Elafibranor for previously treated primary biliary cholangitis

Primary biliary cholangitis (PBC) is a chronic, progressive autoimmune condition that leads to a build-up of bile in the liver. This happens because the body's immune system destroys bile ducts in the liver, causing cholestasis. This means that the flow of bile through the liver and biliary system is impaired or stalled. Over time, chronic cholestasis leads to scarring of the liver (fibrosis and cirrhosis) and liver failure, and can ultimately lead to death. Clinical experts highlighted that early treatment is crucial to prevent irreversible liver damage. The cause of PBC is not known, but it is thought to be a mix of environmental and genetic factors. PBC is typically diagnosed by testing for biochemical indicators of liver function (such as alkaline phosphatase). Many people do not have symptoms until they have significant liver damage. Common symptoms for those who have symptoms include fatigue and itchy skin (pruritus). Around 20,000 people in the UK have PBC, with an annual incidence of 2 to 3 per 100,000. Most people with PBC are women (90%) and over the age of 40 (75%). The patient experts described the challenges of living with PBC such as severe fatigue and severe itching. They emphasised that the chronic symptoms significantly impact the daily lives of people with PBC and their families and carers. The committee recognised that PBC is a progressive condition that significantly impacts the daily lives of people who have it. It also noted the poor prognosis for people who experience irreversible liver damage.

The clinical experts explained that the first-line treatment option for PBC is ursodeoxycholic acid (UDCA). People whose PBC has an inadequate response to UDCA, or people who are intolerant of it, have obeticholic acid (OCA) with or without UDCA as a second-line treatment, as recommended in NICE's technology appraisal guidance on OCA for treating primary biliary cholangitis. They could also continue to have UDCA monotherapy. The patient experts explained that the treatment options are limited for people who cannot have, or whose PBC has not responded to, UDCA or OCA. They highlighted other treatments they have had, including plasmapheresis, nasobiliary drainage, and treatments for itching (such as colestyramine). They explained that some treatments were very intensive, needed very frequent hospital visits, did not work as expected and had side effects. Some people also have limited access to specialist treatment centres because they live far away from them. The clinical experts outlined that PBC treatments aim to slow the progression to end-stage liver disease and to reduce the quality-of-life burden of symptoms. But there is a significant unmet need because for 40% of people, their PBC does not respond to UDCA and for over 30% of people, PBC does not respond to second-line treatments. The patient experts added that people may need a liver transplant if their PBC does not respond to current treatments, or if they cannot tolerate them. But their shared experience was that the waiting list for transplants is very long. They noted that a liver transplant may not treat the fatigue associated with PBC. Also, other symptoms, including itching, may recur after a transplant. The patient experts said that when this happened, it left them feeling helpless. The committee concluded that there was an unmet need for people who cannot tolerate UDCA, or whose PBC has an inadequate response to it. It heard from the clinical and patient experts that existing treatments are limited, do not work for some people, and can have considerable side effects.

The company explained that elafibranor would be used as a second-line treatment for people who cannot tolerate UDCA, or whose PBC has an inadequate response to it. This is the same line of treatment as OCA. The clinical experts highlighted that treatment with OCA may worsen itching in some people and 6% of people stop OCA because of severe itching. Many people with PBC who experience itching avoid treatment with OCA because of the risk of worse itching. The patient experts explained that the main benefit of elafibranor is reducing the burden of itching. The clinical experts agreed, noting that elafibranor is a treatment that does not worsen itching, unlike OCA for some people. The EAG suggested that elafibranor might be used as a third-line treatment too. It understood that there was no evidence for the effectiveness of elafibranor at this treatment line. But it noted the mechanism of action of elafibranor meant that third-line use may be possible. Elafibranor is a peroxisome proliferator-activated receptor (PPAR) agonist, combining the effects of PPAR-alpha and PPAR-delta activation. This is a different mechanism of action to UDCA and OCA. The committee concluded that there was enough evidence to consider elafibranor for second-line use, and that OCA with or without UDCA was an appropriate comparator for elafibranor. It concluded that any third-line use of elafibranor was uncertain.

