Fluocinolone acetonide intravitreal implant for treating chronic diabetic macular oedema

The Committee considered the evidence presented by the manufacturer on the clinical effectiveness of fluocinolone acetonide intravitreal implant. The Committee noted that the main sources of evidence were the FAME A and B randomised controlled trials, which enrolled people with diabetic macular oedema and included a preplanned analysis defined by the median duration of diabetic macular oedema, which had been used to identify a subgroup of people with chronic diabetic macular oedema (that is, of more than 3 years' duration). It noted that in the FAME trials, patients were only treated in 1 eye and that the majority of patients were treated in their worse‑seeing eye. The Committee noted that fluocinolone acetonide intravitreal implant was associated with statistically significant gains in the proportion of patients with chronic diabetic macular oedema who had a more than 15‑letter increase in best corrected visual acuity (BCVA) from baseline compared with the sham injection group. It also noted that mean change in BCVA from baseline at month 36 of the FAME trials was statistically significantly greater in the fluocinolone acetonide intravitreal implant group compared with the sham injection group. The Committee concluded that fluocinolone acetonide intravitreal implant showed greater clinical effectiveness than sham injection in people with chronic diabetic macular oedema.

The Committee discussed the characteristics of the FAME trial population and how these related to the patient group specified in the marketing authorisation for fluocinolone acetonide intravitreal implant and to the group of patients likely to receive treatment in UK clinical practice. The Committee noted that the efficacy data submitted were based on a subgroup defined as having chronic disease with a duration of longer than 3 years, rather than as unresponsive to available therapies as specified in the marketing authorisation. The Committee further noted that some patients included received additional treatments during the trials such as laser photocoagulation and anti‑VEGF injections. The Committee discussed the concerns that this implied that in the FAME trials fluocinolone acetonide intravitreal implant was not being used as specified in the marketing authorisation. The Committee noted the manufacturer's submission and comments on the appraisal consultation document stating that the patients in the FAME trials had poor visual acuity, and that the patients in the comparator arm had disease that was responding poorly to the background therapies when given in the trial. However, the Committee remained concerned that the data from the FAME trials did not specifically reflect a group of patients who had disease that was insufficiently responsive to other therapies because people in the FAME trials did not necessarily have disease that was unresponsive to treatment with anti‑VEGF or laser photocoagulation before randomisation. The Committee considered that this could affect the levels of response observed in clinical practice because one of the mechanisms of action of corticosteroids was to act on VEGF. It noted comments received on the appraisal consultation document that studies of other intravitreal corticosteroids (such as triamcinolone) have shown a beneficial effect on macular oedema when used after anti‑VEGF treatments. The Committee remained concerned that the data from the trials may not be fully representative of the group of people who would be receiving fluocinolone acetonide intravitreal implant in UK clinical practice. However, it concluded that although there is uncertainty about the generalisability of the trial data to UK clinical practice, fluocinolone acetonide intravitreal implant is likely to be clinically effective in this population.

The Committee considered the manufacturer's approach to conducting an indirect comparison of fluocinolone acetonide intravitreal implant and laser photocoagulation using the FAME trials and the DRCR Protocol B study (2008). The Committee was aware of the Evidence Review Group's (ERG's) concerns over the value of the indirect comparison given the absence of a common comparator to link the FAME studies with the identified DRCR Protocol B study. It also noted that the retinopathy in patients included in the DRCR study was not as severe as in the FAME trials; that 40% of the DRCR population was laser‑naive; and that in the DRCR study there was no stipulation on duration of diabetic macular oedema at randomisation. The Committee concluded that the indirect comparison could not be interpreted with confidence and in any case was inappropriate to the scope for this appraisal, which considered fluocinolone acetonide intravitreal implant when diabetic macular oedema has been insufficiently responsive to available therapies.

The Committee considered the evidence for adverse events associated with fluocinolone acetonide intravitreal implant. It was aware that fluocinolone acetonide intravitreal implant was associated with the formation or progression of cataract. It noted that although there were similar rates of pre‑existing cataract between the 2 trial groups (77.9% and 77.0%) among people who were phakic (still had their natural lens) at baseline, cataract surgery was needed by a greater percentage (85.1% versus 36.4%) of these people in the fluocinolone acetonide intravitreal implant group than in the sham injection group. The Committee heard from clinical specialists that the majority of people with chronic diabetic macular oedema would be likely to develop cataracts at some stage but that fluocinolone acetonide intravitreal implant might accelerate this. The Committee also noted that administration of fluocinolone acetonide intravitreal implant was associated with increased intraocular pressure: 5.3% of people in the fluocinolone group needed intraocular pressure‑lowering surgery compared with 0% in the sham injection group. The Committee heard from clinical specialists that intraocular pressure‑lowering surgery was a particular concern and for this reason, in addition to the associated acceleration of cataract development, clinicians would be likely to use fluocinolone acetonide intravitreal implant conservatively and be reluctant to use it too early in the treatment pathway. The Committee concluded that it was appropriate to take account of these adverse events when considering the approach to economic modelling.

