Aflibercept for treating diabetic macular oedema

Clinical effectiveness

4.4 The Committee considered the clinical‑effectiveness data from the VISTA and VIVID trials that compared laser therapy with aflibercept (2 mg intravitreal injections every 4 weeks or every 8 weeks after 5 initial monthly doses). The Committee discussed the concerns of the Evidence Review Group (ERG) about the generalisability of the results to clinical practice in England. The Committee heard that the trials included patients whose mean HbA1c was lower than generally found in clinical practice in England, meaning that people in a clinical setting may not respond to treatment as well as reported in the clinical trials. The Committee heard from the clinical experts that HbA1c values do not normally affect the prognosis or treatment options for people with DMO. Cardiovascular markers, for example hypertension, have a bigger impact on the disease. The Committee concluded that overall the trials were generalisable to clinical practice in England.

4.5 The Committee considered the results of the VISTA and VIVID trials, which reported that aflibercept significantly improved visual acuity compared with laser for the primary outcome of mean change in best corrected visual acuity (BCVA). The results of the secondary outcomes showed that aflibercept was better than laser in all outcomes apart from in the NEI‑VFQ‑25 quality‑of‑life scores in VISTA 2Q8. The Committee concluded that aflibercept was better than laser based on the results presented in the trials. The Committee considered the ERG's subgroup analysis for the clinical effectiveness of aflibercept compared with laser in the CRT 'less than 400 micrometres' group (see section 3.24) and the company's and ERG's subgroup analyses for '400 micrometres or more' group (see sections 3.4 and 3.24 respectively). The ERG used the post‑hoc analysis of the VISTA and VIVID trials, presented by the company, to establish the relative risk of aflibercept compared with laser for the gains and losses in visual acuity for the 2 subgroups. The Committee heard from the ERG that results of the analysis were uncertain because it broke the randomisation, was based on small patient numbers (n=78 in the less than 400 micrometres group and n=208 in the 400 micrometres or more group) and used inappropriately pooled data from the VISTA and VIVID trials. The subgroup analyses showed a statistically significant improvement in visual acuity gains and prevention of visual acuity losses with aflibercept compared with laser in patients with a CRT of 400 micrometres or more. In the CRT less than 400 micrometres group, aflibercept had no significant improvement over laser in all but 1 of the visual acuity outcomes. The Committee noted the comments received during consultation from the comparator companies about the recently published results from the Protocol T study. It was aware that the study compared the relative efficacy and safety of aflibercept with bevacizumab and ranibizumab and that it provided some evidence for the relative effectiveness of aflibercept compared with bevacizumab and with ranibizumab in people with a CRT less than 400 micrometres. The Committee also noted the comments from one of the comparator companies and heard from the ERG that the evidence from the trial for the comparison of aflibercept with ranibizumab was not relevant for this appraisal because the ranibizumab treatment arm was dosed at 0.3 mg pro re nata (PRN), which is not consistent with the dose specified in the summary of product characteristics for ranibizumab (0.5 mg PRN). The Committee acknowledged the limitations of the subgroup analysis on CRT from the Protocol T study and agreed that the results could not be considered in its decision‑making. The Committee also acknowledged the limitations of the subgroup analysis on CRT from the VIVID and VISTA trials but concluded that it was the only clinical data provided on the effectiveness of aflibercept in this group and that the results could be considered for decision‑making.

4.6 The Committee considered the evidence from the network meta‑analysis submitted by the company. It noted that the results indicated a statistically significant difference between aflibercept and ranibizumab for mean change in BCVA but no significant differences in alternative visual acuity outcomes (for example gain of 15 or more Early Treatment of Diabetic Retinopathy Study (ETDRS) letters, gain of 10 or more ETDRS letters, or safety). The Committee considered comments received in consultation by a comparator manufacturer regarding the network meta‑analysis. It noted a comment that the network meta‑analysis provided by the company was misleading because it did not include all the relevant trials for aflibercept (the Da Vinci trial) and ranibizumab (READ‑2 or RESOLVE trials). The Committee was further aware from the ERG that another network meta‑analysis comparing aflibercept with ranibizumab was identified (Haig et al. 2014), which included the Da Vinci, READ‑2 and RESOLVE trials. However 3 of the trials included in the analysis were unpublished. The Committee heard from the clinical experts that because aflibercept and ranibizumab are the same class of drug they would be expected to have similar clinical efficacy. They also noted that DMO in some people may respond better to either aflibercept or ranibizumab but it is not possible to predict in advance the most effective treatment. The Committee concluded that aflibercept is likely to have similar clinical effectiveness to ranibizumab, based on the results of the network meta‑analysis and clinical expert opinion.

