Dexamethasone intravitreal implant for the treatment of macular oedema secondary to retinal vein occlusion

Clinical effectiveness

4.7 The Committee considered the potential of dexamethasone to offer additional health-related benefits compared with currently available treatment options. It heard from the patient experts about the impact of dexamethasone on their quality of life. Patients advised that after the administration of dexamethasone their sight improved and they were able to resume normal daily activities.

4.8 The Committee considered the evidence for the efficacy of dexamethasone intravitreal implant compared with best supportive care from the GENEVA trials. It noted that the primary outcome was the percentage of patients with an improvement of BCVA of 15 letters or more, which represented a gain of three lines on the EDTRS chart (this enables patients to see letters half the height of those they could see before). The Committee heard from clinical specialists that this represented the gold standard for assessing the effect of treatment on visual acuity, although a gain of 10 letters is also considered to be clinically significant.

4.9 The Committee considered the results from the trials. It first considered the pooled primary outcome data from the GENEVA trials for the entire population with macular oedema following RVO. The Committee noted that dexamethasone is associated with a statistically significant improvement in visual acuity (based on percentage of patients with an improvement of BCVA of 15 letters or more) compared with sham treatment at day 30, 60 and 90, but there was no statistically significant improvement at day 180. The Committee also noted that dexamethasone was associated with a statistically significant improvement in mean change in BCVA at day 30, 60, 90 and 180 for all people with macular oedema following RVO. The Committee also considered the pooled primary outcome data from the GENEVA trials for the CRVO and BRVO subgroups. It noted that dexamethasone was associated with a statistically significantly higher proportion of people gaining BCVA of 15 letters or more compared with sham treatment for CRVO (day 30 and 60), BRVO (day 30, 60 and 90), BRVO with macular haemorrhage (day 30, 60 and 90) and at all time points for BRVO with previous laser treatment. These results for the subgroups were similar to those obtained for the combined population with RVO. The Committee concluded that treatment with dexamethasone is clinically effective when compared with best supportive care.

4.10 The Committee considered the evidence for adverse events associated with dexamethasone. These included cataracts, raised intraocular pressure and infection. The Committee noted that evidence was limited by trial duration to adverse events after two treatments with data collected up to 360 days. The Committee discussed the impact of dexamethasone in causing cataracts and the potential issues for people with diabetes, who have a higher risk of developing glaucoma and cataracts. The Committee heard from the clinical specialist that a similar infection risk to that of its comparators would be expected. It also heard that the incidence of raised intraocular pressure was 25% with dexamethasone but that this is usually well managed with eye drops and that 1% of people required laser or surgical procedures for management of elevated intraocular pressure. A higher percentage was likely to need surgery for cataracts according to the number of dexamethasone injections given. The Committee concluded that there were some concerns about the long term safety of dexamethasone treatment because the marketing authorisation is based on a evidence base trial with two re-treatments over 360 days and the manufacturer assumed that up to six treatments would be given and there are limited data on long-term treatment and multiple re-treatment (see section 3.6).

4.11 The Committee considered the evidence for re-treatment from the open-label extension of the GENEVA trials (which was submitted as academic-in-confidence information, as described in section 3.6, and is therefore not presented here). The Committee noted that for the whole population the proportion of patients with an improvement in BCVA of at least 15 letters was higher in the group that received dexamethasone at day 0 and day 180 compared with those who received sham at day 0 and dexamethasone at day 180. The Committee also considered the expected frequency of treatment with dexamethasone at day 180. The Committee heard from the clinical specialist that the criteria for re-treatment were based on the patient's experience with previous treatment (whether their vision had initially improved with treatment but had started to deteriorate), deterioration in visual acuity as assessed by BCVA and the persistence of macular oedema. The Committee was further advised that it is difficult to use a cut-off value for visual acuity that would indicate re-treatment, but that with a loss of five letters it would be appropriate to consider re-treatment. The Committee agreed with the clinical specialist that although the safety data relate to 6-monthly treatment, it is expected that clinicians may re-treat at 4 months in clinical practice, but may not treat more frequently because of the risk of adverse events from the accumulation of dexamethasone in the eye.