The Committee considered the manufacturer's economic model and sensitivity analyses, and discussed the key parameters used in it. The Committee noted the sensitivity of the model to the assumptions about the relationship between a person's treated eye and their overall visual acuity, the number of fluocinolone acetonide intravitreal implant treatments in the first 3 years, assumptions about the benefits of treatment after month 36, the rate of re‑treatment, and the source of health‑related quality of life values. The Committee concluded that the cost‑effectiveness estimates were most sensitive to the source of health‑related quality of life values, the assumption that a person's overall visual acuity related only to their treated eye, and the assumption that all treated eyes were better‑seeing eyes.

The Committee considered the ERG's critique of the manufacturer's original base‑case results. It was aware that the ERG had made a series of explorations using the manufacturer's original base‑case incremental cost‑effectiveness ratio (ICER) of £22,600 per quality‑adjusted life year (QALY) gained as a starting point. The Committee noted that by correcting an error relating to the application of annual discontinuation rates in the model, and by changing the formulae for averaging male and female mortality rates and for calculating the percentage of patients remaining on fluocinolone acetonide intravitreal implant beyond 9 years, the ERG's revisions increased the manufacturer's original base‑case ICER from £22,700 to £26,500 per QALY gained for fluocinolone acetonide intravitreal implant compared with optimised standard of care without the patient access scheme. The Committee agreed that the ERG's initial error corrections to the model were appropriate.

The Committee went on to consider the ERG's further explorations to the original model, which were included in the manufacturer's revised model submitted for the rapid review:

• aligning assumptions about the rate of laser administrations per patient in the optimised standard of care group for year 1 of the model with the trial data

• applying revised unit costs for some adverse events

• revising the assumptions about the extrapolation of benefits beyond 3 years in the model

• applying an adjusted bilateral treatment rate in the fluocinolone acetonide intravitreal implant group of 85.2% of the treatment rate in the optimised standard of care group

• use of a 10‑letters response criterion

• applying the absolute cost and quality of life uplift associated with bilateral treatment in the optimised standard of care group to the patients in the fluocinolone group needing bilateral treatment but for whom it was contraindicated

• applying the cost of blindness when only the treated eye fell below a BCVA of 35 letters rather than both eyes.
  
  The Committee concluded that the above amendments to the assumptions in the economic model submitted by the manufacturer for the rapid review were reasonable.

The Committee discussed the assumption in the model that 20% of patients were treated in their best‑seeing eye, 40% in their worse‑seeing eye and 40% of patients were treated in both eyes. The Committee noted the manufacturer's comment on the appraisal consultation document that in the FAME trials the majority of the patients had their worse‑seeing eye treated because of ethical considerations but that in clinical practice, more patients would be treated in their best‑seeing eye because this is more likely to protect functional vision. The Committee agreed that this could be the case but considered that currently there is a lack of evidence. It concluded that the assumption of 20% of the patients being treated in their best‑seeing eye, 40% in their worse‑seeing eye and 40% of patients receiving bilateral treatment was reasonable.

The Committee discussed the use of second and subsequent fluocinolone acetonide intravitreal implants and the assumption in the model of re‑treatment only after 36 months. It noted that the summary of product characteristics states that an additional implant may be administered after 12 months if the patient experiences decreased vision or an increase in retinal thickness secondary to recurrent or worsening diabetic macular oedema. The Committee heard from the manufacturer that this wording was included in the summary of product characteristics because patients in the FAME trials could receive a further implant after 12 months. However, subsequent research showed that the effect of the corticosteroid should last 36 months. The Committee understood from the manufacturer that it does not promote having the implants more frequently than once every 36 months and that there are currently no data available demonstrating further benefit from reimplantation. The Committee accepted the assumption in the model of no more than 1 implant every 36 months.

