CADTH Reimbursement Review

Brolucizumab (Beovu)

Sponsor: Novartis Pharmaceuticals Canada Inc.

Therapeutic area: Diabetic macular edema

This multi-part report includes:

Clinical Review

Pharmacoeconomic Review

Stakeholder Input

Clinical Review

Executive Summary

An overview of the submission details for the drug under review is provided in Table 1.

Introduction

Diabetic macular edema (DME) is a vision-threatening complication of diabetes mellitus (both type 1 and type 2). The persistent elevation of blood glucose in people with diabetes causes damage to the capillaries, such as those in the eye, resulting in diabetic retinopathy (DR).¹ Some patients with DR and especially those with continually poorly managed blood glucose can experience swelling in the retina, known as DME.² Generally, DME manifests as slowly progressive vision loss. The degree of vision loss can vary considerably and depends on the severity, duration, and location of intraretinal fluid (IRF), among other factors. Signs of DME include blurred vision, retinal hemorrhages, retinal detachment, colours appearing “washed out” or faded, changes in contrast sensitivity, impaired colour vision, gaps in vision (scotomas), and potentially permanent vision loss. Untreated DME is considered to be the leading cause of vision loss, visual disability, and legal blindness in people with DR.^2-4 Patients’ health-related quality of life (HRQoL) and daily functioning will be significantly affected, and indirect costs due to lost productivity are high if patients are left untreated.^5-7 It has been estimated that there are approximately 60,000 adults with DME in Canada who experience vision impairment requiring treatment.^8,9

Current therapies for DME in Canada include non-pharmacological interventions (laser therapy and vitrectomy) and pharmacological interventions (intravitreal anti–vascular endothelial growth factor [VEGF] drugs and intravitreal steroids). Health Canada–approved anti-VEGF drugs for DME treatment include ranibizumab and aflibercept (with bevacizumab used off-label), while approved intravitreal steroids include dexamethasone.

Brolucizumab is a humanized VEGF inhibitor that suppresses endothelial cell proliferation in vitro and reduces neovascularization and vascular permeability.¹⁰ On November 30, 2022, brolucizumab was approved by Health Canada for the treatment of DME. The reimbursement criteria for brolucizumab requested by the sponsor are the same as the Health Canada indication. The recommended dose for brolucizumab is 6 mg (0.05 mL) administered by intravitreal injection every 6 weeks for the first 5 doses. Thereafter, the physician may modify treatment intervals based on disease activity, as assessed by visual acuity and/or anatomical parameters. In patients without disease activity, treatment up to every 12 weeks (3 months) could be considered. In patients with disease activity, treatment every 8 weeks (2 months) could be considered; however, the interval between 2 doses should not be less than every 8 weeks (2 months).

The objective of this report is to perform a systematic review of the beneficial and harmful effects of brolucizumab 6 mg for the treatment of patients with DME.

Stakeholder Perspectives

The information in this section is a summary of the input provided by the patient groups that responded to CADTH’s call for patient input and from the clinical expert consulted by CADTH for the purpose of this review.

Patient Input

Patient input was provided as a joint submission by 5 groups: Fighting Blindness Canada, Canadian Council of the Blind, Canadian National Institute for the Blind (CNIB), Vision Loss Rehabilitation Canada, and Diabetes Canada. The information used to inform the submission was based on an online survey of people in Canada living with DR or DME that was conducted in the first months of 2020 by the submitting organizations. A total of 67 people responded to the survey; many (44.4%) were aged 61 to 80 years (n = 54) and the majority (76.1%) of respondents reported DME or DR in both eyes. Most respondents were either working full time (38.9%) or were retired (33.3%) (n = 54). A separate survey conducted by Canadian Council of the Blind in April 2020 further supported the submission by providing data on the impact of the COVID-19 pandemic on people in Canada who are blind, deaf-blind, or partially sighted (n = 572).

Survey respondents emphasized that DR and DME have substantial and life-altering impacts on daily life, including on reading, driving, and using a phone. In addition to concern for their eyesight worsening, coping with everyday life and general safety when outside of the home were identified as notable concerns of respondents during the preceding month. The results of the Canadian Council of the Blind survey showed that fear, anxiety, loneliness, and other psychosocial impacts were intensified for patients with age-related macular edema and DR during the pandemic.

The majority (56.4%) of respondents indicated they were currently receiving injections for DR or DME; the most common therapeutic options were Lucentis (29.4%), Eylea (24.6%), Avastin (20.2%), and Ozurdex (13.5%). Most respondents (54.5%) indicated they were satisfied with their injections and 63.6% indicated the injections have helped them avoid losing more eyesight (n = 22). Of note, 31.8% of respondents reported missing injections in the last year. According to respondents (n = 6), the reasons for cancelled or delayed appointments in the past included being too busy (50.0%), feeling unwell (33.3%), not being able to find someone to take them to the appointment (16.7%), and fear of injections (16.7%). According to respondents (n = 22), the most difficult part of eye injection appointments was the long wait times (50.0%), finding someone to take them to and from the appointment (31.8%), anxiety or fear about the injection (27.3%), and taking time off work (27.3%). No survey respondents reported experience with brolucizumab.

The submitting organizations indicated that any treatment that reduces the physical, psychological, and logistical strain on patients would be preferred. The submitting organizations suggested this may be addressed by a treatment that is less invasive, or similarly invasive but administered less frequently. Further, the submitting organizations suggested that any treatment that can extend the interval between injections while still minimizing vision loss would be considered advantageous for patients living with vision loss, particularly in the context of the COVID-19 pandemic.

Clinician Input

Input From the Clinical Expert Consulted by CADTH

The clinical expert consulted by CADTH indicated the treatment goals for DME are to delay and, in some cases, reverse the progression of DME and/or DR and to improve vision-related and general quality of life. Considering that most patients are currently required to attend treatment visits once every 1 to 3 months, the clinical expert noted there is an unmet need for treatments that can be given at longer treatment intervals, without recurrence of disease, to reduce the burden on patients and caregivers associated with frequent treatment visits and to increase adherence to treatment regimes.

The clinical expert noted that brolucizumab is expected to have a place in therapy that is similar to that of other anti-VEGFs as a first-line or later line of treatment in patients with DME. Compared with other anti-VEGFs, treatment with brolucizumab is anticipated to reduce the burden of care by increasing the intervals between treatments while still maintaining therapeutic benefit, which could potentially address the unmet need related to frequent treatment visits.

The clinical expert indicated that patients with DR associated with vision loss secondary to centre-involving DME are suitable candidates for brolucizumab. The clinical expert stated that brolucizumab can be used in patients who are treatment-naive or those who require a change in therapy due to inadequate response to other anti-VEGF drugs. Patients who may not be suitable for treatment include those who present with major structural damage to the macular retina (e.g., macular atrophy or fibrosis), according to the expert.

The clinical expert noted that clinical evaluation and optical coherence tomography (OCT) should be performed for prognosis and follow-up at dosing visits. Key assessment outcomes included change in visual acuity and retinal thickness and the presence of retinal fluid. According to the expert, an optimal response to anti-VEGFs is generally achieved 6 to 12 months after initiation of therapy.

The clinical expert indicated that brolucizumab should be discontinued in patients experiencing treatment futility with proof of irreversible anatomic or functional damage, such as macular atrophy (schema) and fibrosis.

Regarding prescribing conditions, the clinical expert recommended retina subspecialty care as the most appropriate treatment setting for the prescription and administration of brolucizumab in urban areas; care by trained comprehensive ophthalmologists with experience and expertise in managing DME would be sufficient in rural settings.

Clinician Group Input

No input was received from any clinician groups for this submission.

Drug Program Input

Input was obtained from the drug programs that participate in the CADTH reimbursement review process. The following were identified as key factors that could potentially impact the implementation of a CADTH recommendation for brolucizumab:

• considerations for initiation of therapy

• system and economic issues.

The clinical expert consulted by CADTH provided advice on the potential implementation issues raised by the drug programs.

Clinical Evidence

Pivotal Studies and Protocol-Selected Studies

Description of Studies

Two studies, KESTREL (N = 566) and KITE (N = 360), met the inclusion criteria for the systematic review section. They were similarly designed phase III randomized controlled trials (RCTs) that evaluated the noninferiority of brolucizumab (6 mg every 8 weeks or every 12 weeks during maintenance, with every 16 weeks as an option after week 72 in the KITE trial in the event of disease stability, i.e., if no disease activity was observed in the last 2 disease activity assessment [DAA] visits) to aflibercept (2 mg every 8 weeks during maintenance). Noninferiority was based on an a priori noninferiority margin of 4 Early Treatment Diabetic Retinopathy Study [ETDRS] letters. The change from baseline in best-corrected visual acuity (BCVA) using ETDRS letters at week 52 in the full analysis set population was a primary end point. The dosing frequency for brolucizumab was determined by disease activity as assessed by visual acuity and/or anatomic parameters. Patient demographic and disease characteristics were generally well balanced across the arms in both trials at baseline. The mean age of enrolled patients at baseline in these studies ranged from 62.2 to 64.4 years, and the majority were male (greater than 58%) and white (greater than 73%). The mean time since the diagnosis of DME was 9.4 to 12.5 months in the KESTREL trial and 9.9 to 10.4 months in the KITE trial; the mean baseline central subfield thickness (CSFT) in these 2 studies ranged from 453 μm to 484 μm, and the mean baseline BCVA score ranged from 63.7 to 66.6 in both studies although, in the KITE trial, there was an imbalance in the number of ETDRS letters: 66.0 (SD = 10.8) in the brolucizumab group and 63.7 (SD = 11.7) in the aflibercept group. All enrolled patients were naive to anti-VEGF therapies. Outcomes included changes in BCVA, anatomical outcomes, DR severity, vision-related function, injection frequency, and safety, with a primary analysis at week 52 and data up to 100 weeks.

Efficacy Results

Key efficacy and safety results from the pivotal trials are presented in Table 2. The results of the KESTREL and KITE trials support the noninferiority of brolucizumab 6 mg (5 times the loading doses of every 6 weeks followed by maintenance injections every 8 weeks or every 12 weeks) versus aflibercept 2 mg (5 monthly loading doses followed by maintenance injections every 8 weeks) for the change in BCVA. The noninferiority of brolucizumab 6 mg to aflibercept 2 mg was demonstrated for the primary end point (change from baseline in BCVA at week 52 for the study eye) using a noninferiority margin of 4 letters (P < 0.001 for noninferiority). The between-group least squares (LS) mean difference for brolucizumab 6 mg versus aflibercept 2 mg in the KESTREL trial was −1.3 letters (95% confidence interval [CI], −2.9 to 0.3); the between-group LS mean difference in the KITE trial was 1.2 letters (95% CI, −0.6 to 3.1). In addition, several sensitivity analyses by the sponsor, as well as a supportive analysis using the per-protocol population, were consistent with the findings of the primary analyses. Results were consistent for change from baseline to week 100. Results of prespecified subgroup analyses (i.e., baseline BCVA categories and CSFT categories) were generally consistent with the overall population at week 52; however, the study was not powered to detect subgroup differences. Noninferiority of brolucizumab 6 mg to aflibercept 2 mg was also demonstrated for the mean change from baseline in BCVA averaged over the period from week 40 through week 52.

The change in retinal thickness (measured as CSFT) from baseline and patients with a CSFT of less than 280 µm were secondary outcomes in the studies. According to the expert, the reduction in retinal thickness correlates well with the improvement in visual acuity. In the KESTREL trial, the LS mean difference in the change from baseline in CSFT between the brolucizumab 6 mg arm and the aflibercept 2 mg arm was −5.1 µm (95% CI, −22.3 µm to 12.2 µm). Over the period from week 40 through week 52, the average LS mean of the change from baseline in CSFT between brolucizumab 6 mg and aflibercept 2 mg was −1.4 µm (95% CI, −17.9 µm to 15.0 µm). In the KITE trial, the LS mean difference in the change from baseline in CSFT to week 52 between the brolucizumab 6 mg arm and the aflibercept 2 mg arm was −32.8 µm (95% CI, −52.5 µm to −13.0 µm). Over the period from week 40 through week 52, the average LS mean of the change from baseline in CSFT between brolucizumab 6 mg and aflibercept 2 mg was −29.4 µm (95% CI, −48.6 µm to 10.2 µm; P = 0.001) in favour of brolucizumab. These differences between brolucizumab and aflibercept were similar at year 2.

The change from baseline in the NEI VFQ-25 composite score, which measures vision-related functions and some aspects of HRQoL, was a secondary outcome in the KESTREL and KITE trials, but not included in the statistical hierarchy. In the KESTREL trial, the LS mean estimates for change from baseline in the composite score were 7.1 for brolucizumab 6 mg and 8.1 for aflibercept 2 mg, with between-group LS mean differences of −1.0 (95% CI, −3.4 to 1.4) at week 52, and LS mean estimates of 6.2 for brolucizumab 6 mg and 6.4 for aflibercept 2 mg, with LS mean differences of −0.2 (95% CI, −2.9 to 2.6) at week 100. In the KITE trial, the LS mean estimates for change from baseline in the composite score were 9.1 for brolucizumab 6 mg and 6.5 for aflibercept 2 mg, with between-group LS mean differences of 2.5 (95% CI, 0.2 to 4.8) for the change from baseline at week 52, and LS mean estimates of 9.3 for brolucizumab 6 mg and 5.9 for aflibercept 2 mg, with LS mean differences of 3.4 (95% CI, 0.8 to 6.1) at week 100.

The ETDRS Diabetic Retinopathy Severity Scale (DRSS) was used to measure disease activity in patients with DME. Regression of DRSS is another clinically meaningful outcome in this study population; however, this outcome was not included in the statistical hierarchy. For most of the results for this outcome at week 52 and week 100, there was no evidence to suggest a difference between brolucizumab and aflibercept. Given the uncertainty in the analysis due to the lack of statistical testing and imprecision in the between-group differences, no definite conclusions on the change in disease severity can be drawn.

The pivotal trials measured the proportions of patients with the presence of IRF and subretinal fluid (SRF) as secondary outcomes. IRF and SRF are indicators of active disease. According to the clinical expert consulted by CADTH, IRF is a more relevant outcome than SRF in patients with DME, noting that SRF is uncommon in DME and is a marker for more severe DME. This outcome was not tested statistically due to a previous failure of the hierarchical testing procedure. A numerically lower proportion of patients treated with brolucizumab 6 mg had the presence of IRF and/or SRF compared with the aflibercept 2 mg group in both studies at week 52 and week 100.

Frequency of injection was noted to be an important outcome of interest by both patients and the clinical expert, as it may have implications on the frequency of adverse events (AEs), HRQoL, the burden of treatment, and patient adherence and, subsequently, can have an impact on the treatment effect. The proportion of patients treated with brolucizumab who were maintained on a schedule of every 12 weeks was reported descriptively in the studies. Among the patients who received treatment with brolucizumab, approximately half maintained a treatment interval of 12 weeks at week 52 in both studies, ||| ||| ||| ||| || |||||||| |||||||||| ||| |||| ||||||| || |||| ||| in the KESTREL and KITE trials, respectively. Among the patients who completed treatment with brolucizumab at week 100, the majority were being treated every 8 weeks (67.1% in the KESTREL trial and 52.5% in the KITE trial).

Harms Results

The safety profile for brolucizumab 6 mg was generally consistent with that of aflibercept 2 mg in the KESTREL and KITE trials. The proportion of patients reporting at least 1 ocular AE in the study eye up to week 100 was comparable across treatment arms in both studies (48.7% and 50.3% in the brolucizumab 6 mg and aflibercept 2 mg group, respectively, in the KESTREL trial; 40.8% and 40.9% in the brolucizumab 6 mg and aflibercept 2 mg group, respectively, in the KITE trial). Overall, the most frequently reported ocular AEs related to brolucizumab in both studies were cataract, conjunctival hemorrhage, vitreous detachment, vitreous floaters, intraocular pressure increased, diabetic retinal edema, dry eye, eye pain, posterior capsule opacification, conjunctivitis, and reduced visual acuity. Cataract was the most commonly reported ocular AE, which was anticipated because of the age of the study populations. Ocular serious AEs (SAEs) were reported with low frequency and were similar between the 2 treatment groups in both studies: 3.7% and 2.7% of patients in the KESTREL trial and 2.8% and 1.7% of patients in the KITE trial in the brolucizumab 6 mg group versus the aflibercept 2 mg group, respectively. The incidence of withdrawals due to AEs (WDAEs) was also similar between the 2 treatment groups: 1.6% and 1.1% of patients in the KESTREL trial and 2.8% and 2.2% of patients in the KITE trial in the brolucizumab group compared with the aflibercept group, respectively. There were 15 deaths in the KESTREL study, 8 (4.2%) in the brolucizumab group and 7 (3.7%) in the aflibercept group. In the KITE study, 13 (7.3%) deaths occurred in the brolucizumab group and 9 (5.0%) in the aflibercept group. According to the sponsor, none of the deaths were related to study treatment.

Table 2: Summary of Key Results From Pivotal and Protocol-Selected Studies

AE = adverse event; ANOVA = analysis of variance; ATE = arterial thromboembolic event; BCVA = best-corrected visual acuity; CI = confidence interval; CSFT = central subfield thickness; DRSS = diabetic retinopathy severity scale; FAS = full analysis set; LS = least squares; NA = not applicable; NEI VFQ-25 = National Eye Institute Visual Functioning Questionnaire–25; NR = not reported; q.8.w. = every 8 weeks; q.12.w. = every 12 weeks; q.16.w. = every 16 weeks; SAE = serious adverse event; WDAE = withdrawal due to adverse event.

^aAnalyzed using an ANOVA model with baseline BCVA categories (≤ 65 letters, > 65 letters), age categories (< 65 years, ≥ 65 years), and treatment as fixed-effect factors.

^bAnalyzed using an ANOVA model with baseline CSFT categories (< 450 μm, ≥ 450 μm to < 650 μm, ≥ 650 μm), age categories (< 65 years, ≥ 65 years), and treatment as fixed-effects factors.

^cAnalyzed using an analysis of covariance model with treatment as a fixed-effects factor and corresponding baseline value of the end point as a covariate.

^dThe 95% CI for binomial proportions is based on the Clopper-Pearson exact method. The statistical model used logistic regression adjusting for baseline DRSS score categories (≤ 4, ≥ 5), age categories (< 65 years, ≥ 65 years), and treatment as fixed-effect factors. The 95% CI for the treatment difference was estimated using the bootstrap method.

^eResults derived from the Kaplan-Meier analyses of the time needed to reach the interval of every 8 weeks that was estimated at week 60 and week 96, i.e., the time of the immediately preceding disease activity assessment visits.

Sources: Clinical Study Reports for KESTREL¹¹ and KITE.¹²

Critical Appraisal

The KESTREL and KITE trials were similarly designed randomized, double-blind, active-controlled, noninferiority phase III trials comparing brolucizumab (6 mg and 3 mg in the KESTREL trial; 6 mg in the KITE trial) with aflibercept (2 mg). The overall designs of the KESTREL and KITE trials were appropriate for the objectives of the studies. There were no major concerns with regard to the method of randomization, stratification, allocation concealment, and masking for randomized assignment. The baseline characteristics of the study population were generally well balanced between treatment arms and across studies, with the exception that patients in the KITE study had a relatively large imbalance in the number of ETDRS letters between the 2 treatment arms and a somewhat thicker retina at baseline, and were more likely to have received prior ocular medications compared with those enrolled in the KESTREL study. However, the clinical expert thought these differences were unlikely to impact the results between the studies.

In the KESTREL and KITE trials, the results of change from baseline in BCVA at week 52 using the per-protocol population were consistent with those in the full analysis set. In both studies, sensitivity analyses were conducted to assess the robustness of the hypothesis testing resulting from the primary analysis. Various methods were used to account for missing data, such as the mixed model for repeated measures (MMRM) modelling assuming a missing-at-random mechanism, or a last observation carried forward (LOCF) approach. The missing-at-random assumption may be a concern, given that the primary reasons for discontinuation from the study included patient decision, death, and AEs. Further, the LOCF method assumes that patient outcomes do not change after patients drop out, which may not hold true in practice. Therefore, performing additional sensitivity analyses that do not assume that missing data are missing at random could be useful. However, the results of the sensitivity analyses confirmed those of the primary analysis, suggesting these approaches in handling missing data were unlikely to introduce bias in the primary end point. The risk of attrition bias due to missing data was of particular concern for the National Eye Institute Visual Functioning Questionnaire–25 (NEI VFQ-25), as the proportion of missing data was greater than 20% for some treatment groups.

