Crizanlizumab for preventing sickle cell crises in sickle cell disease

3 Committee discussion

The appraisal committee considered evidence submitted by Novartis, a review of this submission by the evidence review group (ERG), NICE's technical report, and responses from stakeholders. See the committee papers for full details of the evidence.

The appraisal committee was aware that several issues were resolved during the technical engagement stage, and agreed that:

People are unlikely to have crizanlizumab alongside regular blood transfusions to prevent recurrent VOCs (issue 5, technical report page 7).

The committee discussed the following issues (issues 2, 4, 5, 6 and 7), which were outstanding after the technical engagement stage.

New treatment option

People with sickle cell disease who have recurrent VOCs would welcome a new treatment option

Vaso-occlusive crises happen when sickle-shaped red blood cells block blood vessels (vaso-occlusion) in different parts of the body. It means not enough oxygen is delivered to tissues and organs, causing ischaemic injuries and excruciating pain. If someone has 2 or more crises a year, they are said to have recurrent VOCs. The patient experts explained that, while they have learnt to avoid common trigger factors, VOCs are unpredictable in terms of when they happen and how severe they are. Recovery can take days to weeks, depending on the cause of the crises. The patient experts described how this unpredictability can be emotionally distressing, and how it can suddenly prevent them from being able to work or do other planned activities. They explained that severe episodes can often require treatment in hospital, and the build-up of complications over time and resulting organ damage significantly affects their quality of life. Treatments to prevent VOCs in adults include hydroxycarbamide or regular blood transfusions. One patient expert explained that, although hydroxycarbamide had effectively reduced the severity of their crises, they stopped treatment because of the potential toxicity associated with its long-term use. The clinical expert explained that because hydroxycarbamide is a chemotherapy drug, people often prefer not to take it because they are concerned about serious side effects. Hydroxycarbamide may also affect male fertility and prenatal development during pregnancy. The patient experts explained that there have been no new treatments for sickle cell disease for many years. There is an unmet need for an effective and well-tolerated treatment that can be taken over a lifetime to reduce VOCs. The patient experts explained that fewer VOCs would mean fewer GP visits and hospital admissions, a reduced risk of organ damage, improved mental health and less time off work. The committee concluded that people with sickle cell disease would welcome a new treatment that reduces the frequency of VOCs and improves their quality of life.

Comparators

The relevant comparators are hydroxycarbamide and regular blood transfusions

Standard treatment to prevent VOC is generally best supportive care (for example avoiding trigger factors and maintaining general physical and psychological health) with or without hydroxycarbamide (also known as hydroxyurea). Regular blood transfusions may be considered for a small number of people for whom hydroxycarbamide is inappropriate. The company assumed that people would not have treatment with crizanlizumab alongside regular blood transfusions to prevent VOC. The committee noted that, because crizanlizumab had no effect on haemoglobin or measures of haemolysis in the SUSTAIN trial (the main clinical trial; see section 3.3), combining it with blood transfusions could potentially benefit people with sickle cell disease. The clinical expert highlighted that they were not aware of any data to support combined use. They explained that, because regular blood transfusions substantially reduce the number of sickled red blood cells, this reduces the need for crizanlizumab because of its mechanism of action. The clinical expert described how people can have adverse effects from blood transfusions, such as iron overload, and that clinicians prefer not to combine treatments that may further increase the risk of adverse events. The committee concluded that hydroxycarbamide and regular blood transfusions are the only relevant comparators. It agreed that, although there is a lack of evidence, it was unlikely that people would have crizanlizumab alongside regular blood transfusions to prevent VOCs.

