Odevixibat for treating progressive familial intrahepatic cholestasis

The patient experts explained that the quality of life of a child with PFIC may be extremely poor. They emphasised that the pruritus can be debilitating, and that people can scratch themselves to the point of bleeding and skin damage. The patient experts stressed the profound nature of the pruritus, describing it as "itching from the inside out". Poor growth is a common concern for carers, particularly in PFIC1. The clinical experts explained that children with PFIC eat a specific diet and take fat-soluble vitamin supplements to improve nutrient absorption. One patient expert highlighted that children may need a feeding tube to help manage the condition, which can be traumatic for the children as well as challenging for carers. Children with PFIC often have their education severely disrupted. This can be because of absence through illness and hospital attendances, and disrupted sleep impairing their ability to learn when at school. When compounded by condition-related learning disabilities, the educational attainment and social development of children with PFIC may be significantly affected. Carers explained that they needed to provide constant care to children with PFIC. Commonly, the demands are such that carers cannot work full time, resulting in loss of earnings and implications for career development. One carer explained that she could no longer carry on with her job as her daughter's condition deteriorated, because of the demands of juggling hospital visits and sleepless nights. The patient experts stressed that a diagnosis of PFIC affects the entire family. Siblings can be affected by the large number of hospital visits and the experience of seeing a sibling suffer. The unpredictability of the condition, particularly the speed of progression, along with financial pressures, can cause anxiety and other psychological difficulties for people with PFIC and their families. The committee concluded that PFIC has a significant effect on the quality of life of people with the condition, family members and carers.

The committee noted that there are no medicines licensed for PFIC in the UK. With medicines used off label, such as ursodeoxycholic acid, cholestyramine and rifampicin, the aim is to control the pruritus and delay progression to a liver transplant. However, response to off-label medicines varies, and there is no data from randomised controlled trials to support their clinical effectiveness. The clinical experts explained that cholestyramine is only commonly used in newborns because older children find it hard to tolerate. Surgical options such as partial external biliary diversion (PEBD) are associated with a decrease in serum bile acid levels and increased native liver survival. However, the clinical experts explained that PEBD is rarely used in the NHS and is only an option in a limited group, for example, those who have no liver fibrosis and whose liver disease is not advanced. The committee heard that, for PEBD, an external stoma needs to be created. This can be distressing and can have a significant effect on quality of life. Also, with a stoma, there are risks of complications such as electrolyte disturbance, dehydration, bile leakage and other problems. The clinical experts explained that PEBD is often declined as a therapeutic option by individuals and families. This is because many of them perceive that the adverse effects outweigh the potential benefits. More recently adopted methods of surgical biliary diversion (SBD), such as an internal biliary drainage or internal ileal exclusion, avoid the need for an external stoma bag. But there is a lack of data about their relative benefit. The clinical experts explained that these methods are generally used as a longer-term solution in people whose condition has responded to PEBD but who do not want or cannot tolerate an external stoma bag. For people who do not have SBD, or when pruritus persists despite surgery, a liver transplant is the only remaining option.

The committee heard that a liver transplant is needed for most people with PFIC by age 20 years. This is because of liver disease and uncontrollable pruritus. The patient experts explained that a liver transplant can be successful in resolving pruritus, so significantly improving the quality of life for children with PFIC and their carers. However, transplants are associated with complications such as infection, increased risk of skin or liver cancer and life-threatening complications of graft rejection. Lifelong immunosuppression, frequent hospital visits, regular monitoring for rejection after transplants and the potential for recurrence of pruritus are big concerns for people with PFIC and their families. Consultation comments submitted after the first evaluation meeting stressed that a transplant can negatively affect a child's social development. This is because of lost school days for surgery and the inability to participate in activities or careers associated with high risks of infection. The committee recognised that treatment options for PFIC are currently limited. It concluded that there was an unmet need for a new treatment for this condition.

The clinical experts explained that the current pathway of care for people with PFIC varies depending on the type. They explained that control of pruritus with off-label medicines such as ursodeoxycholic acid is more successful in people with PFIC3 than with PFIC1 or PFIC2. This means that people with PFIC3 are less likely to progress to surgery. They clarified that PEBD is most effective at reducing serum bile acid levels in PFIC2. However, long-term outcomes after the procedure, such as time to transplant, are uncertain because of a lack of data. The clinical experts highlighted that a liver transplant is less likely to be offered to people with PFIC1. This is because of the potential for lasting non-liver complications including severe diarrhoea and pancreatitis and the high risk of recurrent pruritus. The committee concluded that the current pathway of care for PFIC is largely determined by type.

