TA226 · STA

Rituximab for the first-line maintenance treatment of follicular non-Hodgkin's lymphoma

RecommendedMay 2011

Source documents

Intervention

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

Conditions

follicular non-hodgkin's lymphomahaematology · advanced
follicular non-hodgkin's lymphomahaematology · maintenance
follicular non-hodgkin's lymphomahaematology · first-line maintenance

Comparators

NameType Established Committee preferred
observation (no treatment)no treatmentYes
ibritumomab tiuxetanactive drug
observationno treatmentYes
watchful waitingno treatmentYes

Clinical trials

TrialDesignPhasePivotal
PRIMA trial (Primary Rituximab and Maintenance)RCTIIIYes
EORTC 20981 studynot statednot stated
ECOG 1496 studyRCT
SAKK 35/98 studyRCT
PRIMA trialRCTphase IIIYes
PRIMARCTYes
ECOG 1496RCT
SAKK 35/98RCT

Economic model

markov (company)
Time horizon: 6 years (base case), sensitivity analyses at 20 and 30 years
Cycle length: not specified

ICER

Below £20,000 (Rituximab maintenance vs observation) · moderate uncertainty
£20,000–£30,000 (rituximab maintenance versus observation) · moderate uncertainty
£20,000–£30,000 (rituximab maintenance vs observation) · moderate uncertainty
£20,000–£30,000 (rituximab maintenance vs comparator) · moderate uncertainty

Methodological decisions (29)

comparator selection

Ibritumomab tiuxetan was listed as a comparator in the decision problem but was excluded from the economic model

Company: Ibritumomab tiuxetan not considered appropriate because of limited evidence of benefits and limited UK prescribing

ICER impact: uncertain_direction

cost assumption

Economic model did not include disutility values for grade 3 and 4 adverse events or for receiving chemotherapy

Company: Costs associated with grade 3 and 4 adverse events were included but disutility values were not included

ICER impact: uncertain_direction

cost assumption

Inclusion and costing of adverse events

Company: Did not include disutilities for grade 3 and 4 adverse events; assumed equivalent costs

ERG: Noted omission would favour rituximab maintenance because rituximab arm experiences more adverse events; concerned about underestimation of PF2 costs

ICER impact: decreases

model structure

Model structure and flexibility for sensitivity analyses

Company: Used three-state Markov model

ERG: Noted model structure did not allow sufficient flexibility for requested sensitivity analyses; had to adjust outcomes/costs post-hoc rather than adjust parameters

ICER impact: uncertain_direction

mortality assumption

Conversion rate from progression-free survival to overall survival

Company: 89.2% conversion rate in revised base-case

ERG: Not explicitly stated but implied concern about high conversion rate

Committee: Sensitivity analyses with 70%, 80%, 90% conversion rates plausible; clinical specialists indicated expectation of at least 70%

ICER impact: uncertain_direction

other

Early stopping bias adjustment: ERG proposed adjusting hazard ratio from 0.55 to 0.719 to account for PRIMA trial stopping earlier than planned

Company: Not explicitly stated

ERG: Evidence suggests studies that stop earlier often overestimate clinical benefit; hazard ratio should be adjusted for early reporting bias

Committee: Adjusting for early reporting bias is not routinely included in technology appraisals and is not a current requirement in NICE methods guide

ICER impact: increases

population generalisability

Applicability of EORTC 20981 study data for economic modelling

Company: Used EORTC 20981 data to inform adverse event costs

ERG: Questioned reliability because EORTC 20981 participants received different induction treatments; only half received rituximab-containing induction

ICER impact: uncertain_direction

population generalisability

Age adjustment: median age in PRIMA trial was 57 years but UK clinical practice typically sees mean age 60-65 years at start of first-line treatment

Company: Revised base case using mean age of 62.5 years appropriately reflects UK clinical practice

ERG: Manufacturer's age adjustment method did not reflect prognostic importance of incident age; proposed adjusting hazard ratios for specific age groups

Committee: Satisfied that manufacturer's base-case analysis appropriately adjusted for age; ERG's hazard ratio adjustment for different age groups not needed

ICER impact: decreases

population generalisability

Age of trial population versus NHS practice

Company: Median age 57 years at randomisation in PRIMA trial

ERG: Added uncertainty regarding applicability to UK practice where mean age typically 60-65 years

Committee: Manufacturer's revised base-case with mean age 62.5 years at induction appropriately adjusted