Clinical effectiveness evidence

People on crizanlizumab have significantly fewer sickle cell-related pain crises than people on placebo

The clinical evidence came from SUSTAIN, a double-blind, randomised multicentre trial of crizanlizumab compared with placebo. SUSTAIN treatment centres were in the US, Brazil and Jamaica. The trial had a 52‑week follow up, during which treatment was given. Use of hydroxycarbamide alongside crizanlizumab was permitted in both arms of the trial, but patients having regular blood transfusions were excluded. The primary outcome for SUSTAIN was the annual rate of sickle cell-related pain crises. These were defined as acute episodes of pain caused by a VOC, which resulted in a visit to a medical facility and treatment with pain relief medication. The median annual rate of sickle cell-related pain crises was significantly lower for the licensed dose of crizanlizumab (1.63) than for placebo (2.98; p=0.01). Overall and serious adverse event incidence was comparable across arms.

The company's model

The model structure is not appropriate because the baseline health state occupancy does not reflect that recorded in SUSTAIN and transition probabilities cannot be derived

The eligibility criteria in the SUSTAIN trial included 2 to 10 VOCs in the previous 12 months, and patients were randomised based on VOC rate (2 to 4 or 5 to 10) and by concomitant hydroxycarbamide use (yes or no). The company's Markov model included 3 main health states: no VOCs, 1 or 2 VOCs, 3 or more VOCs, and a death health state. But the committee understood that, because of the trial's eligibility criteria, SUSTAIN could only provide information about people who had 2 or more VOCs at baseline. The ERG considered that the baseline health state occupancy in the model should reflect the patient population in the trial. The company's model structure meant that transition probabilities cannot be accurately estimated because it is not known how patients with less than 2 VOCs would transition after the first year in the model, given the 1‑year cycle length and duration of the trial. The company explained that, because it had trial results at 1 year, it wanted to show that there was a gradual change over that year. It did this by applying a half-cycle correction to the 1‑year cycle length. This meant the distribution of patients at baseline did not match SUSTAIN and the committee discussed that a monthly-cycle length may have been more appropriate. Patients were randomly redistributed to each health state at the end of each cycle, but in the same proportions observed in SUSTAIN so the overall distribution remained the same. The committee considered whether this random allocation of patients may break the link between crizanlizumab and its treatment effect. It also considered whether it was plausible that someone would have entirely random fluctuations in annual VOC frequency over time. The committee discussed how, if VOC frequency is not completely random, it's likely that the proportion of patients in the high VOC health state would get smaller over time. This is because it would be expected that the risk of death increases with the number of VOCs. The clinical expert explained that because VOCs are unpredictable it is common for people with moderate disease (around 20% to 30%) to fluctuate between a year when they have one hospital admission and a year when they have multiple admissions. However, the committee heard that, with an effective treatment, the clinical expert would expect a trend towards occupying the lower frequency VOC health states beyond the random fluctuations of crises. They explained that the way in which VOC states are classified in the SUSTAIN trial does not reflect clinical practice. The committee concluded that the company's Markov model structure was not appropriate because the baseline health state occupancy and transition probabilities for all the possible health states cannot be derived. It agreed that the baseline health state occupancy should reflect the population with recurrent VOC outlined in crizanlizumab's marketing authorisation.