The company has positioned odevixibat as a first-line treatment for PFIC. Because no active treatment is routinely commissioned in the NHS for PFIC, the committee agreed that standard care without odevixibat was the appropriate comparator, as listed in the NICE scope. The company considered this included SBD such as PEBD but did not include off-label medicines. This was because people having odevixibat could also have off-label medicines for symptom management and that these medicines have poor clinical effectiveness. The ERG noted that off-label medicines were included in the NICE scope and would form part of standard care without odevixibat. The clinical experts highlighted that, if odevixibat was approved, off-label medicines would be started in the time leading up to diagnosis being confirmed. They also pointed out that off-label medicines would be started in babies younger than 6 months, who are not included in the marketing authorisation for odevixibat. The clinical experts confirmed that odevixibat would most likely be started in people having off-label medicines who had little or no drop in serum bile acid levels. They also stated that odevixibat would likely replace surgical options such as PEBD. However, they thought that PEBD was unlikely to be offered as a subsequent treatment for people whose condition did not respond to odevixibat. This was because both interventions work in similar ways by reducing the amount of bile acids in the gut available for reuptake. So, the likelihood of a response to PEBD in people whose condition does not respond to odevixibat is small. One clinical expert estimated that there would be no response in about 10% of people. The committee concluded that the comparators for odevixibat were off-label medicines and SBD, including PEBD, and that sequential use of odevixibat and PEBD is unlikely in NHS practice.

The patient and clinical experts highlighted that there is an unmet need for treatments specifically targeting PFIC. They emphasised that odevixibat has the potential to improve quality of life, remove the need for SBD and delay the time to transplant for people with PFIC. The committee heard that complete relief of pruritus would represent a successful treatment, but anything to reduce pruritus would be beneficial. The clinical experts noted the need for a treatment that, in addition, both improved growth and preserved liver function. The committee recalled that cholestyramine is effective at lowering serum bile acid levels, but that it can be poorly tolerated (see section 4.3). It concluded that people with PFIC and their families would welcome odevixibat as a treatment for the condition.

The main clinical trial evidence for odevixibat came from a phase 3 completed randomised controlled trial, PEDFIC1, and an ongoing single-arm open-label extension study, PEDFIC2. These trials enrolled people with a clinical diagnosis of PFIC1 or PFIC2 who had elevated serum bile acid levels and cholestatic pruritus:

• PEDFIC1 enrolled children 6 months and older, 23 of whom had odevixibat 40 micrograms/kg/day and 19 of whom had 120 micrograms/kg/day. A further 20 people had placebo. The follow-up period was 24 weeks.

• PEDFIC2 is an ongoing long-term follow-up study of PEDFIC1. It has enrolled 71 people who have had odevixibat 120 micrograms/kg/day. This includes 53 people in cohort 1 who had previously participated in PEDFIC1 (19 who had 40 micrograms/kg/day, 15 who had 120 micrograms/kg/day and 19 who had placebo) and 16 people in cohort 2. Cohort 2 includes people of any age who weighed over 5 kilograms with any type of PFIC who either had not met the eligibility criteria for PEDFIC1 or were eligible for enrolment after PEDFIC1 recruitment had been completed, so had not had odevixibat before. Interim data from week 24 analyses were available from a July 2020 data cut.
  
  The company also provided evidence for odevixibat from a completed exploratory phase 2 study. This study enrolled 20 children with cholestatic pruritus of any cause, who were allocated to odevixibat at doses of 10, 30, 60, 100 or 200 micrograms/kg/day for 4 weeks. The committee noted the wide range of odevixibat doses and that only 10 people in the trial had PFIC (types 1, 2 or 3). The committee concluded that the PEDFIC1 and 2 studies were the most appropriate data sources for odevixibat.

The committee first considered the clinical-effectiveness evidence for odevixibat compared with off-label medicines. It noted that most people in both the odevixibat and placebo arms of PEDFIC1 were having concurrent off-label medicines. So, it agreed that PEDFIC1 provided relevant comparative data because off-label medicines form part of current standard care and are likely to be given alongside odevixibat in clinical practice. The company did not present any data comparing odevixibat with PEBD or other types of SBD. It explained that an indirect comparison is planned that will compare odevixibat with standard care both with and without SBD. Comparative clinical-effectiveness data in the company's model came from NAPPED. This was a natural history cohort study that included 130 people with PFIC1 and 264 people with PFIC2 having standard care. The median follow-up time was 4.1 years (range 1.5 to 12.3 years). During this time, 48% of people with PFIC1 and 23% with PFIC2 had SBD. The committee agreed no evidence had been presented to compare odevixibat with PEBD. It concluded that the most appropriate comparative data source available for off-label medicines was PEDFIC1.

The primary outcome for PEDFIC1 for Europe and the rest of the world was the proportion of people who had a reduction of at least 70% in the serum bile acid level from baseline or levels that reached 70 micromol/litre or less. The primary outcome for PEDFIC2 (Europe and the rest of the world) was the change in serum bile acid levels from baseline over the treatment period. The primary outcome in the US for both PEDFIC1 and 2 was the proportion of positive pruritus assessments over the treatment period. The company measured this using a new observer-reported outcomes (ObsRO) instrument developed for this purpose. The ObsRO instrument captures scratching on a scale of 0 (representing no scratching) to 4 (representing the worst possible scratching) using twice-daily patient and carer questionnaires. A positive pruritus response is defined by the company as an observer-reported scratching score of 1 or below, or a reduction of 1 or more points from baseline. Both studies also collected data on changes in growth, liver function, health-related quality of life, and the number of people having surgery or liver transplants. The patient experts explained that a reduction in pruritus would have the biggest effect on the quality of life of people with PFIC. The clinical experts explained that the relationship between serum bile acid levels and pruritus levels is complex, and that the 2 do not always correlate. Nonetheless, in general, lower serum bile acid levels are associated with improved pruritus and native liver survival. The patient experts highlighted that improvements in growth and liver function tests are important outcomes to people with PFIC. This is because they are generally associated with reduced pruritus, and improved sleep and quality of life. The committee concluded that the main outcomes important to clinicians and people with PFIC and their families were captured in the company's clinical trials.

