Draft guidance consultation

Transthyretin amyloidosis (ATTR) is caused by abnormal transthyretin proteins being produced by the liver and accumulating as deposits in tissues of the body (amyloidosis). Transthyretin amyloidosis with cardiomyopathy (ATTR-CM) is a type of ATTR where most deposits accumulate in the heart. There are 2 types of ATTR-CM:

• Wild-type ATTR-CM, which is more common. It mostly affects older people and is more common in men.

• Hereditary ATTR-CM (also known as familial amyloid cardiomyopathy), which affects people born with inherited mutations in the transthyretin (TTR) gene.

As ATTR-CM progresses it can lead to the loss of mobility and independence, which affects quality of life for people with ATTR-CM and their carers as people become increasingly reliant on carers to do daily activities. In turn, carers may experience fatigue and isolation as they devote more time to caring. The patient expert explained that ATTR-CM significantly affects physical ability, noting that walking even short distances can be challenging. There is an associated financial burden as people, or their carers, may have to retire earlier than expected or travel long distances to specialist treatment centres. Delays in diagnosis can also cause anxiety for people living with symptoms without fully understanding what they are experiencing. In hereditary ATTR-CM, multiple family members can be affected, which can carry considerable psychological burden, such as anxiety and guilt about passing the condition on to children. The committee concluded that ATTR-CM is a debilitating condition that substantially affects both physical and psychological wellbeing.

The clinical experts noted that the diagnosis pathway has changed since the ATTR-ACT trial was done. They explained that most people are now diagnosed using medical imaging, rather than biopsy, and more people are being diagnosed at earlier stages. But the clinical experts noted that many people with ATTR-CM are still not diagnosed. The patient expert also highlighted that even with the improvements seen in diagnosis, without any disease-modifying treatments, early diagnosis is of limited value to people living with ATTR-CM. The committee noted that while the rate of diagnosis of ATTR-CM has increased rapidly, the condition is still likely to be underdiagnosed.

The company considered that the most appropriate comparator for tafamidis in this indication is best supportive care (BSC). This is limited to the management of symptoms as well as supportive care, such as diuretics. A small number of people with ATTR-CM also have polyneuropathy with mixed clinical features. NICE's final scope included treatments for people with mixed phenotype ATTR as comparators. NICE has recommended 3 treatments for polyneuropathy:

• NICE technology appraisal guidance on vutrisiran for treating hereditary transthyretin amyloidosis

• NICE highly specialised technologies guidance on inotersen for treating hereditary transthyretin amyloidosis

• NICE highly specialised technologies guidance patisiran for treating hereditary transthyretin amyloidosis.

In the original NICE technology appraisal guidance on tafamidis for treating transthyretin amyloidosis with cardiomyopathy (from here referred to as TA696), the company gave evidence from 2 trials:

• ATTR-ACT (pivotal): a 30-month, phase 3 double-blind randomised controlled trial. It evaluated how effective, safe, and tolerable tafamidis was compared with placebo in adults with wild-type or hereditary ATTR-CM, who had a staging of 1 to 3 (n=441) based on the classification system from the New York Heart Association (NYHA).

• ATTR-ACT extension study (ATTR-ACT LTE): an open-label extension of ATTR-ACT including people from ATTR-ACT as well as people with ATTR-CM who did not take part in ATTR-ACT (ongoing; number of patients not reported).

The company gave new clinical effectiveness and safety evidence data from the August 2021 data cut of the ATTR-ACT LTE with 84 months of follow up. The company gave updated overall survival (OS) and time-to-treatment discontinuation (TTD) data for tafamidis only, which was censored for heart transplant, cardiac mechanical assist device (CMAD) implantation and loss to follow up. The exact figures are considered commercial in confidence by the company and cannot be given here. The company also provided all-cause mortality data for people in NYHA class 1 and 2 compared with people in NYHA class 3, and adverse events from the longer-term follow up. The committee noted that the other trial outcomes that were specified in the scope, including cardiovascular-related mortality and hospitalisations, had not been included in the company submission. It noted that the ClinicalTrials.gov record for the ATTR-ACT LTE stated that cardiovascular-related mortality and hospitalisations were assessed at month 60. The company stated that only OS and TTD were available at 84 months and other outcomes had not been collected at every interval of the long-term extension study. The committee concluded that it would have liked to see all new data that was available since TA696 on the outcome measures specified in the scope.

Similarly to TA696, 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. The model included 5 health states, 4 defined by NYHA classes (1, 2, 3, and 4) and death. People could move to a more severe health state (decline) or to a less severe one (improve). The company did not do an updated systematic literature reviews for economic evaluations, costs and resource use or utilities. The committee noted that the company used the same model structure as TA696 and considered that the company's model was suitable for decision making. It concluded that it would have preferred for the company to update its systematic literature reviews for economic evaluations, costs and resource use or utilities.

