3 Committee discussion
The appraisal committee considered evidence submitted by Novartis, a review of this submission by the evidence review group (ERG), and submissions from other stakeholders. See the committee papers for full details of the evidence.
3.1 NICE has already produced technology appraisal guidance on aflibercept and ranibizumab in this indication. These treatments are only recommended if all of the following circumstances apply in the eye to be treated:
the best-corrected visual acuity (BCVA) is between 6/12 and 6/96
there is no permanent structural damage to the central fovea
the lesion size is less than or equal to 12 disc areas in greatest linear dimension
there is evidence of recent presumed disease progression (blood vessel growth, as indicated by fluorescein angiography, or recent visual acuity changes).
The committee was aware that the fast track appraisal process only permits recommendations to be made in line with the recommendations of the appraisals for the comparator treatments. So, it understood that any recommendation for brolucizumab would be constrained by these criteria. The company presented a cost-comparison case, in which it proposed that:
the overall health benefits associated with brolucizumab are similar to or greater than those associated with aflibercept and ranibizumab and
the total costs associated with brolucizumab are similar to or lower than those associated with aflibercept and ranibizumab.
The committee understood that aflibercept and ranibizumab are standard treatments for wet age-related macular degeneration in the NHS. So, it concluded that aflibercept and ranibizumab were appropriate comparators for this appraisal.
3.2 The committee was aware that bevacizumab was specified as a comparator in the scope issued by NICE and considered whether the company should have included it as a comparator in its submission. It acknowledged that because bevacizumab has not been appraised by NICE for treating wet age-related macular degeneration it could not be considered as a comparator in the fast track appraisal process. So, it concluded that bevacizumab was not a relevant comparator in this appraisal.
3.3 The company presented the results from the HAWK and HARRIER trials, comparing the safety and effectiveness of brolucizumab (3 mg and 6 mg) with aflibercept (2 mg). HAWK and HARRIER included adults aged 50 years or more with active choroidal neovascularisation caused by age-related macular degeneration and who had not had anti-vascular endothelial growth factor therapy. The primary outcome was change in BCVA from baseline to week 48. People in HAWK and HARRIER had brolucizumab monthly for the first 3 months then every 8 or 12 weeks, or aflibercept monthly for 3 months then every 8 weeks. The HAWK results showed from baseline to week 48, the least squares mean difference (LSMD) in change in BCVA between brolucizumab (3 mg) and aflibercept was -0.6 (95% confidence interval [CI] -2.5 to 1.3). When comparing brolucizumab (6 mg) with aflibercept, the LSMD was -0.2 (95% CI -2.1 to 1.8). The HARRIER trial results showed from baseline to week 48, the LSMD in change in BCVA between brolucizumab (6 mg) and aflibercept was -0.7 (95% CI -2.4 to 1.0). The committee agreed that because there was not a statistically significant difference in the change in BCVA results for brolucizumab and aflibercept the treatments were likely to be similar. But, it questioned whether these results confirmed that brolucizumab could be considered non-inferior to aflibercept. The company explained that there was a pre-defined non-inferiority margin of 4 letters set by the regulator, so the results confirmed brolucizumab was non-inferior to aflibercept. The ERG agreed that the results suggested brolucizumab and aflibercept were similarly effective but questioned whether the frequency and approach to dosing used in HAWK and HARRIER was reflective of clinical practice. It highlighted that most people having treatment for wet age-related macular degeneration in clinical practice would have treatment using a flexible approach such as treat and extend (TREX), pro re nata (PRN; treatment monitored frequently and administered as need), or PRN and extend (PRNX). The committee acknowledged the ERG's concerns but agreed that the results of HAWK and HARRIER were robust. It concluded that brolucizumab is non-inferior to aflibercept.
3.4 To support the HAWK and HARRIER results and estimate the relative effectiveness of brolucizumab compared with ranibizumab, the company presented the results of a network meta-analysis. The network meta-analysis assessed the relative effectiveness of brolucizumab, aflibercept and ranibizumab across 6 outcomes. The network comparing mean change in BCVA from baseline to 1 year contained 13 studies, including HAWK and HARRIER (see section 3.3). The results showed that treatment with brolucizumab (6 mg) every 8 or 12 weeks was similarly effective compared with various aflibercept (2 mg) and ranibizumab (0.5 mg) dosing regimens (every 4, 8 and 12 weeks, PRN, PRNX, and TREX). The committee concluded that the effectiveness of brolucizumab is similar to aflibercept and ranibizumab.