4.7 The Committee considered the safety profile of aflibercept. The Committee noted the clinical experts' agreement that, based on clinical practice and the results of the trials, aflibercept is well tolerated. The Committee accepted that there were no major safety concerns associated with aflibercept.

Cost effectiveness

4.8 The Committee considered the economic model submitted by the company and the ERG critique. The Committee noted that the company model was well structured and accounted for vision loss in both the better seeing eye and worse seeing eye. The Committee concluded that the company model was acceptable for assessing the cost effectiveness of aflibercept.

4.9 The Committee considered the cost of blindness used in the model. The Committee heard from the ERG that the annual cost of blindness had been applied monthly and had not been discounted (see section 3.19), but that this had been corrected in the original ERG base‑case analysis, which the Committee concluded was appropriate.

4.10 The Committee considered the utility values used in the model. The Committee noted that the EQ‑5D trial data could be used in the base case because it reflects the NICE reference case, but the Committee was aware that the directly measured EQ‑5D values may underestimate the effect of ophthalmic conditions on health‑related quality of life and the impact of improvement in BCVA. It considered that the literature‑sourced values from Czoski‑Murray et al. (2009) were not ideal because the values apply only to the bilateral BCVA, which meant that the company had to use an adjustment factor to calculate the utility values of the worse seeing eye. The Committee acknowledged the company's reason for using Czoski‑Murray et al. (2009) utility values in its submission (that is, consistency with other NICE technology eye appraisals). It also acknowledged that sensitivity analyses using the utility values from Brown (1999) and Brown (2000) were included. It concluded that the Czoski‑Murray et al. utility values, although not ideal, were an acceptable basis for its decision‑making.

4.11 The Committee considered the cost‑effectiveness analysis and incremental cost‑effectiveness ratios (ICERs) for aflibercept compared with laser for the whole trial population. It noted that aflibercept dominated (was less costly and more effective than) laser in the company's original base case and continued to dominate laser in all of the company's original sensitivity analyses. However, it noted that in the ERG's original base‑case analysis, when the cost of blindness error was corrected the incremental costs increased from −£2440 (cost saving) to £12,900 and the ICER calculated by the ERG was £33,900 per quality‑adjusted life year (QALY) gained (section 3.20). The Committee considered the company's new evidence and revised cost‑effectiveness analyses provided after consultation on the appraisal consultation document for this comparison (section 3.25–3.28). The Committee noted that the revised company ICER for aflibercept compared with laser in the whole trial population was £21,700 per QALY gained. The Committee considered each of the company's changes to the economic model (see section 3.27) and the ERG's critique of the amendments (see sections 3.29–3.33) in turn (see sections 4.12–4.17).

4.12 The Committee considered the company's rationale for decreasing the number of aflibercept injections in year 2 from 5.45 to 4.00 (see sections 3.26–3.27) and the ERG's critique (see section 3.29). The Committee acknowledged that the summary of product characteristics for aflibercept and ranibizumab states a reduced dosing interval after the first 12 months, and agreed that there is uncertainty around the average number of aflibercept injections that a person would receive after the first 12 months. Given that there is no robust clinical data for estimating the average number of aflibercept injections in year 2, the Committee concluded that the economic modelling of treatment should be based on trial data, and that a sensitivity analysis that included an equalisation of the number of injections of aflibercept and ranibizumab in year 2 was an acceptable basis for its decision‑making.