In PEDFIC1, the proportion of positive pruritus assessments (a reduction of at least 70% in serum bile acid level from baseline or reaching 70 micromol/litre or less) compared with placebo after 24 weeks of treatment was statistically significantly greater in the odevixibat combined treatment arms (33%) than the placebo arm (0%). The results suggested a difference in response for people who had 40 micrograms/kg/day of odevixibat compared with 120 micrograms/kg/day, but this was not statistically significant. Also, the results were based on small patient numbers (the exact proportions are academic in confidence and cannot be reported here). There was a statistically significantly greater proportion of positive pruritus assessments (using the ObsRO instrument) in people in PEDFIC1 who had odevixibat (all doses; 54%) compared with placebo (29%). For people who continued to have odevixibat in PEDFIC2, the improvement in serum bile acid levels and pruritus was maintained. However, the greatest improvements were seen in those people who had not had odevixibat before, that is, people who had placebo in PEDFIC1 or were newly enrolled. The results also suggested some additional serum bile acid response to the 120 micrograms/kg/day dose in people whose condition did not respond to the 40 micrograms/kg/day dose in PEDFIC1. (The exact proportions are academic in confidence and cannot be reported here.) The committee noted that the PEDFIC2 data used to determine the response to up titration included 4 people with a follow up of only 24 weeks. Improvements in growth were also seen in PEDFIC1 for odevixibat compared with placebo and were maintained in people continuing odevixibat in PEDFIC2. The committee concluded that odevixibat was effective in reducing both serum bile acid level and pruritus in PFIC1 and PFIC2.

The committee next considered the clinical effectiveness of odevixibat by PFIC type. It recalled that, in PEDFIC1, only people with PFIC1 and PFIC2 were enrolled. Serum bile acid response rates improved in both types, but the data suggested a potential difference in the response rates by type. However, the committee noted that patient numbers in the subgroups were small, that the trial had not been designed to detect a difference by type, and that no statistical comparisons by type had been done. The committee noted that 5 people in PEDFIC2 had PFIC3 and 1 person had PFIC6. However, there was no data for odevixibat in PFIC4 and PFIC5, even though these are included in the marketing authorisation. At the last data cut, 80% (4 of 5) people with PFIC3 had a serum bile acid response according to the definition in PEDFIC2. At the second meeting, the committee noted that the reduction in serum bile acid levels seen in PEDFIC2 for PFIC6 was smaller than for other subtypes. This result was uncertain because it was based on results from 1 person. The committee concluded that subgroup analyses from PEDFIC2 suggested some serum bile acid reduction for all subtypes enrolled. However, it noted these results were based on small numbers, with very little evidence for PFIC types other than 1 and 2.

In PEDFIC1, the proportion of people who had a treatment-related adverse event was higher for odevixibat (33%, 14 of 32) than placebo (15%, 3 of 20). The committee noted that the proportion of people with any adverse effect during the treatment period was high at 83% (35 of 42) in the odevixibat arm and 85% (17 of 20) in the placebo arm. However, no serious adverse events related to odevixibat were reported in the phase 2 study or PEDFIC1 and 2. The clinical experts explained that odevixibat is well tolerated in clinical practice. The main adverse events are gastrointestinal and may be alleviated in some people by using the lower starting dose. The company stated that no additional safety monitoring is needed for odevixibat, and there are no special precautions or warnings for its use. The committee concluded that odevixibat has an acceptable adverse event profile.

The clinical experts considered that the evidence from PEDFIC1 and 2 was broadly generalisable to the population with PFIC seen in England. However, the committee was aware of several potential differences between the clinical trial populations and NHS clinical practice. To enrol in both PEDFIC1 and 2, people needed to have a serum bile acid level of 100 micromol/litre or more and an average pruritus score of 2 or more on the company's ObsRO instrument. The committee noted that 5 people in PEDFIC1 and 3 people in PEDFIC2 had been excluded because they met the pruritus eligibility criteria but did not have a high enough serum bile acid level. The committee recalled that the aim of treatment is to reduce pruritus, so these people would likely have treatment in clinical practice. PEDFIC1 also excluded people with a previous lack of response to ileal bile acid transporter inhibitors and SBD within 6 months. The ERG flagged that odevixibat may also be used in these people and that they were included in cohort 2 of PEDFIC2. At the second evaluation meeting, the committee noted that the average age in PEDFIC1 was 4.25 years. One clinical expert highlighted that, if odevixibat were recommended, clinicians would treat PFIC from diagnosis. They explained that PFIC1 and PFIC2 are commonly diagnosed in people within the first few months of life. The committee recalled that odevixibat has a marketing authorisation for treating PFIC in people aged 6 months and older. So, the population who had odevixibat in clinical practice may be younger than that included in the company's trials. The clinical experts theorised that, if PFIC was treated with odevixibat earlier, the response could be better than that reported in the trials, although data to support this is lacking. This was because the liver disease would be less advanced and fluid bile acid accumulation causing cholestasis could be prevented. So, there was a possibility that the clinical trial results underestimated odevixibat's treatment effect in clinical practice. The committee recognised that the population included in the company's trials may not fully reflect that in clinical practice. However, given the limited data available, it concluded that data from the full populations of PEDFIC1 and 2 were suitable for decision making.