4.12 The Committee then considered the clinical effectiveness of dexamethasone compared with bevacizumab. It heard from the clinical specialist that bevacizumab is generally considered by ophthalmologists to be efficacious and safe, although there is some uncertainty as to the optimal dosing schedule, relative effectiveness versus dexamethasone and frequency of treatments. The Committee considered the results of the mixed-treatment comparison provided by the manufacturer in response to requests made at the first Appraisal Committee meeting. It noted that the difference in BCVA was marginally favourable for dexamethasone over bevacizumab at day 60, but that at day 180 the difference in BCVA was marginally favourable for bevacizumab over dexamethasone. However, at both time points the confidence intervals were wide and therefore little certainty could be placed in either of the findings. The Committee noted the ERG's misgivings about the lack of completeness of the mixed-treatment comparison and also that the two cost–utility analyses provided by the manufacturer favoured bevacizumab in terms of efficacy. Nevertheless the Committee considered that a clinical efficacy advantage for dexamethasone over bevacizumab could not be concluded with certainty based on current evidence.

4.13 The Committee considered the mode of delivery and adverse events associated with dexamethasone and bevacizumab, including infection and needle phobia. The clinical specialist informed the Committee that dexamethasone was administered with a larger needle than existing treatments and that dexamethasone was associated with increased intraocular pressure and cataracts. However, more injections of bevacizumab were needed and this presented a greater infection risk. The clinical specialist also informed the Committee that with bevacizumab, which had no agreed protocol for use, sterile endophthalmitis had been reported, but was rare. The Committee concluded that the adverse events associated with dexamethasone were expected to be more severe and more difficult to treat than those associated with bevacizumab.

Cost effectiveness

4.14 The Committee considered the cost effectiveness of dexamethasone relative to best supportive care. It noted the manufacturer's revised base-case estimate of the ICER for all people with RVO of £17,600 per QALY gained. It understood that this estimate incorporated an assumption that 90% of people would be treated for macular oedema in their 'worst-seeing eye'. It heard from clinical specialists that in practice this proportion would be between 90% and 97%. It further heard from clinical specialists that the manufacturer's estimate of the percentage (6.5%) of people who require treatment for the other eye (fellow eye involvement) was plausible. Patient experts had highlighted the importance of treating the first eye affected, even though overall acuity depends mostly on the 'better-seeing' eye, because RVO and other eye conditions may later affect the critical second eye. The Committee concluded that in this instance it was appropriate to treat the first eye affected.

4.15 The Committee considered the manufacturer's response to requests made at the first Appraisal Committee meeting. It understood that the revised base-case estimate of the ICER incorporated revised assumptions for the setting for the administration of dexamethasone and the cost of care for people with blindness. It further understood that the manufacturer had provided scenario analyses to demonstrate the effect of different re-treatment rates. The Committee noted that the revised base-case ICER did not incorporate the Committee's request that the extrapolation of data for the observation arm should be based on those trial data between 0 and 6 months rather than 3–6 months. The Committee discussed each of these assumptions, and the results of the sensitivity analyses for re-treatment rates, in turn.

4.16 The Committee considered the setting for the administration of dexamethasone. It noted that the revised base case was based on an assumption of 75% of procedures being performed in a day-case unit and 25% in an outpatient clinic, instead of 100% in an outpatient clinic as requested by the Committee at the first Appraisal Committee meeting. The Committee noted comments from consultees that it was reasonable to expect a higher proportion to be performed in a day-case unit while ophthalmologists are gaining familiarity with dexamethasone administration, but that in time it would be expected that most would be performed in an outpatient clinic. The Committee therefore accepted the proportions used by the manufacturer in the revised base case for the setting for the administration of dexamethasone.