The Committee discussed the fact that the model did not apply quality of life detriments to adverse events such as cataracts, glaucoma or raised intraocular pressure. The Committee noted the manufacturer's rationale that the impact of adverse events such as cataract formation on visual acuity was incorporated in the overall utility measures, that the application of these decrements would be short in duration and therefore have a minimal impact on the ICERs, and that the decrements associated with surgery would depend on when a patient chose to have surgery. Nevertheless, the Committee concluded that if disutilities associated with operations, procedures and hospital attendances (such as cataract removal, glaucoma surgery, retinal detachment surgery, vitrectomy and treatment of endophthalmitis) had been taken into account the ICERs would increase.

The Committee discussed the utility values for the better‑seeing eye used in the economic model. The Committee noted that the manufacturer's original analyses had been based on Brown et al. (2000) utilities. It further noted that the ERG's explorations of the manufacturer's original model retained the Brown et al. (2000) utility values, and also explored the effect of applying utility values from Brown et al. (1999) and from a regression analysis in the manufacturer's submission for NICE technology appraisal guidance 237 on ranibizumab for the treatment of diabetic macular oedema (replaced by technology appraisal guidance 274 on ranibizumab for treating diabetic macular oedema). The Committee agreed that the utility values may vary in their appropriateness in being applied to people with chronic diabetic macular oedema. It accepted that the utility values from the manufacturer's submission for NICE technology appraisal guidance 237 were relatively insensitive to changes in visual acuity and therefore ICERs using these utilities may be numerically too high.

The Committee discussed further the relative merits of Brown et al. (1999) and Brown et al. (2000). It was aware that the Brown et al. (1999) utility values were based on a larger population with a mix of eye diseases whereas the Brown et al. (2000) study was based on a smaller population with age‑related macular oedema (but not diabetic macular oedema). It noted that the data from Brown et al. (2000) included a large drop in utility values between some health states, which may have arisen from the smaller patient numbers in each health state. It also noted that utility values from Brown et al. (2000) for the best health states were higher than would be expected for people with chronic diabetic macular oedema, who are usually older people with comorbidities and this, in turn, increased the uncertainty around the validity of this source of utility values. The Committee heard from the ERG that Brown et al. (2002) found no difference in the reduction in the quality of life associated with similar levels of visual acuity loss depending on the causes of vision impairment. The Committee considered the manufacturer's rationale for using Brown et al. (2000) values based on the higher proportion of patients with oedema compared with Brown et al. (1999). It also noted that the manufacturer restated in its comment on the appraisal consultation document that Brown et al. (2000) represents a more accurate source of utility values for people with diabetic macular oedema. However, the Committee was not persuaded that this outweighed the benefits of the larger sample size and the associated finer delineation of utilities possible with larger patient numbers. The Committee concluded that there were limitations to using the available utility values to model the health‑related quality of life of the group of people with chronic diabetic macular oedema. However, of the available source of utility values, Brown et al. (1999) had advantages because of the size of the sample included in the study. Because there were some uncertainties about the most appropriate source of utility values for this patient population the Committee agreed to consider a range of ICERs based on both Brown et al. (1999) and Brown et al. (2000).

The Committee noted that the manufacturer had also included analyses using utility values from Czoski‑Murray et al. (2009) and Heintz et al. (2012). The Committee discussed the appropriateness of using the values from Heintz et al. (2012) and considered that the very slight differences between utilities for the loss of better‑seeing eye vision relative to worse‑seeing eye vision lacked face validity. The Committee then discussed the utility values presented in Czoski‑Murray (2009). It understood that the source of utility values had been considered for the appraisal of ranibizumab for the treatment of diabetic macular oedema (NICE technology appraisal guidance 237, replaced by NICE's technology appraisal guidance 274 on ranibizumab for treating diabetic macular oedema) and noted that using these utility values resulted in ICERs in between the ones obtained using Brown et al. (1999) and Brown et al. (2000). The Committee agreed to consider Czoski‑Murray et al. (2009) as another source of utility values for this appraisal.