The company base case did not include fibrates as a comparator for elafibranor. Bezafibrate was included in the company's model (see section 3.9 for more information about the model) for treating itching with UDCA and OCA, but not as a stand-alone second-line treatment. Submissions from professional organisations and NHS England identified fibrates as a potential comparator. The company explained that it did not include fibrates because they are used off-label and are not recommended by NICE. It added that fibrates have not been studied to regulatory standards, so there might be long-term safety concerns. The EAG referenced a UK audit that found that some people having second-line treatment for PBC had fibrates. But it noted these may have been used as an add-on treatment for itching, rather than to treat PBC. The clinical experts explained that fibrates are used in combination with second-line treatments to treat itching. They added that fibrates would not be widely used as a second-line treatment for PBC because of toxicity and limited evidence of efficacy. The committee concluded that fibrates were not used primarily to treat PBC and were not an appropriate comparator for elafibranor.

The main source of clinical-effectiveness evidence for elafibranor was the ELATIVE trial. This was a phase 3, randomised, placebo-controlled, double-blind trial. It investigated the efficacy of elafibranor compared with placebo in people aged 18 to 75 years who had PBC that had had an inadequate response to UDCA, or who were intolerant to it. The trial recruited 161 people and, of these, 108 had elafibranor and 53 had placebo. The mean age was 57 years and 96% of people in the trial were female. The trial was carried out in sites around the world, including in the UK. The primary outcome was cholestasis response at week 52. This was defined as:

• an alkaline phosphatase level of less than or equal to 1.67 times the upper limit of the normal range, with a reduction of 15% or more from baseline, and

• total bilirubin levels within the normal range.
  
  At week 52, 50.9% of people in the elafibranor arm had cholestasis response compared with 3.8% for placebo (p<0.0001) in the intention-to-treat population. The committee concluded that ELATIVE was relevant for evaluating elafibranor for treating PBC.

The company did a network meta-analysis (NMA) to generate comparative evidence for elafibranor and OCA. It included the ELATIVE trial data and data from the POISE trial. The POISE trial was similar to the ELATIVE trial but investigated OCA at a 5 mg to 10 mg dose and a 10 mg dose. The company used a random-effects model in its base case, but also used a fixed-effects model in a scenario analysis. The outcomes included in the NMA were the:

• odds of cholestasis response at 12 months

• mean change in alkaline phosphatase levels from baseline at 12 months

• odds of alkaline phosphatase normalisation at 12 months

• mean change in pruritus from baseline at 12 months, according to the 5‑D itch questionnaire score, using the earliest data after starting treatment

• mean change in the PBC‑40 questionnaire score (itch domain) from baseline at 12 months, using the earliest data after starting treatment

• odds of a pruritus treatment-emergent adverse event (any severity) in 12 months

• odds of all-cause discontinuation in 12 months.

• mean change in high-density lipoprotein (HDL)‑cholesterol from baseline at 12 months.
  
  The EAG raised several concerns with the methodology of the NMA, which included the:

• use of odds ratios in the NMA instead of risk ratios (this was because odds ratios tend to overestimate effects with a link between exposure and outcome, if interpreted as risk ratios)

• statistical methods used

• studies that were excluded from the analysis

• transitivity assumption.
  
  The committee concluded that the company's NMA methodology was subject to some limitations. But, overall it was suitable for decision making.

The results of the NMA cannot be reported here because they are confidential. The EAG highlighted that the credible intervals for all outcomes used in the company's economic model were very wide. The company explained that there were issues with convergence in the random-effects model. But it noted that the results were similar with the fixed-effects model. The EAG explored additional analyses with different assumptions and agreed with the company that different approaches to the NMA did not change the overall conclusions. It noted that elafibranor and OCA had statistically significant treatment effects compared with placebo, but there was substantial uncertainty in the size of effects. When they were compared in the NMA, the uncertainty in the treatment effect size for both treatments led to very wide credible intervals. The EAG explained that the results showed that there was substantial underlying uncertainty about the effectiveness of elafibranor compared with OCA. The committee noted that the very wide credible intervals meant that there may be no difference in treatment effect between elafibranor and OCA. It noted that the cost-effectiveness results showed that more quality-adjusted life years (QALYs) were gained with elafibranor compared with OCA, and asked the company if this was plausible. The company explained that the point estimates from the NMA showed a large difference in treatment effect for elafibranor compared with OCA. And it was these point estimates that were used in the modelling. The company added that the credible intervals were wide because of small sample sizes, because PBC is a rare condition. Also, very few events were recorded in the placebo arm. But it added that the uncertainty was mitigated because additional analyses showed that the NMA point estimates were stable. The company considered that the incremental cost-effectiveness ratio (ICER) was also stable because the probabilistic ICERs (which included the wide credible intervals) were similar to the deterministic ICERs.

