According to the cost-effectiveness criteria implemented by the National Institute for Health and Care Excellence (NICE) in conventional Health technology Assessments (HTAs), treatments are recommended for routine use within NHS England if the most plausible incremental cost-effectiveness ratio (ICER) is below £20,000 per quality adjusted life year (QALY) gained. Treatments with an ICER below £30,000 per QALY gained may be recommended depending on the certainty of the ICER estimates, unmet need, and strength of clinical data (NICE, 2012).
In 2009, NICE introduced end-of-life criteria, altering the cost-effectiveness threshold for patients with a short life expectancy (NICE, 2014). In 2013, the Highly Specialised Technology (HST) programme replaced the Advisory Group for National Specialised Services (AGNSS) for the assessment of ultra-orphan drugs (NICE, 2017), and in 2016 the Cancer Drugs Fund (CDF) was revised to provide accelerated access to cancer treatments (NICE, 2018). All three policies altered the criteria for cost-effectiveness for their respective treatments, as shown in Table 1.
This article provides a summary of the three programmes, including the relative number of treatments recommended within these schemes, and how often managed access agreements (MAAs)/ patient access schemes (PASs) are included in the decision. The ethical implications of these policies are discussed in a separate SIRIUS article [link].
Table 1: Disease and treatment eligibility for HST, CDF, and EoL criteria
CDF, Cancer Drugs Fund; EoL, end-of-life criteria; HST, Highly Specialised Technology; QALY, quality adjusted life year; TA, technology appraisal. *The criteria are weighted according to the number of QALYs gained.
Data were taken from the SIRIUS Oncology HTA Database [link], and the SIRIUS Orphan Drug Database. The SIRIUS Oncology HTA Database contains details of all NICE assessments since December 2010, with corresponding data for SMC, HAS, G-BA/ IQWiG, PBAC, and CADTH assessments for the included treatments and indications. The Orphan Drug Database includes SMC, HAS, and G-BA data for all treatments included in the NICE HST programme. Data for end-of-life treatments were taken from the SIRIUS Oncology HTA Database and the NICE website for non-oncology treatments. Information on NICE criteria was obtained from the NICE website.
Rare diseases provide several barriers to market access. The low patient numbers make clinical development problematic, reducing the potential to collect strong clinical evidence. The smaller population also necessitates a higher per-patient cost to recoup the R&D costs, which combined with the weaker clinical evidence reduces the potential for cost-effectiveness. The European Medicines Agency (EMA) created an orphan designation to provide incentives for companies to develop treatments for diseases with fewer than 1 in 2,000 patients (EMA, 2018a), including protocol assistance, fee reductions, grants, and ten-year market exclusivity, among others (EMA, 2018b). However, these factors might not always provide enough assistance for ultra-rare diseases (1 in 50,000) to be considered cost-effective by NICE at the usual threshold. The HST programme was therefore introduced by NICE, which, by the most recent guidelines introduced in April 2017, raised the cost-effectiveness threshold to £100,000 per QALY gained, five times higher than the £20,000 for conventional HTAs (NICE, 2017). Further, there would be a system whereby if the incremental QALYs gained exceeded ten, then the QALYs would receive an additional weighting, allowing treatments with higher clinical benefit greater potential to be cost-effective, as shown in Table 2. The access to treatments for rare diseases in other European countries is discussed in a separate SIRIUS article [link].
Table 2: QALY weighting system for HST
Source: NICE (2017). HST, Highly Specialised Technology; QALY, quality adjusted life year.
As of October 2018, eight treatments have FEDs within the HST programme, as shown in Table 3. Only the most recent assessment for the treatment of severe combined immunodeficiency due to adenosine deaminase deficiency (ADA-SCID) with Strimvelis, was assessed under the above cost-effectiveness criteria. The committee-preferred ICER was £120,506 vs. haematopoietic stem cell transplantation (HSCT) from a matched unrelated donor, and £12,106 vs. a haploidentical donor, with undiscounted QALY gains of 14.0 and 19.6, respectively. Therefore, with the weighting system applied, Strimvelis was cost-effective under the new criteria, and recommended by NICE.
Table 3: NICE HSTs with FEDs
HST, Highly Specialised Technology; MAA, managed access agreement; PAS, patient access scheme; Ph, phase; QALY, quality adjusted life year; RCT, randomised control trial.
Before the introduction of the most recent criteria, only sebelipase alfa was not recommended. The company-estimated 5-year net budget impact was £53,548,573, and sensitivity analyses by the ERG estimated the figure to be between £23,439,245 to £126,845,895. The committee gave no preferred estimate, but deemed the budget impact too high for the uncertain benefits.
Three treatments (elosulfase alfa, ataluren, and asfotase alfa) had 5-year MAAs as clinical benefits and/ or value for money were uncertain, but the committee deemed that further research could address the uncertainty sufficiently to assess value for money. For asfotase alfa and elosulfase alfa, the data are being collected in databases created specifically for their MAAs. These are owned by the manufacturers, and freely accessible to NICE and NHS England. The data for ataluren are stored in the pre-existing NorthStar database for duchenne muscular dystrophy and owned by the NorthStar network, but freely shared with the marketing authorisation holder, NICE, and NHS England. In all three cases, if the treatment is not recommended by NICE at the end of the MAA, then funding for all patients, including those already on treatment, will cease.