The committee recalled that, at the week‑24 data cut in PEDFIC2, the maximum treatment duration with odevixibat was 48 weeks. The ERG noted that changes in long-term outcomes (including survival, reduced transplant rates or delays to a liver transplant with odevixibat) would therefore not have been captured in the evidence base. The effect of treatment on serum bile acid level, pruritus and growth over a longer period was also unknown. The clinical experts explained that people would have odevixibat until they had a lack of response or intolerable side effects, which may be after many years. The committee concluded that the effect of odevixibat on long-term outcomes was uncertain.

The committee recalled that the company's main trial evidence was limited to PFIC1 and PFIC2, and that there was no data for many of the less prevalent types. One clinical expert emphasised the rarity of the condition, estimating that 10 people a year at most were diagnosed with the most common type, PFIC2, in her clinic. Given that PFIC4, PFIC5 and PFIC6 account for a small proportion of all diagnoses, it is unlikely that further data could be collected on the rarer types in clinical trials. The committee agreed that the practical challenges of recruiting people with the rarer types of PFIC to clinical trials made data collection outside of the existing studies implausible. At the second evaluation meeting, the clinical experts stressed that odevixibat inhibits reuptake of bile acids in the colon. So, it is expected to work in all PFIC types with some bile flow out of the liver to the gut. People with PFIC2 with a bile salt export pump protein (BSEP) 3 mutation have a complete absence of the BSEP. So, their condition would not be expected to respond to treatment. However, the committee noted that people with a BSEP3 mutation were excluded from the marketing authorisation for odevixibat, so would not have treatment in the NHS. One clinical expert explained that odevixibat might not be effective in PFIC5. This is because it results in deficient BSEP protein expression and causes unregulated bile acid synthesis in the liver. Bile acid levels are so high that blocking reuptake in the intestine may not resolve the symptoms. The committee recalled that there was no clinical evidence available to show whether odevixibat did or did not work in PFIC5. It was also aware of the rare nature of this subtype. (The company's response to consultation stated that, worldwide, the literature reports PFIC5 in 9 people.) So, the number of people with PFIC5 in the NHS is extremely small. Finally, the committee was aware that the marketing authorisation recommended odevixibat for a general PFIC population. The committee concluded that there was limited data in the less prevalent subtypes of PFIC.

The committee recalled that the marketing authorisation for odevixibat specifies a starting dose of 40 micrograms/kg/day. The dose can be escalated to 120 micrograms/kg/day if there has not been an adequate clinical response after 3 months of continuous therapy. The clinical experts classed an adequate response to odevixibat as improvements in at least 2 of the 3 main PFIC outcomes: serum bile acid levels, pruritus and liver function tests. They acknowledged that a definition of response might vary among clinicians. However, they explained that the dose of odevixibat would likely be increased if little or no improvement in these outcomes was seen. At the second evaluation meeting, the company agreed that this definition was likely to be used in clinical practice to determine the need for dose escalation. The ERG also stated that pruritus is the most clinically important outcome, so would primarily be used to assess response to treatment. The committee noted that the dosage of odevixibat given in the clinical trials was not based on response. People who had 40 micrograms/kg/day in PEDFIC1 and then went into PEDFIC2 had the high dose regardless of the previous response to treatment. Also, people enrolled in the PEDFIC2 cohort 2 started on high-dose odevixibat, whereas they would start on a lower dose in clinical practice. The clinical experts explained that the mechanism underlying response in PFIC was complex but expected the condition in some people to respond to dose escalation. The committee agreed that the dose of odevixibat would be escalated in people whose condition showed no improvement in at least 2 of serum bile acid levels, pruritus and liver function tests.

The company developed a semi-Markov model to estimate the cost effectiveness of odevixibat. The population included in the model was limited to people with PFIC1 and PFIC2, reflecting evidence from the PEDFIC1 study. The model health states included response and loss of response for serum bile acid, response and loss of response to PEBD, a liver transplant, after a liver transplant and death. Only people who had odevixibat could have a serum bile acid response, which the company assumed was always associated with an improvement in pruritus. Following loss of response to odevixibat, people in the model did not have SBD, instead progressing straight to a liver transplant. People having standard care with off-label medicines were assumed not to have a serum bile acid response and entered the model in the serum bile acid loss-of-response health state. They could then progress to a liver transplant from any of the loss-of-response health states, but not from the PEBD response state. Most people remained in the liver-transplant health state for 1 cycle only. However, a small proportion of people in both arms remained for an additional cycle to represent people who had another transplant. The company assumed that people moved up to the higher dose of odevixibat if there was no response after 6 months of continuous treatment at 40 micrograms/kg/day.