3.5 The company explained that the number of ocular and non-ocular adverse events were balanced across brolucizumab and aflibercept treatment groups in HAWK and HARRIER (see section 3.3). It also noted that the results of the network meta-analysis (see section 3.4) showed that adverse events with brolucizumab are similar to aflibercept and ranibizumab. The committee was aware that brolucizumab's summary of product characteristics notes a risk of retinal vasculitis and retinal vascular occlusion. The ERG explained that because these adverse events were rare, it was not likely to affect the view that the overall impact on health of those associated with brolucizumab are similar to those of aflibercept and ranibizumab. The committee concluded that adverse events with brolucizumab are likely to be similar to aflibercept and ranibizumab.
3.6 The committee concluded that because changes in BCVA and reports of adverse events with brolucizumab, aflibercept and ranibizumab were similar, the treatments were also likely to provide similar overall health benefits.
3.7 The company estimated the number of brolucizumab injections in years 1 and 2 from the network meta-analysis, which used pooled data from HAWK and HARRIER and accounted for differences in a random-effects model. The ERG agreed that this approach was appropriate and incorporated the company's estimates in its preferred analysis. The committee concluded that the number of brolucizumab injections in years 1 and 2 should be based on trial data and agreed the company's approach was acceptable.
3.8 To estimate the number of injections for aflibercept and ranibizumab applied in the model in years 1 and 2, the company used a weighted average approach. This approach weighted the number of injections from different treatment regimens by the amount each regimen was proportionally used in practice. The company obtained estimates of proportions from a survey of 50 healthcare professionals who are retinal specialists. The ERG noted that most people in clinical practice followed a flexible treatment regimen such as TREX and PRN. So, it considered that the company's approach, which pooled different regimens, was unnecessary and presented an alternative approach, which separately estimated year 1 and 2 dose frequencies for TREX and PRN regimens. The committee agreed that because TREX was the most commonly used regimen in practice, it preferred analyses based on this. It concluded that for the comparators, the number of injections in years 1 and 2 should be based on TREX regimens.
3.9 The company assumed that the number of injections given for each treatment in year 3 and beyond would be the same as the number of injections given in year 2. It explained that in absence of longer-term data, it was difficult to assume that effectiveness would be maintained if the number of injections reduced. So, it stated that assuming the number of injections was the same as in year 2 was a reasonable and consistent approach. The ERG explained that if the number of brolucizumab injections was increased because of disease activity in HAWK and HARRIER it could not later be decreased. It highlighted that because of this, the number of brolucizumab injections given in the trials was likely to be an overestimate of the number of injections given in clinical practice. It also highlighted that there were no available data to compare injection numbers across treatments for year 3 and beyond. So, it presented an alternative, based on analysis from NICE's technology appraisal guidance on aflibercept, where the number of injections is assumed to be the same for all treatments in year 3 and beyond. The committee agreed that the company's approach could lead to an overestimate of injection numbers. It also agreed that in the absence of longer-term data to inform estimates of injection numbers in year 3 and beyond, it was reasonable to assume it would be equivalent across treatments. It was uncertain if the year 3 injection numbers estimate from NICE's appraisal of aflibercept were accurate but agreed in the absence of more robust alternatives they were acceptable for decision making. So, it concluded it preferred the ERG's approach, which assumed the same number of injections for all treatments in year 3 and beyond from NICE's appraisal of aflibercept.
3.10 The company presented cost-comparison results for brolucizumab compared with aflibercept and ranibizumab. When taking account of the confidential commercial discounts for all treatments, it showed that the total cost associated with brolucizumab was similar or lower than aflibercept and ranibizumab (the exact results are confidential and cannot be reported here). The committee acknowledged these results but recalled that it preferred the following assumptions incorporated in the ERG's preferred analysis:
comparator injection numbers in year 1 and 2 based on TREX regimen estimates
injection numbers in year 3 and beyond is equivalent for all treatments and based on estimates from NICE's appraisal of aflibercept.
Taking account of these assumptions, and the confidential commercial discounts for all treatments, the total cost for brolucizumab was similar or lower than aflibercept and ranibizumab (the exact results are confidential and cannot be reported here). The committee concluded that the criteria for a positive cost comparison were met, because:
the overall health benefits of brolucizumab are similar to those of aflibercept and ranibizumab
the total costs associated with brolucizumab are similar to or lower than those of aflibercept and ranibizumab.
The committee therefore recommended brolucizumab, in line with the previous recommendations for aflibercept and ranibizumab, as a cost-effective use of NHS resources for treating wet age-related macular degeneration in adults.
3.11 The company noted that visual impairment resulting from wet age-related macular degeneration is recognised as a disability. So, the patient population of this appraisal is a protected group under the Equality Act 2010. The committee took the fact that brolucizumab would be used in people with visual impairment into consideration when formulating its recommendations.