TA873 · STA

Cannabidiol for treating seizures caused by tuberous sclerosis complex

Recommended with restrictionsTechnology Appraisal Committee BJanuary 2023

Cannabidiol is recommended as an add-on treatment for seizures caused by tuberous sclerosis complex in people aged 2 years and over, only if seizures are not controlled by 2 or more antiseizure medications (used alone or in combination) or these treatments were not tolerated; seizure frequency is checked every 6 months and cannabidiol is stopped if seizure frequency has not fallen by at least 30% compared with the 6 months before starting treatment; and the company provides cannabidiol according to the commercial arrangement.

Source documents

Intervention

cannabidiol (Epidyolex)
· cannabinoid receptor agonist · oral

Conditions

seizures caused by tuberous sclerosis complexneurology_psychiatry · refractory
seizures caused by tuberous sclerosis complexneurology_psychiatry · chronic
tuberous sclerosis complexneurology_psychiatry

Comparators

NameType Established Committee preferred
usual care including antiseizure medications with or without surgery or vagus nerve stimulationstandard of careYes
usual care with antiseizure medicationsstandard of careYes
usual carestandard of care

Clinical trials

TrialDesignPhasePivotal
GWPCARE6RCTnot_stated_in_chunkYes

Economic model

markov
Time horizon: Lifetime
Cycle length: 1 week
other (company)
Time horizon: lifetime
Cycle length: 1 week

ICER

£20,000–£30,000 (cannabidiol vs usual care) · high uncertainty

Methodological decisions (18)

carer utility

Number of carers and inclusion of carer utility decrements in model

Company: Assumed 2 cumulative carers (1 primary plus others) based on Dravet syndrome data; included utility decrements for carers

ERG: not_stated_in_chunk

Committee: Appropriate to include carer quality of life effects; 1.8 carers appropriate for modelling (concluded at end of section 3.14, not fully shown in chunk)

ICER impact: decreases

comparator selection

ERG highlighted that everolimus may form part of usual care, but committee noted everolimus has different indication and requires restrictive monitoring, so population receiving everolimus at same pathway point would be very small. Committee agreed usual care including antiseizure medications with or without surgery or vagus nerve stimulation was appropriate comparator.

ERG: Questioned whether everolimus should be considered as part of usual care in the population where antiseizure medications are not tolerated

Committee: Usual care including antiseizure medications with or without surgery or vagus nerve stimulation

ICER impact: uncertain_direction

cost assumption

Average dose of cannabidiol for cost modelling

Company: 12 mg/kg/day based on clinical expert advice and German dispensing data

ERG: Concerned that 12 mg/kg/day was not verified by clinical trial data

Committee: 15 mg/kg/day for decision making to account for loss of benefit compared to GWPCARE6 (which used up to 25 mg/kg/day or 50 mg/kg/day)

ICER impact: increases

cost assumption

Healthcare resource use estimates, particularly hospitalisation rates

Company: Used Delphi panel estimates; higher hospitalisation rates reflecting spectrum of severity in tuberous sclerosis complex

ERG: Modelled 50% reduction in hospitalisation rates to align with other cannabidiol appraisals (Dravet and Lennox-Gastaut syndromes)

Committee: 50% reduction from Delphi panel hospitalisation rates should be used for decision making

ICER impact: decreases

cost assumption

Modelling of costs and utility benefits associated with TAND (tuberous sclerosis complex-associated neuropsychiatric disorder) aspects

Company: Included TAND costs and modelled utility benefit and cost reduction for responders (≥50% seizure reduction); conservative estimates at technical engagement using lowest utility benefit and ages 2-6 years only

ERG: Concerned that prevalence rates, costs and utility benefits not from clinical trials; Delphi consensus on response definition incomplete; responder proportion from full trial applied only to 2-6 year-olds

Committee: Appropriate to include TAND costs and benefits despite uncertainties, using company's technical engagement estimates

ICER impact: decreases

cost assumption

Hospital admission rates; whether baseline rates from Delphi panel should be adjusted downward

Company: Used baseline hospitalisation rates from Delphi panel consensus

ERG: Reduced hospitalisation rates by 50% in base case

Committee: Agreed preferred assumptions aligned with ERG base case including 50% reduction in hospitalisation rates

ICER impact: increases

cost assumption

Average dose of cannabidiol to use in cost-effectiveness model

Company: Used 12 mg/kg/day

ERG: Used 15 mg/kg/day in base case

Committee: Preferred assumptions aligned with ERG base case using 15 mg/kg/day. Acknowledged that changing average dose significantly increased cost-effectiveness estimates.

ICER impact: increases

model structure

How to model the proportion of people seizure-free over a 7-day cycle using regression model predictions

Company: Used 6.5-day cut-off based on negative binomial regression model limitations; later provided scenario with 6.61-day cut-off

ERG: Highlighted that 6.5-day cut-off overestimates seizure-freedom; provided scenario using 7-day cut-off

Committee: 6.61-day cut-off is most appropriate for decision making

ICER impact: uncertain_direction

model structure

Committee noted that uncaptured benefits in the company's model included quality-of-life benefits from reducing seizure severity, reduction in hospitalisation rates, physical and mental benefits to the wider family, and mortality benefit from reducing SUDEP related to tuberous sclerosis complex.

