Dabrafenib with trametinib for treating BRAF V600E mutation-positive glioma in children and young people aged 1 year and over

Gliomas are the most common type of brain cancer in children and young people. They develop from the glial cells that support the nerve cells of the brain and spinal cord. Gliomas are classified by how quickly they grow. Most gliomas are grade 1 or 2, referred to as low-grade glioma (LGG), and do not grow or only grow slowly. Grade 3 and 4 gliomas, referred to as high-grade glioma (HGG), grow rapidly. Consequently, HGG is associated with worse outcomes than LGG. BRAF is a gene that encodes the protein B‑Raf, which influences cell growth. People with BRAF V600E mutation-positive LGG live for less time than people with glioma without the mutation. The patient experts emphasised the traumatic nature of a glioma diagnosis for children and young people, their families and caregivers, and the limitations of current treatment. They noted that glioma and its treatment can delay education, restrict socialising, and cause lasting emotional impact. They highlighted that the toxicity associated with conventional chemotherapy can lead to people with glioma and their caregivers choosing to stop treatment. The patient experts also explained that currently available treatments need regular travel to hospital, incur significant costs and need a substantial time commitment. For these reasons, people with glioma and their caregivers would value additional treatment options, particularly those that can be taken at home. The committee understood the comments from the patient experts about the effect of glioma on people who have it, their families and caregivers, and recognised that there is a high burden for people with LGG and HGG.

The clinical experts explained that the aims of treatment for glioma include stopping or delaying progression and improving neurological function and quality of life. Current treatment for LGG includes maximal surgical resection, followed by systemic chemotherapy or radiotherapy at first relapse or progression. At further relapse or progression, other chemotherapy regimens may be used. Current treatment for HGG includes maximal surgical resection followed by chemoradiotherapy (typically with adjuvant temozolomide, see NICE's technology appraisal guidance on carmustine implants and temozolomide for the treatment of newly diagnosed high-grade glioma). At relapse or progression, some people will have further chemotherapy, radiotherapy or resection and some people will have best supportive care. The clinical experts noted that there is an unmet need for treatments for children and young people with BRAF V600E mutation-positive glioma. The clinical experts thought that dabrafenib plus trametinib, as a treatment that is specific for BRAF V600E mutations, represents a step-change in care compared with current treatment. The committee concluded that current treatment for glioma is limited and that dabrafenib plus trametinib offers a new treatment option.

The clinical evidence came from TADPOLE, a multicentre, open-label, phase 2 study done across 20 countries (including the UK). TADPOLE consisted of 2 sub-studies. The LGG cohort was a randomised controlled trial in which people aged 1 to 17 years with BRAF V600E mutation-positive LGG were randomised to have dabrafenib plus trametinib (n=73) or vincristine plus carboplatin (n=37). The HGG cohort was a single-arm prospective cohort study in which people aged 1 to 17 years with BRAF V600E mutation-positive HGG had dabrafenib plus trametinib (n=41). People in both cohorts had treatment until progression, unacceptable toxicity, lack of clinical benefit, start of new treatment or death. The median follow up was 39.0 months in the LGG cohort and 45.2 months in the HGG cohort.

The primary outcome measure was overall response rate. In the dabrafenib plus trametinib arm it was 54.8% (95% confidence interval [CI] 42.7% to 66.5%) and in the vincristine plus carboplatin arm it was 16.2% (95% CI 6.2% to 32.0%). Secondary outcomes included progression-free survival and overall survival. Progression was assessed by investigators and by central independent review. Median progression-free survival by investigator assessment was 46.0 months (95% CI 38.6 months to not estimable) in the dabrafenib plus trametinib arm and 30.8 months (95% CI 7.0 months to not estimable) in the vincristine plus carboplatin arm. Median progression-free survival by independent review was 24.9 months (95% CI 12.9 to 31.6 months) in the dabrafenib plus trametinib arm and 7.2 months (95% CI 2.8 to 11.2 months) in the vincristine plus carboplatin arm. Overall survival data was immature at the end of the study, with 1 death in the vincristine plus carboplatin arm and no deaths in the dabrafenib plus trametinib arm. The committee concluded that the evidence from TADPOLE showed that dabrafenib plus trametinib was more effective than vincristine plus carboplatin for treating LGG.