The Committee discussed the most appropriate adjustments needed to the better‑seeing eye utility values in the model when the worse‑seeing eye or both eyes were treated. The Committee considered the 6 scenario analyses carried out by the ERG (see section 1.38) which varied the health‑related quality of life impact of changes in the vision of the worse‑seeing eye and the resultant QALY gain associated with treatment of the worse‑seeing eye or both eyes. The Committee understood the concerns of the ERG about using the Heintz et al. (2012) data to calculate the amount of gain from treating the worse‑seeing eye, because of the small numbers of patients in some of the visual acuity levels and also because the manufacturer's calculation did not account for visual acuity in the better‑seeing eye being correlated with visual acuity in the worse‑seeing eye. The Committee also discussed the utilities for worse‑seeing eyes in Brown et al. (1999) and heard from the ERG about an additional study, Sahel et al. (2007), which also showed that the visual acuity in the worse‑seeing eye has little impact on health‑related quality of life unless it is severely affected. The Committee was aware that there could be psychological benefits from treating the worse‑seeing eye that had not been captured in the calculation of the QALY. However, the Committee noted that the FAME trials collected data (with a majority of worse‑seeing eyes) on the effect of visual impairment on quality of life using the disease‑specific VFQ‑25 questionnaire, and that this had shown no difference between the groups treated with fluocinolone acetonide intravitreal implant and those treated with sham injection. On balance, the Committee considered scenario analysis 3 (which assumed that changes in vision for people treated in their worse‑seeing eye had 30% of the health‑related quality of life impact of the same change in vision from treating their better‑seeing eye) to be the most appropriate for decision‑making. The Committee was not persuaded that 30% was an unreasonable reflection of the clinical situation for people with chronic diabetic macular oedema and noted that this was also consistent with previous appraisals.

The Committee then considered the cost‑effectiveness results based on data from the FAME trials for all patients with chronic diabetic macular oedema taking into account its concerns (sections 2.9 to 2.18). The Committee noted that the most plausible ICER for fluocinolone acetonide intravitreal implant compared with optimised standard of care in the original guidance was at least £47,600 using the utilities from Brown et al. (2000) and £80,000 using the utilities from Brown et al. (1999) per QALY gained. The Committee noted that with the patient access scheme the ICERs were reduced to £37,600 using the utilities from Brown et al. (2000), £42,700 using the utilities from Czoski‑Murray et al. (2009) and £63,500 using the utilities from Brown et al. (1999) per QALY gained. The Committee considered that the lowest of these estimates remained over £30,000 per QALY gained and therefore outside the range normally considered cost effective (£20,000 to £30,000 per QALY gained). It also noted that there was substantial uncertainty in these estimates, particularly in the extent to which the results of the clinical trial data included in the analyses could be applied to the population of people who would receive treatment in clinical practice. The Committee concluded that fluocinolone acetonide intravitreal implant could not be recommended as a cost‑effective use of NHS resources for treating people with chronic diabetic macular oedema that is insufficiently responsive to available therapies.

The Committee discussed the subgroup of people with chronic diabetic macular oedema who had treatment in an eye with a pseudophakic lens. The Committee acknowledged that this is an identifiable subgroup of people with chronic diabetic macular oedema. The Committee noted that the numbers of patients in the subgroup were approximately half the chronic diabetic macular oedema population in the FAME trials and that the clinical effectiveness in terms of 15‑letter gain in visual acuity in this subgroup was numerically worse than that in the total population (albeit with wide confidence intervals). The Committee considered that the comparatively small numbers of such patients in the FAME trials led to uncertainty in the estimates of clinical effectiveness for this group and thus in the estimates from the economic modelling but accepted that the subgroup proposed was reasonable. The Committee also considered that its concerns about the extent to which the data for the subgroup were representative of the group of people who would receive fluocinolone acetonide intravitreal implants in UK clinical practice remained valid (section 2.6). However, on balance, the Committee concluded that fluocinolone acetonide intravitreal implant had been shown to be clinically effective in this subgroup of people.

The Committee considered the cost effectiveness of fluocinolone acetonide intravitreal implant in the pseudophakic subgroup. It noted that the most plausible ICER for this subgroup in NICE's technology appraisal guidance on fluocinolone acetonide intravitreal implant for treating chronic diabetic macular oedema after an inadequate response to prior therapy was between £29,700 per QALY gained using the utilities from Brown et al. (2000) and £50,600 using the utilities from Brown et al. (1999). The Committee noted that with the patient access scheme, the ICERs presented by the ERG were £30,000 per QALY gained using the utilities from Brown et al. (1999), £21,000 per QALY gained using the utilities from Czoski‑Murray et al. (2009) and £17,500 per QALY gained using the utilities from Brown et al. (2000). The Committee considered that the most plausible estimates of cost effectiveness would be in the upper end of this range, and that there was significant uncertainty around this estimate. The Committee was persuaded that the technology had been shown to meet a clinical need in people whose disease is unresponsive to available therapies. On balance, the Committee concluded that fluocinolone acetonide intravitreal implant could be a cost‑effective use of NHS resources and recommended it as an option for people with chronic diabetic macular oedema that is insufficiently responsive to available therapies and if the implant is to be used in an eye with an intraocular (pseudophakic) lens.