacy Results
All efficacy data are descriptive, and no inferences can be made because there were no hypotheses being tested.
In Study 101, 4 patients were considered non-evaluable, given the inconsistent use of patching as well as the variability of lighting conditions and test-course difficulty. Following vector administration, follow-up mobility testing using the injected eye indicated that 4 of the 8 evaluable patients were able to complete the mobility test at a light level that was at least 1 level darker than baseline.
In Study 102, monocular mobility testing was assessed for change from baseline for the eye injected in Study 102. Of the 11 patients who received voretigene neparvovec, 8 were considered evaluable for mobility testing, while three were not and, therefore, were not included for presentations of this parameter. All 8 evaluable patients whose eyes had been injected in Study 102 completed the mobility test 1 year after injection at a light level that was at least 1 level darker than baseline. Five of the 8 (63%) evaluable patients received the maximum attainable score. Mobility testing results continued to show that, through year 4, all 8 evaluable patients completed the mobility test at a light level that was at least 1 level darker than baseline.
For the eye injected in Study 101 at year 1, 4 of 7 (57%) evaluable patients showed a decrease of 10 dB or more in FST compared with baseline, with a change from baseline in all 7 evaluable patients ranging from −8.1 dB to −33.7 dB. At year 2, 3 of 7 (43%) evaluable patients showed a decrease of 10 dB or more in FST, with a change from baseline ranging from −5.7 dB to −29.1 dB in 6 of the 7 evaluable patients.
At the time of database lock for the CSR for Study 102 (May 19, 2019), the available FST data indicated that the light sensitivity of the eye injected in Study 102 increased after injection for 8 of 11 patients; this increase was greater than the 10 dB cut-off, was considered clinically important in 7 of 11 patients, and was below this cut-off for 1 patient. Light sensitivity remained stable in the other 3 patients.
In Study 101 at year 1, for the eye injected in that study, 7 of the 12 (58%) evaluated patients had a change in LogMAR score of 0.3 or more compared with baseline (corresponding to an improvement of at least 3 lines [15 letters] on the eye chart). At year 2, 4 of 11 (36%) patients had a change in LogMAR score of 0.3 or more compared with baseline. At year 3, 5 of 9 (56%) patients had a change in LogMAR score of 0.3 or more compared with baseline.
In the Study 102 LogMAR scores, based on Holladay scale for off-chart results, 1 (9%) of the 11 patients assessed had a change in LogMAR score of 0.3 or more compared with baseline 1 year after receiving treatment.
Harms Results
All patients in Study 101 and Study 102 experienced at least 1 TEAE.
One SAE was reported in Study 101, where 1 patient experienced an anal fistula that required hospitalization. The SAE was mild in severity and considered related to an underlying diagnosis of inflammatory bowel disease. The event was recovered and/or resolved with no sequelae.
One SAE was reported in Study 102, which was elevated intraocular pressure (grade 4) in the patient’s right eye resulting in hospitalization. The SAE, which was moderate in severity, was deemed related to the use of a depo-steroid injection for a known rare complication of vitrectomy (endophthalmitis).
No deaths or discontinuations from the study due to adverse events were reported in either study.
Critical Appraisal
Studies 101 and 102 are phase I, single-arm, open-label, non-randomized studies with a small sample size and are not considered to provide high-quality evidence to support the efficacy of voretigene neparvovec.
No conclusions can be drawn regarding the clinical efficacy of voretigene neparvovec. Study 101 was a dose-escalating study, and neither Study 101 nor Study 102 were designed or powered to assess the clinical efficacy of voretigene neparvovec. No formal hypothesis testing was conducted, and the within-group changes were not designed to be inferential. Only descriptive statistics were presented.
During the course of Study 101, the mobility test was further refined and standardized, which affected both the number of patients considered evaluable and the interpretation of the results.
The dosage of voretigene neparvovec used in 9 of the 12 patients enrolled in Study 101 was not the dosage approved by Health Canada; hence, the generalizability of the study results to the Canadian patient population is unclear.
Compared with Study 101, Study 301 included a broader population, including younger patients (4 years of age and older in Study 301 versus 8 years of age and older in Study 101) and patients with less advanced disease (VA no better than 20/60 in Study 301 versus VA no better than 20/160 in Study 101). The more stringent criteria introduced in Study 102 for determining the number of viable retinal cells (≥ 3 disc areas of retina without atrophy or pigmentary degeneration within the posterior pole) was also used in Study 301, whereas in Study 101, patients were eligible if they had one or more disc areas of retina that were not involved in complete retinal degeneration. Finally, periocular injection of the various corticosteroids used in Study 101 and Study 102 was discontinued in Study 301 to decrease the incidence and severity of elevated intraocular pressure and cataract formation or progression.