Overall response rate was the primary outcome measure and was 56.1% (95% CI 39.7% to 71.5%) with dabrafenib plus trametinib. Median progression-free survival by investigator assessment was 24.0 months (95% CI 12.5 months to not estimable), and by independent review was 9.0 months (95% CI 5.3 to 20.1 months). Overall survival data was immature at the end of the study, with 17 deaths (41.5%); median overall survival not estimable (95% CI 19.8 months to not estimable). The committee concluded that the evidence from TADPOLE was the best available to show the effectiveness of dabrafenib plus trametinib for treating HGG.

For LGG, dabrafenib plus trametinib is indicated for people who need systemic treatment. The EAG noted that the clinical evidence from the comparative part of TADPOLE was from people who were eligible for first-line systemic treatment. So, there was no comparative evidence or economic analyses of dabrafenib plus trametinib as a treatment for LGG at the second or later lines of treatment. The clinical experts noted that, in current clinical practice, the availability of dabrafenib plus trametinib is restricted by managed access programmes or clinical trials, so it is only used at second and later lines of treatment. They explained that this restriction was purely practical and that most clinicians would prefer to use dabrafenib plus trametinib as a first-line treatment for BRAF V600E mutation-positive LGG. Because of this, the committee concluded that the evidence from TADPOLE for first-line use was sufficient to evaluate dabrafenib plus trametinib, because this is the point in the pathway at which it will likely be used for LGG.

Because the HGG cohort of TADPOLE was single arm, the company did unanchored indirect treatment comparisons (ITCs) to compare dabrafenib plus trametinib with the comparators. For the no prior temozolomide subgroup, the company's systematic literature review identified 2 studies, both of which had used temozolomide. For the prior temozolomide subgroup, the company was unable to identify any studies using best supportive care. The company received clinical advice that after temozolomide has not worked, chemotherapy tends to be ineffective. So, using studies in which people who previously had treatment with temozolomide then had chemotherapy would be a reasonable proxy for best supportive care. The company was able to identify 2 such studies: 1 that used cilengitide and 1 that used bevacizumab. The unanchored ITCs used either matching-adjusted indirect comparison (MAIC) or inverse probability of treatment weighting methods. MAIC was used when only aggregate data was available for the comparator; inverse probability of treatment weighting was used when patient-level data was available. For both subgroups, the ITCs produced progression-free survival and overall survival hazard ratios that were less than 1 and had 95% CIs that did not include 1. This implied that dabrafenib plus trametinib statistically significantly improved overall survival and progression-free survival compared with temozolomide and best supportive care. The exact results of the ITCs are considered confidential by the company so cannot be reported here. The EAG agreed with the methods used for the ITCs but cautioned that they are associated with uncertainty. The EAG noted the small sample sizes, the limited number of covariates adjusted and the lack of data on BRAF V600E mutation status in the comparator studies. It noted that for the no prior temozolomide ITC, the comparator studies were approximately 20 years old. It also noted that for the prior temozolomide ITC, the comparator studies used chemotherapy and so were only proxies for best supportive care. The committee acknowledged the uncertainty with using unanchored ITCs to establish comparative efficacy but concluded that they are acceptable for decision making.

The company developed an individual-based state transition model which included analyses for both LGG and HGG cohorts. There were 3 health states common to both analyses: progression-free after first treatment, progressed and death. The LGG part of the model had an additional health state for when people with LGG transform to HGG (secondary HGG). The model simulates the individual histories of a sample of people aged 1 to 17 years with BRAF V600E mutation-positive glioma over a lifetime horizon. The time that people spent in the various health states was based on time-to-event data from TADPOLE and literature sources. The key events, as described in the model, included:

• 'progression (not because of malignant transformation)

• malignant transformation (in the LGG analysis)

• death (glioma-related death, malignant death or non-glioma related)'.
  