Committee: Committee acknowledged these uncaptured benefits were important for improving quality of life and would take them into account in decision making despite cost-effectiveness uncertainties

other

Number of carers to include in model; whether carer disutilities should be applied additively

Company: Initially 2 carers (1 primary + others) based on Lagae 2019 data from Dravet syndrome; updated to 1.8 carers at committee direction

ERG: Used 1.8 carers in base case, aligning with Dravet and Lennox-Gastaut guidance; concerned about additive application of disutilities

Committee: Agreed that carer disutilities should not be purely additive; 1.8 carers appropriate to align with previous cannabidiol appraisals. While 2 carers have greater total burden than 1, it would not be 2 times greater.

ICER impact: decreases

other

Impact of institutionalisation on carer utility; whether and how much institutionalisation reduces carer disutility

Company: Initially did not model impact on carer utility during institutionalisation. Updated to use ERG approach after committee direction.

ERG: Assumed 50% reduction in carer disutility for institutionalised individuals

Committee: Appropriate to include reduction in carer disutility for institutionalisation, with ERG approach preferable given lack of data supporting specific size of reduction

ICER impact: increases

other

Proportion of people achieving seizure-freedom; use of 6.61-day cut-off for proportion seizure-free over 7 days

Company: Not explicitly detailed in these pages

ERG: Not explicitly detailed in these pages

Committee: Preferred to use 6.61-day cut-off for proportion seizure-free over 7 days

ICER impact: uncertain_direction

population generalisability

Committee noted differences between arms in GWPCARE6 regarding clobazam use, which can increase cannabidiol effectiveness when taken adjunctively. However, clinical expert confirmed clobazam is used in NHS practice similarly to GWPCARE6 and use would not be expected to increase if cannabidiol was recommended.

Committee: GWPCARE6 is broadly generalisable to NHS clinical practice

ICER impact: negligible

stopping rule

Company proposed stopping cannabidiol if seizure frequency does not reduce by 30% from baseline, assessed every 6 months. Committee concluded this was appropriate.

Committee: Stop if seizure frequency has not fallen by at least 30% compared with the 6 months before starting treatment, assessed every 6 months

ICER impact: uncertain_direction

utility source

Health state utility values for the model

Company: Used vignette study with general public using time trade-off method; values kept confidential

ERG: not_explicitly_stated_in_chunk

Committee: Acknowledged uncertainty and differences from other cannabidiol appraisals but considered analyses using company's utility values acceptable

ICER impact: uncertain_direction

utility source

Health state utilities derived from vignette study using time trade-off method by general public rather than from literature or clinical trials

Company: Vignettes from general public appropriate given lack of literature values; utilities considered confidential

ERG: Not explicitly stated, but accepted methodology while noting uncertainty

Committee: Company's health state utilities are uncertain and differ from utilities in other cannabidiol appraisals. The committee would have liked to see data supporting plausibility of vignette values, especially for seizure-free and most severe health states. However, acknowledging high uncertainty, considered analyses using company's utility values in decision making.

ICER impact: uncertain_direction

utility value choice

Discrepancy in health state utilities between tuberous sclerosis complex and Lennox-Gastaut/Dravet syndromes; difficulty comparing seizure severity across conditions due to differences in seizure types and TSC being a multi-organ disease

Company: Differences partly due to different seizure severity categories; TSC multi-organ complications make comparison difficult

ERG: Not explicitly differentiated

Committee: Utility values uncertain due to multi-organ nature of TSC and differences from other cannabidiol appraisals. Committee concerned company model did not adequately capture impact of non-seizure aspects of TSC on quality of life.

ICER impact: uncertain_direction

utility value choice

Seizure-free health state utility for carers; whether to use average 45-year-old utility or adjust for carer ageing

Company: Updated to use utility value for average 45-year-old based on typical parent age of 13-year-old child

ERG: Concerned this overestimated impact of seizure freedom; provided scenarios with lower values; maintained 0.881 value in preferred assumption after consultation

Committee: Value of 0.881 for seizure-free health state for carers is uncertain but appropriate for decision making. Absolute utility decrement unchanged regardless of value used.

ICER impact: negligible

Evidence gaps

short follow upSeizure-free days per week not collected as outcome in GWPCARE6; modelling approach relies on regression from week 16 data
surrogate not validatedNo direct evidence that treatment efficacy at 15 mg/kg/day is identical to that at 25 mg/kg/day or 50 mg/kg/day used in GWPCARE6
no uk dataHealthcare resource use estimates sourced from Delphi panel and German dispensing data rather than UK-specific data
otherHealth state utility values uncertain and differ from other cannabidiol appraisals (Lennox-Gastaut and Dravet syndromes); no literature values available for tuberous sclerosis complex; vignette method used but plausibility not supported by data
otherCarer health state utilities could not be derived from clinical trial; values from vignette study are uncertain
otherTAND prevalence rates, costs and utility benefits not based on clinical trial data; Delphi panel did not reach full consensus on response level needed to delay TAND; potential for double-counting if vignettes already capture TAND aspects
otherSeveral uncaptured benefits of cannabidiol not modelled: quality-of-life benefit from reducing severity of seizures, reduction in hospitalisation from severity reduction, physical and mental benefit to wider family including siblings, mortality benefit from reducing SUDEP
short follow upHealthcare resource use likely overestimated in company model; unclear whether TAND improvements from seizure reduction occur in adults or only early ages

Commercial arrangement

simple discount pas · confidential

Special considerations

End of life considered (not met) Severity modifier applied Innovation acknowledged Equality issues raised

Cross-references

TA423comparator guidance — NICE technology appraisal guidance on cannabidiol with clobazam for Dravet syndrome
TA615comparator guidance — NICE technology appraisal guidance on cannabidiol with clobazam for Lennox-Gastaut syndrome
utility reuse — Fenfluramine for treating seizures associated with Dravet syndrome - utility values and health state approach