TA919 · STA

Rimegepant for treating migraine

Recommended with restrictionsSeptember 2023

Only for people with acute migraine with or without aura in adults where: at least 2 triptans were tried and did not work well enough, OR triptans were contraindicated or not tolerated AND NSAIDs and paracetamol were tried but did not work well enough

Source documents

Intervention

rimegepant (Vydura)
calcitonin gene-related peptide receptor antagonist · CGRP receptor antagonist · oral

Conditions

migraine with auraneurology_psychiatry · acute
migraine without auraneurology_psychiatry · acute
migraineneurology_psychiatry · acute
episodic migraineneurology_psychiatry · acute
chronic migraineneurology_psychiatry · acute

Comparators

NameType Established Committee preferred
placeboplacebo
best supportive carebest supportive careYesYes

Clinical trials

TrialDesignPhasePivotal
BHV3000-301RCTphase 3Yes
BHV3000-302RCTphase 3Yes
BHV3000-303RCTphase 3Yes
BHV3000-310RCTphase 3
BHV3000-201observationalphase 2/3

Economic model

other (company)
Time horizon: 2 years
Cycle length: 48-hour decision tree phase followed by Markov phase

Methodological decisions (10)

comparator selection

Whether placebo is appropriate comparator after triptans have failed or are contraindicated

Company: Placebo represents best supportive care, appropriate comparator given no other treatment options available

ERG: Agreed placebo was most appropriate comparator

Committee: Placebo most appropriate comparator representing best supportive care

ICER impact: negligible

model structure

Baseline monthly migraine days distribution - whether to use observed data from BHV3000-201 or model using Poisson distribution.

Company: Preferred observed data from BHV3000-201 as natural distribution of full range of MMD data.

ERG: Preferred Poisson distribution as it aligned with expected distribution for acute treatment and distribution observed for migraine prevention. Observed data was sporadic.

Committee: Poisson distribution of BHV3000-201 trial data should be used to model baseline MMDs.

ICER impact: uncertain_direction

model structure

Appropriate time horizon for acute migraine treatment model - 2 years vs 20 years.

Company: 20 years more appropriate. Migraine is chronic and recurs across person's life. Acute migraine attacks occur over at least 20 years. Consistency with preventative rimegepant appraisal which used 20 years. 31% of people remained on treatment at 5 years.

ERG: 2 years sufficient. Costs and benefits of acute treatment occur immediately. Should reflect differences in costs and health-related quality of life for each specific attack of short duration.

Committee: 2-year time horizon sufficient. Although migraine is chronic and lifelong disease, 2 years captures all cost and benefit differences of each migraine attack. Effect on cost-effectiveness between 2 and 20 years now small after model corrections.

ICER impact: negligible

mortality assumption

Placebo response trajectory - duration and mechanism. Whether placebo response is only expectation effect or includes natural resolution and other factors.

Company: Placebo response duration of 1-2 years is plausible. Response unlikely for people not on treatment. Base case conservative as no costs for placebo. Model includes natural migraine resolution.

ERG: People on placebo will have some form of treatment and may have response. Not including placebo costs conservative. Scenario with placebo healthcare resource use costs inappropriate as not applied to rimegepant arm.

Committee: No loss of placebo response in model. RCTs designed to identify difference between active treatment and no active treatment. Placebo response includes natural resolution, other medicines, regression to mean - not just expectation effect. Little potential for natural resolution at 2 hours. Effects seen in placebo arm would also be in rimegepant arm.