ICER impact: decreases

proportional hazards

Early reporting bias adjustment for progression-free survival hazard ratio

Company: Used unadjusted progression-free survival HR of 0.55 from PRIMA trial

ERG: Adjusted HR to 0.719 (30.7% increase) to account for early reporting bias based on Bassler et al. 2010 metaregression

ICER impact: increases

proportional hazards

Hazard ratio adjustment for early trial closure bias in PRIMA trial

Company: Hazard ratio of 0.55 for PFS without adjustment

ERG: Early closure may overestimate benefit; hazard ratio should be increased to adjust for bias

Committee: Adjustment for early reporting bias not routinely included in NICE appraisals and not current requirement in NICE methods guide; manufacturer's hazard ratio of 0.55 appropriate

ICER impact: increases

survival extrapolation

Manufacturer used Gompertz function to extrapolate progression-free survival data beyond the end of PRIMA trial (which ended at 25 months median follow-up) to estimate median progression-free survival

Company: Gompertz function provided better fit than alternative functions

ICER impact: uncertain_direction

survival extrapolation

Choice of parametric function for extrapolating progression-free survival beyond trial follow-up (4 years)

Company: Used Gompertz parametric function as base case, which generated highest overall estimate compared with exponential function

ERG: Expressed concerns about use of Gompertz function and noted that if other functions were used, conversion rates would need to be higher for ICER below £30,000

ICER impact: increases

survival extrapolation

Conversion of progression-free survival gain to overall survival gain

Company: Base case assumed 89.2% conversion rate; questioned plausibility of lower rates

ERG: Predicted at least 50% conversion rate needed for ICER below £30,000; explored 70%, 80%, 90% rates in sensitivity analyses

ICER impact: increases

survival extrapolation

Maturity of overall survival data

Company: Used available PRIMA trial data with follow-up to 4 years

ERG: Noted follow-up data not available beyond 4 years and median time to progression/death could not be estimated by group; cautioned data immature with few events

ICER impact: increases

survival extrapolation

Extrapolation of progression-free survival benefit beyond PRIMA trial period; manufacturer assumed proportional increase in survival over time

Company: Proportional increase in survival is appropriate for extrapolation

ERG: Manufacturer's extrapolation assuming proportional increase may not reflect true effect

Committee: Not explicitly stated; deferred to clinical opinion on plausibility of duration assumptions

ICER impact: uncertain_direction

survival extrapolation

Extrapolation method for clinical benefit beyond observed PRIMA trial period

Company: Proportional increase in survival with time assumed in base case

ERG: Method may not reflect true effect

Committee: Acknowledged uncertainty but accepted manufacturer's sensitivity analyses

ICER impact: uncertain_direction

treatment effect duration

Duration of clinical benefit from rituximab maintenance treatment assumed to be sustained over 6 years (4 years beyond the end of treatment in PRIMA trial). After 6 years, risk of disease progression assumed equal in both arms

Company: Assumed 6-year duration based on PRIMA trial results, EORTC 20981 study and expert opinion

Committee: The model was sensitive to assumptions regarding duration of treatment effect; when treatment effect stopped after 47 months instead of 72 months, ICER increased to £21,151 per QALY

ICER impact: increases

treatment effect duration

Duration of clinical benefit of rituximab maintenance treatment: manufacturer assumed 6 years (2 years treatment + 4 years sustained benefit), ERG suggested 28 months based on patient-level data, clinical specialists suggested 3-4 years (1-2 years beyond treatment)

Company: 6 years duration of benefit is plausible, supported by EORTC 20981 study for second-line treatment and clinical opinion

ERG: 28 months based on patient-level data from PRIMA trial after which progression rate no better than observation

Committee: Satisfied that manufacturer's sensitivity analyses (28, 36, and 48 months) presented most plausible range in line with clinical opinion and available data

ICER impact: increases

treatment effect duration

Duration of clinical benefit of rituximab maintenance treatment

Company: 6 years (2 years of treatment and 4 years of sustained benefit)

ERG: 28 months based on patient-level data from PRIMA trial

Committee: Sensitivity analyses with 28, 36 and 48 months considered plausible; clinical specialists indicated 3-4 years (1-2 years beyond treatment) likely, with up to 6 years plausible

ICER impact: increases

treatment effect duration

Duration of clinical benefit from rituximab maintenance treatment

Company: Not explicitly stated in this chunk

Committee: 36 to 48 months, in line with clinical opinion

ICER impact: decreases

treatment effect waning

Duration of clinical benefit from rituximab maintenance therapy

Company: Base case assumed clinical benefit sustained for 72 months (6 years); also presented sensitivity analyses with 28, 36, and 48 months