The clinical experts noted that the NMA may have underestimated the effect of elafibranor in reducing the odds of itching compared with OCA. This was because clinicians were aware that OCA may cause itching in some people, so people already with severe itching would not have entered the POISE trial. But people already with severe itching would have entered the ELATIVE trial because of the expected reduced itching burden. Also, people who had already had second-line treatments such as OCA were included in the ELATIVE trial. So, it may be expected that these people would have worse outcomes for itching than those having second-line elafibranor. The clinical experts also highlighted that the most meaningful measure of treatment effect for clinicians is alkaline phosphatase normalisation. They expected that an NMA on this measure alone would show a less uncertain treatment effect for elafibranor compared with OCA, because a clear benefit is seen in clinical practice. The committee noted that the company had provided an analysis in which alkaline phosphatase normalisation was used as the measure of treatment effectiveness and that this increased the ICER. The committee concluded that the NMA had limitations but was suitable for decision making, noting that the alternative analyses supported the conclusions. But it considered that the very wide credible intervals showed significant uncertainty in the treatment effectiveness of elafibranor compared with OCA.

The committee noted that the key measure of response in ELATIVE was normalisation of alkaline phosphatase and bilirubin, which are biochemical markers of liver function (see section 3.5). It recalled the clinical expert's testimony that the most meaningful measure of treatment effect for clinicians is alkaline phosphatase normalisation (see section 3.7). It also recalled the European Medicines Agency's (EMA's) recent concerns about the conditional marketing authorisation for OCA. The conditional marketing authorisation was initially granted because OCA was shown to reduce levels of alkaline phosphatase and bilirubin, which was considered indicative of an improvement in liver condition. But, the EMA recently considered that the effectiveness of OCA had not been proven with longer-term data from the phase 4 confirmatory COBALT trial. The committee asked the clinical experts whether this indicated that alkaline phosphatase normalisation is not an appropriate surrogate outcome, and whether this was important to consider for elafibranor. The clinical experts explained that alkaline phosphatase normalisation is known to be associated with better outcomes for people with PBC. They reiterated that it is the most meaningful measure for clinicians in assessing PBC changes. They considered that the negative results of the COBALT trial were because of limitations in trial recruitment, unblinding and treatment switching, rather than limitations of alkaline phosphatase normalisation as an outcome measure. The committee acknowledged that the ELATIVE trial results included surrogate outcomes used to monitor PBC progression. Also, that there was uncertainty about the relationship between these measures and long-term outcomes. But it also acknowledged that there were potential limitations of the COBALT trial. The committee concluded that there were uncertainties in defining cholestasis response that were important for it to consider in its decision making.

The company developed a Markov model to estimate the cost effectiveness of elafibranor. The company explained that the model was consistent with the model used in NICE's technology appraisal guidance on OCA for treating primary biliary cholangitis. The modelled health states covered 2 distinct components: PBC biomarkers and liver disease. In the model, people could move between mild, moderate and high-risk biomarker states. These levels corresponded to the risk of progressing to liver disease. People whose PBC responded to treatment, while on treatment, had a higher probability of being in a lower risk biomarker state. When a person stopped treatment, they returned to their original state before second-line treatment. Only people in moderate and high-risk biomarker states could move to liver disease states. Liver disease health states included hepatocellular carcinoma, decompensated cirrhosis, pre liver transplant, liver transplant, post liver transplant and PBC re-emergence. Once in the pre-transplant state, people could move to having a liver transplant and post-transplant states. People could also get PBC re-emergence from these 2 states and transition to death from any health state. The committee concluded that the model accurately reflected the PBC pathway and was suitable for decision making.

