Tafamidis for treating transthyretin amyloidosis with cardiomyopathy

3.1 There are 2 types of transthyretin amyloidosis with cardiomyopathy (ATTR‑CM):

3.2 The patient experts explained that ATTR-CM significantly affected their physical ability. For example, one noted that walking even short distances could be challenging, while another stated that they were no longer able to do their daily physical activities. Another patient expert explained that ATTR-CM made them feel older than they should and limited them to mostly staying seated. The patient experts noted that ATTR-CM also affected psychological wellbeing, for example, symptoms of breathlessness leading to anxiety. Loss of independence and an increased reliance on caregivers was also highlighted as a cause of depression related to the condition. The patient experts explained that psychological effects can be exacerbated for people with hereditary ATTR‑CM because it can affect multiple members of a family across different generations, and there was anxiety about passing it on to children. The committee concluded that ATTR-CM is a debilitating disease which significantly affects mental and physical wellbeing.

3.3 The patient experts explained that getting an accurate diagnosis for ATTR-CM could be challenging. They noted that awareness of the condition, and the type of ATTR-CM a person has (see section 3.1), can vary and lead to delays to diagnosis. One patient expert explained that they waited more than 4 years for an accurate diagnosis, highlighting a general lack of understanding about the condition outside of specialist centres. The company highlighted National Amyloidosis Centre (NAC) data showing that, on average, it took 3 years or more for a person to be accurately diagnosed with ATTR-CM. Two of the clinical experts agreed that there were challenges in diagnosing ATTR‑CM accurately, but noted the developments in recent years. Radionuclide imaging (DPD scan), a test usually used to detect bone abnormalities, is increasingly being used to diagnose ATTR-CM. It is very sensitive to detecting amyloid deposits in the heart. This has improved awareness of ATTR‑CM and increased diagnoses. Two of the clinical experts highlighted that the low cost of DPD scanning meant most hospitals would be able to do them. So, they expected further increases in ATTR‑CM diagnoses. They also noted increased awareness of ATTR‑CM after NICE's highly specialised technologies guidance recommended inotersen and patisiran for treating hereditary transthyretin amyloidosis with polyneuropathy (see section 3.5). The committee concluded that the availability of new diagnostic tests and treatments for the disease had improved awareness of ATTR-CM, but recognised that diagnosis can still take a long time.

3.4 Two of the clinical experts explained that transthyretin amyloid deposits are often an incidental finding in people having DPD scans and may not necessarily be associated with a diagnosis of ATTR-CM. They explained that the population they see in practice has a range of amyloid deposits, sometimes because of older age, for example. Also, there is no defined point at which amyloid deposits become amyloidosis. So, it is unclear why some amyloid deposits progress to amyloidosis and others do not. In addition, because other common comorbidities can lead to increased breathlessness and decreased mobility, reaching a definitive ATTR-CM diagnosis is challenging. At consultation, the company commented that the NAC published a non-invasive diagnostic pathway for ATTR-CM in 2016. It explained that the pathway had been validated and implemented at 17 early access to medicines scheme (EAMS) sites. It further explained that based on this pathway, only people with symptoms and certain clinical features are eligible for DPD scans. So, amyloid deposits identified through the pathway are not incidental. It highlighted that, because of this, amyloid deposits found when doing DPD scans for another condition would not result in a diagnosis of amyloidosis. At the second committee meeting one of the clinical experts confirmed that incidental amyloid deposits would not result in an increase in amyloidosis diagnoses, because other clinical features outlined in the ATTR-CM diagnostic pathway would be taken into account. The committee questioned whether everybody with heart failure and an abnormal DPD scan would be diagnosed with amyloidosis. One clinical expert explained that not everyone with heart failure would have a DPD scan, and only those with cardiac symptoms specific to ATTR-CM (such as thickening of the heart muscle) would be tested. They further explained that people diagnosed with ATTR-CM and having treatment in clinical practice would likely reflect the population from the ATTR-ACT pivotal trial (see section 3.9) although in the trial a biopsy was required. The clinical experts and the NHS England representative explained that when amyloidosis is suspected people are referred to the NAC for more rigorous testing. The committee was aware that the European public assessment report for tafamidis states that there are difficulties in diagnosing people with ATTR‑CM in NHYA class 1, particularly if they do not have heart failure. It also states that an accurate diagnosis cannot be formally established without a number of procedures (such as biopsy and scintigraphy). The committee acknowledged this and agreed that even with DPD scans and a diagnostic pathway, there would still be challenges in diagnosing ATTR‑CM. The committee concluded that it was unclear if the increased availability of DPD scans would lead to an increase in diagnosing amyloidosis or ATTR‑CM.