Patient weight from SUSTAIN should be used in the model

3.1 In the economic model, SUSTAIN trial data was used to inform the treatment effect of crizanlizumab on the frequency of VOC in patients with recurrent VOC. Because of the short duration of the SUSTAIN trial, the Hospital Episodes Statistics (HES) database was used to estimate the risk of acute sickle cell disease-related complications or death associated with varying frequency of VOC. In the company's base case, baseline patient characteristics used in the model for age (mean 37.1) and gender distribution (63% female) were taken from the HES database. Body weights (55 kg for females, 65 kg for males) from NICE's guideline on sickle cell disease, which were based on expert opinion, were used to calculate an average patient body weight of 58.7 kg (adjusted for the proportion of females from the HES data). The company considered that using these data sources was consistent with the natural history data in the model and reflected the characteristics of people with sickle cell disease in the UK. The ERG noted that the HES database and the NICE guideline were representative of all UK patients, not just those who had recurrent VOC. Therefore, the data may include patients who would not have crizanlizumab in clinical practice because they are not eligible according to the marketing authorisation. The ERG considered that patient characteristics should come from SUSTAIN, the main source of treatment efficacy. The committee discussed whether people would have crizanlizumab at a younger age than in the SUSTAIN trial, because this would make them less likely to develop complications later on in life if the treatment was effective. The clinical expert explained that sickle cell disease is an inherited condition that is often symptomatic from childhood. Intervening earlier would prevent longer-term complications and organ damage, and minimise its psychological and social effects. The committee noted that the marketing authorisation excluded people under 16 so recommendations for this group were outside its remit. The clinical expert suggested that the ratio of females to males in people with recurrent VOC is likely to be around 50:50 in clinical practice. They explained that children with sickle cell disease tend to have a lower weight because of their chronic ill health and increased metabolic rate. The clinical expert suggested that people with a higher frequency of VOC might be slightly underweight because of their ill health and more frequent hospital admissions. Additional analyses from US data provided by the company at technical engagement indicated that a small proportion of patients with recurrent VOCs (n=729) had an average body weight of 73.6 kg, which was higher than the company's base case. The committee noted that this was lower than the average body weight for the total population of people with sickle cell disease of 84.2 kg (n=11,788). The committee acknowledged that prolonged ill health may affect patient weight, particularly during VOCs, but agreed that it would still expect people's body weight to be greater than in the company's base case. It understood that a higher average patient body weight (60 kg to 80 kg) would increase the dose and number of vials of crizanlizumab needed and subsequent drug wastage. The committee noted that in the HES database most patients had very few or no VOCs per year and the gender distribution was more evenly split in the subgroup of patients with recurrent VOCs than in the company's model. It discussed how in SUSTAIN the mean patient age was lower, gender distribution ratio was closer to 50%, and body weight was higher than in the company's base case. The committee agreed that using patient characteristics from the SUSTAIN trial would maintain the internal validity of the trial results. It noted that the company considered the trial population to be generalisable to those expected to have crizanlizumab in the NHS. The committee concluded therefore that patient weight from SUSTAIN should be used in the model. It agreed with the ERG's preference to use the age and gender distribution from the HES database to maintain the link between age and gender mix, and complications and mortality estimated from HES data. The committee agreed however that the inputs from the HES database may not reflect the population who would receive crizanlizumab in clinical practice. The committee understood that using patient characteristics from SUSTAIN significantly increased the company's base-case incremental cost-effectiveness ratio (ICER).

Hydroxycarbamide use from the SUSTAIN trial should be used in the model

3.2 In the company's base case, hydroxycarbamide use in the crizanlizumab and standard care arms was assumed to be 14.2%, based on the UK National Haemoglobinopathy Registry (NHR) annual report 2018 to 2019. The ERG noted that the NHR includes all patients with sickle cell disease and that hydroxycarbamide use is likely to be higher in people with recurrent VOCs. It considered that the higher use of hydroxycarbamide from SUSTAIN should be used in the model. The clinical expert explained that the SUSTAIN trial included sites in the US, where hydroxycarbamide use is slightly higher than the UK. The committee discussed how it would be expected that all people with sickle cell disease would have been offered or had hydroxycarbamide for at least 6 months before being considered for crizanlizumab. This is line with guidance from the British Society for Haematology. It is also in line with the company's positioning of crizanlizumab as either an add on to hydroxycarbamide, if it does not adequately reduce VOCs, or as a single treatment if hydroxycarbamide is inappropriate. The clinical expert said that, in people with homozygous sickle cell disease, hydroxycarbamide use is around 30% in people with recurrent VOCs. The committee heard that the clinical expert might expect approximately 50% of people taking hydroxycarbamide to still be on first-line treatment after 6 months. The committee concluded that hydroxycarbamide use from SUSTAIN should be used in the model, but acknowledged that there is uncertainty around the proportion of people who would have concomitant hydroxycarbamide in England. It understood that the proportion of hydroxycarbamide use from the SUSTAIN trial significantly increased the company's base-case ICER.