ICER impact: increases

population generalisability

Applicability of trial population to the proposed NHS population - company proposed narrower population (post-hoc subgroup of people who stopped 2+ triptans) versus ERG preference for modified intention-to-treat (mITT) full trial population

Company: Used post-hoc subgroup analysis (9.3% of pooled RCTs) as main evidence, amended prespecified subgroup to include people who stopped triptans due to both efficacy and intolerability after at least 1 administration route failed

ERG: Preferred modified intention-to-treat (mITT) population because it is larger, includes people who cannot take triptans, and reduces risk of bias from post-hoc subgroup amendment

Committee: mITT population is most appropriate, allows use of all trial data including BHV3000-310 study, reduces risk of bias from post-hoc subgroup definition amendment

ICER impact: decreases

population generalisability

Generalisability of trial results (conducted in people with 2-8 migraines per month) to people with chronic migraines (15+ headache days per month)

Company: No evidence of differences expected between episodic and chronic populations; few medication overuse headache events in long-term study

ERG: Generalisability unresolvable without comparative evidence; concerned chronic migraines are harder to treat due to medication overuse headache risk

Committee: Trial results generalisable to both episodic and chronic populations despite potential differences, acknowledging severity variation over time

ICER impact: uncertain_direction

population generalisability

Inclusion of BHV3000-310 trial (conducted in China/Korea) in analyses versus exclusion due to perceived cultural differences in pain reporting

Company: Trial not included initially because could not extract triptan-failure subgroup; trial did not reflect UK population due to cultural differences in pain reporting

ERG: BHV3000-310 should be included as it provides additional data on approved dispersible tablet formulation; no evidence of cultural differences in pain reporting based on baseline severe pain proportions

Committee: BHV3000-310 should be included; excluding it increases uncertainty; treatment arms within same country means relative effects still informative

ICER impact: decreases

population generalisability

Whether trial results for episodic migraines (2-8 attacks per month) are generalisable to chronic migraines. Clinical experts stated chronic migraines are more treatment-resistant and it may not be appropriate to extrapolate. However, committee accepted results are generalisable to both populations.

Company: No evidence of differences in effectiveness expected between episodic and chronic migraines. Few medication overuse headache events in long-term study BHV3000-201.

ERG: Generalisability unresolvable without comparative evidence. Medication overuse headache is a bigger problem for chronic migraines.

Committee: Trial results are generalisable to both episodic and chronic migraines, acknowledging severity can vary over time and distinction is not clear cut.

ICER impact: uncertain_direction

stopping rule

Whether response to single rimegepant dose should inform subsequent responses. Company model assumed no response to first dose means never respond. Clinical experts said treatment typically tried for 2-3 attacks.

Company: Response to single dose would inform subsequent responses. This is a built-in stopping rule. Treatment anticipated to align with trial design (stop after 1 dose if no response). Full pack costed for those with no response to account for possible wastage.

ERG: Stopping rules based on anticipated practice, but decision to stop based on clinician-patient discussion.

Committee: People and clinicians should consider stopping treatment if no response after 2-3 attacks. Multiple doses would likely be tried in practice before stopping. Formal stopping recommendation not needed.

ICER impact: decreases

treatment effect waning

Whether rimegepant taken as needed for acute treatment causes long-term reduction in monthly migraine days (preventative effect).

Company: When rimegepant taken as needed for acute treatment, there will be long-term reduction in MMDs, based on 1-year follow-up from BHV3000-201.

ERG: Results highly uncertain - post-hoc analysis of uncontrolled study without comparator group, randomisation or blinding. Long-term data lacking.

Committee: Biological plausibility acknowledged but insufficient clinical evidence. Assumption removed from base case. Questioned whether additional preventative effect would occur if patient already on approved CGRP preventative treatment. Should not be included in model but may be considered small uncaptured benefit.

ICER impact: decreases

Evidence gaps

short follow upClinical trials only included people with 2-8 migraines per month, excluding those with chronic migraines (15+ headache days per month), limiting generalisability of effectiveness data to chronic migraine population
single arm evidence onlyNo comparative head-to-head trial data available for subgroup who cannot have triptans; only subgroup analyses available with baseline imbalances for migraine severity, aura and most bothersome symptom
no direct comparisonNo direct evidence of differences in effectiveness between episodic and chronic migraines. Trials only included people with 2-8 migraines per month.
single arm evidence onlyReduced monthly migraine days based on post-hoc analysis of uncontrolled long-term study BHV3000-201 with no comparator group, randomisation or blinding.
short follow upLong-term evidence limited. Lack of long-term data on how response to single attack predicts response for future attacks.

Special considerations

Innovation acknowledged Equality issues raised