3.5 Treatment options for ATTR-CM are limited to managing symptoms and supportive care, such as diuretics. A small proportion of people with ATTR-CM also have polyneuropathy (mixed clinical features). NICE has recommended 2 treatments for polyneuropathy:

3.6 The NYHA functional classification system is commonly used in clinical practice to assess heart failure, and is sometimes used to measure the severity of ATTR‑CM. It groups symptoms into 1 of 4 classes depending on how limited a person's physical activity is. A person whose disease was classed as NYHA 1 would be able to do ordinary physical activity. But someone whose disease was classed as NYHA 4 would be unable to do physical activity without feeling discomfort. The clinical experts explained that although NYHA classification is used in clinical practice, it has limitations. Because it is a patient-reported measure it can vary from day to day, and there can be inconsistencies in people's symptoms. For example, people with the same activity tolerance may classify their level of heart failure differently. One clinical expert noted that variability in how people classify themselves was a common issue with all symptom-based measures. They also highlighted that the NYHA classification system is used as an entry criterion in most heart failure trials. The clinical experts commented that it was difficult to identify whether movement between the NYHA classes was a result of ATTR-CM progressing or changes in other comorbidities. One of the clinical experts suggested that a measure based on cardiac markers such as B-type natriuretic peptide and glomerular filtration rate had potential to identify disease stage and who is benefiting from treatment, but evidence of its use in ATTR-CM in clinical practice was limited. The ERG explained that there were merits to assessing disease severity using objective measures such as cardiac markers. But it commented that in the ATTR-ACT pivotal trial (see section 3.9), unlike the NYHA classification which was measured every 6 months, these cardiac markers were only measured at baseline, month 12, and after stopping treatment. So, they were not measured frequently enough to accurately characterise the disease. The committee concluded that using the NYHA classification in ATTR-CM had limitations, but acknowledged that there was insufficient trial evidence available to consider an alternative. This was because cardiac markers, which could have been used to identify disease stage and who would benefit from treatment, were not measured frequently enough in the trial.

3.7 The marketing authorisation for tafamidis does not specify starting and stopping rules for tafamidis based on the NYHA classification system. The company highlighted that NYHA classifications have been incorporated in previous NICE recommendations to define populations eligible for treatment with heart failure therapies. The committee noted that the marketing authorisation states that tafamidis should be 'started as early as possible in the disease course when the clinical benefit on disease progression could be more evident. Conversely, when amyloid-related cardiac damage is more advanced, such as in NYHA class 3, the decision to start or maintain treatment should be taken at the discretion of a physician knowledgeable in the management of patients with amyloidosis or cardiomyopathy'. The committee recalled that NYHA class 1 means that people can do ordinary physical activity (see section 3.6). It considered if tafamidis would be used for people who are easily able to do the activities of daily living (no functional limitations). The clinical experts explained that they would have reservations about offering treatment to people whose disease is classed as NYHA 1 because they have no functional limitations and might not benefit from treatment. At consultation, the company highlighted that this contradicted tafamidis' marketing authorisation, which states that treatment should be started as soon as possible. The company proposed a stopping rule in which people would stop tafamidis if their disease progressed to NYHA class 4. It explained that there was limited evidence to support using tafamidis in people whose disease was NYHA class 4, who had severe heart failure symptoms, because they were excluded from the ATTR-ACT pivotal trial. Also, the company highlighted that its proposed stopping rule reflected treatment stopping in ATTR-ACT, in which most people stopped tafamidis quickly after progressing to NYHA class 4. It also noted that because tafamidis does not improve symptoms caused by ATTR-CM it would be clinically appropriate to stop treatment when a person's disease is classed as NYHA 4. A clinical expert noted that stopping treatments when the disease progresses to NYHA class 4 was common because at this stage people are very unwell. They explained that people would be unable to travel for treatment, so treatment would likely be stopped and best supportive care offered. Comments from the patient organisation supported this view, stating that making decisions about stopping treatment in advanced disease stages were not uncommon. Conversely, 2 of the clinical experts noted that it would be challenging to stop treatment when disease progressed to NYHA class 4 because no alternative treatments were available. These 2 clinical experts also explained that people's disease often varies between NYHA class 3 and 4 and that this was typical of ATTR-CM. They noted that some people whose disease was classed as NYHA 4 could improve, so could change to NYHA class 3 or better. The ERG noted that improvements shown by changes in NYHA class were also seen in ATTR-ACT. The committee recalled that the company's proposed stopping rule was not specified in tafamidis' marketing authorisation. It agreed, that on balance, it would be difficult for clinicians to implement a stopping rule for tafamidis. This was because the disease can often vary between NYHA class 3 and 4 and the lack of alternative treatments for NYHA class 4 disease meant people would likely prefer to keep taking tafamidis. The committee concluded that using the NYHA classification alone to accurately define the population who were eligible to have tafamidis had limitations. So, it also concluded that it would not consider starting and stopping rules for tafamidis based on the NYHA classification system in its decision making.