There is limited evidence of a prolonged treatment benefit while on treatment and after stopping treatment with crizanlizumab

3.3 The company assumed a constant lifetime treatment effect while on treatment with crizanlizumab in its base case. The committee concluded that no evidence was presented to show a prolonged treatment benefit with crizanlizumab after the 52‑week trial duration. Duration of treatment efficacy was a key driver of the cost-effectiveness results, particularly if any waning of the treatment benefit is taken into consideration. The committee noted that, because there are limited long-term data on the efficacy of crizanlizumab, the uncertainty around a prolonged treatment effect could not be resolved. The company's base case also assumed that crizanlizumab's treatment benefit continues for 2 years in all people after stopping treatment. This was based on results from SUCCESSOR, a retrospective chart review of patients who completed the SUSTAIN trial, which reported data from the 52 weeks after SUSTAIN finished, when patients did not have crizanlizumab. The company highlighted that patients who had the licensed dose of crizanlizumab in the SUSTAIN trial, who were then followed up in SUCCESSOR, had a similar mean annual VOC rate to patients who had the same dose of crizanlizumab in the SUSTAIN trial. The ERG considered that a gradual waning of treatment effect over the 2 years was more likely, and that this would be the same as 1 year of full treatment effect post-discontinuation. In its base case, it reduced the post-discontinuation benefit to 1 year (in people who completed 1 year of treatment) to align with the data available from SUCCESSOR. The committee discussed how the SUCCESSOR data were uncertain because of the small number of patients who had the licensed crizanlizumab dose (n=15). The data were also uncertain because the patients who agreed to further follow up could have been those who had the best outcomes on treatment. Both of these could bias the results in favour of crizanlizumab. It noted that the SUCCESSOR study included the per-protocol population of the SUSTAIN trial, who had at least 12 of the 14 planned study doses of crizanlizumab. The clinical expert suggested that, because crizanlizumab may reduce inflammation of the endothelium, its treatment effect might continue after stopping the drug. However, they said that this was speculative and would depend on how long the drug stays in the blood after stopping treatment. The committee discussed how the evidence from SUCCESSOR did not provide direct evidence of an ongoing treatment effect after stopping treatment. It acknowledged that it may be possible for a short carryover of treatment effect based on the pharmacokinetic and pharmacodynamic data from the SUSTAIN trial. The committee concluded that, although it was possible that there is some treatment benefit after stopping treatment with crizanlizumab, there is very little evidence to support this.

A single utility value from SUSTAIN should be used for all VOC health states, with per event utility decrements applied

The utility values for the VOC health states in the company's model were from an unpublished analysis of a 3‑year US registry study (LEGACY) with an additional utility decrement applied for individual VOCs and complications of sickle cell disease. The value for the utility decrement for VOC events was from a longitudinal study of health-related quality of life in people in the UK with sickle cell disease (Anie et al. 2012). Utility decrements for other sickle cell disease complications were taken from a range of published sources. The company considered that, because health-related quality of life data in SUSTAIN was collected at set time points and not specifically when a VOC occurred, the data may have not fully captured the expected impact of VOCs on quality of life. The company also considered that the trial was too short to show an overall change in health-related quality of life related to sickle cell disease complications and long-term organ damage. It therefore preferred to use the LEGACY study, with its longer follow up. The ERG noted that the company's approach to estimating utility values may overestimate the impact of individual VOCs on health-related quality of life. It considered that it was not appropriate to account for sickle cell disease complications and long-term organ damage through the health state utility value for the VOC groups. This was because they are already accounted for separately in per event utility decrements. Because patients are randomly distributed between health states the committee discussed how it would be unlikely that patients moving from a more severe to less severe health state would recover from long-term organ damage. The committee understood that the utility values derived from the SF‑36 data of patients who completed SUSTAIN were similar for each health state but noted slightly higher utility values for states with more VOCs. The committee did not think this was plausible, so agreed with the ERG's approach of applying a single utility value across all 3 VOC health states (as a weighted average from SUSTAIN), with per event decrements for individual VOCs and complications. This significantly increased the company's base-case ICER. Most of the SF‑36 questionnaires were administered outside of a recall period that included a VOC event. However, the committee noted that some patients in SUSTAIN had a VOC within the recall window of the SF‑36 survey so the company could have estimated the VOC utility decrement from SUSTAIN. The committee concluded that the ERG's method for estimating utility values was more appropriate than the company's method. This is because it reduced the risk that VOC events are double counted and better represented health-related quality of life in people with recurrent VOCs.