The EAG highlighted concerns with liver-disease-free survival and overall survival predicted by the company's model. It highlighted that liver-disease-free survival had likely been underpredicted in the company model and that overall survival predictions had not been validated by experts or literature. The EAG compared liver disease risk predictions over 5, 10 and 15 years from the ELATIVE trial and the company model. It showed that the model estimated lower transplant-free survival than the trial. The exact figures are confidential and cannot be reported here. The EAG explained that the underprediction could be because of the following features of the model:

• people could transition from the moderate-risk PBC biomarker state to liver disease, unlike in the model used for NICE's technology appraisal guidance on OCA for treating primary biliary cholangitis

• the increase in mortality from the general population mortality for high-risk PBC biomarker states may have been too high and miscalculated (the EAG changed this in its base case)

• the immediate increase in PBC biomarker risk after treatment discontinuation

• people having UDCA could not move from moderate to low-risk states after 12 months

• the uncertainty in long-term transition probabilities.
  
  The EAG also highlighted uncertainty with overall survival predictions by comparing median survival predictions in the model with estimates from the literature. The model predicted a median survival for hepatocellular carcinoma of 1.5 years, and 4 years for decompensated cirrhosis. This was compared with survival estimates from the literature for hepatocellular carcinoma after 5 years that ranged from 43% to 69%. The EAG suggested that the uncertainty could be reduced if the modelled survival predictions were validated by clinical experts and external literature. The EAG base case changed how the increase in mortality from the general population mortality was applied for high-risk PBC biomarker states. The company base case added the 1.2% excess mortality to the 2% general population mortality for a total of 3.2% mortality. The EAG base case applied the increase multiplicatively instead, for a total mortality rate of 2.02%. The committee noted the modelled survival benefit of elafibranor and asked the clinical experts if it was plausible. The clinical experts explained that a survival benefit of elafibranor was plausible and expected. This is because elafibranor leads to superior alkaline phosphatase normalisation compared with OCA, which in turn is expected to lead to better survival. The committee concluded that the modelled survival predictions for elafibranor and OCA were plausible but uncertain. This was because of underlying uncertainty with the NMA results (see section 3.7) and surrogate outcomes (see section 3.8), and also because the survival predictions had not been validated. The committee accepted the EAG's approach to calculating mortality for the high-risk PBC biomarker state.

OCA can be taken for a person's lifetime and the company's model included treatment discontinuation for OCA over the full-model time horizon. The company estimated OCA discontinuation by applying the 12‑month risk ratio for discontinuation from the NMA (see section 3.6) to elafibranor discontinuation from ELATIVE. The company originally extrapolated the treatment discontinuation for OCA beyond 12 months by assuming an exponential distribution. But the EAG raised that this assumed a constant rate of discontinuation, which did not agree with the clinical experts' opinion. The clinical experts explained that discontinuation mostly occurs early in the treatment course for OCA. So, the company updated the extrapolation distribution to a lognormal distribution. It also assumed a lifetime difference in discontinuation between elafibranor and OCA. The EAG commented that the approach to discontinuation was different from the approach used in NICE's technology appraisal guidance on OCA for treating primary biliary cholangitis. In that model, treatment discontinuation was only considered for the first year of treatment. The EAG noted that the proportion of people still on treatment with OCA at 5 years in the real-world UK data was higher than the estimate from the company's model. The exact figures are confidential and cannot be reported here. The clinical experts acknowledged that the real-world UK data was in line with their experiences, but people add other treatments instead of stopping altogether. The EAG base case also used a lognormal distribution for treatment discontinuation but only assumed a 1‑year difference in discontinuation between elafibranor and OCA. The EAG explained that people who stopped OCA plus UDCA in the model had UDCA alone. The EAG added that this costs less and leads to worse clinical outcomes, but the reduced costs outweighed the reduced outcomes. So, assuming greater discontinuation for OCA made elafibranor less cost effective compared with OCA. The committee concluded that there was uncertainty in the treatment discontinuation for OCA. But it noted that using either the company's or the EAG's preferred approach had a very small impact on the cost-effectiveness results.