  The committee was satisfied with the company's modelling approach.

The company and the EAG disagreed on whether to use independent-assessed or investigator-assessed progression-free survival from TADPOLE to model progression. As noted in section 3.6 and section 3.7, there were large differences between the results of these assessments of progression-free survival. The company used investigator-assessed progression-free survival in its base case. It received clinical advice that, in clinical practice, investigator-assessed progression-free survival is a more accurate reflection of the timing of progression and of the decision to stop treatment than independent review of progression-free survival. The EAG disagreed and noted that independent review of progression-free survival is more widely used in clinical trials. Its preferred approach would be to use investigator-assessed progression-free survival for treatment discontinuation and independent-assessed progression-free survival for health state occupancy. But, it noted that the company's model was not flexible enough to allow this. So, it suggested a pragmatic approach which used independent-assessed progression-free survival for both. The clinical experts explained that the difference between the investigator and independent estimations of progression-free survival was because independent reviewers only see the tumour scan in isolation. In contrast, investigator assessment accounts for the scan, as well as information directly gained from the person with glioma and their caregiver. The committee agreed with the company and clinical experts that investigator assessment is a more accurate reflection of when progression occurred. It concluded that for this evaluation, investigator-assessed progression-free survival should be used.

The company's base case used a piecewise hybrid approach to extrapolate progression-free survival over the entire time horizon. The company justified the piecewise approach on the basis that parametric extrapolations did not fit the observed Kaplan–Meier (KM) curves well. The cut-off point (next-to-last observed event) was chosen because of the low number of people at risk after 2 years and because it was aligned to the expected treatment duration of dabrafenib plus trametinib in clinical practice. The EAG felt that the piecewise approach was not appropriate because the choice of cut-off point was highly arbitrary. It also noted that the company had not adequately justified using the same rate of progression for both arms in the extrapolation phase of LGG, nor using a constant hazard ratio for lifetime in HGG. The EAG's preferred approach was to fit independent parametric curves and extrapolate across the entire time horizon. The committee considered that the company's piecewise approach had limitations. It noted that there was only a small number of people in TADPOLE, and that this meant that there was not much data used to generate the KM curve. The committee thought that this may limit the generalisability of the KM curves from TADPOLE to people with glioma in NHS practice. It also agreed with the EAG that the choice of cut-off point was arbitrary. So, the committee concluded that it preferred the EAG's approach of the 2 methods presented for extrapolating progression-free survival.

The marketing authorisations for dabrafenib and trametinib state that treatment should continue until 'disease progression or until the development of unacceptable toxicity'. They also state that 'there are limited data in patients older than 18 years of age with glioma, therefore continued treatment into adulthood should be based on benefits and risks to the individual patient as assessed by the physician'. In the company's base case for LGG, the KM data was used to model progression until week 193 (about 3.7 years). This aligned with clinical advice given to the company that suggested that treatment with dabrafenib plus trametinib would be stopped around 2 to 5 years, in the absence of progression, because of cumulative toxicity. For HGG, clinical advice to the company suggested that clinicians would be more reluctant to stop treatment, given the lack of alternative options and the poor prognosis of HGG. Nevertheless, some people with glioma that has a maintained response to dabrafenib plus trametinib may stop treatment because of cumulative toxicity. In the company HGG base case, the model assumes an informal stopping rule of 12.5 years if there is no progression. The EAG noted that the treatment duration of dabrafenib plus trametinib is uncertain and that removing the stopping rule makes it less cost-effective. The clinical experts considered that the likely duration of dabrafenib plus trametinib is difficult to predict. This is because it is likely that there will always be some people whose glioma responds well to treatment and who experience minimal side effects. Furthermore, when stopping dabrafenib plus trametinib, people with glioma may experience high relapse rates and then need retreatment. The committee concluded that treatment duration should reflect the marketing authorisations and a stopping rule should not be included for either LGG or HGG.