4.17 The Committee considered the revised assumption for the cost associated with blindness in the revised base case. It understood that the manufacturer had accepted the rationale for applying this cost to only those people who have both eyes in the worst health state. However, the Committee noted that in the revised base-case estimate, the cost associated with blindness was reduced as a proxy for adjusting the proportion of people who fall into this health state of severe visual impairment in both eyes. The manufacturer adjusted the cost associated with blindness to 25% of the full cost (that is, £1490 instead of £5963), and increased this every 6 months by 10% of the reduced cost (that is, an increase of £149 for each cycle). The Committee accepted this to be a reasonable reduction and concluded that although it would have been preferable to have adjusted the assumptions around the number of people with severe visual impairment in both eyes, the manufacturer's adjustment went some way towards meeting the Committee's request and was considered acceptable.

4.18 The Committee considered the extrapolation of health-state data beyond the trial for the observation arm. It noted the manufacturer's assertion that it was inappropriate to include 0- to 3-month data for BRVO because spontaneous resolution occurs in approximately a quarter of people during that period. The Committee also heard from the manufacturer that although it may be appropriate to include 0- to 3-month data for CRVO, the ERG's exploratory analysis used aggregate data rather than transition probabilities based on individual patient data. The manufacturer stated that this had led to an overestimation of the ICER for CRVO by the ERG. The Committee accepted that spontaneous resolution occurs in BRVO and therefore it was not appropriate to include the first 3 months of data for this subgroup. It further concluded that using individual patient data for the first 6 months may have been relevant for CRVO, but that the increased ICER for CRVO from the ERG's analysis (from £16,500 to £25,300 per QALY gained including a decrease from 0.29 to 0.22 of net total QALYs and an increase in total cost from £4732 to £5469) may be an overestimate. The Committee concluded that 0- to 3-month data for the observation arm of the GENEVA trials for people with CRVO would not have a substantial impact on the overall ICER and therefore accepted the manufacturer's use of 0- to 3-month data in the revised base case.

4.19 The Committee considered the sensitivity analyses around re‑treatment rates provided by the manufacturer in response to requests made by the Committee at the first meeting. The Committee understood that the manufacturer had combined the revised assumptions relating to the setting for administering dexamethasone (see section 4.16) and the costs associated with blindness (see section 4.17) with three scenario analyses for different re-treatment rates. The Committee noted that the re‑treatment scenarios covered the extreme positions of both the base-case estimates (in which re-treatment rates were based on clinical opinion) and those re-treated at day 180 in the GENEVA trials. The Committee noted that for all people with RVO, the ICER varied between £17,600 (including incremental costs of £3698 and incremental QALYs of 0.40) and £34,700 (including incremental costs of £8041 and incremental QALYs of 0.44) per QALY gained for these scenarios respectively. The Committee further noted the manufacturer's scenario analysis in which the re-treatment rates were at the mid-points between the two extremes. With this assumption the ICER for all people with RVO was £26,300 per QALY gained. The Committee accepted the assumption of mid‑point re-treatment rates.

4.20 On the basis of previous discussions (see sections 4.17 to 4.21), the Committee accepted the revised manufacturer's base-case estimate using the mid-point re-treatment rates from the manufacturer's scenario analysis as the most appropriate estimate of the ICER for its consideration. It therefore concluded that the decision regarding the cost effectiveness of dexamethasone compared with best supportive care should be based on the manufacturer's ICER of £26,300 per QALY gained for all people with RVO. The Committee further concluded that this represented an acceptable level of cost effectiveness in this case and that dexamethasone intravitreal implant for the treatment of RVO represents a cost-effective use of NHS resources when compared with best supportive care.