The clinical experts highlighted that the model did not capture treatment differences for other types of PFIC, for example, that people with PFIC3 are less likely to have SBD (see section 4.5). They also highlighted that improvements in growth and liver function were important outcomes to people with PFIC and their families but had not been included in the company's modelling. The company assumed that people entered the model at the age of 4.25 years (the average age in PEDFIC1). However, the committee recalled that people with PFIC1 and PFIC2 may start treatment at a younger age in clinical practice. The ERG noted that the modelled age represented the average for all PFIC subtypes, some of which are not commonly diagnosed in newborns, However, it provided a scenario analysis in which people entered the model at a lower age of 3 years. The committee concluded that the basic model structure was appropriate for decision making, but that people may start odevixibat younger than assumed in the company's model.

The company used data from PEDFIC1 to populate the patient characteristics and serum bile acid response to odevixibat for people having the 40 micrograms/kg/day dose in the economic model. The company calculated the proportion of people having high-dose odevixibat in the model using the ratio between the people with a response at the low dose and those with a response at any dose. For people having high-dose odevixibat, the model used the serum bile acid response at week 24 in PEDFIC2 for people whose condition had not responded to low-dose odevixibat in PEDFIC1. The committee noted that the company's assumptions about high-dose odevixibat were calculated using data from few people. For example, week‑24 data at the cut-off was only available to inform the response rates for 4 people whose condition did not respond to 40 micrograms/kg/day. At the second meeting, the clinical experts estimated that around 30% of people would have high-dose odevixibat in clinical practice. The committee noted that this proportion was similar to the assumption in the company model. It concluded that the company's assumptions about high-dose odevixibat were associated with uncertainty but acceptable for decision making.

In people whose condition had responded to odevixibat, the company modelled loss of serum bile acid and pruritus response using the stopping rate from PEDFIC2. The ERG noted that people in the PEDFIC2 study who stopped odevixibat did so because of adverse effects, not because of a lack of serum bile acid response. This meant that the loss-of-response rate is likely to be higher in clinical practice than that modelled by the company. The clinical experts explained that people would be keen to keep having odevixibat if it improved pruritus. They thought people would only likely stop treatment if they had unbearable side effects or progression of liver disease. For this reason, the stopping rate in clinical practice was likely to be low and was therefore comparable to that in PEDFIC2. One clinical expert estimated that about 30% of people would stop odevixibat over time. The committee concluded that further data on the long-term effectiveness of odevixibat would be useful.

For the standard care arm, the probabilities for having PEBD and subsequent progression to a liver transplant were taken from the NAPPED natural history study. The committee noted that NAPPED was a global study. This meant that the rates of SBD reported (48% of people with PFIC1 and 23% with PFIC2) were likely higher than those in England, where this surgery is rarely done. The clinical experts highlighted that geographical variations in PEBD rates were due to differences in PFIC subtype prevalence and clinician preference. They estimated that, before the availability of odevixibat in a clinical trial, PEBD was used in around 25% to 30% of people with PFIC in the UK. The company assumed that, in 5% of people, the response to PEBD would be lost in each cycle. This was based on clinical advice to the company. This advice was that the loss of response for PEBD would be slightly higher than that for odevixibat because of the complications associated with surgery. At the second committee meeting, the clinical experts explained that if someone had had odevixibat, it was unlikely that they would go on to have PEBD. This is because response to any PFIC treatment depends on the liver retaining some ability to transport bile acids into the gut. When bile acid transport out of the liver becomes inadequate (because of uncorrected liver disease or loss of bile acid transport receptor expression), response is lost. Therefore, if odevixibat treatment eventually fails, PEBD is unlikely to be effective. The clinical experts explained that response to PEBD is unpredictable. The committee noted that, after consultation, the company and ERG base cases included PEBD in the odevixibat and standard care arms at the rate reported in the NAPPED study. It agreed that both the proportion of people who had PEBD and those whose condition subsequently lost response in the company's model were uncertain. It considered company and ERG scenarios that varied these parameters. For the standard care arm, in the absence of further data sources, the committee accepted the company's assumptions for PEBD. However, in the intervention arm, the committee concluded PEBD rates had been overestimated because people who had had odevixibat were unlikely to go on to have PEBD.

The company calculated the probability of a liver transplant in people who had not had PEBD in both arms using data from native liver survival curves in NAPPED. The ERG flagged that this data included people whose condition both did and did not show a serum bile acid response to treatment. So, transplant rates for odevixibat would likely be higher than was modelled. In its base case, the ERG assumed equal rates of liver transplants in the health states for serum bile acid loss of response and PEBD loss of response. The committee concluded that, in the absence of further data sources, the ERG's probability of a liver transplant was most appropriate for people who had not had PEBD.