Statistical hierarchy was used for multiplicity adjustment for selected outcomes. Some important outcomes were not included in the hierarchy, such as vision-related HRQoL assessed by NEI VFQ-25, change in retinal thickness, and change in DR severity. In addition, the statistical hierarchy failed relatively early on at some point. In the KESTREL trial, the noninferiority of brolucizumab 3 mg to aflibercept 2 mg was not achieved at week 52. As per the study protocol, confirmatory testing did not proceed to assess the superiority of brolucizumab versus aflibercept for the following outcomes in the hierarchical testing procedure, which limits drawing definite conclusions on these outcomes.

Based on the patients’ baseline characteristics, the study population in the KESTREL and KITE trials may not fully represent the typical population with DME in Canada who would be receiving anti-VEGF therapy. The inclusion criteria for the KESTREL and KITE trials were reasonable and reflective of the eligibility criteria for anti-VEGF treatment in clinical practice. Although all patients enrolled in the KESTREL and KITE trials were naive to anti-VEGF treatment, and exhaustive exclusion criteria were used in the 2 studies, the clinical expert consulted by CADTH indicated that brolucizumab can be used in a broader population, such as those with a blood glucose level that is poorly controlled, or those who had received previous anti-VEGF therapy.

In the 2 pivotal studies, patients in the brolucizumab group could have their dosing interval extended, reduced (once patients on brolucizumab dropped back to every 8 weeks because of disease activity, they could not extend the treatment interval for the rest of the study, which may contradict clinical practice), or maintained postrandomization, based on the assessments of disease activity. Changes in treatment interval and dosage were not allowed for the treatment with aflibercept, and these patients remained on a fixed interval of every 8 weeks during the maintenance phase; this is contrary to clinical practice, as the product monograph for aflibercept states the treatment interval can be extended after the first year of treatment. According to the clinical expert, patients in the real world may not receive as many loading doses as in the clinical trial and it is possible that the outcomes observed in practice could differ from those shown in the clinical trials.

Indirect Comparisons

Description of Studies

The sponsor-submitted indirect treatment comparisons (ITCs) provided indirect evidence on the efficacy and safety of brolucizumab relative to other anti-VEGFs for adult patients with DME. The active comparators for brolucizumab included aflibercept, ranibizumab, and bevacizumab. Relevant RCTs were identified through a systematic literature search. Forty-three RCTs were included in the network meta-analysis (NMA). Outcomes of change in BCVA, retinal thickness, change in disease activity, study discontinuation, and safety were evaluated in the study population. A Bayesian NMA approach was used for data synthesis.

Efficacy Results

The sponsor-submitted NMA provided indirect comparative evidence for brolucizumab versus other anti-VEGF drugs. After including 43 trials in an NMA, none of the treatments were favoured when brolucizumab was compared with other active treatments for the treatment of DME, such as aflibercept, ranibizumab, or bevacizumab, in improving visual acuity and lessening disease severity. For most comparisons, the effect estimate was too imprecise (i.e., wide 95% credible intervals [CrIs]) to draw a conclusion about the comparative effects. Treatment with brolucizumab was associated with a greater reduction in retinal thickness than bevacizumab or ranibizumab. In addition, the ITC results suggested that patients treated with brolucizumab may receive fewer injections compared with other anti-VEGFs, though these results were derived from a naive comparison rather than an NMA and, in particular, should be interpreted with caution. The key limitations for the ITC are significant heterogeneity (in study design and patient characteristics) across the included RCTs, and imprecision around the effect estimates (i.e., wide 95% CrIs), which precluded drawing a conclusion for most outcome comparisons. This limits the conclusions that can be drawn from this ITC.

Harms Results

The risk of ocular AEs, nonocular AEs, and study discontinuation were evaluated in the NMA. The results suggested that none of the treatments were favoured for reduction in the risk of ocular or nonocular AEs. For all comparisons, the effect estimate was too imprecise (i.e., wide 95% CrIs) to draw a conclusion about the comparative effects. Limitations to the NMA preclude making firm conclusions about the relative risks of harm for brolucizumab compared with other anti-VEGFs.

Critical Appraisal

In the sponsor-provided ITC, the degree of heterogeneity (such as the patients’ disease characteristics at baseline) between the included studies was difficult to assess because of incomplete reporting of study characteristics. Description of trial design, sample size, and disease duration were reported. However, the ITC failed to report information related to the methods used for handling missing data. There was considerable variability in study design, year of conduct, sample size, and treatment regimen. The risk of bias in the included trials was assessed using a checklist from National Institute for Health and Care Excellence (NICE), but further details on how the risk-of-bias assessment was carried out were not provided.

Similarly, inadequate information about and variability in the baseline patient characteristics reported contribute to heterogeneity in the studies included in the ITC. Clinical trial eligibility criteria were described for the trials that were ultimately included in the NMA. However, many individual studies failed to report or inadequately reported patient characteristics, resulting in gaps in the extracted ITC data. There was a lack of information about key baseline characteristics, such as the presence of significant diabetic macular ischemia, patient’s previous treatment and response, presence of IRF, and presence of systemic comorbidities, including hypertension, chronic kidney disease, obesity, or cardiac conditions.

Most of the patients’ baseline characteristics were presented graphically. Even though some of these characteristics were comparable, such as age (which ranged from 58 to 66 years) and hemoglobin A1C level (which ranged from 7.3 to 8.7), heterogeneity still exists. The mean time since diagnosis of DME ranged from 1.2 to 3.4 years. The duration of diabetes ranged from 10 to 18 years in the included studies. Based on data from 26 trials, the mean BCVA scores ranged from 33 to 71 letters. Based on data from 25 trials, the mean retinal thickness at baseline ranged from 321 µm to 596 µm, and the majority of studies included patients with a retinal thickness of more than 400 µm. There was also heterogeneity in the reporting of methods for measuring retinal thickness and in the results of changes in retinal thickness. The apparent heterogeneity, based on the factors that were reported in combination with the inability to assess those that were not reported, means there is considerable uncertainty as to whether the assumptions related to homogeneity were met. The treatment effect of the study drug could differ by patient characteristics at baseline. Despite acknowledging the degree of heterogeneity, the technical report did not provide information on the assessments of heterogeneity (e.g., graphic representation of baseline characteristics, statistical tests) sufficient to fully understand the sources of heterogeneity. Therefore, it is plausible that the potential for heterogeneity could have influenced the comparative efficacy and safety estimates, and it is not possible to quantify or identify the direction of the bias. Several assumptions were made when defining treatment node assignment, for example: that a different dose did not impact the outcomes, that the ranibizumab 0.3 mg and 0.5 mg doses had similar efficacy, and that there were no significant differences in the effect of different regimens (aflibercept every 4 weeks, every 8 weeks, or as needed); it is uncertain whether these assumptions are valid.

For injection frequency, the ITC examined brolucizumab every 8 weeks and every 12 weeks. Data were pooled without conducting an NMA. Although drug administration with fixed treatment intervals according to a specific protocol is commonly observed in clinical trials, the clinical expert consulted by CADTH indicated that, in the real world, the treatment regimen could be more flexible, based on the patient’s response. Therefore, the findings from clinical trials may not reflect clinical practice.

Other Relevant Evidence

This section includes a summary of 1 additional relevant study, KINGFISHER,¹³ which was included in the sponsor’s submission to CADTH, as it was considered by the sponsor to provide further information on the safety of brolucizumab in patients with DME. The study compared the efficacy and safety of brolucizumab versus aflibercept in patients with DME, but the treatment intervals used for both drugs were shorter than the recommended intervals (beyond the loading phase) in the respective Health Canada–approved monographs.^14,15

The frequency of dosing for brolucizumab was selected based on previous studies that have suggested that, for some patients with DME, frequent dosing (i.e., every 4 weeks) with anti-VEGF therapy may be required to improve and maintain functional and anatomical outcomes.¹³ The clinical expert consulted by CADTH did not consider the efficacy results for the treatment intervals used in the KINGFISHER study to be relevant or generalizable to clinical practice; therefore, they are not included in this summary. However, given the frequency of administration, the clinical expert suggested that the safety data may provide information on whether intraocular inflammation and retinal vasculitis are idiosyncratic AEs related to brolucizumab itself rather than to the frequency of intravitreal injections.

Description of Study

One multicentre, randomized, double-blinded, active-controlled, parallel-group, prospective, phase III study, KINGFISHER,¹³ was conducted to evaluate the efficacy and safety of brolucizumab versus aflibercept in the treatment of adult patients with visual impairment due to DME. The primary objective was to demonstrate that brolucizumab was noninferior to aflibercept with respect to change in visual acuity from baseline up to week 52.

A total of 346 and 171 patients were randomized to brolucizumab 6 mg administered every 4 weeks and aflibercept 2 mg administered every 4 weeks, respectively. The 48-week double-blind treatment period was followed by a 4-week follow-up period up to week 52. Patients were evaluated every 4 weeks for the duration of the study. Only 1 eye was selected as the study eye and treated with the study drug. Study discontinuation rates were 10.1% and 8.8% in the brolucizumab and aflibercept arms, respectively. A total of 189 patients (54.6%) in the brolucizumab arm and 94 patients (55.0%) in the aflibercept arm received all 13 injections following the regimen of every 4 weeks.

The inclusion and exclusion criteria used in the KINGFISHER study¹³ were generally consistent with the eligibility criteria used in the pivotal KESTREL¹¹ and KITE¹² studies. The KINGFISHER study included adult patients with type 1 or type 2 diabetes who were diagnosed with visual impairment due to DME involving the centre of the macula and had a hemoglobin A1C level of 12% or less at screening. Of note, patients could either be treatment-I or could have previously received anti-VEGF therapy but could not have received any anti-VEGF therapies or undergone any intraocular surgery or laser photocoagulation within the 3-month period before baseline.

The mean age of patients was 60.9 years (SD = 10.59) in the brolucizumab arm and 60.2 years (SD = 9.31) in the aflibercept arm. There was a higher proportion of males than females in both arms (56.1% in the brolucizumab arm and 61.4% in the aflibercept arm). The mean time since diagnosis with DME was 20.6 months (SD = 29.94) and 18.2 months (SD = 25.60) in the brolucizumab and aflibercept arms, respectively. The mean BCVA score was 61.3 letters (SD = 10.14) and 60.5 letters (SD = 11.27 μm) in the brolucizumab and aflibercept arms, respectively. The mean CSFT was 514.1 μm (SD = 138.94 μm) and 511.2 μm (SD = 156.29 μm) in the brolucizumab and aflibercept arms, respectively. Most patients in both treatment arms had mild to moderately severe nonproliferative DR: 75.6% in the brolucizumab arm (n = 167) and 76.3% in the aflibercept arm (n = 90). Prior anti-VEGF treatment in the study eye was reported in 95 patients (27.5%) and 52 patients (30.4%) in the brolucizumab and aflibercept arms, respectively.

Harms Results

A total of 105 patients (30.3%) in the brolucizumab arm and 59 patients (34.5%) in the aflibercept arm reported at least 1 ocular AE. The most common ocular AE reported in the brolucizumab arm was vitreous detachment in 10 patients (2.9%). A total of 3 patients (0.9%) in the brolucizumab arm reported at least 1 serious ocular AE: vitreous hemorrhage in 2 patients (0.6%), and cataract subcapsular and retinal vasculitis in 1 patient (0.3%) each. No patients in the aflibercept arm reported any serious ocular AEs. | ||||| || | |||||||| |||||| ||| | |||||||| |||||| |||||||| |||| ||||| ||||||||| || ||| |||||||||||| ||| ||||||||||| ||||| ||||||||||||| ||| || || |||||| ||| |||| ||| |||| |||||| |||||| ||||| |||||||||| || ||||||| || || ||| |||||||| || |||||||| |||||| ||| | |||||||| ||||||| ||||||||||||||

Intraocular inflammation was reported in 14 patients (4.0%) in the brolucizumab arm versus 5 patients (2.9%) in the aflibercept arm. Retinal vasculitis was reported in 3 patients (0.9%) in the brolucizumab arm versus 1 patient (0.6%) in the aflibercept arm. Retinal vascular occlusion was reported in 1 patient in each arm (0.3% versus 0.6% in the brolucizumab and aflibercept arms, respectively). |||||||| |||||||||||||| |||||| |||| |||||||| || || |||||||| |||||| || ||| |||||||||||| ||| |||||| | |||||||| |||||| || ||| ||||||||||| |||| ||||||||||| |||||||| ||||||||| ||||||||| ||| |||||||| || | |||||||| |||||| || ||| |||||||||||| ||| |||||| | |||||||| |||||| || ||| ||||||||||| ||||. No reports of endophthalmitis were recorded.

Critical Appraisal

The trial was at low risk of bias due to the randomization; the 2 arms were generally balanced with respect to baseline demographic and disease characteristics. The trial was double-blind; however, there was some potential for unmasking because the personnel providing the injections were aware of the assigned treatment. The likelihood of unmasking and potential for bias in the reporting of subjective outcomes (i.e., some harms) is uncertain. The trial was powered for the safety assessment according to FDA recommendations. Attrition was relatively low and balanced across the arms, suggesting a low risk of attrition bias.

The clinical expert consulted by CADTH for this review advised that the results of the KINGFISHER study would not be generalizable to the patient population and clinical practice in Canada because the frequency of administration, every 4 weeks, is not a relevant treatment interval. Further, aflibercept is rarely administered every 4 weeks in clinical practice.

Conclusions

Brolucizumab 6 mg (every 8 weeks or every 12 weeks during maintenance therapy), was found to be noninferior to aflibercept 2 mg every 8 weeks for the mean change in BCVA from baseline after 1 year of treatment in anti-VEGF-naive patients with DME, based on evidence from 2 double-blind phase III RCTs (the KESTREL and KITE trials). Results for mean change in BCVA after 100 weeks of treatment were generally consistent with the 1-year results. The results of other BCVA outcomes, retinal thickness, and presence of IRF and/or SRF did not contradict the primary end point findings, but their interpretation is limited by the lack of a noninferiority margin and lack of adjustment for multiple testing. The reduction from baseline in retinal thickness after 1 year of treatment was greater with brolucizumab versus aflibercept in the KITE trial; in the KESTREL trial, the effect estimate was too imprecise (i.e., wide 95% CIs) to draw a conclusion. The results for presence of IRF and/or SRF suggested that brolucizumab may have been favoured over aflibercept, but a firm conclusion cannot be drawn due to the lack of adjustment for multiplicity. Approximately half of the brolucizumab group in each trial maintained the dosing interval of every 12 weeks after 52 weeks of treatment, though conclusions cannot be drawn due to the lack of adjustment for multiplicity and issues regarding the generalizability of the treatment regimens.

The safety profile for brolucizumab was generally comparable with that of aflibercept in the KESTREL and KITE trials. Results from a supportive study, KINGFISHER, which used a more frequent dosing regimen (brolucizumab 6 mg every 4 weeks), showed a safety profile for brolucizumab every 8 weeks or every 12 weeks that was similar to that observed in the KESTREL and KITE trials.

There is no direct comparative evidence on brolucizumab versus any anti-VEGFs other than aflibercept. Evidence from 1 NMA suggested that for change from baseline in BCVA at 12 months, none of the treatments were favoured when brolucizumab 6 mg was compared with aflibercept, ranibizumab, and bevacizumab; however, imprecision around the effect estimates (i.e., wide 95% CrIs) precluded drawing a conclusion for most comparison outcomes. In addition, the NMA results suggested that treatment with brolucizumab may be favourable compared with ranibizumab and bevacizumab for reducing retinal thickness. However, the presence of heterogeneity in the study design and patient characteristics limits the conclusions that can be drawn from the NMA. The results of a naive ITC suggested that treatment with brolucizumab may be related to fewer injections compared with other anti-VEGF drugs, but a conclusion cannot be drawn.

Introduction

Disease Background

DME is a vision-threatening complication of diabetes mellitus (both type 1 and type 2). The persistent elevation of blood glucose in people with diabetes causes damage to the capillaries, such as those in the eye, resulting in DR.¹ Some patients with DR and especially those with continually poorly managed blood glucose can experience swelling in the retina, known as DME.² Generally, DME manifests as slowly progressive vision loss. The degree of vision loss can vary considerably and depends on the severity, duration, and location of IRF, among other factors. Clinically significant macular edema can be defined by retinal thickening at or within 500 µm of the centre of the macula.^16,17 Signs of DME include blurred vision, retinal hemorrhages, retinal detachment, colours appearing “washed out” or faded, changes in contrast sensitivity, impaired colour vision, gaps in vision (scotomas), and potentially permanent vision loss. Untreated DME is considered to be the leading cause of vision loss, visual disability, and legal blindness in people with DR.^2-4

Prevalence of macular edema in patients with type 1 diabetes, patients with type 2 diabetes treated with insulin therapy, and patients treated with antihyperglycemic therapies have been estimated at 11%, 15%, and 4%, respectively.¹⁸ A Canadian retrospective study using records from the Southwestern Ontario database estimated the prevalence of DME in adults with diabetes to be 15.7% and the prevalence of vision loss due to DME to be 2.56%.¹⁹ In this study, more than 50% of patients with DME experiencing vision loss were older than 60 years and more than 22% of patients with DME experiencing vision loss were patients within a First Nations community.¹⁹ Indigenous populations in Canada are disproportionally affected by diabetes,¹⁶ with higher prevalence rates of DR compared with the general population,^20,21 although accurate data on vision loss in this population is limited.¹⁶ Based on the Ontario study’s estimates,¹⁹ and a 2020 Statistics Canada estimate of 2.3 million adults in Canada with diabetes, there are approximately 60,000 adults with DME in Canada who experience vision impairment requiring treatment.^8,9 The incidence and prevalence of diabetes in Canada are projected to increase in the coming years in tandem with an aging population and rising rates of obesity, and this rise in diabetes cases is expected to lead to corresponding increases in DR and DME.¹⁶

Generally, vision loss is associated with significant morbidity, including increased falls, hip fractures, and mortality.²² In addition, it has been suggested that amputation and vision loss due to DR are independent predictors of early death among patients with type 1 diabetes.²³ Progressive visual impairment typically results in significant decrements in daily functioning and quality of life, and indirect costs due to lost productivity are high if left untreated;^5-7 therefore, early detection and treatment of DME is vital.^24,25

Standards of Therapy

Current therapies for DME in Canada include non-pharmacological interventions (laser therapy and vitrectomy) and pharmacological interventions (intravitreal anti-VEGF drugs and intravitreal steroids). Health Canada–approved anti-VEGF drugs for DME treatment include ranibizumab and aflibercept (with bevacizumab used off-label), while approved intravitreal steroids include dexamethasone.

Macular laser photocoagulation (including focal, grid laser, or panretinal) therapy for DME was the standard of care for more than 25 years before the introduction of anti-VEGF drugs and is still widely used either alone or in combination with anti-VEGF treatment.¹⁷ Laser therapy has been shown to slow and/or stabilize vision loss, but has been minimally effective in restoring vision.²⁶ Clinical studies have shown robust efficacy and safety for frequent (e.g., monthly or bimonthly) anti-VEGF injections for the treatment of DME patients.^27-29 The results from these trials have demonstrated that treatment with anti-VEGF drugs substantially improves visual and anatomic outcomes compared with laser photocoagulation, and avoids the ocular side effects associated with laser treatment. Canadian evidence-based guidelines and clinical treatment algorithms recommend anti-VEGF injections as therapy (alone or in conjunction with focal laser therapy) for most patients with clinically significant DME involving central macular thickening. Cases without central macular thickening are recommended to receive focal laser treatment, while eyes with vitreomacular traction and macular edema are recommended as candidates for vitrectomy.¹⁶

The first of the anti-VEGF drugs to be approved in Canada for the treatment of DME was ranibizumab (a humanized recombinant monoclonal antibody fragment with anti-VEGF activity).³⁰ The recommended dose of ranibizumab is 0.5 mg injected intravitreally once a month and continued until maximum visual acuity is achieved, confirmed by stable visual acuity for 3 consecutive monthly assessments performed while on the treatment.³⁰ Other anti-VEGF therapies include aflibercept at the recommended dose of 2.0 mg administered by intravitreal injection monthly for the first 5 consecutive doses, followed by 1 injection every 2 months.¹⁵ After the first year, injections of aflibercept may be extended by increments of up to 2 weeks based on disease activity, although data on intervals longer than 4 months are limited.¹⁵ Bevacizumab, an anti-VEGF drug, is also used in the first line in clinical practice but does not have any Health Canada–approved indications for the treatment of DME and is reimbursed only in certain jurisdictions.