3.8 At technical engagement, the company highlighted that introducing tafamidis reduced delays to ATTR-CM diagnoses. It noted that the availability of tafamidis would result in ATTR-CM being detected earlier because of more awareness among cardiologists. The company also noted that since tafamidis became available in the EAMS, the proportion of people whose disease was diagnosed in NYHA class 1 or 2 has increased from 75% to 86%. The committee understood that the short-term observational EAMS data were presented to support the assumption that introducing tafamidis could reduce diagnosis delays. It acknowledged that although the EAMS data were informative, they can only show that diagnosis delays were reducing when tafamidis was available. They cannot show that the delays were reducing because of tafamidis. A statement from NHS England suggested that if NICE recommended tafamidis, and awareness was increased through educational campaigns, diagnosis rates may improve further. The committee recalled that average time to diagnosis at the NAC was 3 years or more (see section 3.3). The company noted that 1 in 3 patients were diagnosed in less than 6 months, while 40% waited for 4 years or more. The committee acknowledged that reducing the average time to diagnosis to less than 6 months would represent a substantial improvement. But, it recognised these diagnoses were made at a specialist centre and questioned if such reductions could be achieved at other centres in clinical practice. The ERG highlighted that the trend of earlier diagnosis seen in the EAMS could be explained by improvements in diagnostic tools since the ATTR-ACT pivotal trial (see section 3.3). It noted that when ATTR-CM is suspected the diagnostic pathway may lead to quicker diagnoses, but when it is not suspected, substantial diagnosis delays may still occur. The committee acknowledged this, and agreed it was not possible to attribute reductions in diagnosis times at EAMS sites to the availability of tafamidis. It also noted web comments at consultation which highlighted evidence of an increasing number of ATTR-CM diagnoses at the NAC before tafamidis was available through EAMS. There were also comments from the EAMS centres' cardiac group that a major contributor to the shorter diagnosis times was the changes in diagnostic tools and algorithms. The committee also recalled that awareness of ATTR-CM had increased (see section 3.3) and questioned whether recommending tafamidis would further increase awareness and reduce diagnosis times. The committee concluded that there was not enough evidence provided to support the assumption that introducing tafamidis would reduce ATTR-CM diagnosis delays.

3.9 The clinical evidence came from 2 trials:

3.10 The primary analysis from the ATTR-ACT pivotal trial compared the results of a pooled tafamidis (20 mg and 80 mg doses) treatment group with the placebo group. The primary outcome, a combined measure of all-cause mortality and cardiovascular-related hospitalisations, was assessed in a hierarchical analysis using the Finkelstein-Schoenfeld method. At month 30, 186 people (70.5%) were alive in the tafamidis group compared with 101 people (57.1%) in the placebo group. Of those alive at month 30, people who had tafamidis had fewer annual cardiovascular-related hospitalisations (0.297) on average than those who had placebo (0.455). Tafamidis statistically significantly reduced all-cause mortality and frequency of cardiovascular-related hospitalisations compared with placebo. The committee considered that although the parts of the primary outcome were clinically relevant to patients and clinicians, it questioned whether the combined measure would be considered in clinical practice. It also considered the secondary outcome results from ATTR-ACT and noted that at month 30 compared with placebo, tafamidis was associated with statistically significant reductions in:

3.11 The committee considered the predefined subgroup analyses from the ATTR‑ACT pivotal trial, specifically those examining the effectiveness and safety of tafamidis in:

3.12 Quality of life was measured in the ATTR-ACT pivotal trial using 3 scales: the Kansas City Cardiomyopathy Questionnaire (KCCQ), EQ‑5D‑3L and EQ‑5D visual analogue scale. The company explained that the KCCQ is a valid and reliable measure of health status for people with heart failure. It measures physical function, symptoms (frequency and severity), social function and quality-of-life domains, and calculates an overall summary score with lower scores showing worse impairment. The committee noted that the KCCQ overall summary score results from ATTR-ACT showed that from baseline to month 30, people taking tafamidis had a slower decline in quality of life than people taking placebo (least squares mean difference compared with placebo, 13.65 [p<0.0001]). The committee also noted the results measured by the EQ‑5D‑3L and EQ‑5D visual analogue score (these are considered confidential by the company and cannot be reported here). The committee concluded that compared with placebo, tafamidis slowed the decline in quality of life for people with ATTR-CM.

3.13 Most of the adverse events of treatment seen in the ATTR-ACT pivotal trial were mild to moderate in severity, with fewer in the tafamidis treatment groups. The company highlighted that the proportion of people reporting serious events was higher in the placebo group. The committee concluded that tafamidis was generally safe and well tolerated.

3.14 The company modelled the costs and benefits for tafamidis using a cohort-level Markov state‑transition model. To capture the natural disease progression of ATTR-CM, model health states were based on the NYHA classification system (see section 3.6). The model included 5 health states, 4 defined by NYHA classes (1, 2, 3, and 4) and 5 being death. People could move to a more severe health state (decline) or to a less severe one (improve). The company explained that because NYHA classification captured aspects of functional limitation and symptom severity it was suitable to model changes in ATTR-CM. It also highlighted that NYHA classification predicted health‑related quality of life and survival well, and it had been widely used in cost‑effectiveness models. The committee recalled its concerns about the NYHA classification system (see section 3.6), but concluded that because there was no available alternative the company's model could be considered for decision making.

3.15 Both the company's and ERG's analyses after technical engagement included a stopping rule for tafamidis, which assumed that people would stop treatment if their disease progressed to NYHA class 4. The committee noted the limitations of using the NYHA classification system in clinical practice and the lack of evidence about tafamidis' effectiveness beyond NYHA class 1 and 2 (see sections 3.6, 3.7 and 3.11). The committee recalled that the NYHA classification is widely used to define populations eligible for heart failure treatments and has been used in previous NICE recommendations (see section 3.7). But, it noted that although the tafamidis marketing authorisation states that there were limited clinical data in patients whose disease was classed as NYHA 4, it did not specify that it should be stopped (see section 3.7). The committee also recalled that it would be difficult for clinicians to implement a stopping rule for tafamidis. This was because the disease can often vary between NYHA class 3 and 4 and the lack of alternative treatments for NYHA class 4 disease meant people would likely prefer to keep taking tafamidis (see section 3.7). The committee was aware that when considering treatment continuation rules, it needed to take additional factors into account as well as the cost-effectiveness estimates based on continuing treatment only in those whose disease had a specified response (see NICE's guide to the methods of technology appraisal). These factors include:

3.16 After technical engagement, the company included a treatment stopping rule for people in the NYHA class 1 to 3 health states. During technical engagement, a clinical expert explained that it was unlikely people would stop tafamidis in NYHA class 1, 2, or 3. This was because it was unclear when a person's disease changes from one NYHA class to another, and that people usually prefer to remain on treatment if there are no alternatives. Also, because tafamidis is well tolerated and easy to take, a high rate of adherence would be expected. After technical engagement, the company acknowledged this, but included a treatment stopping rule for people in the NYHA class 1 to 3 health states in its revised analysis. The ERG explained that in the economic model, people in the NYHA class 1 to 3 health states who stop treatment with tafamidis are assumed to benefit from treatment indefinitely without any treatment costs. The clinical experts who responded to technical engagement suggested it was unreasonable to assume that treatment benefits would be maintained indefinitely after treatment stops. A clinical expert at the second committee meeting also noted that no evidence was available to support such an assumption. The company noted that the relative treatment effect for tafamidis was estimated from the ATTR-ACT pivotal trial and incorporated the treatment effects for people who remained on treatment and those who stopped. So the effect of stopping treatment is already accounted for in its treatment effectiveness estimates. It also suggested that it is clinically plausible for benefits to continue after treatment stops because tafamidis' mechanism of action reduces cumulative amyloid exposure. So, the longer a person is on treatment, the more their cumulative exposure is reduced. Research group comments received at consultation suggested that ATTR-ACT had not revealed anything about how tafamidis works. The group stated that a recent publication had shown that disease stabilisation does not necessarily inhibit amyloid formation, so the mechanism underlying tafamidis' proposed benefit is unclear. The ERG reiterated that the limitations of the company's approach were acknowledged at technical engagement. It highlighted that the company failed to make use of longer-term data in its extrapolation of treatment stopping. The committee concluded that assuming continued treatment benefits without a cost was overly optimistic and would lead to an underestimated incremental cost‑effectiveness ratio (ICER).

3.17 The ERG presented 2 alternative analyses that used different assumptions about continued treatment benefits in the NYHA class 1 to 3 health states:

3.18 After technical engagement the company modelled overall survival beyond the observed trial period, using generalised gamma extrapolation functions for tafamidis and best supportive care. The ERG explained that the company had changed the extrapolation function it used from log‑normal to generalised gamma, which had more favourable results for tafamidis. The ERG highlighted that although the company acknowledged that the extrapolations based on generalised gamma were optimistic, it had not given a reason for revising this aspect of its analysis at technical engagement (for example, statistical goodness‑of‑fit or external clinical validation). At consultation the company changed the extrapolation to reflect a new ATTR-ACT extension study data cut, which suggested that survival benefits increased the longer people were on tafamidis. It explained that because the estimated survival benefits from the ATTR‑ACT extension study included people who had swapped from placebo to tafamidis, mortality benefits were likely to be underestimated. So, because of this, and because the 2 extrapolation functions were similar in statistical and visual fit, it suggested it was appropriate to use the more optimistic of the two. The ERG noted that according to the Bayesian information criterion the generalised gamma extrapolation function used by the company was the worst fit. The committee concluded that modelling overall survival using generalised gamma extrapolation functions was not fully justified. It agreed to consider only the log‑normal extrapolation functions in its decision making.

3.19 After technical engagement, the company provided analyses that used 3 new assumptions about early ATTR-CM diagnosis. It assumed that introducing tafamidis:

3.20 After technical engagement, the ERG included potential tafamidis drug wastage in its estimate of costs. The committee considered if drug wastage would be substantial for tafamidis. The NHS England representative explained that they did not have any data on tafamidis wastage, so did not know whether it would be significant or not. The committee agreed with including drug wastage in the analysis and acknowledged that including it had little effect on the cost‑effectiveness estimates. So, the committee concluded that it was appropriate to include drug wastage costs in its preferred analysis because some wastage was likely to happen in clinical practice.

3.21 The company derived health state utility values using EQ‑5D‑3L data from the ATTR-ACT pivotal trial. The committee recalled that the company had also collected quality-of-life data using the KCCQ measure, which included function domains (see section 3 12). However, it agreed that the EQ‑5D‑3L data were more suitable for the economic model because they were in line with the reference case. The committee recalled its concerns about using health states based on NYHA classes in the model, but agreed that they could be considered because there was no available alternative (see section 3.14). The committee concluded that the company's health state utility values were appropriate for decision making.

3.22 Before technical engagement, the company's analysis did not adjust health state utility values to account for the effect of increasing age. The ERG highlighted that this meant utility values for tafamidis and best supportive care were better than for the age-equivalent general population. The clinical experts involved in technical engagement explained that it was not plausible that someone with ATTR-CM could have a better quality of life than someone of a similar age and sex from the general population. The company adjusted its utility values to take account of increasing age after the observed trial period in its revised analyses after technical engagement. The committee concluded that using age-adjusted utility values was appropriate.