The company modelled mortality rates using a variety of sources, which applied to both the odevixibat and standard care arms in the model. For the acute post-transplant mortality rates (applied in the year of transplant in the model), the company used a meta-analysis of mortality rates from 10 PFIC studies reported in the literature. For the long-term mortality rates, applied in the model from the second year after transplant, the company used data from survival curves from 4 of these studies. It fitted an exponential distribution to this data. This gave an acute post-transplant mortality of 11.31% and a long-term post-transplant mortality of 1.94%. The ERG's analysis, which corrected several errors in the company model, and adjusted the meta-analysis output, produced an acute post-transplant mortality rate of 10.92% and long-term rate of 1.42%. The committee agreed with the ERG's corrections and considered its mortality rates most appropriate for decision making.

At consultation, the company provided scenario analyses that assumed higher rates of post-transplant mortality in people who had a second transplant compared with rates after a first transplant. The scenario analyses applied hazard ratios reported in a paper by Watt et al. (2010) to the proportion of people assumed to have a second transplant in the model. The paper reported a lower risk of death for people who had a second transplant within 1 year of the first transplant compared with people having a transplant later than this. The company presented 2 scenarios:

• the first assumed that, after the first operation, all retransplants occurred within 1 year (applying a hazard ratio [HR] of 1.52 in the model for the first year only)

• the second assumed all retransplants occurred after 1 year (applying an HR of 4.79 from 2 years onwards).
  
  The ERG noted that Watt et al. (2010) was based on liver transplants occurring between 1990 and 1994. So, the rates reported may not be relevant because retransplantation procedures have improved. It also highlighted inconsistencies in the reporting and statistical analyses. One clinical expert explained a retransplant is needed by 10% to 20% of people with a liver transplant for PFIC. Most of these are for people with a BSEP3 mutation (excluded from odevixibat's marketing authorisation). Also, most occur in the first 3 months after the initial operation because of surgical complications and infections. The clinical experts estimated mortality of about 50% within 1 year for people needing a second transplant. As time goes on, fewer people need a retransplant, but the individual risk of dying increases because retransplant becomes more difficult. This is because of scar tissue build up in the liver and PFIC-specific complications including fat deposits around the graft. The committee noted that these mortality rates were higher than those estimated by the ERG's clinical experts, who predicted an additional 30% mortality for retransplant at any timepoint. Although the committee had not identified retransplant mortality as an issue in the first meeting, it acknowledged that the risk of death after the second transplant was likely to be higher than the first. However, it considered that the true effect on mortality of a second transplant lay between the company's 2 scenarios, because:

• the first scenario did not capture the increased risk of death from a later retransplant

• the second scenario did not capture that the increased chance of retransplant was mainly within the first year after initial surgery.
  
  To account for this, the committee preferred to apply the higher hazard ratio of 4.79 to the proportion of people having more than 1 transplant, but only in the first year in the model.

The company applied the costs of odevixibat in the serum bile acid response state and for 6 weeks in the first cycle of the serum bile acid loss-of-response health state. Dosing was based on the average weight by age up to a weight of 55.5 kilograms. The company also applied a normal distribution to calculate the proportion in each weight category. The costs of off-label medicines were included in the loss-of-response health states for both arms because the company assumed that they would be used alongside odevixibat. Because there were no serious adverse events related to odevixibat in PEDFIC1 and 2, the company did not include costs for adverse events in its base case. It did, however, include costs for carers' lost productivity for everyone younger than 18 years in the model. It stated that odevixibat was expected to have a cost saving beyond the NHS and personal social services (see section 4.37). The committee agreed with the ERG that the inclusion of productivity costs was outside the NICE reference case. It preferred the ERG's analyses, which excluded productivity costs and included costs for commonly occurring treatment-emergent adverse events in PEDFIC1.

The committee noted uncertainty in the company's costs for PEBD. The ERG noted that the company's costs for PEBD included repeated surgeries for 67% of people, with equal costs applied to each surgery (same cost as initial procedure). The ERG stated that these assumptions were likely to be overestimates, so the cost of PEBD surgery in clinical practice would likely be lower. The ERG presented a scenario that used lower costs for PEBD. The committee agreed that the company's costs were uncertain and considered both the company's base case and ERG's scenario in its decision making.

PEDFIC1 and 2 collected Pediatric Quality of Life Inventory (PedsQL) data at baseline and week 24. This was mapped to EQ‑5D‑3L using a mapping algorithm from Khan et al. (2014). However, data was only available for a few people, so the company chose to use utility values from the literature in its base case. The company sourced utility values for odevixibat response from a study by Kamath et al. (2015). For loss of response, it used utility values of 0.91 from healthy children to represent serum bile acid response, and 0.83 from children with chronic intrahepatic cholestasis of any cause (of whom 51% had genetically confirmed PFIC). The ERG noted that, because of ongoing complications (including extra-hepatic features) and symptoms of PFIC, people whose condition has responded to odevixibat are unlikely to have the same quality of life as a healthy child. So, the company's utility values were higher than would be expected in clinical practice. The ERG preferred to use the utility values from the company's mapping study in its base case (0.858 for serum bile acid response and 0.697 for serum bile acid loss of response). The committee agreed that the company's utilities were likely to be high and that values derived directly from the clinical trial were preferred.