Although anti-VEGF therapies are widely accepted as the standard of care for patients with DME, they require frequent injections (8 to 12 per eye per year) to achieve desirable outcomes, creating a high treatment burden for patients and caregivers. Anti-VEGF therapies are also associated with an increased risk of cerebrovascular and cardiovascular events such as thromboembolic events; therefore, they may not be appropriate in all DME patients, especially in patients with prior stroke or other cardiovascular comorbidities.^10,15,30-32 Some patients have an inadequate response to anti-VEGF treatment, although the frequency of suboptimal response is unclear. According to the clinical expert consulted for this review, around 10% of patients may have an inadequate response, while some studies have reported a suboptimal response in as high as 25%³³ to 40%³⁴ of patients on anti-VEGF therapy, depending on how suboptimal response was defined.³⁴ There is limited evidence of the benefit and risks of continuous anti-VEGF injections among patients who did not respond well to prior anti-VEGF therapy.³⁴

In Canada, intravitreal dexamethasone implants are indicated for use in DME patients who are pseudophakic. Triamcinolone acetonide monotherapy administered as an intravitreal steroid injection is considered for off-label use in Canada for the treatment of macular edema in pseudophakic patients, according to the clinical expert consulted for this CADTH review. These intravitreal corticosteroids are generally reserved for patients who do not respond well to anti-VEGF therapy.

Drug

Brolucizumab is a humanized VEGF inhibitor that binds to vascular endothelial growth factor A (VEGF-A) isoforms (e.g., VEGF110, VEGF121, and VEGF165), thereby preventing the binding of VEGF-A to its receptors, VEGFR-1 and VEGFR-2. By inhibiting VEGF-A binding, brolucizumab suppresses endothelial cell proliferation in vitro and reduces neovascularization and vascular permeability.¹⁰

On November 30, 2022, brolucizumab was approved by Health Canada for the treatment of patients with DME. The reimbursement criteria for brolucizumab requested by the sponsor are the same as the Health Canada indication.

As per the product monograph, the recommended dose for brolucizumab is 6 mg (0.05 mL) administered by intravitreal injection every 6 weeks for the first 5 doses. Thereafter, the physician may modify the treatment intervals based on disease activity, as assessed by visual acuity and/or anatomical parameters. In patients without disease activity, treatment up to every 12 weeks (3 months) could be considered. In patients with disease activity, treatment every 8 weeks (2 months) could be considered; however, the interval between 2 doses should not be less than every 8 weeks (2 months).¹⁰

A table describing key characteristics of commonly used anti-VEGF treatments for DME is presented in Table 3.

Table 3: Key Characteristics of Brolucizumab, Aflibercept, Ranibizumab, and Bevacizumab

ATE = arterial thromboembolic event (includes nonfatal stroke, nonfatal myocardial infarction, or vascular death); DME = diabetic macular edema; IOP = intraocular pressure; mAb = monoclonal antibody; PIGF = placental growth factor; q.4.w. = every 4 weeks; q.6.w. = every 6 weeks; q.8.w. = every 8 weeks; q.12.w. = every 12 weeks; q.16.w. = every 16 weeks; VA = visual acuity; VEGF = vascular endothelial growth factor; VEGF-A = vascular endothelial growth factor A.

^aBevacizumab is used off-label in the treatment of DME.

^bHealth Canada–approved indication.

^cBased on expert opinion.

Sources: Product monographs for Beovu,¹⁰ Eylea,^15,31 Lucentis,³⁰ and Avastin.³²

Stakeholder Perspectives

Patient Group Input

This section was prepared by CADTH staff based on the input provided by patient groups. The full original patient input(s) received by CADTH have been included in the stakeholder section at the end of this report.

Patient input was provided as a joint submission by 5 groups: Fighting Blindness Canada, Canadian Council of the Blind, CNIB, Vision Loss Rehabilitation Canada, and Diabetes Canada. The information used to inform the submission was based on an online survey of people in Canada living with DR or DME that was conducted in the first months of 2020 by the submitting organizations. A total of 67 people responded to the survey; many (44.4%) were between the ages of 61 and 80 years (n = 54) and the majority (76.1%) of respondents reported DME or DR in both eyes. Most respondents were either working full time (38.9%) or were retired (33.3%) (n = 54). A separate survey conducted by Canadian Council of the Blind in April 2020 further supported the submission by providing data on the impact of the COVID-19 pandemic on people living in Canada who are blind, deaf-blind, or partially sighted (n = 572).

Survey respondents emphasized that DR and DME have substantial and life-altering impacts on daily life, including on reading, driving, and using a phone. In addition to the concern for their eyesight worsening, coping with everyday life and general safety when outside of the home were identified as notable concerns of respondents during the preceding month. The results of the Canadian Council of the Blind survey showed that fear, anxiety, loneliness, and other psychosocial impacts were intensified for patients with age-related macular edema and DR during the pandemic.

The majority (56.4%) of respondents indicated they were currently receiving injections for DR or DME; the most common therapeutic options were Lucentis (29.4%), Eylea (24.6%), Avastin (20.2%), and Ozurdex (13.5%). Most (54.5%) respondents indicated they were satisfied with their injections and 63.6% indicated the injections have helped them avoid losing more eyesight (n = 22). Of note, 31.8% of respondents reported missing injections in the last year. According to respondents (n = 6), the reasons for cancelled or delayed appointments in the past included being too busy (50.0%), feeling unwell (33.3%), not being able to find someone to take them to the appointment (16.7%), and fear of injections (16.7%). According to respondents (n = 22), the most difficult part of eye injection appointments was the long wait times (50.0%), finding someone to take them to and from the appointment (31.8%), anxiety or fear about the injection (27.3%), and taking time off work (27.3%). No survey respondents reported experience with brolucizumab.

The submitting organizations indicated that any treatment that reduces the physical, psychological, and logistical strain on patients would be preferred. The submitting organizations suggested this may be addressed by a treatment that is less invasive, or similarly invasive but administered less frequently. Further, the submitting organizations suggested that any treatment that can extend the interval between injections while still minimizing vision loss would be considered advantageous for patients living with vision loss, particularly in the context of the COVID-19 pandemic.

Clinician Input

Input From Clinical Experts Consulted by CADTH

All CADTH review teams include at least 1 clinical specialist with expertise regarding the diagnosis and management of the condition for which the drug is indicated. Clinical experts are a critical part of the review team and are involved in all phases of the review process (e.g., providing guidance on the development of the review protocol; assisting in the critical appraisal of clinical evidence; interpreting the clinical relevance of the results; and providing guidance on the potential place in therapy). The following input was provided by 1 clinical specialist with expertise in the diagnosis and management of DME.

The clinical expert consulted by CADTH indicated that the treatment goals for DME with current therapies are to delay and, in some cases, reverse the progression of DME and/or DR, and to improve vision-related and general quality of life. While current anti-VEGF treatments for DME have been useful for treating DME over the past 10 to 15 years, they need to be given intravitreally by trained clinicians at treatment visits once every 1 to 3 months on an ongoing basis, often for years. This frequent administration poses a significant burden on patients and caregivers, especially in Canada where travel distances can be long and more challenging in winter. The clinical expert noted that longer-acting treatments would fill a significant unmet medical need by improving the convenience of the treatment regimen and reducing the burden on patients and caregivers. It could also improve outcomes by, in part, increasing adherence to treatment regimes. Additionally, the expert noted that not all patients respond to available treatments and, in some cases, a patient’s disease may become refractory to current treatment options.

Place in Therapy

According to the clinical expert, brolucizumab is expected to have a place in therapy that is similar to other anti-VEGFs as a first-line or later line of treatment in patients with DME. Treatment with brolucizumab is anticipated to reduce the burden of care by increasing the intervals between treatments while still maintaining therapeutic benefit compared with other anti-VEGFs, which could potentially address the unmet need related to frequent treatment visits.

In the clinical expert’s opinion, there are no clinical reasons to make it mandatory that patients first try other treatments before initiating brolucizumab. Brolucizumab is expected to be prescribed as a first-line (or later-line) treatment for DME and, as with any of the existing treatments, earlier initiation is important to achieve the best clinical outcomes.

Patient Population

Patients with DR associated with vision loss secondary to centre-involving macular edema are the best candidates for brolucizumab, according to the clinical expert. The clinical expert indicated that brolucizumab can be used in patients who are treatment-naive or require a change in therapy due to an inadequate response to other anti-VEGF drugs. Patients with better baseline visual acuity, centre-involving edema of recent onset, and better control of diabetes and comorbid conditions may be more likely to benefit from treatment. Patients who present with major structural damage to the macular retina (e.g., macular atrophy or fibrosis) may not be suitable candidates for treatment. Suitability for treatment would be assessed using clinical exam, IV fluorescein angiography, and OCT, potentially with the addition of OCT angiography. As OCT is used in current clinical practice, misdiagnosis is unlikely to occur. OCT is used not only for diagnosis but also for prognosis and follow-up.

Assessing Response to Treatment

The clinical expert noted that clinical evaluation and OCT should be performed at almost every dosing visit to assess treatment response with a treat-and-extend approach to achieve the longest sustainable interval without recurrence, and that monitoring between dosing visits would not be required. Key assessment outcomes include change in visual acuity, retinal thickness, injection frequency, and the presence of retinal fluid.

The clinical expert reported that optimal response to anti-VEGFs is generally achieved 6 to 12 months after initiation of therapy. In their experience, the majority of patients can achieve stabilized vision and improved quality of life, and about 50% to 65% of patients can achieve improvement in visual acuity.

The clinical expert noted that when assessing the magnitude of change in visual acuity, it is crucial to keep in mind that patients with better vision at baseline generally have less room for improvement than those with poor baseline vision. As such, the clinical expert reported there is no agreed-upon threshold indicative of a clinically meaningful change in visual acuity in patients with DME.

The clinical expert noted that the presence of SRF or IRF is an indicator of active disease that prompts the modification of the treatment plan, often involving a reduction in the injection interval.

Discontinuing Treatment

The clinical expert indicated that brolucizumab should be discontinued in patients experiencing treatment futility with proof of irreversible anatomic or functional damage, such as macular atrophy (ischemia) and fibrosis.

Prescribing Conditions

The clinical expert recommended retina subspecialty care as the most appropriate treatment setting for the prescription and administration of brolucizumab, especially in urban areas. In rural settings, trained comprehensive ophthalmologists with experience and expertise in managing DME may also be able to provide care.

Clinician Group Input

No input was received from any clinician groups for this submission.

Drug Program Input

The drug programs provide input on each drug being reviewed through CADTH’s reimbursement review processes by identifying issues that may impact their ability to implement a recommendation. The implementation questions and corresponding responses from the clinical experts consulted by CADTH are summarized in Table 4.

Table 4: Summary of Drug Plan Input and Clinical Expert Response

BCVA = best-corrected visual acuity; CDEC = CADTH Canadian Drug Expert Committee; CSFT = central subfield thickness; DME = diabetic macular edema; SD-OCT = spectral domain optical coherence tomography; VEGF = vascular endothelial growth factor.

Clinical Evidence

The clinical evidence included in the review of brolucizumab is presented in 3 sections. The first section, the systematic review, includes pivotal studies provided in the sponsor’s submission to CADTH and Health Canada, as well as those studies that were selected according to an a priori protocol. The second section includes indirect evidence from the sponsor and indirect evidence selected from the literature that met the selection criteria specified in the review. The third section includes additional relevant sponsor-submitted studies that were considered to address important gaps in the evidence included in the systematic review.

Systematic Review (Pivotal and Protocol-Selected Studies)

Objectives

To perform a systematic review of the beneficial and harmful effects of brolucizumab 6 mg/0.05 mL solution for intravitreal injection for the treatment of DME.

Methods

Studies selected for inclusion in the systematic review included pivotal studies provided in the sponsor’s submission to CADTH and Health Canada, as well as those meeting the selection criteria presented in Table 5. Outcomes included in the CADTH review protocol reflect outcomes considered to be important to patients, clinicians, and drug plans.

Of note, the systematic review protocol presented subsequently was established before the granting of a Notice of Compliance by Health Canada.

The literature search for clinical studies was performed by an information specialist using a peer-reviewed search strategy according to the PRESS Peer Review of Electronic Search Strategies checklist.³⁵

Published literature was identified by searching the following bibliographic databases: MEDLINE All (1946–) through Ovid and Embase (1974–) through Ovid. All Ovid searches were run simultaneously as a multifile search. Duplicates were removed using Ovid deduplication for multifile searches, followed by manual deduplication in Endnote. The search strategy comprised both controlled vocabulary, such as the National Library of Medicine’s MeSH (Medical Subject Headings), and keywords. The main search concept was Beovu (brolucizumab). Clinical trial registries were searched: the US National Institutes of Health’s clinicaltrials.gov, WHO’s International Clinical Trials Registry Platform (ICTRP) search portal, Health Canada’s Clinical Trials Database, and the European Union Clinical Trials Register.

No filters were applied to limit the retrieval by study type. Retrieval was not limited by publication date or by language. Conference abstracts were excluded from the search results. Refer to Appendix 1 for the detailed search strategies.

The initial search was completed on July 28, 2022. Regular alerts updated the search until the meeting of the CADTH Canadian Drug Expert Committee (CDEC) on November 23, 2022.

Grey literature (literature that is not commercially published) was identified by searching relevant websites from the Grey Matters: A Practical Tool For Searching Health-Related Grey Literature checklist.³⁶ Included in this search were the websites of regulatory agencies (FDA and European Medicines Agency). Google was used to search for additional internet-based materials. Refer to Appendix 1 for more information on the grey literature search strategy.

These searches were supplemented by reviewing bibliographies of key papers. In addition, the sponsor of the drug was contacted for information regarding unpublished studies.

Two CADTH clinical reviewers independently selected studies for inclusion in the review based on titles and abstracts, according to the predetermined protocol. Full-text articles of all citations considered potentially relevant by at least 1 reviewer were acquired. Reviewers independently made the final selection of studies to be included in the review, and differences were resolved through discussion.

Findings From the Literature

A total of 245 citations were identified from the literature search and 3 potentially relevant citations were identified from other sources. After screening, 4 reports of 2 unique studies were included in the systematic review (Figure 1). The included studies are summarized in Table 6.

Figure 1: Flow Diagram for Inclusion and Exclusion of Studies

Table 6: Details of Included Studies

ADA = antidrug antibody; AE = adverse event; BCVA = best-corrected visual acuity; CSFT = central subfield retinal thickness; DM = diabetes mellitus; DME = diabetic macular edema; DR = diabetic retinopathy; DRSS = Diabetic Retinopathy Severity Scale; ETDRS = Early Treatment Diabetic Retinopathy Study; IVT = intravitreal treatment; NEI VFQ-25 = National Eye Institute Visual Functioning Questionnaire–25; PDR = proliferative diabetic retinopathy; q.4.w. = every 4 weeks; q.6.w. = every 6 weeks; q.8.w. = every 8 weeks; q.12.w. = every 12 weeks; q.16.w. = every 16 weeks; RCT = randomized controlled trial; SD-OCT = spectral domain optical coherence tomography; VEGF = vascular endothelial growth factor.

Note: 1 additional report was included (submission).³⁸

^aInformation was obtained from Clinicaltrials.gov.

Sources: Clinical Study Reports for KESTREL¹¹ and KITE.¹²

Description of Studies

Two studies were included in the systematic review: KESTREL¹¹ and KITE.¹² They were identically designed phase III, multicentre, randomized, double-blind, active-controlled, noninferiority trials that aimed to evaluate the efficacy and safety of brolucizumab compared with aflibercept in DME patients. Only treatment-naive patients were eligible for the studies.

The KESTREL trial (N = 566) was a 3-arm study designed to evaluate the efficacy and safety of brolucizumab 6 mg and 3 mg compared with aflibercept 2 mg in the study population. Patients were randomized using interactive response technology to 1 of the treatment arms in a ratio of 1:1:1, stratified by Japanese ethnicity. The dose of 3 mg was not recommended in the product monograph; therefore, results related to this dose were not included in this review.¹⁰ The KITE trial (N = 360) was a 2-arm study designed to evaluate the efficacy and safety of brolucizumab 6 mg compared with aflibercept 2 mg in patients with DME. Eligible patients were randomized using interactive response technology to 1 of the 2 treatment arms in a ratio of 1:1, stratified by systemic exposure sampling. All patients, investigators, and selected staff from the sponsor who had contact with patients or investigators or those who were involved in the direct conduct of the study until the final database lock were masked to the treatment assignment. In both studies, to ensure masking during study conduct, patients in both the brolucizumab arm and aflibercept arm received sham or active injections at every visit (except weeks 20, 28, and 100), and the investigational sites had masked and unmasked staff. The primary efficacy outcome in the 2 studies was change the from baseline in BCVA at week 52.

Both trials consisted of a screening period of up to 2 weeks followed by a 96-week double-blind phase. During the screening phase, patients were assessed for study eligibility based on prespecified inclusion and exclusion criteria.

Inclusion and Exclusion Criteria

The key inclusion criteria for both studies included patients aged 18 years or older with type 1 or 2 diabetes with macular thickening secondary to DME involving the centre of the fovea (CSFT ≥ 320 µm on spectral domain OCT), a hemoglobin A1C level of 0% or less, and BCVA scores of 78 to 23 letters using the ETDRS protocol (20/40 to 20/320 Snellen equivalent). Patients were excluded if they had active proliferative DR or had received previous treatment with any anti-VEGF drug or dexamethasone intravitreal implant. They were also excluded from the study if they had any active intraocular or periocular infection, active intraocular inflammation, or uncontrolled glaucoma in the study eye. The key inclusion and exclusion criteria for the trials are shown in Table 6. Only 1 eye was assigned as the study eye in the studies. If both eyes were eligible, the eye with the worse BCVA at baseline was selected (unless the other eye was deemed by investigators to be more suitable for treatment).

Baseline Characteristics

A summary of the baseline characteristics of the full analysis set population in both studies is shown in Table 7. Overall, the baseline demographic and ocular characteristics of patients were balanced between the treatment arms within each study. The mean age of patients enrolled was 62.2 to 64.4 years. The majority of the patients were male (58% to 67%) and white (73% to 84%). At baseline, the level of hemoglobin A1C was 7.4% to 7.7%, on average. The mean time since the diagnosis of DME was 9.4 to 12.5 months in the KESTREL trial and 9.9 to 10.4 months in the KITE trial; the mean baseline CSFT in these 2 studies ranged from 453 μm to 484 μm. The mean baseline BCVA score was approximately 64 letters in both studies although, in the KITE trial, there was an imbalance in the number of ETDRS letters: 66.0 letters (SD = 10.8) in the brolucizumab group and 63.7 letters (SD = 11.7) in the aflibercept group. IRF was present in almost all patients, while SRF was present in only a small portion of the study population (31% to 37%). The majority of the patients were categorized as having either mild or severe nonproliferative DR. In the KESTREL trial, the mean CSFT was 453.1 μm (SD = 123.4 µm) in the brolucizumab 6 mg arm and 475.6 μm (SD = 135.8 µm) in the aflibercept 2 mg arm. In the KITE trial, the mean CSFT was 481.1 µm (SD = 132.5 µm) in the brolucizumab 6 mg arm and 484.4 µm (SD = 134.6 µm) in the aflibercept 2 mg arm. Only a small number of patients had received prior ocular medications (2% to 3% in the KESTREL trial and 9% to 11% in the KITE trial). Concomitant use of ocular medications was reported in 31% to 37% of the patient population.

Interventions

In both studies, patients randomized to the brolucizumab groups (brolucizumab 3 mg or 6 mg in the KESTREL trial, and brolucizumab 6 mg in the KITE trial) received 5 loading doses every 6 weeks followed by dosing every 12 weeks in the maintenance phase, with optional adjustment to every 8 weeks if disease activity was identified at predefined assessment visits. In the KITE trial, patients had the option to extend the treatment interval by 4 weeks at week 72.