4.13 The Committee considered the company's rationale (see sections 3.26–3.27) and the ERG's critique (see section 3.30) for increasing the cost of a laser administration from £139 to £194. The Committee concluded that it was appropriate to have an equal cost for both a laser and intravitreal injection administration and agreed to increase the cost of laser administration from £139 to £194.

4.14 The Committee considered the company's rationale (see section 3.27) and the ERG's critique (see section 3.29) for increasing laser monitoring visits in the model from 4 to 12 in year 1. The Committee was also aware that 12 monitoring visits per year was not consistent with professional guidance. The Committee was aware of the importance of modelling according to trial protocol when considering treatment and that it was not necessary when considering monitoring (because trials have to collect regular data). The Committee concluded that it was not appropriate to increase the number of monitoring visits for laser in year 1 in the economic model.

4.15 The Committee considered the company's rationale (see section 3.27) and the ERG's critique (see section 3.30) for decreasing the fellow eye involvement at baseline from 85% to 46%. The Committee noted that the company's proposed amendment, in which the Committee was given to assume that all of the 46% of people would be treated, was very similar to the actual value used in the ERG's original base case and would have little impact on the ICER. The Committee concluded that it was not necessary to decrease the rate of fellow eye involvement.

4.16 The Committee considered the company's rationale (see section 3.27) and the ERG's critique (see section 3.29) for using the Preis et al. (2005) mortality multiplier instead of the ERG's preferred multiplier from Mulnier et al. (2006). The Committee agreed that using a mortality rate from Preis et al. rather than from Mulnier et al. would have little impact on the ICER because it would be combined with the mortality multiplier from Hirai et al. (2008). The Committee concluded that it was not necessary to include the mortality multiplier from Preis et al. in the cost‑effectiveness analysis.

4.17 Using patient‑level data from the VIVID and VISTA trials to inform the transition probabilities in the economic model: The Committee considered the company's rationale (see section 3.27) and the ERG's critique (see section 3.29) for the use of patient‑level data. The Committee accepted that the patient‑level data may improve the ICER for aflibercept compared with laser but was concerned that it had not been critiqued by the ERG. Furthermore the Committee heard that the company had not used these data for any analysis of the patient group with a CRT of less than 400 micrometres. The Committee concluded that it was not necessary to include individual patient data from the VISTA and VIVID trials instead of the relative risks from the network meta‑analysis in the cost‑effectiveness analysis.

4.18 The Committee then considered the revised base‑case ICER for aflibercept compared with laser in the whole trial population that incorporated the Committee's preferred assumption of an increased cost of laser administration. The Committee noted that the ICER was £33,100 per QALY gained. The Committee also considered the sensitivity analysis applying 4.0 aflibercept injections in year 2 and noted that the ICER was £30,800 per QALY gained. The Committee agreed that it was appropriate to use the company's revised base‑case analysis and the resulting ICER for aflibercept compared with laser of £33,100 per QALY gained. However it was aware from the clinical experts that in clinical practice the choice of treatment depends on the CRT and so it agreed that it should consider separately the cost effectiveness of aflibercept compared with laser in people with a CRT of less than 400 micrometres and in people with a CRT of 400 micrometres or more.

4.19 The Committee considered the ICERs for aflibercept compared with laser in the subgroup of people with a CRT of less than 400 micrometres (for whom ranibizumab is not recommended; NICE's technology appraisal guidance on ranibizumab for treating diabetic macular oedema). The ICER for the whole trial population had suggested that aflibercept compared with laser in this group was not a cost‑effective use of NHS resources. The Committee noted that the ICER for the less than 400 micrometres CRT subgroup was £49,400 per QALY gained for the comparison of aflibercept with laser. The Committee noted that the company had not submitted any new evidence for this subgroup in its response to consultation on the appraisal consultation document. It considered the revised ICER using the increased cost of laser administration, which was £48,300 per QALY gained (see section 3.30). The Committee heard from the clinical experts that the use of laser as the standard of care for treatment of DMO is declining. In this subgroup, clinicians sometimes prefer to use bevacizumab outside its marketing authorisation (when available) or adopt a watch‑and‑wait strategy until CRT increases and the person becomes eligible for treatment with ranibizumab. The Committee noted that it had not been presented with evidence on the cost effectiveness of aflibercept compared with bevacizumab or with a 'watch‑and‑wait' strategy. The Committee concluded that, based on its consideration of all the evidence (including no evidence of its cost effectiveness against other treatment strategies), aflibercept is not a cost‑effective use of NHS resources compared with laser treatment for people with a CRT of less than 400 micrometres and is therefore not recommended.

