The Appraisal Committee reviewed the data available on the clinical and cost effectiveness of ranibizumab, having considered evidence on the nature of visual impairment caused by macular oedema secondary to retinal vein occlusion (RVO) and the value placed on the benefits of ranibizumab by people with the condition, those who represent them, and clinical specialists. It also took into account the effective use of NHS resources.
The Committee heard from patient experts about the problems associated with visual impairment caused by macular oedema. It heard that the loss of vision has a significant effect on the independence of people with the condition. The patient experts also stated that the condition affects ability to work and hobbies such as reading and gardening. The patient experts acknowledged that although people may be worried about having an injection in the eye, they are willing to receive injections in order to keep their sight. The Committee agreed that loss of vision caused by macular oedema secondary to retinal vein occlusion seriously impairs health‑related quality of life.
The Committee heard from clinical specialists that the current standard treatment for visual impairment caused by macular oedema secondary to branch RVO (BRVO) is grid laser photocoagulation but that in interim guidelines from the Royal College of Ophthalmologists it is only recommended after a period of 3 to 6 months following the initial event (obstruction of retinal veins) and following the absorption of the majority of the haemorrhage. The Committee heard from the clinical specialists that grid laser photocoagulation is not an option for people with central RVO (CRVO), because CRVO does not respond to grid laser photocoagulation (as outlined in the Royal College of Ophthalmologists' interim guidelines on RVO) and the current standard treatment is dexamethasone or anti‑vascular endothelial growth factor (VEGF) drugs such as bevacizumab.
The Committee considered the comparators for the appraisal, and specifically bevacizumab intravitreal injection. It was aware that bevacizumab does not have a UK marketing authorisation for the treatment of RVO and heard from patient experts that they were concerned about using unlicensed treatments for which there was no formal post‑marketing surveillance, particularly if there were alternatives that have a UK marketing authorisation. The Committee noted that a marketing authorisation is not a prerequisite for a comparator in a NICE technology appraisal. It noted that NICE's guide to the methods of technology appraisal, in recommending comparison with technologies that are 'best practice' or in 'routine use', is not intended to be restrictive but to emphasise the need for comparison with all relevant comparators; any medicine in routine use or considered to be best practice should be considered a comparator. The Committee was minded to conclude that bevacizumab fulfils the requirements for inclusion as a comparator but noted the advice from the NICE Board that this decision should be based on a careful consideration of its use in clinical practice for the condition concerned and, critically, a thorough assessment of its efficacy and safety.
The Committee considered the information in the Decision Support Unit (DSU) report on the product quality of reformulated bevacizumab. It was aware of consultation comments on the DSU report raising concerns on quality and reports of endophthalmitis and uveitis with intravitreal bevacizumab, although it was not clear how the number of reports compared with those observed with ranibizumab. The Committee also noted the comments from consultation on the DSU report that reformulation of pharmaceutical products is not an unusual practice and is routinely performed in many other circumstances under a 'specials' licence. The Committee noted that reformulating bevacizumab for intravitreal use requires a 'specials' licence from the Medicines and Healthcare products Regulatory Agency (MHRA) and that this means manufacturers must adhere to a range of conditions and inspections. The Committee was satisfied that there is some level of good manufacturing practice in place when pharmaceutical products are reformulated under a 'specials' licence and that such practice is not exceptional.
The Committee considered the evidence relating to the safety of bevacizumab as reported by the DSU. It was aware that in the 22 randomised controlled trials (RCTs) identified for age‑related macular degeneration (AMD), diabetic macular oedema and RVO, adverse events were few compared with sham injection, laser photocoagulation and other intravitreal treatments. The Committee noted the pooled analysis from the IVAN and CATT trials (both in patients with AMD), which showed a statistically significantly higher rate of systemic adverse events in the bevacizumab group than in the ranibizumab group. However, it also noted observational data from a large study by Curtis et al. suggesting no difference in the risk of adverse events between bevacizumab and ranibizumab. It also noted a population‑based case–control study, including over 90,000 patients, that reported no relationship between the risk of systemic events such as myocardial infarction, venous thromboembolism, stroke or congestive heart failure and the administration of intravitreal bevacizumab or ranibizumab. The Committee concluded that the evidence base relating to the safety of bevacizumab was sufficient to inform judgement of whether bevacizumab is an appropriate comparator.