In both studies, at the beginning of the maintenance phase, patients in the brolucizumab group were initially on a treatment schedule of every 12 weeks. The patients remained on that schedule as long as the masked investigator did not identify DME disease activity that, in their opinion, required more frequent anti-VEGF treatment. If a need for a regimen of every 8 weeks was determined to be appropriate, the patients were assigned to receive injections every 8 weeks thereafter, up to the end of the study. In the KITE study, a disease stability assessment was carried out at week 72 (i.e., no disease activity detected during the last 2 DAA visits). If disease stability was observed at week 72, injection intervals could be extended by 4 weeks (i.e., from every 12 weeks to every 16 weeks or from every 8 weeks to every 12 weeks).

In the 2 studies, patients who were randomized to the comparator group (aflibercept 2 mg) received 5 loading doses every 4 weeks followed by fixed dosing every 8 weeks in the maintenance phase. No treatment frequency adjustments were permitted in the aflibercept 2 mg arm.

In the KESTREL and KITE trials, if patients needed 8-week interval dosing at a previous DAA visit, rescue treatment was allowed with laser (focal and/or grid) photocoagulation, along with study treatment, from week 36 onward if DME worsened and caused a loss of 10 or more letters at 2 consecutive visits or 15 or more letters at 1 visit compared with the best previous measurement, with BCVA not better than baseline. Panretinal photocoagulation was permitted at any time during the study as deemed necessary by the investigator, and the patient was allowed to continue the study with the assigned study treatment.

Dose adjustments and/or interruptions were not permitted in either study unless the interruptions were warranted by an AE. Treatment frequency adjustments (from every 12 weeks to every 8 weeks in the KESTREL trial, or from every 16 weeks or every 12 weeks to every 8 weeks in the KITE trial) were permitted in the brolucizumab arms if DME disease activity was identified by the masked investigator in the prespecified DAA visits. No treatment frequency adjustments were permitted in the aflibercept 2 mg arm.

Outcomes

A list of the efficacy and harm end points identified in the CADTH review protocol that were assessed in the clinical trials included in this review is provided in Table 8. These end points are further summarized subsequently. A detailed discussion and critical appraisal of the outcome measures is provided in Appendix 3.

Change From Baseline in Visual Acuity

The change from baseline in BCVA (ETDRS letters) averaged at week 52 was the primary end point in both studies. Secondary end points of change in visual acuity included:

• change from baseline in BCVA at each visit up to week 100

• proportion of patients gaining 5, 10, 15, or more ETDRS letters in BCVA from baseline at each visit up to week 100

• proportion of patients avoiding loss in BCVA of 5, 10, 15, or more ETDRS letters from baseline at each visit up to week 100.

These outcomes were also assessed over time (i.e., at all assessment time points through week 100).

The BCVA score was measured based on the ETDRS visual acuity chart assessed at a starting distance of 4 m. ETDRS charts consist of 70 letters that are distributed across 14 rows. Each row contains a series of 5 letters of equal difficulty, with standardized spacing between the letters and rows. The level of difficulty increases between successive rows as the size of the characters decreases. The BCVA score corresponds to the number of letters 1 can read from the ETDRS chart. The visual acuity letter score is equal to the total number of letters read correctly at 4 m plus 30. If fewer than 20 letters are read correctly at 4 m, the visual acuity letter score is equal to the total number of letters read correctly at 4 m plus the total number of letters in the first 6 lines read correctly at 1 m. The ETDRS letter score could result in a maximum score of 100, with higher scores indicating better visual acuity. Reading more lines (i.e., more letters) indicates better visual acuity.^39,40 In terms of a minimal important difference (MID), no data were identified in patients with DME. For more information regarding the ETDRS refer to Appendix 3.

ETDRS results can be converted to Snellen fractions, another common measure of visual acuity, in which the numerator indicates the distance at which the chart was read, and the denominator indicates the distance at which a person may discern letters of a particular size. A larger denominator indicates worsening vision; for example, a person with 20/100 vision can read letters at 20 feet that a person with 20/20 vision could read at 100 feet.

Retinal Thickness

The retinal thickness of the study eye was measured using spectral domain OCT. This is a key anatomic parameter of the central macula and is defined as the average thickness of a 1 mm circular area centred around the fovea and measured from the Bruch membrane to the internal limiting membrane, inclusive. The change from baseline in CSFT from baseline at each visit up to week 100 was a secondary outcome in both studies. A reduction in CSFT is considered a favourable outcome in the treatment of DME; however, an MID has not been established. For more information on the use of OCT to measure changes in retinal thickness, refer to Appendix 3.

HRQoL and Vision-Related Function

Both studies measured the change from baseline in the NEI VFQ-25 composite score at week 52 (and over time) as a secondary end point. The NEI VFQ-25 was administered from baseline up to week 100 by masked site staff. The NEI VFQ-25 is a questionnaire developed to measure vision-targeted quality of life. The questionnaire consists of 25 items relevant to 11 vision-related constructs as well as a single-item, general-health component. The overall composite score ranges from 0 to 100, with 0 representing worst vision-related function and 100 representing best vision-related function. In addition, there are 12 subscale scores (e.g., near vision, distance vision, driving).⁴¹ The questionnaire has a reported MID of between 3.3 and 6.13 points for the overall composite score. A psychometric validation study of the NEI VFQ-25 specifically in patients with DME found that the MID for each NEI VFQ-25 domain ranged from 8.80 (general vision) to 14.40 (role difficulties) and produced a composite score MID of 3.33 to 6.13 points, depending on the approach used to estimate the MID.⁴² For more information on the properties of the NEI VFQ-25, refer to Appendix 3.

Blindness (Legal)

Legal blindness is defined as a BCVA of 20/200 or less in both eyes, measured with a Snellen chart, and/or a visual field of 20 degrees or narrower.⁴³ This outcome was not assessed in either study.

Change From Baseline in DR Severity

The severity of DR was evaluated using the ETDRS DRSS score assessed by the central reading centre based on colour fundus photography images and fluorescein angiography in the study eye. The ETDRS DRSS is a scale that consists of 13 levels of graded photographic characteristics that were defined to categorize the severity of DR for individual eyes, ranging from no retinopathy to severe vitreous hemorrhage. Each of the 13 levels on the scale is defined by a set of criteria based on the presence and/or severity of abnormalities, with scores increasing in order of severity from 10 to 85. Higher scores on the scale indicate worsening of DR. Step progression refers to an increase in photographic level that can be used to describe the change (improvement or worsening) in DR over time.⁴⁴ In the ETDRS, the proportion of eyes that progressed 2 or more levels at follow-up was relatively similar among all severity categories at the 1-year follow-up time point, establishing 2-step progression as a reasonable outcome measure for all baseline retinopathy levels.⁴⁴ For more information regarding the DRSS, refer to Appendix 3. The end points related to DR were derived from the ETDRS DRSS score assessed by the central reading centre based on colour fundus photography images in the study eye at different visits. The proportion of patients with an improvement of 2 or more steps in DR severity from baseline on the ETDRS DRSS at week 52 was a secondary end point in both studies. Other end points included a 2-step or greater or 3-step or greater improvement or worsening from baseline over time on the ETDRS DRSS.

When the ETDRS DR severities were evaluable, they were categorized on the original scale with scores varying from 10 (DR absent) to 85 (very advanced proliferative DR). All DRSS values were then converted into a 12-level scale, allowing the derivation of the 2-step and 3-step change from baseline for each postbaseline assessment (weeks 28, 52, 76, and 100). All DRSS analyses presented in this report were based on the 12-level scale. The following table illustrates how the severity of DR was defined based on the original DRSS and a modified 12-point scale (Table 9).

Table 9: Definition of DRSS Using the Original DRSS and the 12-Point Scale

DR = diabetic retinopathy; DRSS = Diabetic Retinopathy Severity Score; NPDR = nonproliferative diabetic retinopathy; PDR = proliferative diabetic retinopathy.

Sources: Clinical Study Reports for KESTREL¹¹ and KITE.¹²

Proportion of Patients With Presence of SRF or IRF

The proportion of patients with an absence of IRF and/or SRF at week 52 and over time were measured as secondary end points. SRF and IRF located specifically in the central subfield (within a 1 mm diameter of the centre of the macula) were of interest.

Frequency of Injections

The proportion of patients in the brolucizumab arm on a treatment interval of every 8 weeks, every 12 weeks, or every 16 weeks at 1 year and 2 years as well as the treatment intervals in this arm over time were secondary outcomes of the studies. The time to first need for treatment every 8 weeks in the study population was reported in both studies, which identified disease activity that, according to the masked investigator, required more frequent anti-VEGF treatment.

Harm Outcomes

In both studies, the safety analysis included the incidence and severity of ocular and nonocular AEs occurring throughout the study period. The occurrence of AEs was assessed at all assessment time points.

Statistical Analysis

Sample Size Calculation

In the KESTREL trial, a sample size of 160 patients per arm was required to demonstrate noninferiority (using a noninferiority margin of 4 ETDRS letters) between brolucizumab 6 mg or 3 mg versus aflibercept 2 mg in the full analysis set population with respect to the change in BCVA from baseline at week 52, with 90% power at a 1-sided alpha level of 0.025, assuming equal means and a common SD of 11 letters. Considering a dropout rate of 10%, a total of 534 patients (178 per arm) were planned to be randomized.

Similarly, in the KITE trial, a sample size of 160 patients per arm was required to demonstrate noninferiority (using a noninferiority margin of 4 ETDRS letters) between brolucizumab 6 mg versus aflibercept 2 mg in the full analysis set population with respect to the change in BCVA from baseline at week 52, with 90% power at a 1-sided alpha level of 0.025, assuming equal means and a common SD of 11 letters. Considering a dropout rate of 10%, a total of 356 patients (178 per arm) were planned to be randomized.

Statistical Analysis for Efficacy Outcomes

In the KESTREL and KITE trials, the primary and relevant secondary efficacy end points were assessed using a hierarchical approach for multiple testing. The objective of the analyses on the primary and first key secondary end points was to demonstrate noninferiority of brolucizumab to aflibercept with respect to change from baseline in BCVA, using a margin of 4 ETDRS letters. Superiority testing for secondary end points was performed only if noninferiority was demonstrated for the change in BCVA from baseline to week 52, or from baseline to week 40 through to week 52.

In the KESTREL trial, the following noninferiority hypotheses are related to a noninferiority margin of 4 letters:

• H₁. Brolucizumab 6 mg is noninferior to aflibercept for change in BCVA from baseline to week 52.

• H₂. Brolucizumab 6 mg is noninferior to aflibercept for change in BCVA from baseline averaged over weeks 40 to 52.

• H₃. Brolucizumab 3 mg is noninferior to aflibercept for change in BCVA from baseline to week 52.

• H₄. Brolucizumab 3 mg is noninferior to aflibercept for change in BCVA from baseline averaged over weeks 40 to 52.

These 4 hypotheses were tested sequentially in the order of their numbering, i.e., confirmatory testing of the second, third, or fourth hypothesis required the rejection of each preceding null hypothesis. In this setting, each hypothesis was assessed at a 1-sided significance level of 0.025.

In the KITE study, the following noninferiority hypotheses are related to a noninferiority margin of 4 letters:

• H₁. Brolucizumab 6 mg is noninferior to aflibercept for change in BCVA from baseline to week 52.

• H₂. Brolucizumab 6 mg is noninferior to aflibercept for change in BCVA from baseline averaged over weeks 40 to 52.

These 2 alternative hypotheses were tested sequentially in the order of their numbering, i.e., confirmatory testing of the second hypothesis required rejection of the first null hypothesis. The average change from baseline in CSFT over the period from week 40 through week 52 was derived as the average of the changes from baseline to weeks 40, 44, 48, and 52. In this setting, each hypothesis was assessed at a 1-sided significance level of 0.025, while keeping the global type I error rate at 0.025.

Continuous variables were summarized using the number of observations, mean, standard deviation (SD), standard errors (SE), median, quartiles, and minimum and maximum values. Categorical variables were summarized with number of observations, the number of observations for each category, and the corresponding percentage. Where appropriate, 2-sided 95% CIs for point estimates of the mean or proportion were provided. For the between-treatment difference for brolucizumab versus aflibercept, point estimates and 95% CIs were provided as appropriate, unless otherwise specified.

Primary Outcome of the Studies

In both studies, the primary efficacy end point was the change from baseline in BCVA in the study eye at week 52. The first key secondary end point was the average change from baseline in BCVA in the study eye over the period from week 40 through week 52. For each patient, this end point was defined as the average of the changes from baseline to weeks 40, 44, 48, and 52.

The primary analysis of the primary and first key secondary end points was based on the full analysis set, with LOCF imputation of missing or censored BCVA values. An analysis of variance (ANOVA) model was used in data analyses. The model included treatment, baseline BCVA (≤ 65 letters to > 65 letters) and age category (< 65 years to ≥ 65 years) as factors. The 2-sided 95% CI for the LS mean difference (brolucizumab minus aflibercept) was presented in letters. Noninferiority was considered established if the lower limit of the corresponding 95% CI was greater than −4 letters. P value for treatment comparison (2-sided) and P value for noninferiority (4-letter margin) (1-sided) were presented.

Noninferiority Margin

In the KESTREL and KITE studies, a noninferiority margin of 4 ETDRS letters was used in the primary outcome analysis, where noninferiority would be demonstrated if the lower limit of the corresponding 95% CI of the difference in change in adjusted mean BCVA from baseline between brolucizumab and aflibercept was greater than −4 ETDRS letters. According to the clinical study reports, this noninferiority margin “provided assurance that both absolute efficacy and efficacy relative to the active comparator… could be established and any proof of noninferiority only occurred if the observed treatment differences were of no clinical relevance.”^11,12

Secondary Outcomes of the Studies

Secondary efficacy end points in the pivotal studies are related to visual acuity (e.g., change from baseline in BCVA at each visit up to week 100), dosing regimen (e.g., proportion of patients maintained at every 12 weeks up to week 100 for brolucizumab treatment arms only), anatomy (e.g., CSFT determined by spectral domain OCT, proportion of patients with presence of SRF and/or IRF at each assessment visit) or status of DR (e.g., proportion of patients with an improvement or worsening of 2 or more or 3 or more steps from baseline in the ETDRS DRSS score at each assessment visit).

All secondary efficacy end points were summarized and presented descriptively based on the full analysis set, with LOCF imputation for missing or censored data. The continuous secondary end points related to BCVA and CSFT were analyzed using ANOVA models. The estimates of LS mean for each treatment and for the treatment differences for brolucizumab minus aflibercept, including 95% CIs for the treatment differences, were presented. For categorical variables (binary end points), frequency tables (count and percentage) were provided by time point. In addition, proportions and treatment differences in proportions along with 95% CIs were presented for each time point using logistic regression, with treatment, the corresponding baseline status (similar to the ones specified for the ANOVA models), and age categories as fixed effects.

Type I Error Control

In both studies, confirmatory hypothesis testing for the additional secondary end points was performed in case the proof of noninferiority related to BCVA was successful for the 4 hypotheses specified previously (for the KESTREL trial) for the primary and first key secondary end points (corresponding to H₁, H₂, H₃, and H₄), or for the 2 hypotheses specified previously (for the KITE trial) for the primary and first key secondary end points.

In the KESTREL trial, the additional hypotheses were linked to the following end points when the proof of noninferiority related to BCVA was successful for the first 4 hypotheses specified previously:

• H₅. Average change from baseline in CSFT over the period from week 40 through week 52 in the study eye

• H₆. Absence of fluid in the study eye at week 52 (“no” indicates absence of SRF and IRF)

• H₇. Change from baseline in CSFT at week 4 in the study eye

• H₈. Average change from baseline in BCVA over the period from week 40 through week 52 in the study eye

All tests were 1-sided tests for superiority of brolucizumab 6 mg versus aflibercept 2 mg only if:

• each of the first 4 null hypotheses was rejected at a 1-sided significance level of 0.025

• the entire alpha was distributed between the null hypotheses related to the superiority testing of H₅ (90% of 0.025 = 0.0225) and H₆ (10% of 0.025 = 0.0025).

The familywise type I error rate was controlled at the 1-sided 2.5% level across the tested null hypotheses using the closed testing procedure specified in Figure 2, using the graphical method of Bretz et al.⁴⁵

Figure 2: Multiple Testing Strategy in the KESTREL Study

H = hypothesis.

Source: Clinical Study Report for KESTREL.¹¹

In the KITE study, the additional efficacy hypotheses were linked to the following end points:

• A. Average change from baseline in CSFT over the period from week 40 through week 52 in the study eye

• B. Average change from baseline in BCVA over the period from week 40 through week 52 in the study eye

• C. Fluid status “yes/no” in the study eye at week 52 (“no” indicates absence of SRF and IRF)

All tests were 1-sided tests for superiority of brolucizumab versus aflibercept. The alternative hypotheses were tested hierarchically in order (A, then B, then C), i.e., confirmatory testing of the hypothesis requires rejection of the previous null hypothesis. In this setting, each hypothesis was assessed at a 1-sided significance level of 0.025, while keeping the global type I error rate at 0.025.

Sensitivity Analyses

In the KESTREL and KITE trials, sensitivity analyses were conducted (in the full analysis set only) to assess the robustness of the hypothesis testing resulting from the primary analysis using the MMRM assuming missing at random. Other sensitivity analyses on the primary end point were also considered, such as a tipping point analysis or multiple imputation.

Subgroup Analyses

Subgroup analyses were performed in the year 1 Clinical Study Report, for the primary and key secondary efficacy variables only, to assess the consistency of treatment effect across various subgroups of interest. The prespecified subgroups in the KESTREL and KITE trials were:

• age (< 65 years, ≥ 65 years)

• gender (male, female)

• diabetes type (type 1, type 2)

• baseline hemoglobin A1C (< 7.5%, ≥ 7.5%)

• baseline BCVA (≤ 65 letters, > 65 letters)

• duration of DME since the primary diagnosis (≤ 3 months, > 3 months to < 12 months, ≥ 12 months)

• DME type (focal, diffuse), as per central reading centre

• baseline CSFT (< 450 μm, ≥ 450 μm to < 650 μm, ≥ 650 μm)

• baseline status of IRF (presence, absence)

• baseline status of SRF (presence, absence)

Ethnicity (Japanese, not Japanese) was another subgroup that was explored in the KESTREL trial. Only data for the subgroups defined in the CADTH review protocol are presented in this report.

Data Imputation Methods

In the KESTREL and KITE studies, the primary analysis of the primary and first key secondary end points was based on the full analysis set, with LOCF imputation of missing or censored BCVA values.

Missing data for all the secondary efficacy end points were imputed using the LOCF method.

Analysis Populations

The analysis populations in the KESTREL and KITE trials were defined using the same methods.

The full analysis set consisted of all randomized patients who received at least 1 intravitreal injection of the study treatment. This was the primary analysis set for all efficacy analyses. Patients were analyzed according to the treatment they were assigned to at randomization.

The per-protocol set was a subset of the full analysis set and excluded or censored patients with important protocol deviations and analysis restrictions that were expected to significantly affect the validity of the assessment of efficacy and/or safety at week 52, such as lack of compliance, missing data, prohibited concomitant medication, and deviations from the inclusion or exclusion criteria. The per-protocol set was not used in any analysis presented in the year 2 KESTREL and KITE Clinical Study Reports.

The safety analysis set (SAF) included all patients who received at least 1 intravitreal injection of the study drug. Patients in the SAF were analyzed according to the treatment arm from which they received the majority of treatments up to and including week 48. The SAF was used in the analysis of all safety variables.

Results

Patient Disposition

In the KESTREL study, of the 873 patients who were screened, 189 and 187 were randomized to the brolucizumab 6 mg group and aflibercept 2 mg group, respectively. Prior to week 100, 35 patients (18.5%) in the brolucizumab group and 34 patients (18.2%) in the aflibercept group withdrew prematurely from the study.

In the KITE study, out of 480 screened patients, 179 and 181 were randomized to the brolucizumab 6 mg group and aflibercept 2 mg group, respectively. Before week 100, the proportion of patients who discontinued the study before week 100 was higher in the brolucizumab group (36 patients; 20.1%) than in the aflibercept group (25 patients; 13.8%).