Five out of the seven recommended HSTs are for life-long treatments, and six out of seven of the associated diseases have early onset (from birth to juvenile) on average, with the seventh, Fabry disease, described as having variable age of onset. The annual budget for high-cost specialised technologies is £156 million. In HST1, with the addition of 20 new patients receiving eculizumab every year, the estimated cost rose from £58 million in year one to £82 million by year five. Similarly, the cost for ataluren rose from £8.6 million in year one to £16.0 million by year five due to the expected number of patients receiving treatment rising from 35 to 65.
Notably, four out of the seven currently approved HSTs have PASs, which could ameliorate costs in these instances, and both ataluren and elosulfase alfa have PASs during the MAA.
The CDF is designed to increase public access to cancer treatments. Treatments within the CDF must be potentially cost-effective at the £20-30,000 threshold, but are allowed to present with additional uncertainty that can be addressed during an MAA (NICE, 2018), thus allowing the public to access potentially beneficial treatments while the uncertainty is addressed.
The CDF is a managed access scheme, and all treatments within the CDF have associated MAAs, usually with a maximum of 2 years, but with no mandated limit, often linked to the completion of the main (phase II-III) clinical trial. At time of writing, there are 19 published MAAs. Four of these are for the use of pembrolizumab in different indications, three are for the use of nivolumab in different indications, and 12 are for other treatments, as shown in Table 4. Two of the 19 MAAs have been successfully completed (TA446 and TA447), with subsequent recommendation by NICE outside the CDF. Fifteen (of the 19) published MAAs reported continuing data collection from the main clinical trials until they reached data maturity/trial completion. While 13 specified this as the primary source, all MAAs used multiple data sources, including the (mandated) Systemic Anti-Cancer Therapy (SACT) dataset, and Blueteq database (14 of 19), owned by Public Health England and NHS England, respectively. Five MAAs reported using SACT as the primary data source. One HTA for pembrolizumab in Hodgkin lymphoma (TA540) included only a subpopulation of the indication (those unable to have autologous stem cell transplants) within the CDF.
No treatment within the CDF has a PAS, as the treatments must prove potentially cost-effective at the list price. If the treatment is accepted after the MAA with a PAS, then the manufacturer could be asked to refund the NHS the difference in treatment cost for the duration of the MAA. However, some appraisals, such as TA505 for ixazomib, have commercial access agreements making the treatment cheaper while it is available on the CDF.
Table 4: Source of data during the MAA for CDF
NICE, National Institute for Health and Care Excellence; SACT, Systemic Anti-Cancer Therapy. *Calculated from date of issue for MAA and end of MAA. **Primary source was SACT and other Public Health England datasets.
The introduction of end-of-life criteria in January 2009 reflected that life extension for patients with very short life-expectancy might carry additional benefits to those captured by the QALY metric, and therefore additional weight should be applied to QALYs gained for EoL treatments. However, the NICE appraisal committee received no specific instruction with regard to the magnitude of the additional weight, and over time, practice has led to the application of a maximum weight of 2.5 from a starting point of £20,000 per QALY (NICE, 2014). This effectively results in a threshold of £50,000 per QALY if diseases have less than 24 months life expectancy and the treatment could be shown to increase life span by at least 3 months.
As of October 2018, 61 single technology appraisals were found to meet the EoL criteria since December 2010. Of these, only one treatment was outside oncology (TA386), ruxolitinib for the treatment of disease-related splenomegaly or symptoms in adults with myelofibrosis, which was recommended by NICE. Two HTAs included more than one population within the same appraisal (making a total of 63 appraisals for EoL):
TA451 included two diseases (acute lymphoblastic leukaemia and chronic myeloid leukaemia).
TA517 included patients in first and second-line (with Merkel cell carcinoma).
Both populations in TA517 met EoL criteria, as did patients with acute lymphoblastic leukaemia in TA451. However, only two out of three of the chronic myeloid leukaemia subpopulations were considered to meet EoL criteria (in TA451). All treatment indications in TA451 and TA517 were recommended by NICE. Seven HTAs had mixed responses from NICE, with specific subpopulations meeting EoL criteria. Of these, six were recommended, while ramucirumab received a negative recommendation (TA403). In two HTAs, the committees deemed that there was insufficient evidence to assess EoL criteria, and both treatments (TA416 and TA510) were recommended within the CDF. Forty-six of the 63 appraisals included PASs.
A comparison between oncology HTAs meeting EoL criteria with those that do not suggests that a higher proportion of those meeting EoL criteria receive positive recommendations, as shown in Figure 1.
Figure 1: NICE recommendations according to EoL criteria
HTA, Heath & Technology Assessment; Mixed, only a subpopulation meets EoL criteria or is positively recommended; UTC, unable to conclude. Numbers of HTAs are shown above their respective bars; percentages are shown below the graph (may not equal 100% because of rounding).
Eight treatments have currently received FEDs within the HST programme, including seven positive recommendations. Only Strimvelis has been assessed according to the latest cost-effectiveness criteria. Three treatments have MAAs and four have PASs.
Nineteen treatments have published MAAs within the CDF. All treatments recommended within the CDF have MAAs, and no treatment has a PAS. Two treatments have completed the MAAs and have since been recommended by NICE outside the CDF.
As of October 2018, 61 treatments have been found to meet EoL criteria, 60 of which were for oncology indications. 93% (51) of EoL treatments were recommended by NICE, compared with 59% of non-EoL treatments. Forty-six of the 63 appraisals included a PAS, and two EoL treatments included an MAA within the CDF.