The Committee considered the evidence base for the efficacy of bevacizumab in treating visual impairment caused by macular oedema secondary to BRVO and CRVO. It noted the small trials (2 in BRVO, 3 in CRVO) identified in the DSU report, of which 3 were published only as abstracts. The Committee agreed that all trials reported significant mean improvements in best corrected visual acuity (BCVA) for bevacizumab compared with sham injections, but because no direct comparisons of ranibizumab with intravitreal bevacizumab are currently available, a mixed treatment comparison would be needed to answer the question of relative effectiveness between the 2 treatments. The Committee noted that the DSU was not asked to address this question specifically. The Committee concluded that the available evidence was limited with small sample sizes and differences in study populations but on balance sufficient to inform judgement of whether bevacizumab is an appropriate comparator.
Having noted the available evidence and comments from consultation on the safety, efficacy and quality of intravitreal bevacizumab, the Committee concluded that bevacizumab is an appropriate potential comparator in this appraisal. However, the Committee further concluded that there is currently insufficient evidence to make robust comparisons with ranibizumab needed for a cost‑effectiveness analysis.
The Committee considered the evidence presented by the manufacturer on the clinical effectiveness of ranibizumab. It noted that the main sources of evidence came from the BRAVO and CRUISE RCTs, which included patients with macular oedema secondary to BRVO and CRVO respectively. It also noted the evidence from the 12‑month open‑label extension of both trials, the HORIZON study. The Committee noted that the decision problem for the appraisal included people with or without retinal ischaemia. However it was aware that most patients with retinal ischaemia were excluded from the BRAVO and CRUISE trials, because patients with a brisk afferent pupillary defect (which equates to severe retinal ischaemia) were excluded. It heard from the clinical specialists and the manufacturer that this meant that there was no evidence of ranibizumab for treating visual impairment caused by RVO in patients with significant ischaemia.
The Committee noted that in both BRAVO and CRUISE ranibizumab was associated with statistically significant mean gains in BCVA in the treated eye compared with sham injection for the 6‑month treatment phase. It also noted that ranibizumab was associated with sustained gains in BCVA at 12 months in both BRAVO and CRUISE, and that these were statistically significant (p<0.01 and p<0.001 respectively). The Committee was aware that ranibizumab could be used as needed in both arms of both trials from 6 months. In addition, the Committee was aware that in the BRAVO trial, grid laser photocoagulation was permitted after 3 months in both the sham group and the ranibizumab group, confounding the results of the treatment phase from month 3 onwards. Despite this, the Committee concluded that ranibizumab is a clinically effective treatment for visual impairment caused by macular oedema secondary to BRVO and CRVO compared with sham injection, if there is no significant retinal ischaemia.
The Committee considered the evidence for adverse effects associated with ranibizumab. It noted that the safety of ranibizumab had been shown previously in patients with wet AMD (NICE's technology appraisal guidance on ranibizumab and pegaptanib for the treatment of age-related macular degeneration) and diabetic macular oedema (NICE's technology appraisal guidance on ranibizumab for treating diabetic macular oedema). The Committee also noted that the overall frequency of adverse effects in the BRAVO and CRUISE trials at month 6 was low. It agreed that ranibizumab was safe and well tolerated in patients with macular oedema secondary to RVO.
The Committee noted that the BRAVO and CRUISE trials collected data on the effect of visual impairment on quality of life using the National Eye Institute Visual Function Questionnaire‑25 (NEI VFQ‑25) questionnaire. It noted that both trials reported a statistically significant difference in NEI VFQ‑25 score at month 6 between the ranibizumab and sham injection groups. The Committee concluded that treating patients with ranibizumab improves the quality of life of people with visual impairment caused by macular oedema secondary to RVO.
The Committee considered the manufacturer's original economic model and the critique and exploratory analyses performed by the Evidence Review Group (ERG). It noted the manufacturer's original base‑case incremental cost‑effectiveness ratios (ICERs) of £8,600 and £20,500 per quality‑adjusted life year (QALY) gained for ranibizumab compared with standard care for CRVO and BRVO respectively and the ICERs for ranibizumab compared with dexamethasone in CRVO and BRVO which were £7,200 and £5,500 per QALY gained respectively. The Committee broadly accepted the model structure, but was concerned by some of the uncertainties highlighted by the ERG around the assumptions used by the manufacturer. In particular, the Committee did not accept:
the assumption that all patients would be treated in their 'better‑seeing eye', having heard from clinical specialists that this is not the case in clinical practice, and that most patients in the CRUISE and BRAVO trials were treated in their 'worse‑seeing eye' (see section 3.22)
the manufacturer's use of 'better‑seeing eye' utility values from the Brown study, without age adjustment (see section 3.27)
the absence of a mortality risk associated with RVO in the model (see section 3.25)
the use of pooled transition probabilities during months 7 to 12 of the BRAVO trial, which overestimated the efficacy of ranibizumab compared with sham injection (see section 3.23)
the potential bias in the indirect comparison between ranibizumab and dexamethasone (both BRVO and CRVO), with different exclusion rules for ischaemia, patients with different durations of macular oedema and different severities in the trials of each drug (see section 3.24).