In both trials, the main reason for study discontinuation was “patient decision”: 10.1% and 7.5% in the brolucizumab 6 mg group and aflibercept 2 mg group, respectively in the KESTREL study; 7.8% and 3.9% in the brolucizumab 6 mg group and aflibercept 2 mg group, respectively, in the KITE study. Other reasons were AEs, death, and loss to follow-up.

Details of patient disposition are shown in Table 10.

Table 10: Patient Disposition

FAS = full analysis set; NR = not reported; PP = per protocol.

Sources: Clinical Study Reports for KESTREL¹¹ and KITE.¹²

Exposure to Study Treatments

The extent of exposure to the study drugs was calculated as the number of injections received for both active treatments and sham.

The mean number of active intravitreal treatment injections from baseline to week 96 was 10.6 weeks (SD = 3.0) in the brolucizumab 6 mg group and 13.0 weeks (SD = 3.2) in the aflibercept 2 mg group in the KESTREL study, and 10.3 weeks (SD = 2.8) in the brolucizumab 6 mg group and 13.2 weeks (SD = 3.1) in the aflibercept 2 mg group in the KITE study (Table 11). The median number of active injections of brolucizumab 6 mg was 11.0 in the KESTREL study and 10.0 in the KITE study, while the median number of active injections of aflibercept 2 mg was 15.0 in both studies.

Table 11: Summary of Study Treatment Exposure in the Study Eye (SAF Population)

SAF = safety analysis set; SD = standard deviation.

Sources: Clinical Study Reports for KESTREL¹¹ and KITE.¹²

Efficacy

Only those efficacy outcomes and analyses of subgroups identified in the review protocol are reported subsequently.

Change From Baseline in Visual Acuity

Change From Baseline in BCVA (Primary End Point)

In the KESTREL study, for the change from baseline in BCVA at week 52, the LS mean difference in ETDRS letters between the brolucizumab 6 mg group and the aflibercept 2 mg group was −1.3 letters (95% CI, −2.9 to 0.3; P < 0.001). In the KITE study, the LS mean difference in ETDRS letters between the brolucizumab 6 mg group and the aflibercept 2 mg group at week 52 was 1.2 letters (95% CI, −0.6 to 3.1; P < 0.001). Both studies met the primary end point of noninferiority in the full analysis set population using a noninferiority margin of 4 letters. Supportive analyses using the per-protocol analysis sets in both the KESTREL and KITE studies were consistent with the primary analyses. In sensitivity analyses in both the KESTREL and KITE studies, the results of the sensitivity analyses (using the full analysis set and MMRM assuming missing at random) in the change from baseline in BCVA at week 52 for the study eye were consistent with the primary analysis. Results of prespecified subgroups of interest were generally consistent with those in the overall population at week 52 (data not shown).

Noninferiority of brolucizumab 6 mg to aflibercept 2 mg was also demonstrated for the average change from baseline in BCVA over the period from week 40 through week 52 in both studies. In the KESTREL trial, the LS mean difference in ETDRS letters between the brolucizumab 6 mg group and the aflibercept 2 mg group was −1.5 letters (95% CI, −3.0 to 0; P < 0.001). This outcome was not tested for superiority due to earlier failure in the testing hierarchy. In the KITE trial, the LS mean difference in ETDRS letters between the brolucizumab 6 mg group and the aflibercept 2 mg group was 0.9 letters (95% CI, −0.9 to 2.6; P < 0.001). While this outcome was tested for superiority in the KITE trial, the difference did not reach statistical significance.

Similar results were found for year 2. In the KESTREL trial, for the change from baseline in BCVA at week 100, the LS mean difference in ETDRS letters between the brolucizumab 6 mg group and the aflibercept 2 mg group was −1.7 letters (95% CI, −3.8 to 0.4). In the KITE trial, the LS mean difference in ETDRS letters between the brolucizumab 6 mg group and the aflibercept 2 mg group at week 100 was 2.6 letters (95% CI, 0.2 to 4.9).

Proportion of Patients With a Gain in BCVA of 15 or More Letters From Baseline or Gain of 84 Letters or More

The between-group differences in the adjusted proportions of patients who gained 15 or more ETDRS letters in BCVA from baseline, or reached a BCVA of 84 letters or more at week 100 with brolucizumab 6 mg versus aflibercept 2 mg were −3.0% (95% CI, −12.5% to 6.3%) in the KESTREL trial and 13.6% (95% CI, 3.3% to 23.5%) in the KITE trial.

Proportion of Patients With a Loss in BCVA of 15 or More Letters From Baseline

The between-group differences in the adjusted proportions of patients with a loss of 15 or more ETDRS letters in BCVA from baseline at week 100 with brolucizumab 6 mg versus aflibercept 2 mg were 1.1% (95% CI, −1.6% to 3.5%) in the KESTREL trial and −1.3% (95% CI, −4.8% to 2.0%) in the KITE trial.

Details of the change in visual acuity from baseline in the 2 studies are available in Table 12.

Table 12: Efficacy — BCVA-Related Outcomes, LOCF

BCVA = best-corrected visual acuity; CI = confidence interval; FAS = full analysis set; LOCF = last observation carried forward; LS = least squares; NI = noninferiority; NR = not reported; PP = per protocol; SE = standard error.

^aAnalyzed using an analysis of variance model with baseline BCVA categories (≤ 65 letters, > 65 letters), age categories (< 65 years, ≥ 65 years), and treatment as fixed-effect factors.

^bThe statistical model used logistic regression adjusting for baseline BCVA categories (≤ 65 letters, > 65 letters), age categories (< 65 years, ≥ 65 years), and treatment as fixed-effect factors. The 95% CI for the treatment difference was estimated using the bootstrap method.

Sources: Clinical Study Reports for KESTREL¹¹ and KITE.¹²

Change From Baseline in Retinal Thickness

Change From Baseline in CSFT

In the KESTREL trial, patients in the brolucizumab 6 mg group had similar reductions in CSFT from baseline at week 52 compared with the aflibercept 2 mg group. The LS mean difference in the change from baseline in CSFT between the brolucizumab 6 mg arm and the aflibercept 2 mg arm was −5.1 µm (95% CI, −22.3 µm to 12.2 µm). At week 4, the LS mean difference in the change from baseline in CSFT between the 2 treatment groups was −2.1 μm (95% CI, −18.7 µm to 14.4 µm). Over the period from week 40 through week 52, the average LS mean for the change from baseline in CSFT between the brolucizumab 6 mg and aflibercept 2 mg groups was −1.4 µm (95% CI, −17.9 µm to 15.0 µm).

In the KITE trial, the LS mean difference in the change from baseline in CSFT to week 52 between the brolucizumab 6 mg arm and the aflibercept 2 mg arm was −32.8 µm (95% CI, −52.5 µm to −13.0 µm). At week 4, the LS mean difference in the change from baseline in CSFT between the 2 treatment groups was −15.4 μm (95% CI, −35.0 µm to 4.1 µm). Over the period from week 40 through week 52, the average LS mean for the change from baseline in CSFT between the brolucizumab 6 mg and aflibercept 2 mg groups was −29.4 µm (95% CI, −48.6 µm to −10.2 µm; P = 0.001) in favour of brolucizumab.

These differences between brolucizumab and aflibercept were similar at year 2. At week 100, the LS mean difference in the change from baseline in CSFT between the brolucizumab 6 mg arm and the aflibercept 2 mg arm was −2.9 µm (95% CI, −21.1 µm to 15.3 µm) in the KESTREL trial, and −29.2 µm (95% CI, −51.6 µm to −6.8 µm) in the KITE trial.

Proportion of Patients With a CSFT of Less Than 280 μm

In both the KESTREL and KITE trials, the proportion of patients at week 52 with CSFT less than 280 μm was higher in the brolucizumab 6 mg group compared with the aflibercept 2 mg group. The LS mean difference in the proportion of patients with CSFT less than 280 μm between the brolucizumab arm and the aflibercept arm was 13.4% (95% CI, 4.9% to 23.7%) in the KESTREL trial and 16.3% (95% CI, 5.7% to 25.9%) in the KITE trial.

At week 100, the proportion of patients with CSFT less than 280 μm was higher in the brolucizumab 6 mg group compared with the aflibercept 2 mg group. The LS mean difference in the proportion of patients with CSFT less than 280 μm between the brolucizumab arm and the aflibercept arm was 11.6% (95% CI, 2.3% to 21.6%) in the KESTREL trial and 14.7% (95% CI, 4.2% to 24.9%) in the KITE trial.

Details of retinal thickness–related outcomes are presented in Table 13.

Table 13: Efficacy — CSFT-Related Outcomes, FAS (LOCF)

CI = confidence interval; CSFT = central subfield thickness; FAS = full analysis set; LOCF = last observation carried forward; LS = least squares; SE = standard error.

^aAnalyzed using an analysis of variance model with baseline CSFT categories (< 450 μm, ≥ 450 μm to < 650 μm, ≥ 650 μm), age categories (< 65 years, ≥ 65 years), and treatment as fixed-effect factors.

^bThe statistical model used logistic regression adjusting for baseline CSFT categories (< 450 μm, ≥ 450 μm to < 650 μm, ≥ 650 μm), age categories (< 65 years, ≥ 65 years), and treatment as fixed-effect factors. The 95% CI for the treatment difference was estimated using the bootstrap method.

Sources: Clinical Study Reports for KESTREL¹¹ and KITE.¹²

Health-Related Quality of Life

No HRQoL tools other than the NEI VFQ-25 were used (Table 14).

Blindness (Legal)

Blindness was not assessed in the KESTREL or KITE trials.

Change From Baseline in Vision-Related Function

This outcome was measured using the NEI VFQ-25 composite score. In the KESTREL trial, the between-group LS mean differences for the change from baseline were −1.0 (95% CI, −3.4 to 1.4) at week 52 and −0.2 (95% CI, −2.9 to 2.6) at week 100. In the KITE trial, the between-group LS mean differences for the change from baseline were 2.5 (95% CI, 0.2 to 4.8) at week 52 and 3.4 (95% CI, 0.8 to 6.1) at week 100.

Details of the NEI VFQ-25 results are reported in Table 14.

Table 14: Efficacy — NEI VFQ-25 Outcomes, FAS (Observed)

CI = confidence interval; FAS = full analysis set; LS = least squares; NEI VFQ-25 = National Eye Institute Visual Functioning Questionnaire–25.

^aAnalyzed using an analysis of covariance model with treatment as a fixed-effect factor and corresponding baseline value of the end point as a covariate.

Sources: Clinical Study Reports for KESTREL¹¹ and KITE.¹²

Change From Baseline in DR Severity

Proportion of Patients With an Improvement From Baseline of 2 or More Steps in the DRSS Score for the Study Eye

At week 52 in the KESTREL trial, the difference in the proportion of patients who achieved a 2-step or greater improvement on the ETDRS DRSS from baseline in the brolucizumab group compared with the aflibercept group was 6.7% (95% CI, 0.6% to 12.9%). In the KITE trial, the difference in the proportion of patients who achieved a 2-step or greater improvement on the ETDRS DRSS from baseline in the brolucizumab group compared with the aflibercept group was 1.1% (95% CI −5.6% to 7.8%). At week 100, the difference in the proportion of patients who achieved a 2-step or greater improvement on the ETDRS DRSS from baseline in the brolucizumab group compared with the aflibercept group was 2.2% (95% CI, −4.0% to 8.4%) in the KESTREL trial and −4.5% (95% CI, −1.7% to 10.8%) in the KITE trial.

Proportion of Patients With an Improvement of 3 or More Steps From Baseline in the DRSS Score for the Study Eye

At week 52, the difference in the proportion of patients who achieved an improvement of 3 or more steps on the ETDRS DRSS from baseline in the brolucizumab group compared with the aflibercept group was 3.9% (95% CI, −2.2% to 10.5%) in the KESTREL trial and −0.6% (95% CI, −7.1% to 5.7%) in the KITE trial. At week 100, the difference in the proportion of patients who achieved a 3-step or greater improvement on the ETDRS DRSS from baseline in the brolucizumab group compared with the aflibercept group was 0.4% (95% CI, −5.7% to 6.8%) in the KESTREL trial and 3.9% (95% CI, −2.3% to 10.0%) in the KITE trial (Table 15).

Table 15: Efficacy — ETDRS DRSS–Related Outcomes, FAS (LOCF)

CI = confidence interval; DRSS = Diabetic Retinopathy Severity Scale; ETDRS = Early Treatment Diabetic Retinopathy Study; FAS = full analysis set; LOCF = last observation carried forward.

^aThe 95% CI for binomial proportions is based on the Clopper-Pearson exact method. The statistical model used logistic regression adjusting for baseline DRSS score categories (≤ 4, ≥ 5), age categories (< 65 years, ≥ 65 years), and treatment as fixed-effect factors. The 95% CI for the treatment difference was estimated using the bootstrap method.

Sources: Clinical Study Reports for KESTREL¹¹ and KITE.¹²

Presence of SRF or IRF

Proportion of Patients with Presence of SRF and/or IRF

At week 52, the difference in the adjusted proportion of patients in the brolucizumab group compared with the aflibercept group with SRF and/or IRF present was −13.2% (95% CI, −23.2% to −3.8%) in the KESTREL trial and −18.4% (95% CI, −28.5% to −8.3%) in the KITE trial. In both the KITE and KESTREL trials, this outcome was not tested statistically due to previous failure of the hierarchical testing procedure.

At week 100, the difference in the adjusted proportion of patients in the brolucizumab group compared with the aflibercept group with SRF and/or IRF present was −12.4% (95% CI, −22.8% to −2.1%) in the KESTREL trial and −16.2% (95% CI, −26.4% to −5.9%) in the KITE trial.

Proportion of Patients with Leakage on Fluorescein Angiography

At week 100 in both studies, the difference in the adjusted proportion of patients with leakage on fluorescein angiography was −14.2% (95% CI, −24.7% to −3.4%) in the KESTREL trial and −19.1% (95% CI, −29.1% to 8.2%) in the KITE trial.

Results of presence of SRF and/or IRF and leakage on fluorescein angiography are shown in Table 16.

Table 16: Efficacy — SRF and IRF Outcomes, FAS (LOCF)

CI = confidence interval; FA = fluorescein angiography; FAS = full analysis set; IRF = intraretinal fluid; LOCF = last observation carried forward; SRF = subfield fluid.

^aThe 95% CI for binomial proportions is based on the Clopper-Pearson exact method. The statistical model used logistic regression adjusting for baseline fluid status (SRF and/or IRF), age categories (< 65 years, ≥ 65 years), and treatment as fixed-effect factors. The 95% CI for the treatment difference was estimated using the bootstrap method.

^bThe 95% CI for binomial proportions is based on the Clopper-Pearson exact method. The statistical model used logistic regression adjusting for baseline leakage on FA status, age categories (< 65 years to ≥ 65 years), and treatment as fixed-effect factors. The 95% CI for the treatment difference was estimated using the bootstrap method.

Sources: Clinical Study Reports for KESTREL¹¹ and KITE.¹²

Frequency of Injections

In the 2 studies, weeks 32 and 36 were the 2 DAA visits for the initial cycle of every 12 weeks (after the loading phase) for the brolucizumab arms. The proportion of patients in the brolucizumab arms and aflibercept arm with time to first need for treatment every 8 weeks at each DAA visit was summarized in Table 17.

In the KESTREL trial, for the brolucizumab 6 mg treatment group, the proportion of patients who maintained a treatment interval status of every 12 weeks was 52.0% (95% CI, 43.7% to 59.6%) up to week 52, and 44.1% (95% CI, 35.7% to 52.1%) up to week 100. Among the patients who completed the treatment study period, 67.1% in the brolucizumab 6 mg arm were on an 8-week treatment interval at week 100.

In the KITE trial, the proportion of patients treated with brolucizumab 6 mg who maintained a treatment interval of every 12 weeks was 45.5% (95% CI, 37.7% to 53.0%) up to week 52; 36.8% (95% CI, 29.1% to 44.5%) of patients maintained a treatment interval of either every 12 weeks or every 16 weeks through week 100. Among the patients who completed the treatment study period, 52.5% in the brolucizumab 6 mg arm at week 100 were on a treatment interval of every 8 weeks.

Harms

Only those harms identified in the review protocol are reported subsequently. Refer to Table 18 for detailed harms data.

Adverse Events

The proportion of patients reporting at least 1 ocular AE in the study eye up to week 100 was comparable across treatment arms in both studies (48.7% and 50.3% in the brolucizumab 6 mg group and aflibercept 2 mg group in the KESTREL trial, respectively; ||||| ||||| ||| ||||| || ||| |||||||||||| | || ||| ||||||||||| | || ||||||| ||||||||||||||.

The most common ocular AEs (3% or greater in any treatment group) reported in both studies were as follows. In the KESTREL trial, the proportion of patients in the brolucizumab and aflibercept groups, respectively, who reported an AE were 8.5% and 7.0% for cataract conjunctival hemorrhage, 5.8% and 1.6% for increased intraocular pressure, 5.3% and 1.6% for vitreous detachment, 5.3% and 3.2% for vitreous floaters, 4.8% and 2.1% for diabetic retinal edema, 3.2% and 0.5% for conjunctivitis, 3.2% and 2.7% for dry eye, 3.2% and 2.7% for eye pain, 3.2% and 1.6% for posterior capsule opacification, and 1.6% and 4.8% for reduced visual acuity. || ||||| ||| |||||||||| || |||||||| || ||| |||||||||||| ||| ||||||||||| ||||||| ||||||||||||| ||||||||| |||| || ||| |||| ||| ||||| ||| ||||||||| |||| ||| |||| ||| |||||||||||| ||||||||||| |||| ||| |||| ||| ||||||||| ||||||||||| ||||||||| |||| ||| |||| ||| ||||||||||||||| |||| ||| |||| ||| ||| |||| |||| ||| |||| ||| ||| ||||| ||| |||| ||| |||| ||| ||||||| |||||| |||||||.

In both the KESTREL and KITE trials, a comparable proportion of patients reported at least 1 nonocular AE between treatment arms: 77.2% and 76.5% for the KESTREL trial and ||||| ||| |||||| for the KITE trial in the brolucizumab 6 mg and aflibercept groups, respectively.

Ocular and nonocular AEs reported for the study eye were mainly mild and moderate in severity in each treatment arm in both studies.

Serious Adverse Events

Ocular SAEs were reported with low frequency in both studies: 3.7% and 2.7% in the KESTREL trial and 2.8% and 1.7% in the KITE trial among patients in the brolucizumab 6 mg group and aflibercept 2 mg group, respectively. Eye disorders, such as cataract, were the main ocular SAEs reported in the studies.

In KESTREL, the incidence of nonocular SAEs was 28.0% in the brolucizumab 6 mg group and 28.9% in the aflibercept group. The proportion of patients who experienced a specific nonocular SAE for the brolucizumab 6 mg and aflibercept groups, respectively, was: 6.9% and 9.1% for cardiac disorders; 9.0% and 7.5% for infections and infestations; 3.7% and 2.1% for metabolism and nutrition disorders; 3.2% and 1.6% for neoplasms; 3.2% and 7.0% for nervous system disorders; 4.2% and 2.1% for vascular disorders;3.7% and 4.8% for renal and urinary disorders, 2.6% and 3.2% for injury, poisoning, and procedural complications; and 3.7% and 3.2% for respiratory, thoracic, and mediastinal disorders.

|| ||||| ||| ||||||||| || |||||||||| |||| ||| ||||| || ||| |||||||||||| ||||| ||| ||||| || ||| ||||||||||| ||||| ||| |||||||||| || |||||||| ||| ||||||||||| | |||||||| |||||||||| ||| ||| ||| |||||||||||| ||| ||||||||||| ||||||| ||||||||||||| |||| |||| ||| |||| ||| ||||||| |||||||||| |||| ||| |||| ||| |||||||||| ||| ||||||||||||| |||| ||| |||| ||| |||||||||||||||| |||||||||| |||| ||| |||| ||| |||||||||| |||| ||| |||| ||| ||||||| |||||| |||||||||| ||| |||| ||| |||| ||| ||||| ||| ||||||| ||||||||||

Withdrawals Due to Adverse Events

The incidence of withdrawals due to AEs was 1.6% and 1.1% in the brolucizumab 6 mg and aflibercept groups, respectively, in KESTREL and |||| ||| |||| || ||| |||||||||||| ||| ||||||||||| ||||||| ||||||||||||| || |||||. The most common reason for discontinuing treatment was infections and infestations, such as endophthalmitis and uveitis.