3.23 The company estimated health state utility values separately for each NYHA class (see section 3.21) and treatment included in the model. It explained that different health state utility values between tafamidis and best supportive care may reflect differences in hospitalisations and adverse events associated with each treatment. The committee recalled that the NYHA classification system was unlikely to be sensitive to changes in ATTR-CM (see section 3.6). The ERG noted that the company modelled substantially different on- and off-treatment utility values in the NYHA class 4 health state. It also explained that estimates of NYHA class 4 utility values were based on very few observations. The company highlighted that the health state utility values were derived from EQ‑5D‑3L data from the ATTR-ACT pivotal trial and were the most appropriate data for the economic analysis. The ERG noted that in ATTR-ACT quality-of-life data were collected only during the on-treatment period, and that in the trial, most people stopped treatment before their disease progressed to NYHA class 4. The ERG explained that the estimated NYHA class 4 utility value for tafamidis could be affected by informative censoring, because the quality of life of anyone who stopped tafamidis in NYHA class 4 was not captured. To account for this, the ERG's analysis after technical engagement assumed that the estimated best supportive care utility value applied to everyone in the NYHA class 4 health state. After technical engagement the company accepted that it was appropriate to apply the best supportive care utility value in NYHA class 4 and it used this assumption in its revised analysis. The committee agreed that it had concerns about using treatment-dependent health state utility values from relatively few observations and the potential for informative censoring to bias these estimates. It concluded that the treatment-dependent utility values were reasonable in NYHA class 1 to 3, and that the best supportive care utility value should be applied in the NYHA class 4 health state.

3.24 The company's preferred analysis estimated that the ICER for tafamidis compared with best supportive care, in the full population, was less than £30,000 per QALY gained (including the company's confidential commercial arrangement). It included the following assumptions:

3.25 The most common transthyretin variants associated with hereditary ATTR-CM are Val122Ile, which is common in people of African and Caribbean family origin, and Thr60Ala, which is common in people with Irish ancestry. The committee acknowledged that ATTR-CM disproportionally affected people from certain ethnic groups, but agreed this was not something that could be addressed in its recommendation. A statement received at consultation noted that the recommendation would deny older people with ATTR-CM access to tafamidis, which could maintain their health and reduce morbidity, while those in the EAMS could have tafamidis. The committee was also aware that evidence suggested ATTR-CM prevalence in women may be underestimated, because women may be less likely to have the red flag symptoms that trigger referral. The committee considered that its recommendation did not disproportionately disadvantage women or certain age groups, and that access to tafamidis through the EAMS was not in the scope of their recommendation.

3.26 The company considered that tafamidis is a breakthrough treatment for ATTR‑CM. It noted that it is a step change in managing the condition, and that it will benefit people with ATTR-CM and their carers. It also highlighted that it was the first treatment for ATTR-CM to reduce mortality and morbidity and reduce cardiovascular-related hospitalisations. It suggested that the high QALY gains seen in the cost-effectiveness analysis represented a major change in managing ATTR-CM. The clinical expert explained that new research had changed their understanding of the way that tafamidis treats ATTR-CM. The committee recalled that a recent publication had shown that disease stabilisation does not necessarily inhibit amyloid formation, so the mechanism underlying tafamidis' proposed benefit is unclear (see section 3.16). The committee acknowledged that there is an unmet need for an effective treatment for ATTR-CM, but considered that the relevant benefits of tafamidis were captured in the economic model.

3.27 The committee recognised that ATTR-CM is a debilitating and progressive condition which has a substantial effect on a person's quality of life (see sections 3.1 and 3.2). It noted that awareness of ATTR-CM was improving but getting a definitive diagnosis was complicated and can take a long time (see section 3.3). Without validated and objective measures for assessing ATTR-CM, identifying people who need treatment and those who are benefiting from treatment will continue to be a challenge (see section 3.6). It acknowledged that tafamidis was more effective than placebo in the outcomes assessed in the ATTR-ACT pivotal trial. It recalled that there was considerable uncertainty about the modelling of stopping tafamidis and if treatment effects continue after this (see sections 3.15 and 3.16). It also recalled that there was a high degree of uncertainty about whether introducing tafamidis would reduce diagnosis delays and result in any additional benefits or cost savings that could be attributed to tafamidis (see sections 3.8 and 3.19). All this considered, the committee's most plausible range of ICERs was substantially above the range that NICE usually considers an acceptable use of NHS resources (see section 3.24). So, the committee did not recommend using tafamidis in the NHS for treating ATTR-CM.