For response and loss of response to PEBD, the company used the utility for healthy children from Kamath et al. (2015). However, it applied a utility multiplier of 0.722 to represent the quality-of-life effect of having a stoma bag. This was taken from a study of adults with ulcerative colitis by Arseneau et al. (2006). For the PEBD loss-of-response health state, the company applied an additional disutility of 0.977 for short stature, reported in a study of children with chronic kidney disease by Al-Uzri et al. (2013). This resulted in utilities of 0.659 for PEBD response and 0.599 for PEBD loss of response. The company also presented scenario analyses using a stoma bag utility multiplier of 0.945 from a colorectal cancer study by Hornbrook et al. (2011) and its own utility elicitation study. (The exact value is academic in confidence and cannot be reported here.) The committee noted that most people in the colorectal cancer study were over 70 years old, so it was unlikely to be comparable to the population with PFIC. It also heard that the company's vignette study only used data from 2 carers of children with PFIC, so was not considered sufficiently robust to capture all stoma bag-related issues by the ERG. At the first committee meeting, the ERG chose to use a disutility multiplier of 0.833 in its base case. This was calculated by averaging the disutilities derived from the colorectal cancer and ulcerative colitis studies, and was preferred by the committee at the time. However, clinical expert feedback at consultation was that the disutility of a stoma bag for PEBD was expected to be comparable to that for ulcerative colitis. So, the ERG included the lower value of 0.722 in its updated base case. The clinical experts explained that the stoma-related effect on quality of life is significant, especially in older children. This is because the disutility may be larger for them compared with other age groups, and they often refuse an external biliary diversion. One clinical expert also highlighted that stoma-related quality of life was likely to be better for someone with colorectal cancer or ulcerative colitis than for someone with a stoma bag collecting bile. This is because the irritant nature of bile at the stoma site can cause problems including infection, which often needs treating with antibiotics and other interventions. At the second meeting, clinical experts also flagged the large volume of fluid loss with a PEBD stoma bag, sometimes up to 1 litre per day. In comparison, stoma bags for ulcerative colitis or colorectal cancer, which are located lower down the gastrointestinal tract, are associated with less fluid loss. The clinical experts agreed that literature utility multipliers from ulcerative colitis and colorectal cancer likely underestimated the quality-of-life effect of a stoma bag. One clinical expert stated that a utility multiplier derived from an infant with a stoma bag for necrotising fasciitis, which also has a high volume of fluid loss, would be more comparable to a PEBD. At the first committee meeting, the patient experts highlighted that people with PFIC and carers have a very negative attitude to having a stoma bag, and that sometimes the invasive surgery may not resolve the pruritus. Consultation comments supported this view, describing a stoma bag as "a great discomfort" and "shameful" for people with PFIC. The committee agreed that the disutility of living with a stoma bag was likely to be lower than the utility multipliers derived from both the ulcerative colitis and colorectal cancer studies. It noted that the utility multiplier from the company's elicitation study was considerably lower than the alternative values but recalled the small sample size informing the results. At the second meeting, the committee concluded that, in the absence of alternative sources, the utility multiplier derived from ulcerative colitis was most appropriate for decision making.

In the model, the company assumed that most people who had a liver transplant did so because of uncontrolled pruritus. For this reason, both the company and ERG used a utility value of 0.710 in their base cases for liver transplantation, which was derived from people with severe pruritus. To represent the quality of life for people with PFIC post-transplant, the company used a value of 0.850, mapped from PedsQL data in a systematic review of children having a liver transplant. The ERG chose to use a lower value of 0.798 for this health state. There was no utility for after a liver transplant from the company's mapping study. So, it calculated the ratio of the utilities for after a liver transplant and for odevixibat response from the literature. This ratio was then applied to the odevixibat response utility from the mapping study. The committee agreed that utilities mapped from the clinical trial were most appropriate. So, it concluded that the ERG's utility value for the post-liver-transplant health state were the most preferrable.

The company and ERG included a carer disutility of -0.05 in the PEBD response, serum bile acid loss of response and post-liver-transplant health states and a disutility of -0.1 for the PEBD loss-of-response health state. The committee recalled that the burden on carers could be substantial because children with PFIC often needed a significant amount of carer support. However, it noted that the disutility for carers had been sourced from NICEs technology appraisal guidance on nusinersen for treating spinal muscular atrophy and dupilumab for treating moderate to severe atopic dermatitis. These conditions manifest in different ways to PFIC. The committee concluded that carer disutilities should be included in the modelling, but that the extent of any carer disutility in PFIC is uncertain.

The committee understood that NICE's interim process and methods of the highly specialised technologies programme (2017) specifies that a most plausible incremental cost-effectiveness ratio (ICER) of below £100,000 per quality-adjusted life year (QALY) gained for a highly specialised technology is normally considered an effective use of NHS resources. For a most plausible ICER above £100,000 per QALY gained, judgements about the acceptability of the highly specialised technology as an effective use of NHS resources must take account of the size of the incremental therapeutic improvement. This is seen through the number of additional QALYs gained and by applying a 'QALY weight'. It understood that a weight of between 1 and 3 can be applied when the QALY gain is between 10 and 30 QALYs. The committee considered that there was uncertainty in both the company's and ERG's analyses. However, it concluded that the undiscounted QALY gains for the scenarios incorporating its preferred assumptions did not meet the criteria for applying a QALY weight.