In TA696, the committee preferred the log-normal model to extrapolate OS for tafamidis. For this evaluation, the company used the longer-term data from ATTR-ACT LTE to fit parametric survival extrapolations for OS for tafamidis. The company's preferred OS extrapolation for tafamidis was the generalised gamma model. The company explained that the non-parametric hazard profiles from ATTR-ACT LTE showed a clear peak at around 2 years, followed by a decline. It noted that only the log-logistic, log-normal and generalised gamma models are able to predict a local peak hazard as seen in the trial and that the generalised gamma model had the closest agreement with the non-parametric hazard estimators. The generalised gamma model also showed high goodness of fit to the observed data from ATTR-ACT LTE, because it had the lowest Akaike information criterion (AIC) value of all the candidate parametric survival models. The EAG highlighted that although the AIC value was low, the generalised gamma model had a higher Bayesian information criterion (BIC) value than the log-normal model, which suggests a worse fit to the observed data. The generalised gamma model also predicted the highest long-term survival estimates for tafamidis of all the candidate parametric survival models. The committee noted that the OS rate at 20 years for people having tafamidis predicted by the company's preferred generalised gamma model was relatively high, given that the average cohort age would be around 95 years at this time point. The exact survival rate is considered commercial in confidence by the company so cannot be reported here. The committee questioned the clinical plausibility of using the generalised gamma model. The company confirmed that survival estimates were capped to general population life tables. The committee noted that the choice of extrapolation model for OS did not have a large effect on the cost-effectiveness results. It concluded that it preferred to use the log-normal model to extrapolate OS for tafamidis, because it appeared to fit the observed data better and had more plausible long-term survival estimates than the generalised gamma model.

The EAG noted that, after the observed trial period, the company assumed that an increasing number of surviving people stopped having tafamidis but continued to have sustained treatment benefit without gaining any further treatment costs. It considered that this assumption was unlikely to be clinically plausible. The EAG base case included the same assumption but also gave 2 scenarios:

• No further stopping of tafamidis after the observed trial period with continued treatment benefit and costs.

• People continuing to stop tafamidis after the observed trial period, with costs applied only to people still having tafamidis. Transition probabilities, adverse event rates, cardiovascular-related hospitalisations, survival and utilities for the BSC group were applied to people who stopped tafamidis, from the point that they stopped.

In the company submission the utility values for people in the NYHA class 1 (in both tafamidis and BSC arms) and class 2 (tafamidis arm) were higher than the UK general population age-matched average. The exact utility values are considered commercial in confidence by the company and cannot be given here. The company considered that people in NHYA class 1 typically do not have symptoms and physical activity limitations and people in NYHA class 2 only experience slight difficulties with physical activity. But the general population of the same age may have other conditions that affect their quality of life. The committee noted that the starting age in the economic model was 74 years. The EAG base case included capped utility values to the age-matched equivalent. The clinical experts explained that it was not plausible that someone with ATTR-CM would have a better quality of life than someone of a similar age from the general population. So the committee concluded that capped utility values for the NYHA class 1 and NYHA class 2 health states were appropriate for decision making.

In TA696, the committee concluded that BSC values (treatment-independent) and utility values should be applied for the NYHA class 4 health state because:

• there was a substantial difference in utility values between arms in NYHA class 4, while utility values for tafamidis and BSC in the other NYHA classes were similar and

• NYHA class 4 utility values were based on a low number of observations.

NICE's manual on health technology evaluation notes that above a most plausible ICER of £20,000 per quality-adjusted life year (QALY) gained, decisions 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 evidence presented, but will also take into account other aspects including uncaptured health benefits. The committee noted that:

• there are currently no disease-modifying treatments available for people with ATTR-CM in England, so there is substantial unmet need in this population (see section 3.4)

• the company had given 84-month follow-up data for people having tafamidis, but no new comparative data was available since TA696 (see section 3.6)

• there is uncertainty about the outcomes in people who stopped treatment with tafamidis, but that the committee's preferred assumptions are likely to be conservative (see section 3.9).

The company cost-effectiveness estimates are considered to be commercial in confidence by the company so cannot be reported here. The company's preferred analysis estimated that the deterministic and probabilistic ICERs for tafamidis compared with BSC were at or above the upper end of the range normally considered a cost-effective use of NHS resources (including the company's confidential commercial arrangement). It included the following assumptions:

• using a generalised gamma parametric distribution function to model OS for tafamidis (see section 3.8)

• using treatment-dependent utility values for the NYHA class 4 health state (see section 3.10).

The committee noted comments from the clinical and patient experts that the hereditary form of ATTR-CM disproportionally affects people with certain variants (such as Val122lle), which are more prevalent in people of African Caribbean and Hispanic ethnicities. In these populations, ATTR-CM is often diagnosed later and has worse outcomes than in other people with ATTR-CM. Val122lle is not associated with polyneuropathy so people with this variant do not have access to disease-modifying therapy. The professional group submissions also highlighted the fact that prescribing tafamidis might be restricted to specialist centres and that people with ATTR-CM are often older and could experience difficulties with travelling long distances to have treatment. The committee took these factors into account in its decision making but agreed that this was not something that could be addressed in its recommendation.

The committee noted that there are no disease-modifying treatments for ATTR-CM available in England and acknowledged the substantial unmet need in this population. It also noted that there is uncertainty about the outcomes in people who have stopped treatment with tafamidis, but that the committee's preferred assumptions are likely to be conservative. So, an acceptable ICER based on this set of assumptions would be towards the upper end of the range normally considered a cost-effective use of NHS resources (£20,000 to £30,000 per QALY gained). The cost-effectiveness estimates that included the committee's preferred assumptions were above the upper end of range that NICE usually considers an acceptable use of NHS resources. So, tafamidis is not recommended.