TA137 · STA

Rituximab for the treatment of relapsed or refractory stage III or IV follicular non-Hodgkin's lymphoma (review of technology appraisal guidance 37)

Recommended with restrictionsJanuary 2007

Source documents

Intervention

rituximab (MabThera)
chimeric monoclonal antibody · targets the CD20 surface antigen of mature B-cell lymphocytes · intravenous infusion

Conditions

relapsed or refractory stage iii or iv follicular non-hodgkin's lymphomahaematology · relapsed_refractory
follicular non-hodgkin's lymphomahaematology · relapsed_refractory
follicular lymphomahaematology · relapsed_refractory

Comparators

NameType Established Committee preferred
chop chemotherapyactive drugYes
best supportive carebest supportive care
observation onlyno treatment
fludarabine-containing regimensactive drug
rituximab maintenance therapyactive drug
no use of rituximabno treatment

Clinical trials

TrialDesignPhasePivotal
EORTC trialRCTYes
GLSG-FCM trialRCTYes
GLSG trialRCT

Economic model

markov (company)
Time horizon: 30 years
Cycle length: not specified

ICER

£20,000–£30,000 (rituximab induction and maintenance vs rituximab maintenance only) · moderate uncertainty
£20,000–£30,000 (rituximab in combination with chemotherapy as induction therapy) · moderate uncertainty

Methodological decisions (16)

comparator selection

Selection of appropriate comparator for dual-use strategy

Company: Comparison with chemotherapy alone

Committee: Appropriate comparator is single use of rituximab as maintenance therapy, as maintenance therapy is expected to become standard

ICER impact: increases

cost assumption

Administration costs for rituximab infusions

Company: Assumed outpatient setting at £86 per visit

ERG: More appropriate for day-case setting at £504 per visit

Committee: Accepted ERG recalculation

ICER impact: uncertain_direction

cost assumption

Post-progression treatment costs

Company: Used average costs from trial, including wide variation

ERG: Aggregated treatments into meaningful categories (chemotherapy, rituximab, other) to avoid skewing from expensive outliers

Committee: Accepted ERG approach

ICER impact: decreases

cost assumption

Cost of rituximab administration

Company: Not specified in detail

Committee: Cost as day-case procedure rather than outpatient visit, reflecting current rapid administration practice

ICER impact: decreases

cost assumption

Aggregation of post-progression treatment costs

Company: Individual itemisation of treatments

ERG: Aggregate treatments into categories

Committee: Aggregate treatments into categories

ICER impact: decreases

model structure

Treatment of event-free period in model fitting

Company: Lack of initial period of non-zero hazards for groups randomised at maintenance

ERG: Should account for protocol-driven event-free period

Committee: Accepted ERG correction; excluding event-free period changed model fit

ICER impact: decreases

model structure

Treatment of event-free period when fitting survival distributions

Company: Excluded protocol-driven event-free period when fitting Weibull model

ERG: Event-free period should be included to avoid changing goodness-of-fit and influencing cost-effectiveness estimates

Committee: Event-free period should not be excluded; excluding it can change distribution fit and bias results

ICER impact: decreases

proportional hazards

Whether proportional hazards assumption holds for comparing treatment groups

Company: Estimated three parameters assuming common parameters across patient groups

ERG: Assumption not substantiated; should estimate four parameters independently

Committee: Accepted ERG concern; independent fitting resulted in lower ICER of £15,775

ICER impact: decreases

proportional hazards

Assumption that treatment pairs share common parameters in Weibull model

Company: Shared common parameters across patient groups, estimating three parameters rather than four

ERG: Proportional hazards assumption not substantiated; should fit functions independently

Committee: Independent fitting of Weibull functions without proportional hazards assumption

ICER impact: decreases

proportional hazards

Manufacturer initially assumed proportional hazards with only treatment effect differences between fitted distributions. Committee noted this strong assumption was not substantiated by RCT data and could overestimate clinical and cost effectiveness.

Company: Assumed proportional hazards such that only treatment effect differed between distributions

Committee: Relax the proportional hazards assumption and independently fit Weibull functions, resulting in ICER of £15,775 per QALY gained

ICER impact: decreases

survival extrapolation

Choice of Weibull model for extrapolating survival data to 30 years

Company: Used Weibull function to extrapolate hazard rates from trial data (limited to 1500 days) to 30 years

ERG: Weibull choice not sufficiently justified; proportional hazards assumption not substantiated; proposed limiting extrapolation to Kaplan-Meier estimates to 1500 days only

Committee: Accepted ERG concerns but allowed exploration of multiple survival models; manufacturer provided additional analyses with exponential and independently fitted Weibull functions

ICER impact: increases

survival extrapolation

Choice of parametric model for extrapolating survival beyond trial follow-up

Company: Weibull model with proportional hazards assumption across groups

ERG: Weibull model choice not justified; proportional hazards assumption not substantiated by trial data

Committee: Exponential model shown to be best fit; independent fitting of Weibull without proportional hazards assumption

ICER impact: decreases

survival extrapolation

Manufacturer initially excluded the protocol-driven event-free period when fitting parametric distributions to trial data. Committee rejected this approach as excluding the event-free period could change goodness-of-fit and influence cost-effectiveness results.

Company: Excluded the protocol-driven event-free period when initially fitting distributions

Committee: Include the event-free period when fitting distributions; additional manufacturer analysis using Weibull and exponential models resulted in ICERs of £21,379 (Weibull) and £16,183 (exponential) respectively

treatment effect duration

Duration of benefit from rituximab treatment

Company: Assumed benefit equivalent to baseline risk after 5 years

ERG: Suggested limiting to observed trial period (1500 days)

Committee: Sensitivity analyses conducted with varying durations (2-5 years)

ICER impact: increases

treatment effect duration

Duration of treatment benefit assumed in extrapolation

Company: Not explicitly stated

Committee: 5-year duration of treatment benefit over lifetime horizon is reasonable

ICER impact: uncertain_direction

treatment effect duration

Duration of treatment benefit assumption in survival extrapolation model

Committee: 5 years treatment benefit duration, which was considered reasonable

ICER impact: uncertain_direction

Evidence gaps

no direct comparisonEORTC trial not designed or powered to evaluate relative efficacy of rituximab maintenance in patients who had or had not received rituximab for induction
single arm evidence onlyNo new evidence available for last-line use of rituximab as monotherapy when all alternatives exhausted; original TA37 recommendation based on prospective case series
short follow upLimited trial follow-up requiring model extrapolation beyond observed data with inherent uncertainty

Special considerations

Innovation acknowledged

Cross-references

TA37precedent — This appraisal is a review of technology appraisal guidance 37 issued in March 2002
TA110same drug — Rituximab for the treatment of follicular lymphoma
TA65same drug — Rituximab for aggressive non-Hodgkin's lymphoma