Mortality

There were 15 deaths that occurred in the KESTREL study, 8 (4.2%) in the brolucizumab 6 mg group and 7 (3.7%) in the aflibercept group. || ||| |||| |||||| || |||||| |||||| |||||||| || ||| |||||||||||| ||||| ||| | |||||| || ||| ||||||||||| ||||||

The cause of death in the 2 studies were cardiac disorders, neoplasms, and infections and infestations. According to the sponsor, none of the deaths were suspected by the investigator to be related to study treatment.

Notable Harms

Ocular hemorrhage was the most commonly reported notable harm prespecified in the protocol of this review, occurring in 10.6% and 11.8% of patients in in the brolucizumab group and the aflibercept group, respectively in KESTREL, and |||| ||| |||| || |||||||| || ||| |||||||||||| ||||| ||| ||||||||||| |||||| |||||||||||| || |||||. Other common notable harms (greater than 5%) included increased intraocular pressure, arterial thromboembolic events, and vitreous floaters.

There were no reports of traumatic cataract, photophobia, or hypersensitivity reactions, which were specified as notable harms in the research protocol of this review.

Table 18: Summary of Harms — Safety Set, Up to Week 100

AE = adverse event; ATE = arterial thromboembolic event; NR = not reported; SAE = serious adverse event; WDAE = withdrawal due to adverse event.

^aFrequency > 3%.

^bFrequency > 5%.

Sources: Clinical Study Reports for KESTREL¹¹ and KITE.¹²

Critical Appraisal

Internal Validity

The KESTREL and KITE trials were similarly designed randomized, double-blind, active-controlled, noninferiority phase III trials comparing brolucizumab (6 mg and 3 mg in the KESTREL trial; 6 mg in the KITE trial) with aflibercept (2 mg). The overall designs of the KESTREL and KITE trials were appropriate for the objectives of the studies. There were no major concerns with regard to the method of randomization, stratification, allocation concealment, and masking for randomized assignment. The baseline characteristics of the study population were generally well balanced between treatment arms and across studies, with the exception that patients in the KITE study had a relatively large imbalance in the number of ETDRS letters between the 2 treatment arms and a somewhat thicker retina at baseline, and more likely to have received prior ocular medications compared with those enrolled in the KESTREL study. However, the clinical expert thought these differences were unlikely to impact results between the studies.

In these 2 studies, by the end of the 2-year treatment, approximately 14% to 20% of the patients had discontinued the study earlier. The dropout rates were comparable in the KESTREL study, but somewhat imbalanced in the KITE study (20% in the brolucizumab group and 14% in the aflibercept group). There is some risk of attrition bias, which may have an impact on the study findings, such as the NEI VFQ-25, which was completed through a patient interview. The main reason for early withdrawal was “patient decision.” No further details were provided for this reason. The COVID pandemic had a significant impact on the study conduct (e.g., centre closed, missed treatments, and missed assessments). In the 2 studies, sensitivity analyses on COVID-19 exposure and impact were conducted for the primary and secondary end points. In addition, ocular and nonocular AEs and SAEs were analyzed based on COVID-19 exposure or impact subgroups. The results in the COVID-19–impacted and nonimpacted subgroups were generally consistent with those in the overall population.

As per the study designs, a different dosing schedule was used between the treatment arms in both the loading and maintenance phases. In the loading phase, the treatment schedule for brolucizumab was 1 dose every 6 weeks for 5 doses; for aflibercept, the treatment schedule was once a month for 5 months. In the maintenance phase in the KESTREL and KITE trials, the treatment interval for the brolucizumab groups could be either every 8 weeks or every 12 weeks. At week 72, patients in the KITE study also had the option to extend the treatment interval to every 16 weeks (from every 12 weeks) or to every 12 weeks (from every 8 weeks), based on disease activity and stability assessments. Intervals in the aflibercept arm could not be adjusted and were given at fixed 8-week intervals.

The studies concluded the noninferiority of brolucizumab 6 mg to aflibercept 2 mg based on primary outcome analyses in the full analysis set population. The noninferiority margin of 4 ETDRS letters for the primary end point, which was determined by the sponsor based on prior clinical trial data and clinical reasoning, aligned with recommended approaches. The patient populations in the KESTREL and KITE trials were similar to those of the confirmatory trials for aflibercept. The clinical rationale was considered reasonable by the clinical expert consulted by CADTH.

The enrolled sample sizes were adequate for the assessment of the primary outcome. Subgroup analyses were prespecified although no conclusion related to subgroup effects can be drawn due to the lack of sample size considerations, control for multiplicity, and statistical testing for treatment-by-subgroup interaction.

In the KESTREL and KITE trials, the results of change from baseline in BCVA at week 52 using the per-protocol population were consistent with those in the full analysis set. In both studies, sensitivity analyses were conducted to assess the robustness of the hypothesis testing resulting from the primary analysis. Various methods were used to account for missing data, such as the MMRM modelling assuming a missing-at-random mechanism, or LOCF approach. The missing-at-random assumption may be a concern, given that the primary reasons for discontinuation from the study included patient decision, death, and AEs. Further, the LOCF method assumes that patient outcomes do not change after they drop out, which may not hold true in practice.⁴⁶ Therefore, performing additional sensitivity analyses that do not assume that missing data are missing at random could be useful. However, the results of the sensitivity analyses confirmed those of the primary analysis, suggesting these approaches in handling missing data were unlikely to introduce bias for the primary end point. The risk of attrition bias due to missing data was of particular concern for the NEI VFQ-25, as the proportion of missing data was greater than 20% for some treatment groups.

Statistical hierarchy was used for multiplicity adjustment for selected outcomes. Some important outcomes were not included in the hierarchy, such as vision-related HRQoL assessed by NEI VFQ-25, change in retinal thickness, and change in DR severity. In addition, the statistical hierarchy failed during testing of the BCVA outcomes in the KESTREL trial and during superiority testing for BCVA in the KITE trial. In the KESTREL trial, the noninferiority of brolucizumab 3 mg to aflibercept 2 mg was not achieved for the outcome of change in BCVA from baseline to week 52. As per the study protocol, there was no confirmatory testing to assess the superiority of brolucizumab versus aflibercept for the subsequent outcomes in the hierarchical testing procedure, which limits drawing definite conclusions for these outcomes.

External Validity

The study population in the KESTREL and KITE trials may not fully represent the typical population in Canada with DME who would be receiving anti-VEGF therapy. Based on patients’ baseline characteristics, there was an overrepresentation of males (DME being approximately evenly distributed between male and female patients in practice, according to the clinical expert) and a very low proportion of Black patients; also, there was no information for Indigenous patients. The inclusion criteria for the KESTREL and KITE trials were reasonable and reflective of the eligibility criteria for anti-VEGF treatment in clinical practice. Although all patients enrolled in the KESTREL and KITE trials were naive to treatment with anti-VEGF therapies and exhaustive exclusion criteria were used in the 2 studies, the clinical expert consulted by CADTH indicated that brolucizumab can be used in a broader population, such as those with a blood glucose level that is poorly controlled (patients with a hemoglobin A1C level higher than 10% were excluded from the KESTREL and KITE trials) or those who had previously received anti-VEGF therapy.

In both studies, the selection of outcome measures and study end points was reasonable. The validity of the outcome measures used in the KESTREL and KITE trials (such as BCVA assessed by number of ETDRS letters, disease status assessed by the ETDRS DRSS, retinal thickness assessed using OCT, and vision-related HRQoL assessed using the NEI VFQ-25) was evaluated. These were relevant clinical outcomes used in the clinical trials for DME; however, an MID has not yet been established in the study population for some of these outcomes (e.g., NEI VFQ-25). The 2-year study duration can be considered sufficient for the efficacy and safety assessment for brolucizumab, although the consequences of reduced injection frequency, such as treatment adherence and associated clinical benefit, need to be evaluated longer term.

In the 2 pivotal studies, patients in a brolucizumab group could have their dosing interval extended, reduced (once patients on brolucizumab dropped back to every 8 weeks because of disease activity, they could not extend the treatment interval for the rest of the study, which may contradict clinical practice), or maintained postrandomization based on the assessments of disease activity. Changes in treatment interval and dosage were not allowed for the treatment with aflibercept, and these patients remained on a fixed interval of every 8 weeks during the maintenance phase; this is contrary to the product monograph for aflibercept, which states that the treatment interval can be extended after the first year of treatment. The expert noted that patients in clinical trials normally receive more injections than they would in practice. Patients in the real world may not receive many loading doses and it is possible that the outcomes observed in practice could differ from those shown in the clinical trials.

Indirect Evidence

Objectives and Methods for the Summary of Indirect Evidence

As there was a lack of studies directly comparing brolucizumab with treatments other than aflibercept for patients with DME, a review of indirect evidence was undertaken.

CADTH conducted a literature search to identify potentially relevant ITCs in this patient population. A focused literature search for ITCs dealing with DME was run in MEDLINE All (1946–) on July 28, 2022. No limits were applied to the search. Titles, abstracts, and full-text articles were screened for inclusion by 1 reviewer based on the population, intervention, comparator, and outcome criteria outlined in Table 5. However, no published ITCs describing the relative efficacy and/or safety of brolucizumab versus other active therapies for patients with DME were identified.

The sponsor submitted an ITC in patients with DME.⁴⁷

The objective of this section is to summarize and critically appraise the indirect evidence available in the sponsor-submitted ITC.

Description of Sponsor-Submitted ITC

The sponsor’s ITC included a systematic review of the literature to identify trials investigating brolucizumab and comparator interventions in adult patients with DME, and a corresponding NMA that compared brolucizumab with other active treatments.

In the sponsor-submitted ITC, brolucizumab was compared with other anti-VEGFs, corticosteroids, laser photocoagulation, any of the aforementioned drug therapies plus vitrectomy, anti-VEGF plus angiopoietin-2 (faricimab), and placebo and/or best supportive care. RCTs (blinded and open label) were included. Changes in visual acuity, retinal thickness, presence of SRF or IRF, and DR-related outcomes were evaluated. In addition, injection frequency, HRQoL, and safety outcomes were examined in this ITC.

The patient population, intervention and comparators, and outcome measures for study selection in this ITC are presented in Table 19.

Table 19: Study Selection Criteria and Methods for the Sponsor-Submitted ITC

ang-2 = angiopoietin-2; DME = diabetic macular edema; HRQoL = health-related quality of life; ITC = indirect treatment comparison; NEI VFQ-25 = National Eye institute Visual Functioning Questionnaire–25; RCT = randomized controlled trial; SLR = systematic literature review; VEGF = vascular endothelial growth factor.

Source: Sponsor-submitted ITC.⁴⁷

Methods of Sponsor-Submitted ITC

Objectives

The objective of this ITC was to compare the treatment efficacy and safety of brolucizumab relative to currently existing therapies for the treatment of DME.

Study Selection Methods

The eligible studies that were used to inform the ITC were identified through a systematic literature search conducted by the ITC authors. A search strategy was developed according to the predefined protocol. The searches were conducted in Embase, MEDLINE, and Cochrane Library in March 2020. Conference abstracts were searched to retrieve the latest clinical studies. Clinical trial registries were searched to identify relevant in-progress RCTs for the interventions of interest (although their results were not included in the synthesis) and the reference lists of included reports and relevant systematic reviews were screened. Thirty-two studies were identified from this search. Studies publishing results after March 2020 (KITE, KESTREL, YOSEMITE, and RHINE) were also included. An update of the systematic literature review was conducted in October 2021, and 7 additional studies were retrieved.

The population of interest was adult patients (> 18 years) with DME. The intervention of interest was brolucizumab (6 mg intravitreal injection every 8 weeks or every 12 weeks). Any of the following therapies listed as monotherapy or combination therapy were considered:

• anti-VEGFs (ranibizumab, aflibercept, bevacizumab, brolucizumab)

• corticosteroids (fluocinolone, dexamethasone)

• laser photocoagulation

• any of the previously specified plus vitrectomy procedures

• anti-VEGF and anti–angiopoietin-2 (faricimab)

Relevant comparators also included placebo, best supportive care, and observation. The efficacy outcomes of interest included visual acuity outcomes (measured using BCVA scores, EDTRS letters, logMAR, and Snellen charts), anatomic outcomes (including change in retinal thickness and presence of SRF or IRF), and DR-related outcomes. Other evaluated outcomes were injection frequency (reported as the mean or median number of injections given at year 1 and year 2), safety, and HRQoL, which was measured with the NEI VFQ-25. The sponsor’s ITC reported outcomes at year 1 and year 2.

Study selection and data extraction were conducted by 2 independent reviewers, with arbitration by a third reviewer, as needed, to resolve disagreements. Details of the methods used to extract data from the included studies were described. The risk-of-bias assessment of each included RCT was conducted using a checklist recommended in the NICE guidelines.⁴⁸ The number of reviewers involved in the risk-of-bias appraisals, whether they worked independently, and any mechanisms in place to resolve disagreements were not reported. The checklist contains questions related to the appropriateness of randomization, the treatment allocation, masking, the similarity of the prognostic factors between groups, any imbalance in the number of dropouts between groups, data reporting, and the appropriate use of statistical methods.

ITC Analysis Methods

A feasibility assessment was conducted to evaluate the suitability of available DME trials for NMAs. The objective of the feasibility assessment was to evaluate clinical heterogeneity and determine the suitability of conducting NMAs for the efficacy and safety outcomes of interest. To investigate potential between-trial heterogeneity, a qualitative assessment of heterogeneity was conducted based on trial design, eligibility criteria, baseline patient characteristics, trial-specific outcome definitions, and the definition of the treatment contrasts required for generating the network.

A number of effect modifiers for visual acuity outcomes were identified in the sponsor’s ITC. The frequency of injections, age, BCVA at baseline, and retinal thickness at baseline were considered related to BCVA outcomes. A covariate of interest was plotted versus BCVA, and a regression line (weighted by the sample size) was presented along with the adjusted R² value. The results of the regression analyses suggested considerable clinical heterogeneity in the reporting of some outcomes in the available evidence. The plots may highlight variables that can be considered for covariate adjustment in the NMA analyses and therefore improve model fit.

Several assumptions were made when defining treatment node assignment:

• The administered dose does not impact the outcomes, i.e., similar efficacy was assumed between the ranibizumab 0.3 mg and 0.5 mg doses and between the bevacizumab 1.25 mg and 1.5 mg doses.

• Previous studies that examined the effect of different regimens (aflibercept every 4 weeks, every 8 weeks, or as needed) assumed there were no significant differences observed among the VEGF Trap-Eye treatment group.

• It is unlikely that adding laser photocoagulation to an anti-VEGF regimen would make a difference to the BCVA outcomes, retinal thickness, or rates of discontinuation.

The relative efficacy of brolucizumab compared with other treatments for the patients with DME was evaluated using Bayesian NMAs for key efficacy outcomes: change from baseline in BCVA (continuous or categories), injection frequency, AEs, and all-cause discontinuation. Both fixed- and random-effect models were fit to all outcomes. Given the heterogeneity in the patient characteristics across the trials, the authors of this ITC expected random effects to be the most appropriate model for the data, unless there was compelling evidence in favour of the fixed-effects model (i.e., deviance information criterion [DIC] ≥ 3 in favour of the fixed-effect model). The total residual deviance was used to assess the model fit, considering that models with a total residual deviance similar to the number of data points often fit the data adequately. The models were run for 3 chains, each with 10,000 iterations after a burn-in of 10,000. Studies were classified as reporting results at year 1 if the absolute difference was less than 5 weeks from 52 weeks, or at year 2 if the absolute difference was less than 5 weeks from 104 weeks. Time points that did not meet these criteria were excluded from the analyses.

The safety and tolerability profile of brolucizumab was qualitatively compared with that of other treatments.

The NMA was conducted using JAGS 4.3.0. All other data analyses and plots were created using the statistical program R and its associated packages.

Methods of the ITC analyses are summarized in Table 20.

Table 20: ITC Analysis Methods

AE = adverse event; BCVA = best-corrected visual acuity; d = distribution; DEX = dexamethasone; DIC = deviance information criterion; dnorm = normal distribution; dunif = uniform distribution; ITC = indirect treatment comparison; LP = laser photocoagulation; NA = not applicable; NMA = network meta-analysis; SD = standard deviation.

^aPrior on baselines: mu ~ dnorm(0,0.0001).

^bOutcomes that were used to inform the pharmacoeconomic analysis for the submission.

Source: Sponsor-submitted ITC.⁴⁷

Results of Sponsor-Submitted ITC

Summary of Included Studies

In total, 43 RCTs were identified and included in the NMA.

Sources of Heterogeneity Identified Across the Included RCTs

Study Design and Patient Characteristics at Baseline

The majority of the included studies were phase I, II, or III trials, and 3 studies were phase IV. Thirteen studies did not report the study phase. Eighteen studies were double-blinded, 8 were single-blinded, 1 was triple-blinded, and 7 were open-label studies. Blinding was unclear for the remaining 6 studies. The number of patients or study eyes ranged from 37 to more than 1,000 across the included trials. The length of follow-up varied from 48 weeks to 168 weeks, although the time points of interest for most of the outcomes were year 1 and year 2.

Most of the patients’ baseline characteristics were presented graphically. In general, across the included studies, patients’ baseline characteristics were comparable for mean age (ranged from 58 to 66 years) and hemoglobin A1C level (ranged from 7.3 to 8.7). The mean time since diagnosis of DME ranged from 1.2 to 3.4 years. The duration of diabetes ranged from 10 to 18 years in the included studies. Based on data from 26 trials, the mean BCVA scores ranged from 33 to 71 letters, indicating a visual acuity of between 20/40 and 20/320. Based on data from 25 trials, the mean retinal thickness at baseline ranged from 321 µm to 596 µm, and the majority of studies included patients with a retinal thickness of more than 400 µm.

Limited information was available on the number of patients with prior DME treatment. Among the studies that reported this information, 4 studies included only treatment-naive patients (100%). The proportion of patients who had received previous anti-VEGF therapy ranged from 5% to 43%.

Table 31 in Appendix 2 provides a description of trial characteristics and key patient characteristics at baseline across the included trials.

Outcome Measures

In the included studies, visual acuity was measured with a variety of methods, such as BCVA scores, EDTRS letters, logMAR, and Snellen charts. BCVA data were captured as the proportion of patients with a change in the number of ETDRS letters (gain of ≥ 5, ≥ 10, and ≥ 15 letters; and loss of ≥ 15, ≥ 10, ≥ 5, < 5, ≥ 0, ≥ 30, or < 10 letters), proportion of patients at a particular BCVA level, or mean change from baseline.

There was heterogeneity in the way retinal thickness was defined in the included trials, such as central retinal thickness, CSFT, central macular thickness, and central foveal thickness. However, the majority of the studies used similar definitions to measure retinal thickness. There was additional heterogeneity in the type of measurement used for retinal thickness.

Change in disease severity from baseline across the included studies was reported as proportion of patients who showed worsening, improvement (≥ 1 step, ≥ 2 steps, or ≥ 3 steps), or no change by steps on the ETDRS DRSS from baseline to specific time points.

Injection frequency was reported as the mean or median number of injections required at the end of year 1 and year 2.

Treatment Regimens

Treatments for DME vary considerably in molecule, dose, and treatment regimen (Table 32 in Appendix 2). Some collapsing and/or simplification of the treatment regimens was required to construct connected networks.

Results

Risk of Bias

Among the included RCTs, the majority of studies (66%) were at low risk of bias due to the randomization. Patient baseline characteristics were judged to be comparable in 80% of the studies, imbalanced in 18% of studies, and not reported in 2% of studies. Masking was judged to be adequate in 36% of studies and unclear in 30% of studies. Thirty-four percent of studies were judged to be at high risk of bias for masking (either open-label or the outcome assessor was unmasked). For withdrawals, 63% of studies were considered at low risk of bias. Thirty percent of studies were at unclear risk of bias and 7% were at high risk of bias in this domain. The risk of selective reporting was low in 80% of studies, unclear in 18% of studies, and high in 2% of studies. The details of the statistical analyses were clearly reported in 66% of studies and unclear in 30% of studies. Two studies (4%) used a per-protocol analysis and were judged at high risk of bias for this item.