The Committee concluded that the most plausible ICERs for ranibizumab compared with standard care and dexamethasone intravitreal implant, based on the manufacturer's base case modified appropriately by the ERG, were likely to range from £31,100 to £52,000 per QALY gained and would therefore be well in excess of £20,000 to £30,000 per QALY gained. It also noted that there was additional uncertainty about the cost‑effectiveness estimates for ranibizumab in people with BRVO because grid laser photocoagulation was permitted after 3 months in both the sham and the ranibizumab groups, confounding the results of the treatment phase from month 3 onwards. The Committee proceeded to consider the revised model submitted by the manufacturer.
The Committee noted the manufacturer's revisions to its economic model submitted in response to consultation. It first noted the amendment to reflect the fact that most patients (90%) would be treated in their 'worse‑seeing eye'. The Committee considered that this was consistent with the BRAVO and CRUISE trials and clinical practice, and concluded that this amendment was appropriate.
The Committee next considered the manufacturer's revised approach to deriving utilities for the 'better‑seeing eye' from Czoski‑Murray et al. (2009) for use in the economic model. It understood that these utilities would only apply to 10% of the people in the revised economic model which now assumed that 90% of people would be treated in their 'worse‑seeing eye' (section 4.14). The Committee noted that the manufacturer had applied a regression equation from Czoski‑Murray et al. (2009) to produce a finer degradation of the utilities. The Committee noted that the ERG accepted the manufacturer's use and implementation of the utilities as applied using the Czoski‑Murray equation and further noted the provisional guidance from the rapid review of NICE technology appraisal guidance 237 (now published as NICE's technology appraisal guidance on ranibizumab for treating diabetic macular oedema) in which the range of utility values was accepted to lie somewhere in between those estimated by Czoski‑Murray and those from the Brown study. The Committee concluded that although uncertain, the use of utilities as applied using the Czoski‑Murray equation was acceptable.
The Committee understood that the manufacturer's submission initially assumed that all people in the economic model would be treated in their 'better‑seeing eye' and therefore did not apply a utility gain associated with treating the 'worse‑seeing eye'. It noted that in the revised economic model (submitted in response to the appraisal consultation document) it was assumed that most people (90%) would be treated in their 'worse‑seeing eye' in line with the ERG's suggestion. The Committee therefore considered the manufacturer's revised assumption of applying a 0.3 utility gain associated with treating the 'worse‑seeing eye'. The Committee heard from the ERG that the manufacturer's assumption was based on an extrapolation of evidence from Brown et al. (2001). It understood that the ERG had used a gain of 0.1 from the Brown study in which utility values were collected separately for the 'worse‑seeing' and 'better‑seeing' eyes. The Committee considered that a 0.3 utility gain associated with treating the 'worse‑seeing eye' seems high given that utility is driven primarily by the 'better‑seeing eye', and therefore lacked face validity. It further noted that the manufacturer had originally suggested that no gain in utility would be obtained from treating the 'worse‑seeing eye'. The Committee was also aware of the results of an analysis from NICE's technology appraisal guidance on dexamethasone intravitreal implant for the treatment of macular oedema secondary to retinal vein occlusion the details of which are commercial in confidence. The Committee concluded that a utility gain of 0.1 associated with treating the 'worse‑seeing eye' was appropriate.
The Committee considered evidence supplied by the manufacturer during consultation relating to excess cardiovascular mortality associated with RVO, that is, the additional risk caused by cardiovascular complications associated with RVO, compared with the general population. The Committee noted that the excess mortality risk incorporated in the ERG base case was based on a paper from 2000 (Tsaloumas), which suggested that a person with RVO would have 1.6 times that of the general population. It noted that, of the papers referenced by the manufacturer in response to the original consultation, none suggested that overall mortality was lower for RVO patients than for the population at large. Some suggested it was greater. But the Committee was also aware that excess cardiovascular mortality had not been applied in NICE's technology appraisal guidance on dexamethasone intravitreal implant for the treatment of macular oedema secondary to retinal vein occlusion. The Committee concluded that the evidence on the risk of cardiovascular mortality associated with RVO was unclear, and therefore it need not be included in the base‑case model to the degree applied in the original ERG report. However, it would remain an uncertainty in the analysis.