The ELATIVE trial collected EQ‑5D‑5L quality-of-life data, but the data was not used in the company model for the PBC biomarker-risk health state. The company explained that a linear mixed-effects model using the ELATIVE trial data led to a utility decrement between moderate and high-risk states that was lower than expected. It suggested this was because of small sample sizes of people with high-risk PBC biomarkers. The company explained that it considered the quality-of-life results from ELATIVE unreliable, so it used values from the literature instead. The values used in the company base case were 0.84 for mild and moderate risk and 0.55 for high risk. The value for mild and moderate risk was for cholestatic disease from Younossi et al. (2000). The value for high risk was for compensated cirrhosis taken from NICE's technology appraisal guidance on sofosbuvir for treating chronic hepatitis C. The utility values from ELATIVE are confidential and cannot be reported here. The EAG agreed with the company that the small sample size added uncertainty, but it still considered the data informative. So, it used the utility values from ELATIVE in its scenario analyses. The EAG base case used more recent utility values from the literature for the high-risk PBC biomarker state. It used 0.717 for compensated cirrhosis from Saeed et al. (2020). The committee concluded that there was uncertainty in the utility values, but it noted that this had a very small impact on the cost-effectiveness results.

The EAG's base case included the following changes to the company's base case:

• 13 minor errors were fixed

• a constant hazard ratio was used for OCA outcomes instead of a risk ratio

• the pre liver transplant health state in the model was excluded

• the excess mortality for the high-risk PBC biomarker state was added multiplicatively instead of additively (see section 3.10)

• a difference in treatment discontinuation between OCA and elafibranor maintained for 1 year was used instead of lifetime use (see section 3.11)

• a high-risk state utility value of 0.717 was used instead of 0.55 (see section 3.12)

• itching was removed as an adverse event in the model because it was captured by the PBC‑40 itch domain outcome

• a compliance rate of 93.6% was used for OCA and elafibranor.
  
  The cost-effectiveness results included confidential discounts for comparator treatments so the exact ICERs cannot be reported here. The company's probabilistic base case ICER (with errors fixed by the EAG) and the EAG's base case ICER were within the range that NICE normally considers a cost-effective use of NHS resources. The committee considered that the most plausible ICER was aligned with the EAG base case but it was subject to considerable uncertainty. It reviewed the incremental cost-effectiveness plane and cost-effectiveness acceptability curve for the EAG base case and noted that they portrayed uncertainty in the results. The spread of probabilistic cost-effectiveness results was very wide and covered all 4 quadrants of the plane.

NICE's manual on health technology evaluations notes that, above a most plausible ICER of £20,000 per 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. But it will also take into account other aspects including uncaptured health benefits. The committee was mindful that PBC is a rare condition, which can make evidence generation particularly difficult. It noted the very high level of uncertainty, specifically that:

• the NMA results included very wide credible intervals that crossed the threshold of no effect (see section 3.7)

• the relationship between alkaline phosphatase and bilirubin normalisation and outcomes for people with PBC was uncertain (see section 3.8)

• the modelled survival predictions had not been validated by clinical experts or external literature (see section 3.10)

• the long-term efficacy of OCA and elafibranor was unknown (see section 3.8)

• redefining treatment effectiveness as alkaline phosphatase normalisation alone led to a considerably higher ICER (see section 3.7)

• the distribution of individual probabilistic ICERs on the incremental cost-effectiveness plane was very wide and covered all 4 quadrants of the plane (see section 3.13).
  
  So, the committee concluded that an acceptable ICER would be around £20,000 per QALY.

Submissions from the company, patient and professional organisations and clinical experts identified potential equality issues for consideration. The issues identified were:

• Approximately 90% of people with PBC globally are women and the incidence rate is 5 to 6 times higher in women than men.

• People who are diagnosed with PBC under the age of 50 experience more severe and progressive PBC and poorer treatment response than people who are aged 50 and over at diagnosis.

• Men are at greater risk of more advanced PBC at diagnosis and poorer treatment response compared with women.

• People with PBC are the most likely to die out of all people waiting for liver transplants because of priority given to others.

• Some evidence suggests smoking, nail polish, hair dyes, hormone replacement and toxic waste are linked to PBC.
  
  The committee considered the concerns raised about access to liver transplants. It understood these concerns but noted that they were outside of its remit. The committee noted the possibility of different treatment outcomes for some groups of people. Age and sex are protected characteristics under the Equality Act 2010. But because its recommendation does not restrict access to treatment for some people over others, the committee agreed that these were not potential equality issues.

The committee recalled the high uncertainty associated with the company's model and long-term outcomes for elafibranor, and that the EAG's and company's base cases were associated with uncertainty. But it noted that the most plausible cost-effectiveness estimates were below the committee's acceptable ICER, even when taking into account the high uncertainty. So, it recommended elafibranor as an option for treating PBC.