The company and NHS England have agreed a confidential commercial discount for odevixibat. All cost-effectiveness results of the economic analysis incorporating this discount, along with any comparator discounts, are confidential, so the ICERs cannot be reported here.

After consultation, the committee noted that the company's and ERG's updated base cases included the same assumptions. However, the ICER was above the threshold considered to provide value for money in the context of a highly specialised service when the confidential discounts for odevixibat and comparators were applied. The committee noted that the ERG's scenario using a start age of 3 years reduced the ICER. It also recalled that people were expected to start odevixibat in clinical practice at a younger age than that assumed in the company's model. Scenarios that assumed a higher mortality after a retransplant also lowered the ICER. The committee concluded that both the company's and ERG's base-case cost-effectiveness results were likely higher than would be expected in clinical practice and that this ICER was likely to be conservative.

At the second committee meeting, the committee considered the following assumptions to be the most appropriate for decision making:

• including PEBD in the standard care arm only using rates from the NAPPED data

• using a start age of 4.25 years (the average age in PEDFIC1), although it recognised the age might be lower than this (see section 4.19)

• using the same probability of a liver transplant for odevixibat and PEBD loss-of-response health states

• using the utility value from the ulcerative colitis study

• using mortality rates for the acute and long term after a liver transplant from the ERG's analyses

• applying a hazard ratio of 4.79 in the first cycle only to the proportion of people with a second transplant

• excluding carer productivity costs

• including costs of common adverse events from PEDFIC1

• applying a 3.5% discount for costs and benefits, with no additional QALY weighting.
  
  Using these assumptions, the cost-effectiveness results for odevixibat compared with standard care were considerably lower than the company's and ERG's base cases. However, they remained somewhat higher than the threshold normally considered an effective use of NHS resources in a highly specialised technology.

The committee also considered that there was some uncertainty surrounding the cost effectiveness of odevixibat for people with PFIC. The committee recognised that:

• it had been presented with very limited data for people with PFIC types other than PFIC1 and PFIC2

• there was no data for odevixibat when used before or compared directly with PEBD

• the long-term effectiveness of odevixibat on survival, time to a liver transplant and use of SBD was unclear

• the proportion of people whose condition stopped responding to treatment and the response rates to high-dose odevixibat were uncertain

• there was no evidence that used the dose escalation schedule in the marketing authorisation that would be used in NHS practice.
  
  The committee acknowledged that some of these uncertainties could be resolved with data collection. It was aware that the PEDFIC2 study was ongoing and could provide further data on survival outcomes, liver transplant rates and alternative utility values for people having high-dose odevixibat. It would also provide further data in PFIC3 and PFIC6, including results for 2 additional people with PFIC6 currently unreported. The committee was aware that the company's planned indirect comparison would provide data on the effectiveness of odevixibat compared with PEBD. It also noted that a global registry had been requested by the regulator that:

• is expected to include some people from the UK

• would provide further data on the time to a liver transplant, SBD rates, survival and safety outcomes.
  
  The committee concluded that additional data for odevixibat that would reduce the clinical-effectiveness uncertainty was expected in the near future.

The company stated that treatment with odevixibat would be started and supervised by clinicians experienced in managing PFIC. It highlighted that the only additional monitoring needed with odevixibat is to determine response, and that no additional safety monitoring is needed. The committee noted that PFIC is currently managed in 3 specialist centres in England. The representative from NHS England confirmed that odevixibat would be started at specialist centres, with the potential to consider monitoring by local healthcare providers if safe and useful. The representative confirmed that additional infrastructure or staff training would not be needed to introduce odevixibat in England. The committee concluded that, if approved, odevixibat would be administered at specialist centres under the existing arrangements for people with PFIC.

The company stated that it considered odevixibat to be a step change in treating PFIC. This was because there are currently no licensed treatments for the condition, and current options have a high failure rate and can be invasive. The company highlighted that odevixibat is easy to administer in capsule form and can be sprinkled onto food for younger children. The clinical experts agreed that odevixibat was innovative because it is the first drug to both improve pruritus and limit progression of liver disease. They also flagged that the improvements in growth in people having odevixibat are important. The committee noted that odevixibat has a novel mechanism of action, no drug interactions and manageable side effects. It recalled that odevixibat was an oral drug that could remove the need for invasive PEBD and the trauma associated with a stoma bag. It also considered that surgical procedures such as a liver transplant and SBD were limited NHS resources that would be released if odevixibat were available. The committee recalled that there was high unmet need in this population. It also noted that odevixibat statistically significantly reduced pruritus and serum bile acid levels in the randomised controlled trial compared with standard care. The committee recognised that odevixibat was innovative.

The committee noted that the population for which odevixibat is indicated includes children and young people. It discussed the need to balance the importance of improving the lives of children and their families with fairness to people of all ages. It noted the principles that guide the development of NICE guidance and standards. This emphasises the importance of considering the distribution of health resources fairly within society as a whole, and factors other than relative costs and benefits alone. The committee acknowledged and considered the nature of the population as part of its decision making.