4.21 The Committee considered the cost effectiveness of dexamethasone relative to bevacizumab. It understood that the manufacturer had presented a cost-minimisation approach for CRVO and that the cost assumptions from this approach had been used in both cost–utility analyses (that is, the mixed-treatment comparison, and the indirect comparison using a different anti-VEGF treatment as a proxy for bevacizumab) (see sections 3.14 and 3.15). The Committee therefore considered that its discussion on the relative cost effectiveness of dexamethasone compared with bevacizumab should focus on the assumptions within the manufacturer's cost-minimisation analysis.

4.22 The Committee considered the assumptions in the manufacturer's cost-minimisation analysis between dexamethasone and bevacizumab. It noted comments from the ERG that the cost of bevacizumab had been modelled on the higher of the two estimates from the two suppliers of bevacizumab for eye treatment (that is, £150 per vial rather than £50 per vial). The ERG further highlighted that the manufacturer's estimate of the average number of doses of bevacizumab was 13.75 in the first 2 years, but that the Pan American Collaborative Retina Study Group had found an average of 7 doses over this time period. The Committee heard from the manufacturer that this study was a small cohort follow-up study and may not be representative of UK clinical practice. In addition the manufacturer highlighted to the Committee that the recently reported HORIZON study found that 3.8 injections per year of ranibizumab were insufficient to maintain the clinical effects seen in the first year (ranibizumab, another anti-VEGF, was given monthly in the CRUISE study). Therefore the manufacturer considered 7 doses of bevacizumab over 2 years to be lower than UK clinical practice. The Committee understood that the cost-minimisation analysis also assumed that the setting for the administration of both bevacizumab and dexamethasone would be the same (that is, 75% of procedures would be carried out in a day-case unit and 25% would be carried out in an outpatient clinic). The Committee noted the ERG's view that all anti-VEGF injections are currently administered in an outpatient clinic. The Committee concluded, on the one hand, that the cost of bevacizumab may have been overestimated, and on the other, that there was uncertainty around the number of injections of bevacizumab. The Committee concluded that the assumptions used in the manufacturer's cost-minimisation analysis between dexamethasone and bevacizumab may overestimate the cost of bevacizumab, but considered the cost may not be as low as suggested by the ERG.

4.23 The Committee considered the results of the cost-minimisation analysis. It noted that the manufacturer estimated the cost of dexamethasone treatment to be approximately £4500 less per patient than bevacizumab, but that with the ERG's amendments, dexamethasone could cost up to £2100 more per patient than bevacizumab. On the basis of previous discussion (see section 4.22) the Committee accepted that there was some uncertainty around the assumptions used to calculate the cost of bevacizumab and concluded that it was as likely that dexamethasone might cost more than bevacizumab as it was that dexamethasone might cost less. The Committee therefore concluded that a cost advantage of dexamethasone compared with bevacizumab had not been conclusively demonstrated.

4.24 In summary, the Committee considered the costs and clinical effectiveness of dexamethasone. It considered that comparisons with best supportive care and bevacizumab were both relevant to the NHS. It acknowledged that an incremental analysis in which dexamethasone, bevacizumab and best supportive care were simultaneously assessed was not available. The Committee recognised the difficulties within the evidence base for bevacizumab and commended the manufacturer's attempts to provide a comparison of the relative clinical and cost effectiveness of bevacizumab and dexamethasone. It considered that a clinical efficacy advantage for dexamethasone over bevacizumab could not be concluded with certainty based on presently available evidence and that the adverse events associated with dexamethasone were expected to be more severe and more difficult to treat than those associated with bevacizumab. It also accepted that there was uncertainty around the assumptions used in the cost minimisation comparing dexamethasone with bevacizumab. Based on the manufacturer's revised ICER of £26,300 per QALY gained compared with best supportive care for all people with RVO, the Committee concluded that dexamethasone intravitreal implant represents a cost-effective use of NHS resources and should be offered as an option for the treatment of macular oedema secondary to RVO.

4.25 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 that required addressing in the guidance.

Summary of Appraisal Committee's key conclusions