The Committee considered the manufacturer's response to the Committee's concerns about the use of pooled transition probabilities. The Committee was aware that the revised model used data from the ranibizumab arm of BRAVO to inform the transitions of all BRVO patients from month 7 to 24 which, in the view of the manufacturer, was a more conservative approach. The Committee noted that this had the effect of boosting the model's mean efficacy of the ranibizumab group after 6 months of treatment, because it assumes the same response to treatment for a person previously treated with ranibizumab as for a person naïve to ranibizumab treatment. It also noted that patients treated with grid laser photocoagulation or dexamethasone will experience the same benefit as patients treated with ranibizumab who are moving onto treatment with ranibizumab given as needed. The Committee heard from the ERG that because of the decline in efficacy of ranibizumab when given as needed in the extension arm of the trial, compared with monthly ranibizumab, it was unclear if the manufacturer's assumption was conservative. The Committee noted that the manufacturer's approach had minimal impact on the revised ICER for ranibizumab compared with dexamethasone in BRVO. The Committee acknowledged that there were advantages and disadvantages to the manufacturer and ERG's approaches. But it concluded that, given the lack of clear data, the approach taken by the manufacturer was appropriate.
The Committee discussed the manufacturer's revised assumption relating to adverse events. The Committee noted that the revised model included updated adverse event rates in year 2, which included iris neovascularisation as an adverse event for ranibizumab and dexamethasone and an updated estimate of the rate of cataract development for dexamethasone (based on 12‑month outcomes from the GENEVA studies). The Committee noted the ERG's concern that it is not clear how the rate of iris neovascularisation was calculated for year 2. Although it acknowledged the ERG's concerns with the methods used to estimate adverse events in year 2, it cautiously accepted the updated safety data in the model.
The Committee considered the manufacturer's consultation response to the ERG's opinion that its exploratory economic comparison of ranibizumab with dexamethasone was biased in favour of ranibizumab. The Committee considered the manufacturer's points that the duration of macular oedema may have been underestimated by up to 1 month and that the rates of ischaemic disease were higher in the ranibizumab studies, and that the comparison at 3 months does not take into consideration the declining efficacy of dexamethasone. The Committee heard from the ERG that there may still be greater bias in favour of ranibizumab. This was because dexamethasone's efficacy starts declining at 2 months (which was incorporated in the analysis at 3 months) so in the first cycle of treatment dexamethasone's efficacy was in between that of the sham and ranibizumab treatment and was assumed to be equivalent to best supportive care in months 2 to 6, after which the same efficacy as ranibizumab was assumed. The Committee accepted that the relative effectiveness of ranibizumab and dexamethasone was uncertain and concluded that it was difficult to quantify any bias.
Having discussed the assumptions in the manufacturer's revised base‑case model the Committee went on to discuss the ICERs produced from this model. It was aware that the manufacturer's revised model included the patient access scheme as revised in the context of NICE technology appraisal 274. It noted the ICERs for BRVO of £23,100 and £2,400 per QALY gained for ranibizumab compared with standard care (grid laser photocoagulation) and with dexamethasone respectively, and the base‑case ICERs for CRVO of £13,900 and £7,000 per QALY gained for ranibizumab compared with best supportive care and dexamethasone respectively. The Committee noted that the ERG had accepted the manufacturer's assumptions relating to 90% of patients being treated in the 'worse‑seeing eye', use of the 'better‑seeing eye' utilities from Czoski‑Murray et al. (2009) as modelled by the manufacturer, excess mortality associated with visual impairment in the 'worse‑seeing eye', updated adverse events for year 2, and a lifetime time horizon. However, the Committee was aware that the ERG had undertaken an exploratory analysis on the revised model in which a maximum utility benefit of 0.1 from treating the 'worse‑seeing eye', instead of the manufacturer's value of 0.3, had been applied. The Committee understood that this was the only difference in the calculation of the ICERs between the analyses. On the basis of its discussions relating to the maximum possible gain in utility from treating the 'worse‑seeing eye' (section 4.16), the Committee concluded that its decision should be made on the basis of the ERG's adjustment to the manufacturer's calculations.