The company's base case assumed that dabrafenib plus trametinib was dosed in line with the schedule in TADPOLE, in which both age and weight determined the dose. The marketing authorisation simplifies this by only dosing by weight. So, the company provided a scenario analysis in which the dose was determined by weight. The committee considered that dosing by weight would be how dabrafenib plus trametinib would be used in NHS clinical practice. It concluded that dosing by weight should be used in the model.

NICE's manual on health technology evaluations notes that, above a most plausible incremental cost-effectiveness ratio (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 noted that there were several sources of uncertainty, specifically that:

• the comparative efficacy of dabrafenib plus trametinib in HGG was based on indirect comparison (see section 3.9)

• the progression-free survival extrapolations were uncertain and based on KM data from a small number of people (see section 3.12)

• the likely duration of treatment with dabrafenib plus trametinib that would be used in NHS clinical practice is unclear (see section 3.13)

• the utility decrements used in the model were sourced from adults (see section 3.15).
  
  But the committee also recalled the statements from the clinical and patient experts that children and young people with BRAF V600E mutation-positive glioma would highly value a new treatment option. It also recalled that they would particularly welcome a treatment that would allow them to live a less restricted life with fewer visits to the hospital. It also noted that because of the rarity of BRAF V600E mutation-positive glioma, the decision risk to the NHS was low. So, the committee concluded that an acceptable ICER would be around £30,000 per QALY.

NICE's manual on health technology evaluations states that the committee's preferred cost-effectiveness estimates should be derived from a probabilistic analysis when possible unless the model is linear. In this evaluation, the EAG noted that the model was linear and so deterministic analyses were acceptable for decision making. In the LGG analysis, because of confidential commercial arrangements for other treatments in the model, the exact cost-effectiveness estimates are confidential and cannot be reported here. Both the company's and EAG's base case ICERs were within the range that NICE considers to be a cost-effective use of NHS resources, but the EAG's was lower. In the HGG analysis, the company's base case deterministic ICER with the severity weighting applied, was £28,624 in the no prior temozolomide subgroup, and £29,072 in the prior temozolomide subgroup. The EAG's base case deterministic ICER, with the severity weighting applied, was £27,500 in the no prior temozolomide subgroup, and £21,512 in the prior temozolomide subgroup.

The committee's preferred assumptions included:

• using investigator-assessed progression-free survival from the TADPOLE study (see section 3.11)

• extrapolating progression-free survival by fitting independent curves to the dabrafenib plus trametinib and comparator KM data and extrapolating over the entire time horizon (see section 3.12)

• not including a stopping rule for modelling treatment duration for either LGG or HGG (see section 3.13)

• using dosing of dabrafenib and trametinib based on weight (see section 3.14)

• applying a severity weighting of 1.2 to LGG QALYs and 1.7 to HGG QALYs (see section 3.16).

During scoping, consultees noted that the population in the marketing authorisation is restricted to 'paediatric patients aged 1 year and older' and that this contributes to inequality based on age. Age is a protected characteristic under the Equality Act 2010. Because the committee is only able to make recommendations within the marketing authorisation, it concluded that this restriction did not represent an equality issue in its evaluation of clinical and cost effectiveness.

The committee considered whether dabrafenib plus trametinib was innovative. The clinical experts explained that because dabrafenib plus trametinib is a targeted treatment for BRAF V600E mutation-positive glioma, it represents a step-change in current treatment. The patient experts noted that because dabrafenib plus trametinib is an oral treatment, it is more convenient, and that people with glioma and their caregivers would value having fewer hospital visits for intravenous treatment. They noted that dabrafenib plus trametinib allowed children and young people to spend more time with family and take part in more recreational activities. The committee noted the benefits associated with better school attendance and increased socialising that would result from fewer visits to the hospital which would not be captured in adult utility values (see section 3.15). For these reasons, the committee concluded that there were uncaptured benefits in the QALY calculations, which it would account for in its decision making.