BCVA Outcomes

Year 1

For the outcome of BCVA at year 1, 37 trials were included in the analysis. Results of a random-effects model (DIC = 282.78; total residual deviation of 76.17) are presented, as this model had a smaller DIC value compared with the fixed-effects model (DIC = 335.35; total residual deviation of 149.28). A graphic representation of the evidence network is presented in Figure 3. Most trials compared active treatments; the network did not include 1 closed loop involving brolucizumab as an intervention.

No treatment was favoured for the outcome of mean change from baseline on BCVA letters (95% CrIs contained the null) when brolucizumab was compared with aflibercept, bevacizumab, or ranibizumab. For the comparisons of brolucizumab with bevacizumab and brolucizumab with ranibizumab, the 95% CrIs were imprecise, precluding conclusions regarding potential differences in effects.

Figure 3: Network Diagram for the Outcome of Change in BCVA at Year 1

AFL = aflibercept; BCVA = best-corrected visual acuity; BEO = Beovu (brolucizumab); BEV = bevacizumab; DEX = dexamethasone; FAR = faricimab; LP = laser photocoagulation; PRP = panretinal photocoagulation; RAN = ranibizumab.

Source: Sponsor-submitted indirect treatment comparison.⁴⁷

Year 2

For the outcome of BCVA at year 2, 14 trials were included in the analysis. Results of a random-effects model (DIC = 112.48; total residual deviation of 29.55) are presented, as this model had a similar DIC value compared with the fixed-effects model (DIC = 117.01; total residual deviation of 39.87). A graphic representation of the evidence network is presented in Figure 4. Most trials compared active treatments; the network did not include 1 closed loop involving brolucizumab as an intervention.

No treatment was favoured for the outcome of mean change from baseline on BCVA letters (95% CrIs contain the null) when brolucizumab was compared with aflibercept, ranibizumab, and bevacizumab for the outcome of BCVA. For all comparisons, the 95% CrIs were imprecise, precluding conclusions regarding potential differences in effects.

A bivariate model was fitted to the data to jointly analyze outcomes at year 1 and year 2, when a correlation between these 2 years was assumed. Results of a random-effects model were reported. Owing to imprecision in the effect estimates (i.e., wide 95% CrIs) for all comparisons, no conclusions could be drawn regarding potential differences in effects.

Figure 4: Network Diagram for the Outcome of Change in BCVA at Year 2

AFL = aflibercept; BCVA = best-corrected visual acuity; BEO = Beovu (brolucizumab); BEV = bevacizumab; DEX = dexamethasone; LP = laser photocoagulation; PRP = panretinal photocoagulation; RAN = ranibizumab.

Source: Sponsor-submitted indirect treatment comparison.⁴⁷

The comparative results of change in BCVA outcomes are outlined in Table 21.

The relative risks of a gain or loss of 15 or more letters at year 1 and year 2 were evaluated in the NMA. No treatment was favoured for these outcomes on BCVA (95% CrIs contain 1) when brolucizumab was compared with aflibercept, bevacizumab, and ranibizumab, except that treatment with brolucizumab may be favourable (95% CrIs did not include 1) compared with ranibizumab for the outcome of gain of 15 letters or more at year 1; however, the 95% CrIs were very close to the null. For all comparisons, the 95% CrIs were imprecise, precluding conclusions regarding potential differences in effects (Table 22).

Diabetic Retinopathy Severity Score

The relative risks of a 2-step or greater improvement at year 1 and year 2 were evaluated in the NMA. For the results at year 1, 12 trials were included in the analysis. The results of the fixed-effects model (DIC = 396.98; total residual deviation of 28.7) are presented. For year 2, 5 trials were included in the analysis. The results of the fixed-effects model (DIC = 93.46; total residual deviation of 9.3) are presented.

No treatment was favoured for these outcomes on disease severity (95% CrIs contain 1) when brolucizumab was compared with aflibercept, bevacizumab, and ranibizumab. For all comparisons, the 95% CrIs were imprecise, precluding conclusions regarding potential differences in effects (Table 23).

Table 21: NMA Results — Mean Change in BCVA as Letters at Year 1 and Year 2

BCVA = best-corrected visual acuity; CrI = credible interval; NMA = network meta-analysis; RCT = randomized controlled trial.

Source: Sponsor-submitted indirect treatment comparison.⁴⁷

Table 22: NMA Results — Relative Risk of Gain or Loss of 15 or More Letters in BCVA at Year 1 and Year 2

BCVA = best-corrected visual acuity; CrI = credible interval; NMA = network meta-analysis; RCT = randomized controlled trial; RR = relative risk.

Note: Bolded numbers indicate that the 95% CrI excludes the null.

Source: Sponsor-submitted indirect treatment comparison.⁴⁷

Table 23: NMA Results — Relative Risk of an Improvement of 2 or More Severity Steps in DRSS at Year 1 and Year 2

CrI = credible interval; DRSS = diabetic retinopathy severity score; NA = not available; NMA = network meta-analysis; RCT = randomized controlled trial; RR = relative risk.

Source: Sponsor-submitted indirect treatment comparison.⁴⁷

Change in Retinal Thickness

The relative mean differences in the change in retinal thickness at year 1 and year 2 were evaluated in the NMA. Brolucizumab was favoured when compared with bevacizumab and ranibizumab (95% CrIs did not contain 0). When compared with aflibercept, no treatment was favoured (Table 24).

Table 24: NMA Results — Change in Retinal Thickness at Year 1 and Year 2

CrI = credible interval; NMA = network meta-analysis; RCT = randomized controlled trial; RMD = relative mean difference.

Note: Bolded numbers indicate that the 95% CrI excludes the null.

Source: Sponsor-submitted indirect treatment comparison.⁴⁷

Serious Ocular and Nonocular AEs

The risks of serious ocular and nonocular AEs up to 2 years were evaluated in the NMA. None of the treatments was favoured when brolucizumab was compared with aflibercept, bevacizumab, or ranibizumab (95% CrIs contain 1). For all comparisons, the 95% CrIs were imprecise, precluding conclusions regarding potential differences in effects (Table 25).

Table 25: NMA Results — Risk of Serious Adverse Events

AE = adverse event; CrI = credible interval; HR = hazard ratio; NMA = network meta-analysis; RCT = randomized controlled trial.

Source: Sponsor-submitted indirect treatment comparison.⁴⁷

Study Discontinuation

The risks of study discontinuation were evaluated in the NMA. No treatment was favoured when brolucizumab was compared with aflibercept and bevacizumab (95% CrIs contain 1). For these comparisons, the 95% CrIs were imprecise, precluding conclusions regarding potential differences in effects. Ranibizumab was favoured when it was compared with brolucizumab (Table 26).

Table 26: NMA Results — Risk of Study Discontinuation

AE = adverse event; CrI = credible interval; HR = hazard ratio; NA = not available; NMA = network meta-analysis; RCT = randomized controlled trial.

Note: Bolded numbers indicate that the 95% CrI excludes the null.

Source: Sponsor-submitted indirect treatment comparison.⁴⁷

Injection Frequency

Injection frequencies were not analyzed using an NMA model but assigned to a particular node and estimated using baseline pooling for both year 1 and year 2 time points. To obtain the pooled mean injection frequency, 2 approaches were considered in the ITC, 1 that obtains the pooled mean using fixed- and random-effects meta-analyses and 1 that uses a weighted average. Given the heterogeneity within the studies, it may be appropriate to use the random-effects estimate of mean injection frequency.

Injection frequencies differ substantially, primarily due to the different treatment regimens used and the study and locations. The average frequency of injection for brolucizumab was 6.9 (SD = 1.3) by the end of the first year, and 10.5 (SD = 2.9) by the end of the second year of treatment. As expected, as-needed and treat-and-extend regimens have fewer injections. It is possible that increased injection frequencies result in increased absolute BCVA gain; however, there was no evidence of treatment effect modification.

Table 27: Number of Injections Reported in the Sponsor-Submitted ITC

ITC = indirect treatment comparison; PRN = pro re nata (as needed); q.4.w. = every 4 weeks; q.8.w. = every 8 weeks; q.12.w. = every 12 weeks; SD = standard deviation; T and E = treat and extend.

Source: Sponsor-submitted ITC.⁴⁷

Critical Appraisal of Sponsor-Submitted ITC

The ITC was based on a systematic literature review that identified studies according to prespecified inclusion criteria. A comprehensive and transparent approach to the systematic review was provided, including a comprehensive search. The literature search was well reported, with a complete copy of the search strategy included in the report. Study selection and data extraction were performed by 2 reviewers independently and any disagreements were resolved through consensus, so the risk of bias and error in study selection and data extraction were minimized. A risk-of-bias evaluation for the studies included in the systematic literature review was performed, based on a tool that considers the appropriateness of randomization and allocation concealment, similarity at baseline across treatment groups in prognostic factors, masking, imbalances in dropouts, outcomes reporting, and intention-to-treat analysis. It was unclear whether the risk-of-bias assessments were performed in duplicate, so the potential for bias and error in the appraisals is uncertain. The results of the risk-of-bias assessment for the studies were included in the NMA, but it was not reported how the results of the quality appraisal factored into the NMA (e.g., sensitivity analyses excluding studies rated with a high risk of bias).

The inclusion criteria would allow a population that is relevant for settings in Canada. The comparisons reported in this ITC have generally incorporated relevant treatments for settings in Canada, including treatments that have extensive clinical use but lack a formal indication from Health Canada such as bevacizumab, which is used in clinical practice in Canada. However, not all outcomes considered important to patients, clinicians, and the drug plans were investigated according to the CADTH protocol, such as HRQoL and change in vision-related function.

The degree of heterogeneity (such as the patients’ disease characteristics at baseline) between the included studies was difficult to assess because of the incomplete reporting of study characteristics. Description of trial design, sample size, and disease duration were reported. However, the ITC failed to report information related to the methods used for handling missing data. There was considerable variability in study design (phase I through IV trials were included), year of conduct, sample size, and treatment regimen. This ITC report indicates that these trials included arms that were pooled for the NMA (various doses of a particular drug were lumped into a single node) for increased network connectivity. The degree of similarity between these trials was not reported. A sensitivity analysis addressing the effect of pooling these doses was not presented. Trials in different phases were included; however, a sensitivity analysis addressing the effect of pooling these types of trials was not presented.

Similarly, inadequate information about and variability in the baseline patient characteristics reported contribute to heterogeneity in the studies included in the ITC. Clinical trial eligibility criteria were described for the trials that were ultimately included in the NMA. However, many individual studies failed to report or inadequately reported patient characteristics, resulting in gaps in the extracted ITC data. There was a lack of information about key baseline characteristics, such as the presence of significant diabetic macular ischemia, the patient’s previous treatment and response, presence of IRF, and systemic comorbidities, including hypertension, chronic kidney disease, obesity, or cardiac conditions.

Most of the patients’ baseline characteristics were presented graphically. Even though some of the patients’ baseline characteristics were comparable, such as age (which ranged from 58 to 66 years) and hemoglobin A1C level (which ranged from 7.3 to 8.7), heterogeneity still exists. The mean time since diagnosis of DME ranged from 1.2 to 3.4 years. The duration of diabetes ranged from 10 to 18 years in the included studies. Based on data from 26 trials, the mean BCVA scores ranged from 33 to 71 letters. Based on data from 25 trials, the mean retinal thickness at baseline ranged from 321 µm to 596 µm, and the majority of studies included patients with a retinal thickness of more than 400 µm. There was also heterogeneity in the reporting of methods for measuring retinal thickness and in the results of changes in retinal thickness. The heterogeneity based on the factors that were reported in combination with the inability to assess those that were not reported means there is considerable uncertainty as to whether the assumptions related to transitivity were met. The treatment effect of the study drug could differ by patient characteristics at baseline. Despite acknowledging the degree of heterogeneity, the technical report did not provide information on the assessments of heterogeneity (e.g., graphic representation of baseline characteristics, statistical tests) that was sufficient to fully understand the sources of heterogeneity. Therefore, it is plausible that the potential for heterogeneity could have influenced the comparative efficacy and safety estimates, and it is not possible to quantify or identify the direction of the bias. Several assumptions were made when defining treatment node assignment, for example: that a different dose did not impact the outcomes, that the ranibizumab 0.3 mg and 0.5 mg doses had similar efficacy, and that there were no significant differences in the effect of different regimens (aflibercept every 4 weeks, every 8 weeks, or as needed); it is uncertain whether these assumptions are valid.

The analytical method used for the ITC was well reported. The authors provided a description of which studies were included in each of the analyses. The study outcomes in the ITC are of interest to the CADTH systematic review protocol and are clinically relevant. The analysis of the extracted data followed the framework suggested by NICE.⁵⁰ The sponsor’s ITC reported on the number of burn-ins and convergence characteristics.

Additional limitations to the ITC include the following:

There was a weak connection between brolucizumab and the rest of the network: brolucizumab was only connected to the network through aflibercept through the KESTREL and KITE studies; this may contribute to uncertainty in the models. The networks consider a large number of interventions; therefore, collapsing and/or simplifying the treatment regimens was required. Although there are some closed loops for some networks, overall, the nodes were connected by few trials. The geometry of the networks likely contributed to uncertainty in the estimates for models (by limiting the ability to assess consistency) and the level of imprecision in certain comparisons as evidenced by wide CrIs. For almost all comparisons and outcomes (with a few exceptions), no conclusion could be drawn about the comparative effects because the 95% CrIs included multiple possible effects (i.e., appreciable benefit, appreciable harm, and little-to-no difference).

For injection frequency, brolucizumab every 8 weeks and every 12 weeks were examined in the ITC. The results of injection frequency were derived from data pooling without conducting an NMA. Although drug administration with fixed treatment intervals according to a specific protocol is commonly observed in clinical trials, the clinical expert consulted by CADTH indicated that, in the real world, the treatment regimen could be more flexible, based on the patient’s response. Therefore, the findings from the clinical trials may not reflect clinical practice.

Other Relevant Evidence

This section includes a summary of 1 additional relevant study, KINGFISHER,¹³ which was included in the sponsor’s submission to CADTH, as it was considered by the sponsor to provide further information on the safety of brolucizumab in patients with DME. The study compared the efficacy and safety of brolucizumab versus aflibercept in patients with DME, but the treatment intervals used for both drugs were shorter than the recommended intervals (beyond the loading phase) in the respective Health Canada–approved monographs.^14,15

The frequency of dosing for brolucizumab was selected based on previous studies that have suggested that, for some patients with DME, frequent dosing (i.e., every 4 weeks) with anti-VEGF therapy may be required to improve and maintain functional and anatomical outcomes.¹³ The clinical expert consulted by CADTH did not consider the efficacy results for the treatment intervals used in the KINGFISHER trial to be relevant or generalizable to clinical practice; therefore, they are not included in this summary. However, given the frequency of administration used in the KINGFISHER trial, the clinical expert suggested the safety data reported would be of value. In particular, the clinical expert suggested that the safety data may provide information on whether intraocular inflammation and retinal vasculitis are idiosyncratic AEs related to brolucizumab itself rather than to the frequency of intravitreal injections.

KINGFISHER Study

Methods

One multicentre, randomized, double-blinded, active-controlled, parallel-group, prospective, phase III study, KINGFISHER,¹³ was conducted to evaluate the efficacy and safety of brolucizumab versus aflibercept in the treatment of adult patients with visual impairment due to DME. The primary objective was to demonstrate that brolucizumab was noninferior to aflibercept with respect to change in visual acuity from baseline up to week 52. The prespecified noninferiority margin was 4 ETDRS letters. Patients were screened and randomized between 2019 and 2020 and the last patient completed the last visit in 2021. It was unclear whether sites in Canada were included in the KINGFISHER study.

Patients who met all inclusion and none of the exclusion criteria during the 2-week screening period were randomized using interactive response technology at a 2:1 ratio to 1 of 2 treatment arms: brolucizumab 6 mg administered every 4 weeks or aflibercept 2 mg administered every 4 weeks. Randomization was stratified by baseline BCVA scores (≤ 34 and > 34 letters read). The 48-week double-blind treatment period was followed by a 4-week follow-up period up to week 52. Patients were evaluated every 4 weeks for the duration of the study.

Of note, only 1 eye was selected as the study eye and treated with the study drug. If both eyes were eligible, the eye with worse visual acuity was selected; however, the investigator could select the eye with better visual acuity for medical reasons or due to local ethical requirements. The unmasked investigator and staff administered the injections and performed postinjection safety assessments, while the masked investigator and staff performed study assessments.

Populations

Inclusion and Exclusion Criteria

The inclusion and exclusion criteria used in the KINGFISHER study¹³ were generally consistent with the eligibility criteria used in the pivotal KESTREL¹¹ and KITE¹² studies. The KINGFISHER study included adult patients (≥ 18 years of age) with type 1 or type 2 diabetes who were diagnosed with visual impairment due to DME. Visual impairment due to DME was defined as the eye with a BCVA score between 73 and 23 ETDRS letters, inclusive, and DME involving the centre of the macula with a CSFT of 320 μm or greater. Of note, patients could either be treatment-I or could have previously received anti-VEGF therapy but could not have received any injections within 3 months before baseline. Patients needed to have a hemoglobin A1C of 12% or less and mild to moderate proliferative DR; these characteristics may be associated with the need for frequent dosing of anti-VEGF therapy.

Patient Characteristics

A summary of baseline characteristics in the KINGFISHER study¹³ is available in Appendix 2.

The mean age of patients was 60.9 years (SD = 10.59) in the brolucizumab arm and 60.2 years (SD = 9.31) in the aflibercept arm. The proportion of patients who were male was 56.1% (n = 194) and 61.4% (n = 105) in the brolucizumab and aflibercept arms, respectively. The majority of patients were white in both treatment arms and the respective proportion was similar in the brolucizumab (83.8%; n = 290) and aflibercept (84.8%; n = 145) arms. The baseline diabetes characteristics were similar between the treatment arms, with 94.5% (n = 327) and 94.7% (n = 162) of patients documented with type 2 diabetes in the brolucizumab and aflibercept arms, respectively. The mean hemoglobin A1C was also similar between treatment groups: 7.84% (SD = 1.476) in the brolucizumab arm versus 7.98% (SD = 1.582) in the aflibercept arm.

The mean duration of diagnosis with DME was 20.6 months (SD = 29.94) and 18.2 months (SD = 25.60) repoIted in the brolucizumab and aflibercept arms, respectively. The study eye was the left eye in 173 patients (50.0%) in the brolucizumab arm and 92 patients (53.8%) in the aflibercept arm. The mean BCVA score was 61.3 letters (SD = 10.14) and 60.5 letters (SD = 11.27) in the brolucizumab and aflibercept arms, respectively. The mean CSFT was 514.1 μm (SD = 138.94) and 511.2 μm (SD = 156.29) in the brolucizumab and aflibercept arms, respectively. Most patients in both treatment arms had mild to moderately severe nonproliferative DR: 75.6% (167 out of 221) and 76.3% (90 out of 118) in the brolucizumab and aflibercept arms, respectively). Prior anti-VEGF treatment in the study eye was reported in 95 patients (27.5%) and 52 patients (30.4%) in the brolucizumab and aflibercept arms, respectively.

Interventions

The investigational drug was brolucizumab 6 mg (0.05 mL) solution for intravitreal injection. The comparator drug was aflibercept 2 mg (0.05 mL) solution for intravitreal injection. Treatments were administered every 4 weeks beginning in week 1 and up to and including week 48.

Panretinal photocoagulation was allowed at any time during the study as required, according to the investigator. Topical ocular corticosteroids were also allowed in the study eye as required. If cataract surgery or yttrium aluminum garnet laser was required in the study eye, it was scheduled such that it did not disturb the schedule of the study treatment. If the fellow eye developed visual impairment due to DME or any other disease during the study, the eye could also be treated with standard of care and other treatments according to clinical practice, at the discretion of the investigator.

Outcomes and Statistical Analysis

The primary end point was the change from baseline in BCVA at week 52 in the study eye using ETDRS-like charts (measure of the visual function of the retina).

A sample size of 357 patients (238 in the brolucizumab arm and 119 in the aflibercept arm) was required to assess noninferiority with respect to BCVA change from baseline at week 52, with 90% power at a 2-sided alpha of 0.05, assuming equal means and a common SD of 11 letters. To meet the FDA recommendation for safety assessment, data from at least 300 patients with 1 year of exposure to brolucizumab 6 mg administered every 4 weeks were required. Consequently, the total sample size was increased to 450 patients with a 2:1 ratio (300 in the brolucizumab arm and 150 in the aflibercept arm), resulting in a statistical power of 95% for assessing noninferiority. Finally, a total of 495 patients (330 in the brolucizumab arm and 165 in the aflibercept arm) were planned to be randomized to account for a dropout rate of 9%.