Immature SUSTAIN trial data

Limitations in the trial data including small sample size and short duration mean cost effectiveness is uncertain

3.4 The company submission highlighted the limitations of the SUSTAIN trial, including the small sample size (n=65 for placebo, n=67 for 5 mg/kg crizanlizumab dose) and short duration (52‑week treatment phase and 6‑week follow up evaluation phase). This did not allow differences in long-term outcomes such as mortality or rare events that occur with low annual frequency, such as acute chest syndrome, to be determined. The SUSTAIN trial also did not provide information on the effect of crizanlizumab in people who do not seek medical treatment for VOCs and may instead manage them at home. The committee noted that the effectiveness of crizanlizumab was highly sensitive to changes in certain parameters in the company's base case, which were associated with uncertainty. These included age, weight and gender distribution, hydroxycarbamide use, duration of treatment effect during and after stopping treatment, and the choice of utility values. The company believes that ongoing data collection could help to address some of these uncertainties. A conditional marketing authorisation has been granted, which requires further data to be collected. The committee noted that the European public assessment report (EPAR) for crizanlizumab was not available at the time of the meeting, and therefore it could not consider this in its decision-making. The committee noted that the company's response to technical engagement does not make it clear if the quality of life data to be collected is enough to inform utility values in the economic model. The committee concluded that it was not clear how collecting more data would reduce uncertainty around crizanlizumab's clinical and cost effectiveness. This is because of the current uncertainty around the model structure and the parameters used. It also concluded that it was not clear how the cost of crizanlizumab to the NHS could be managed in a way that accurately reflects the decision uncertainty.

Cost-effectiveness estimates

The company's base-case ICER is above £100,000 per QALY gained

NICE's guide to the methods of technology appraisal notes that above a most plausible ICER of £20,000 per quality-adjusted life year (QALY) gained, judgements about the acceptability of a technology as an effective use of NHS resources will take into account the degree of certainty around the ICER. The committee will be more cautious about recommending a technology if it is less certain about the ICERs presented.

The committee agreed that an acceptable ICER would be within the range normally considered to be a cost-effective use of NHS resources (£20,000 to £30,000 per QALY gained). The company's deterministic base-case ICER for crizanlizumab compared with standard of care was substantially higher than £100,000 per QALY gained.

The committee's preferred ICER is substantially above £100,000 per QALY gained

3.5 The committee's preferred assumptions included:

With these assumptions, the committee's preferred ICER greatly exceeded the company's base case. The company's base case was substantially more than £100,000 per QALY gained, and substantially above the range NICE considers to be an acceptable use of NHS resources. The committee therefore could not recommend crizanlizumab as treatment option for recurrent VOCs in people with sickle cell disease.

Equality considerations

There are no equality issues that can be addressed in this technology appraisal

The patient experts explained that sickle cell disease is not widely understood, including among healthcare professionals, which often results in poor hospital care and a stigma around seeking pain relief for crises. The committee heard how the condition is more common in people of African or African-Caribbean family origin and that as a group these people tend to have poorer health outcomes than other ethnicities. The committee concluded that issues related to differences in prevalence or incidence of a disease cannot be addressed in a technology appraisal.

Other factors

The benefits of crizanlizumab are captured in the cost-effectiveness analysis

3.6 The company considers crizanlizumab to be innovative because it is well tolerated and an effective treatment for preventing VOCs in people with sickle cell disease. The committee agreed that these are important benefits. But it concluded that it had not been presented with evidence of any additional benefits that could not be captured in the QALY.