4.20 The Committee considered the ICERs for aflibercept compared with laser in the subgroup of people with CRT 400 micrometres or more. The ICER for people with a CRT of 400 micrometres or more was £22,000 per QALY gained. The Committee noted that the company had not submitted any new evidence for this group in response to consultation. It considered the revised ICER by the ERG using the increased cost of laser administration, which was £21,400 per QALY gained (see section 3.30). However, the Committee was aware that the main comparator for this population was ranibizumab, but no comparison with ranibizumab was included in the cost‑effectiveness evidence for the CRT subgroups (section 4.21).

4.21 The Committee considered the cost‑effectiveness analysis and ICERs for aflibercept compared with ranibizumab for the whole trial population. The Committee was aware of the actual discount agreed in the patient access scheme for ranibizumab (this is commercial in confidence and therefore cannot be reported). It noted that the range of analyses (0%–100% discount from the list price) undertaken by the company and the ERG included the discount agreed in the patient access scheme for ranibizumab. The Committee noted that when the exact discount agreed in the patient access scheme for ranibizumab was taken into account, the ICERs for aflibercept compared with ranibizumab from the company's base‑case analysis and from the ERG's analysis were within the range normally considered to be a cost‑effective use of NHS resources (up to £20,000 per QALY gained). However, the Committee noted that ranibizumab is currently only recommended by NICE for people with DMO whose CRT is 400 micrometres or more (NICE's technology appraisal guidance on ranibizumab for treating diabetic macular oedema). The Committee considered that the whole population analysis taken together with clinical expert testimony justified a conclusion that, for people with a CRT of 400 micrometres or more where ranibizumab is the comparator treatment, aflibercept is a cost‑effective use of NHS resources for treating people with DMO.

4.22 The Committee considered the potential cost effectiveness of sequential treatment with anti‑VEGF agents. The Committee was not presented with any cost‑effectiveness data on the sequential use of anti‑VEGF treatments. The Committee raised concerns that potentially multiple treatments of ranibizumab could be followed by multiple treatments of aflibercept (if the person's disease does not adequately respond to ranibizumab). The Committee noted that because an increase in costs of treatment in this circumstance would not be matched by a similar gain in QALYs it is unlikely that sequential treatment with ranibizumab followed by aflibercept would be cost effective. The clinical experts explained that because anti‑VEGF treatments (NICE's technology appraisal guidance on ranibizumab for treating diabetic macular oedema) have only been available for a limited amount of time, there is no established best practice on the optimal conditions for switching anti‑VEGF treatments. The Committee concluded that, in the absence of evidence, no recommendations could be made on the cost effectiveness of sequential treatment with anti‑VEGFs.

4.23 The Committee discussed how innovative aflibercept is in its potential to make a significant and substantial impact on health‑related benefits. It noted the clinical expert views that aflibercept was a useful addition to the anti‑VEGF products available.

4.24 The Committee considered whether it should take into account the consequences of Pharmaceutical Price Regulation System (PPRS) 2014, and in particular the PPRS payment mechanism, when appraising aflibercept. The Appraisal Committee noted NICE's position statement in this regard, and accepted the conclusion 'that the 2014 PPRS Payment Mechanism should not, as a matter of course, be regarded as a relevant consideration in its assessment of the cost effectiveness of branded medicines'. The Committee heard nothing to suggest that there is any basis for taking a different view with regard to the relevance of the PPRS to this appraisal of aflibercept. It therefore concluded that the PPRS payment mechanism was irrelevant for the consideration of cost effectiveness of aflibercept.

Summary of Appraisal Committee's key conclusions