ERG: Cautioned that if gain in progression-free survival progressively declines, ICER could substantially increase; suggested benefit may last only 28 months based on cumulative hazard plots

ICER impact: increases

treatment sequencing

Modelling of post-progression treatments

Company: Used post-progression treatments from PRIMA trial

ERG: Stated that treatments were in line with UK practice but unsure whether timing in trial reflected routine practice

ICER impact: uncertain_direction

utility source

Health-related quality-of-life utilities derived from Pettengel et al. 2008 study (UK-based, 215 patients with follicular lymphoma) rather than EQ-5D questionnaires from PRIMA trial itself

Company: PRIMA trial participants did not routinely complete EQ-5D questionnaires; used Pettengel study instead with values: 0.88 (disease free/PF1), 0.79 (complete response/PF2), 0.62 (active disease/PD)

ICER impact: uncertain_direction

utility source

Health utility values for progression-free health states

Company: Used 0.88 for PF1 (disease free) and 0.79 for PF2 (complete response to therapy) from Pettengel et al. 2008

ERG: Considered same utility value should be used in both states because both represent remission/full response; conducted sensitivity analysis using 0.79 for both

ICER impact: increases

utility source

Health-related quality of life differences between treatment arms

Company: Model includes utility values 0.88 and 0.79 for PF1 and PF2 health states

ERG: Sensitivity analyses on utility values

Committee: Changes to utility gains had marginal effect on ICER; ICER largely driven by overall survival gains

ICER impact: negligible

utility source

Inclusion of utility associated with delaying chemotherapy in the model

Company: Model did not include utility of delaying chemotherapy

Committee: Committee acknowledged that inclusion of utility from delaying chemotherapy would decrease the ICER

ICER impact: decreases

utility value choice

Conversion rate from progression-free survival to overall survival benefit: manufacturer assumed 89.2% conversion in revised base case

Company: 89.2% conversion rate is appropriate; conversion rate was not a direct model input but required artificial modification of parameters for sensitivity analysis

ERG: Should have sought patient registry or observational data to validate conversion rate; rate of 89.2% may be overestimated

Committee: Satisfied that sensitivity analyses with 70%, 80%, and 90% conversion rates provided plausible range

ICER impact: decreases

utility value choice

Utility gain from delaying chemotherapy not included in model

Company: Unable to incorporate because of limitations in model structure

ERG: Not stated

Committee: Acknowledged omission would decrease ICER if included (improve cost effectiveness)

ICER impact: increases

Evidence gaps

immature overall survivalAfter 36 and 38 months median follow-up in PRIMA trial, statistically significant difference in overall survival could not be established between rituximab maintenance and observation arms because of low number of deaths
short follow upPRIMA trial primary outcome met at interim analysis after median follow-up of 25 months; trial closed but observational follow-up continued to 36-38 months. Direct data extends only to 38 months, requiring extrapolation for lifetime horizon
surrogate not validatedManufacturer extrapolated progression-free survival using Gompertz function beyond the trial data to estimate long-term overall survival outcomes, but overall survival benefit not yet demonstrated in trial
short follow upFollow-up data not available beyond 4 years in PRIMA trial; median time to progression and death could not be estimated by treatment group
immature overall survivalFew events observed in PRIMA trial (up to 38 months data analysis); large proportion of patients censored by 1600 days (70% rituximab, 50% observation)
surrogate not validatedAssumption that 50-100% of progression-free survival gain translates to overall survival gain lacks strong trial evidence
immature overall survivalInsufficient deaths in PRIMA trial to estimate overall survival benefit; required extrapolation beyond trial period
short follow upPost-study observational follow-up median of 38 months but concerns about potential confounding from other therapies received after trial
immature overall survivalSmall number of deaths during PRIMA trial period; overall survival associated with rituximab maintenance could not be estimated
surrogate not validatedConversion rate from progression-free survival to overall survival not verified from literature or clinical experience
short follow upPost-study observational follow-up had median of 38 months; uncertainty about benefit duration beyond this

Commercial arrangement

none

Special considerations

Innovation acknowledged

Cross-references

TA110precedent — Rituximab for treatment of follicular lymphoma; recommends R-CVP for first-line induction treatment of advanced follicular NHL
TA137same condition — Rituximab for treatment of relapsed or refractory stage III or IV follicular non-Hodgkin's lymphoma (review of TA37); covers rituximab maintenance after second-line treatment
TA110same drug — Rituximab for the treatment of follicular lymphoma