The Committee considered the ICERs for ranibizumab for CRVO calculated in the ERG's exploratory analyses. It noted that in this incremental analysis dexamethasone was extendedly dominated (that is, dexamethasone is dominated by a combination of 2 other alternatives, in this case best supportive care and ranibizumab) and therefore can be discounted from the analysis. Therefore, the Committee went on to consider the comparison of ranibizumab with best supportive care. It noted that, incorporating the patient access scheme (as revised in the context of NICE technology appraisal 274), ranibizumab was associated with an ICER of £26,200 per QALY gained compared with best supportive care in CRVO. The Committee was aware of remaining uncertainties regarding the possible confounding in the data resulting from both groups in the CRUISE trial receiving ranibizumab as needed from month 7 (section 4.10). It was also aware of the remaining uncertainty because of the absence of a direct comparison with dexamethasone, however on balance the Committee considered that the most plausible ICER for ranibizumab for visual impairment caused by macular oedema secondary to CRVO was between the £20,000 and £30,000 per QALY gained thresholds. It could therefore be considered a cost‑effective use of NHS resources. The Committee therefore concluded that ranibizumab should be recommended as an option for treating visual impairment caused by macular oedema following CRVO. However, there remained uncertainties because of the absence of a direct comparison with dexamethasone.
The Committee considered the ICERs calculated in the ERG's exploratory analyses for ranibizumab for BRVO. The Committee noted that the key comparison was standard care with laser photocoagulation, rather than with dexamethasone, which is only recommended when laser treatment is inappropriate (NICE's technology appraisal guidance on dexamethasone intravitreal implant for the treatment of macular oedema secondary to retinal vein occlusion). The Committee therefore considered the ERG's exploratory ICER of £44,800 per QALY gained for ranibizumab compared with standard care. The Committee was aware that people receiving ranibizumab in the BRAVO trial could receive grid laser photocoagulation from month 3 and that this represented a significant confounding factor in both the manufacturer's and the ERG's calculations of the ICER for BRVO compared with standard care. It therefore considered that this ICER would be an underestimate of the most plausible ICER. The Committee concluded that the most plausible ICER for ranibizumab compared with standard care in treating BRVO was in excess of £44,800 per QALY gained. It further concluded that ranibizumab could not be recommended as an option for treating visual impairment caused macular oedema following BRVO when laser photocoagulation is an appropriate treatment option.
The Committee next considered the population with BRVO for whom grid laser photocoagulation is not an option. It considered that this population would include both those for whom grid laser photocoagulation has not been beneficial and those for whom grid laser photocoagulation is not a suitable treatment. It was aware that NICE's technology appraisal guidance on dexamethasone intravitreal implant for the treatment of macular oedema secondary to retinal vein occlusion had accepted the extent of macular haemorrhage as the definition for the subgroup of people for whom grid laser photocoagulation is not a suitable treatment option. Noting the Royal College of Ophthalmologists' 2010 Interim Guidelines on Retinal Vein Occlusion in which grid laser photocoagulation is only recommended following absorption of the majority of haemorrhage (section 4.3), the Committee accepted the extent of macular haemorrhage as the definition of the group of people for whom grid laser photocoagulation is not an option. The Committee understood that, when grid laser photocoagulation is not an option, the comparator in this analysis would be dexamethasone. It considered the ERG's exploratory analysis in which the ICER was £4,100 per QALY gained for ranibizumab compared with dexamethasone. Consistent with its previous conclusions it recognised that this ICER would be subject to uncertainty because of the absence of a direct comparison between ranibizumab and dexamethasone and because of confounding in the BRAVO trial. However, the Committee remained satisfied that the most plausible ICER would be below £20,000 per QALY gained. The Committee therefore concluded that ranibizumab should be recommended as an option for treating visual impairment caused by macular oedema following BRVO when grid laser photocoagulation has not been beneficial or is not suitable because of the extent of macular haemorrhage.
The Committee discussed how innovative ranibizumab is in its potential to make a significant and substantial impact on health‑related benefits. It agreed that anti‑VEGF treatments were a substantial improvement over previous treatments, but considered that this improvement applied to the class of drugs, including bevacizumab. It stated that the innovation was in the scientific step forward, not the act of licensing. In addition the Committee was not aware of any substantial benefits of ranibizumab over its comparators that would not be already captured into the QALY estimation in the modelling.
The Committee considered whether there were any equalities considerations affecting population groups protected by equality legislation and concluded that there were no equality issues relating to this appraisal in the guidance.