Safety analyses were descriptive only and the number and proportion of patients with ocular and nonocular AEs were summarized according to treatment arm. No imputations were performed for any missing values, with the exception of imputation for partial dates of AEs.

The SAF included all patients who received at least 1 intravitreal injection of the study drug and was used in the analysis of safety variables; patients were analyzed according to the treatment arm from which they received the majority of treatments, up to and including week 48.

Patient Disposition

A summary of patient disposition in the KINGFISHER study¹³ is provided in Table 28.

Of the 765 patients screened, 346 patients were randomized to the brolucizumab arm and 171 patients were randomized to the aflibercept arm. Study discontinuation rates were 10.1% and 8.8% in the brolucizumab and aflibercept arms, respectively. Discontinuation rates related to lost to follow-up were 2.9% in both treatment arms and discontinuation rates related to AEs were 0.9% and 0.6% in the brolucizumab and aflibercept arms, respectively.

Table 28: Summary of Patient Disposition

FAS = full analysis set; PPS = per-protocol set; SAF = safety analysis set.

Source: Clinical Study Report for KINGFISHER.¹³

Exposure to Study Treatment

A total of 189 patients (54.6%) in the brolucizumab arm and 94 patients (55.0%) in the aflibercept arm received all 13 injections following the regimen of every 4 weeks (safety set). The mean number of injections was 11.5 (SD = 2.74) in the brolucizumab arm and 11.6 (SD = 2.40) in the aflibercept arm.

Harms

Only those harms identified in the review protocol are reported subsequently. A summary of harms is provided in Table 29.

Ocular Adverse Events

A total of 105 patients (30.3%) in the brolucizumab arm and 59 patients (34.5%) in the aflibercept arm reported at least 1 ocular AE. ||| |||| |||||| |||||| ||||| |||||||| || ||| |||||||||||| ||| ||| |||||||| |||||||||| || || |||||||| ||||||| | ||||| || | |||||||| |||||| || ||| |||||||||||| ||| |||||||| || ||||| ||| ||||||| |||||| ||||||| |||||| |||||||| ||||||||||| || | |||||||| ||||||| ||| |||||||| ||||||||||| ||| ||||||| |||||||||| || | ||||||| |||||| ||||. No patients in the aflibercept arm reported any serious ocular AEs. A total of 7 patients (2.0%) and 3 patients (1.8%) withdrew from study treatment in the brolucizumab and aflibercept arms, respectively, due to an ocular AE, with the most common ocular event documented as related to an eye disorder, which was reported in 6 patients (1.7%) in the brolucizumab arm and 3 patients (1.8%) in the aflibercept arm.

Nonocular Adverse Events

A total of 209 patients (60.4%) in the brolucizumab arm and 96 patients (56.1%) in the aflibercept arm reported at least 1 nonocular AE. The most common nonocular event reported in the brolucizumab arm was COVID-19 and hypertension in 19 patients (5.5%) each. A total of 69 patients (19.9%) in the brolucizumab arm and 36 patients (21.1%) in the aflibercept arm reported at least 1 serious nonocular AE. The most common serious nonocular AEs were related to infections and infestations in 29 patients (8.4%) reported in the brolucizumab arm. A total of 10 patients (2.9%) and 5 patients (2.9%) withdrew from study treatment in the brolucizumab and aflibercept arms, respectively, due to a nonocular AE, with the most common nonocular event documented as related to infections and infestations, which were reported in 4 patients (1.2%) in the brolucizumab arm and 2 patients (1.2%) in the aflibercept arm.

Mortality

Deaths were reported in 7 patients (2.0%) in the brolucizumab arm, with the most common cause of death recorded as cardiac disorders in 2 patients (0.6%) and infections and infestations in 3 patients (0.9%). Deaths were reported in 5 patients (2.9%) in the aflibercept arm, with the most common cause of death recorded as cardiac disorders and infections and infestations in 2 patients (1.2%) each.

Notable Harms

Intraocular inflammation was reported in 14 patients (4.0%) in the brolucizumab arm versus 5 patients (2.9%) in the aflibercept arm. Retinal vasculitis was reported in 3 patients (0.9%) in the brolucizumab arm versus 1 patient (0.6%) in the aflibercept arm. Retinal vascular occlusion was reported in 1 patient in each arm (0.3% versus 0.6% in the brolucizumab and aflibercept arms, respectively). Arterial thromboembolic events were reported in 15 patients (4.3%) in the brolucizumab arm versus 9 patients (5.3%) in the aflibercept arm. Transient increased intraocular pressure was reported in 3 patients (0.3%) in the brolucizumab arm versus 2 patients (1.2%) in the aflibercept arm. No reports of endophthalmitis were recorded.

Table 29: Summary of Harms, Safety Set

AE = adverse event; SAE = serious adverse event; WDAE = withdrawal (of treatment) due to adverse event.

Note: AEs with a start date on or after the date of administration of the first study treatment were included. AEs that started after the patient discontinued the study treatment and started an alternative treatment for diabetic macular edema in the study eye were censored.

Deaths on or after the date of study treatment administration were included. Deaths after the patient had discontinued from the study treatment and started an alternative treatment for diabetic macular edema in the study eye were included.

^aFrequency ≥ 2% of patients in any treatment arm in the study eye.

^bFrequency > 1 patient in any treatment arm.

^cA patient who experienced retinal vasculitis may have experienced an intraocular inflammation and, therefore, may have also been included under intraocular inflammation.

Source: Clinical Study Report for KINGFISHER.¹³

Critical Appraisal

Internal Validity

The KINGFISHER study was a randomized, double-blind, active-controlled study. Randomization, stratified by baseline BCVA, was achieved using interactive response technology, which likely kept the allocation concealed until assignment to a treatment arm. The baseline demographic and disease characteristics were generally balanced between the arms. Since unmasked personnel administered the injections and performed the postinjection safety assessments, the likelihood of unmasking or revealing knowledge of the treatment received and the potential for introducing bias in subjective outcomes, such as reporting of subjective harms, is uncertain. Treatment groups were generally well balanced in terms of the discontinuation rates (10.1% in the brolucizumab arm versus 8.8% in the aflibercept arm); thus, the risk of attrition bias is probably low. Since the sample size calculation considered the FDA recommendation for safety assessment that requires data from at least 300 patients with 1 year of exposure to brolucizumab 6 mg administered every 4 weeks, the trial was powered for the safety assessment.

External Validity

The clinical expert consulted by CADTH for this review advised that the KINGFISHER study would not be generalizable to the patient population and clinical practice in Canada because the frequency of administration, every 4 weeks, is not a relevant treatment interval. Further, aflibercept is rarely administered every 4 weeks in clinical practice.

Discussion

Summary of Available Evidence

This report summarizes the evidence on brolucizumab for DME from 2 pivotal phase III RCTs (KESTREL and KITE), 1 other phase III RCT (KINGFISHER), and 1 ITC.

Two studies, KESTREL (N = 566) and KITE (N = 360), met the inclusion criteria for the systematic review section. They were similarly designed phase III RCTs that evaluated the noninferiority of brolucizumab (6 mg every 8 weeks or every 12 weeks during maintenance) to aflibercept (2 mg every 8 weeks during maintenance) through the change from baseline in BCVA (ETDRS letters) at week 52 in the full analysis set population as a primary end point. The mean age of enrolled patients at baseline in these studies ranged from 62.2 to 64.4 years, and the majority were male (more than 58%) and white (more than 73%). The median time since the diagnosis of DME was 9.4 to 12.5 months in the KESTREL trial and 9.9 to 10.4 months in the KITE trial; the mean baseline CSFT in these 2 studies ranged from 453 μm to 484 μm; and the mean baseline BCVA scores ranged from 63.7 to 66.6 letters in both studies, although, in KITE, there was an imbalance in the number of ETDRS letters: 66 (SD = 10.8) in the brolucizumab group and 63.7 (SD = 11.7) in the aflibercept group. All enrolled patients were naive to anti-VEGF therapies. Outcomes included changes in BCVA, anatomical outcomes, DR severity, vision-related function, injection frequency, and safety, with a primary analysis at week 52 and data up to 100 weeks.

The KINGFISHER trial (N = 517) was a phase III RCT evaluating the efficacy and safety of brolucizumab 6 mg every 4 weeks versus aflibercept 2 mg every 4 weeks in the treatment of adult patients with visual impairment due to DME. The treatment intervals used for both drugs were shorter than the recommended intervals (beyond the loading phase) in the respective Health Canada–approved monographs. The primary objective was to demonstrate that brolucizumab was noninferior to aflibercept with respect to change in visual acuity from baseline up to week 52. In this review, CADTH focused on the safety results of brolucizumab when a frequent dosing regimen (every 4 weeks) was adopted in the KINGFISHER trial.

One sponsor-submitted ITC was summarized and critically appraised. The sponsor performed an NMA to estimate the effectiveness and safety of brolucizumab in patients with DME compared with other anti-VEGFs (aflibercept, bevacizumab, ranibizumab), dexamethasone intravitreal implants, laser therapy, and placebo or sham. The outcomes of the NMA included change from baseline in BCVA, proportion of patients gaining or losing 10 or more or 15 or more ETDRS letters, retinal thickness, injection frequency, treatment discontinuation, and ocular and nonocular AEs.

Interpretation of Results

Efficacy

The results of the KESTREL and KITE trials support the noninferiority, but not superiority, of brolucizumab 6 mg (5 times the loading dose of every 6 weeks followed by maintenance injections every 8 weeks or every 12 weeks) versus aflibercept 2 mg (5 monthly loading doses followed by maintenance injections every 8 weeks) for the change in BCVA. The noninferiority of brolucizumab 6 mg to aflibercept 2 mg was demonstrated for the primary end point (change from baseline in BCVA at week 52 for the study eye) using a noninferiority margin of 4 letters. In addition, several sensitivity analyses by the sponsor as well as a supportive analysis using the per-protocol population were consistent with the findings of the primary analyses. Results were consistent for change from baseline to week 100. Results of prespecified subgroup analyses (i.e., baseline BCVA categories and CSFT categories) were generally consistent with the overall population at week 52; however, the study was not powered to detect subgroup differences. There was superiority testing in the KITE trial (but not the KESTREL trial) for change in BCVA from baseline averaged from week 40 to week 52 and the between-group difference did not reach significance. The results of the visual acuity-related efficacy outcomes, other than the change from baseline to week 52 and change from baseline averaged over week 40 to week 52 (e.g., patients whose ETDRS scores showed a gain or loss of 15 letters or greater and changes from baseline to week 100), need to be interpreted with caution, since they were not included in the statistical hierarchy.

The change in retinal thickness (measured as CSFT) from baseline and patients with a CSFT of less than 280 µm were secondary outcomes in the studies. According to the expert, the reduction in retinal thickness correlates well with the improvement in visual acuity. Reduction from baseline in retinal thickness averaged over week 40 to week 52 was greater with brolucizumab versus aflibercept (mean difference of 29 µm) in the KITE trial, with no evidence for a difference in the KESTREL trial. Therefore, the finding of superiority of brolucizumab over aflibercept for this outcome was inconsistent between the studies.

The change from baseline in the NEI VFQ-25 composite score, which measures vision-related functions and some aspects of HRQoL, was a secondary outcome in the KESTREL and KITE trials, but not included in the statistical hierarchy. Many subscales of NEI VFQ-25 reflected vision-related functions that were noted in the patient group input to be highly relevant to the functioning of patients with DME (i.e., general vision, mental health, social functioning, dependency, and driving). In the KESTREL and KITE studies, improvements in the composite score were observed in all treatment arms (improvement in score of 6.5 to 9.1 from baseline to week 52 per arm). Due to the lack of statistical testing for this outcome and the risk of attrition bias due to missing data (more than 20% of data were missing for some treatment groups), definitive conclusions on the change from baseline in vision-related functions cannot be made.

ETDRS DRSS was used to measure disease activity in patients with DME. Regression of DRSS is another clinically meaningful outcome in this study population; however, this outcome was not included in the statistical hierarchy. For most of the results for this outcome at week 52 and week 100, there was no evidence to suggest a difference between brolucizumab and aflibercept. Given the uncertainty in the analysis due to the lack of statistical testing and imprecision in the between-group differences, no definite conclusions on the change in disease severity can be drawn.

The pivotal trials measured the proportions of patients with presence of IRF or SRF as secondary outcomes. IRF and SRF are indicators of active disease. According to the clinical expert consulted by CADTH, IRF is a more relevant outcome than SRF in patients with DME, noting that SRF is uncommon in DME and is a marker for more severe DME. In both studies at week 52 and week 100, a numerically lower proportion of patients treated with brolucizumab had IRF and/or SRF present compared with the aflibercept group; however, this outcome was not tested statistically due to a previous failure of the hierarchical testing procedure.

The frequency of injection was noted to be an important outcome of interest by both patients and the clinical expert, as it has implications on the frequency of AEs, HRQoL, the burden of treatment, and patient adherence and, subsequently, has an impact on the treatment effect. The proportion of patients treated with brolucizumab maintained on a schedule of every 12 weeks was reported descriptively in the studies. Among the patients who received treatment with brolucizumab, approximately half maintained a treatment interval of 12 weeks at week 52 in both studies, and 44% and 37% of patients maintained the regimen of every 12 weeks at week 100 in the KESTREL and KITE trials, respectively. Among patients who completed treatment with brolucizumab at week 100, the majority of them were on the every 8 weeks schedule (67.1% in the KESTREL trial; 52.5% in the KITE trial). Differences in the number and frequency of injections between study arms must be considered in light of the study design, as patients in the brolucizumab group could have their treatment interval changed based on their DAAs (although once patients on brolucizumab dropped back to every 8 weeks because of disease activity, they could not extend the treatment interval for the rest of the study); however, changes in treatment interval and dosage were not allowed for the treatment with aflibercept and these patients remained on a fixed interval of every 8 weeks during the maintenance phase. Treatment with aflibercept in the trials was contrary to clinical practice, as the product monograph of aflibercept states that the treatment interval can be extended after the first year of treatment.³¹ Also, once patients on brolucizumab dropped back to every 8 weeks because of disease activity, they could not extend the treatment interval for the rest of the study, which also may not reflect clinical practice. Further, the clinical importance of the between-group difference in the number of injections is difficult to determine. It is unknown whether the number of injections that were avoided in 1 or 2 years would have a meaningful impact for patients on either HRQoL or on reducing disease burden. However, the expert noted that differences in injection frequency were expected to be small in the first year, given the standard loading doses, and greater differences might not be seen until later years of use.

The sponsor-submitted NMA provided indirect comparative evidence for brolucizumab versus other anti-VEGF drugs. After including 43 trials in an NMA, none of the treatments was favoured, in terms of improving visual acuity and lessening disease severity, when brolucizumab was compared with other active treatments, such as aflibercept, ranibizumab, and bevacizumab, for the treatment of DME; however, for most comparisons, the effect estimates were too imprecise (i.e., the 95% CrIs were wide) to draw a conclusion about comparative efficacy or harms. Treatment with brolucizumab was associated with a greater reduction in retinal thickness compared with bevacizumab and ranibizumab. In addition, patients treated with brolucizumab are likely to receive fewer injections compared with other anti-VEGFs, though these results were derived from data pooling without conducting an NMA and, in particular, should be interpreted with caution. The key limitations for the ITC is significant heterogeneity (in study design and patient characteristics) across the included RCTs and imprecision around the effect estimates (i.e., wide 95% CrIs), which precluded drawing a conclusion for most outcome comparisons. This limits the conclusions that can be drawn from this ITC.

Harms

The safety profile for brolucizumab 6 mg was generally consistent with that of aflibercept 2 mg in the KESTREL and KITE trials. The proportion of patients reporting at least 1 ocular AE in the study eye up to week 100 was comparable across treatment arms in both studies (48.7% and 50.3% in the brolucizumab 6 mg and aflibercept 2 mg group in the KESTREL trial, respectively; 40.8% and 40.9% in the brolucizumab 6 mg and aflibercept 2 mg group in the KITE trial, respectively). Overall, the most frequently reported ocular AEs related to brolucizumab in both studies were cataract, conjunctival hemorrhage, vitreous detachment, vitreous floaters, increased intraocular pressure, diabetic retinal edema, dry eye, eye pain, posterior capsule opacification, conjunctivitis, and reduced visual acuity. Cataract was the most commonly reported ocular AE, which was anticipated because of the age of the study populations. Ocular SAEs were reported with low frequency and the incidence of ocular SAEs was similar between the 2 studies (3.7% and 2.7% in the brolucizumab and aflibercept groups in the KESTREL trial, respectively, and 2.8% and 1.7% of patients in the brolucizumab and aflibercept groups in the KITE trial, respectively). In both studies, the incidence of withdrawals due to AEs was also similar between the 2 treatment groups (1.6% and 1.1% in the brolucizumab and aflibercept groups in the KESTREL trial, respectively, and 2.8% and 2.2% of patients in the brolucizumab and aflibercept groups in the KITE trial, respectively). There were 15 deaths in the KESTREL study, 8 (4.2%) in the brolucizumab group and 7 (3.7%) in the aflibercept group. In the KITE study, 13 (7.3%) deaths occurred in the brolucizumab group and 9 (5.0%) in the aflibercept group. According to the sponsor, none of the deaths were considered to be related to study treatment.

In the KINGFISHER study, with dosing every 4 weeks, a total of 105 patients (30.3%) in the brolucizumab arm and 59 patients (34.5%) in the aflibercept arm reported at least 1 ocular AE. A total of 3 patients (0.9%) in the brolucizumab arm reported at least 1 serious ocular AE. No patients in the aflibercept arm reported any serious ocular AE. A total of 7 patients (2.0%) and 3 patients (1.8%) withdrew from study treatment in the brolucizumab and aflibercept arms, respectively, due to an ocular AE. Intraocular inflammation was reported in 14 patients (4.0%) in the brolucizumab arm versus 5 patients (2.9%) in the aflibercept arm. Retinal vasculitis was reported in 3 patients (0.9%) in the brolucizumab arm versus 1 patient (0.6%) in the aflibercept arm. The results suggested that, when brolucizumab was administered using a more frequent dosing regimen, the safety outcomes were similar to those for the recommended treatment intervals, and no unusual safety signal was observed from this study. The clinical expert found it reassuring that SAEs of retinal vasculitis were not reported more frequently with the dosing of brolucizumab every 4 weeks.

The risk of ocular AEs, nonocular AEs, and study discontinuation were evaluated in the NMA. The results suggested that no treatment was favoured for reduction in the risk of ocular or nonocular AEs; however, all effect estimates were too imprecise (i.e., the 95% CrIs were wide) to draw a conclusion about the relative harms of the various treatments. Patients treated with brolucizumab were associated with a higher risk of all-cause study discontinuation compared with ranibizumab. Limitations to the NMA preclude making firm conclusions about relative risks of harm for brolucizumab compared with other anti-VEGFs.

Conclusions

Brolucizumab 6 mg every 8 weeks or every 12 weeks during maintenance therapy was found to be noninferior to aflibercept 2 mg every 8 weeks for the mean change in BCVA from baseline after 1 year of treatment in anti-VEGF-naive adult patients with DME, based on evidence from 2 double-blind phase III RCTs (KESTREL and KITE). Results for mean change in BCVA after 100 weeks of treatment were generally consistent with the 1-year results. The results of other BCVA outcomes, retinal thickness, and presence of IRF and/or SRF, did not contradict the primary end point findings, but their interpretation is limited by lack of a noninferiority margin and lack of adjustment for multiple testing. Reduction from baseline in retinal thickness after 1 year of treatment was greater with brolucizumab versus aflibercept in the KITE trial; in the KESTREL trial, the effect estimate was too imprecise (i.e., wide 95% CIs) to draw a conclusion. The results for presence of IRF and/or SRF suggested that brolucizumab may have been favoured over aflibercept, but a firm conclusion cannot be drawn due to the lack of adjustment for multiplicity. After 52 weeks of treatment, approximately half of the brolucizumab group in each trial maintained the dosing schedule of every 12 weeks, although conclusions cannot be drawn due to the lack of adjustment for multiplicity and issues regarding the generalizability of the treatment regimens.