TA874 · STA

Polatuzumab vedotin in combination for untreated diffuse large B-cell lymphoma

Recommended with restrictionsAppraisal Committee CNovember 2022

Recommended for untreated DLBCL with an IPI score of 2 to 5, only if the company provides it according to the commercial arrangement

Source documents

Intervention

polatuzumab vedotin (Polivy)
antibody-drug conjugate · CD79b-directed antibody-drug conjugate · intravenous

Condition

diffuse large b-cell lymphomaoncology · early

Comparators

NameType Established Committee preferred
r-chopactive drugYesYes
r-chpactive drugYes

Clinical trials

TrialDesignPhasePivotal
POLARIXRCTphase 3Yes
GOYARCTphase 3

Economic model

partitioned survival (company)
Time horizon: lifetime
Cycle length: weekly

ICER

£20,000–£30,000 (polatuzumab vedotin with R-CHP vs R-CHOP) · high uncertainty
Confidential (PAS-dependent) (polatuzumab vedotin with R-CHP vs R-CHP) · low uncertainty

Methodological decisions (16)

cost assumption

Patient weight distribution to use for vial costing. Trial population mean weight 75.92 kg lower than NHS Health Survey (78.75 kg).

Company: Weight from full POLARIX population (75.92 kg) appropriate; supported by O'Brien et al. showing 5% weight loss in DLBCL patients at diagnosis

ERG: O'Brien et al. population not generalisable to NHS DLBCL patients

Committee: Weight from Western subgroup of POLARIX (80.1 kg) more generalisable to UK population than full trial; unconvinced by weight loss assumption

ICER impact: increases

cost assumption

Progressed disease supportive care costs estimation

Company: Updated approach based on survey of 3 clinicians about second-line treatment resource use; applied same costs every weekly cycle

ERG: Company's approach assumes indefinite supportive care costs; should account for response to subsequent treatments and periods of lower intensity follow-up; estimated 50% reduction in company costs was appropriate based on time to progression estimates

Committee: Company costs likely overestimated; appropriate reduction is between 25% to 50%, with 30% used in final base case

cost assumption

Company's progressed disease supportive care costs likely overestimated; ERG assumed 50% reduction which was likely underestimate

Company: Original progressed disease costs

ERG: 50% reduction in progressed disease costs

Committee: Reduction in progressed disease costs by between 25% to 50% (updated base case used 30%)

ICER impact: decreases

model structure

Technical error in progression-free survival modelling where PFS capped to OS extrapolation. Company made correction but appropriateness unclear.

Company: Correction ensures PFS does not exceed OS by capping at OS extrapolation

ERG: Correction provides counter-intuitive results and inflexible model; unable to adequately scrutinise validity

Committee: Unclear if correction appropriate; issue remained uncertain despite company clarification

ICER impact: uncertain_direction

model structure

Economic model structure for cost-effectiveness analysis

Company: Used 3-state partitioned survival model with health states: progression-free, progressed disease, and death

Committee: Partitioned survival model is standard for cancer drugs and appropriate

population generalisability

Whether to include IPI 0-1 population. Company provided evidence only for IPI 2-5, excluding IPI 0-1.

Company: Restrict to IPI 2-5 as trial excluded IPI 0-1

Committee: Exclude IPI 0-1 as appropriate in line with trial evidence; IPI 0-1 has good outcomes and small proportion of DLBCL population

ICER impact: uncertain_direction

population generalisability

Patient weight distribution from trial vs NHS population

Company: Initially used full POLARIX population weight (75.92 kg); later argued Western subgroup weight (80.1 kg) and applied 5% weight loss assumption based on O'Brian et al. study suggesting generalisability to full population

ERG: Noted O'Brian et al. was done in a non-generalizable population to NHS DLBCL patients

Committee: Western subgroup weight distribution (80.1 kg) is most likely generalizable to NHS DLBCL population; rejected the 5% weight loss assumption based on single non-generalizable study

population generalisability

Committee preferred to consider full POLARIX population rather than IPI 3-5 subgroup scenario analysis

Company: Provided scenario analysis for IPI 3 to 5 subgroup

Committee: Full POLARIX population

ICER impact: uncertain_direction

survival extrapolation

Use of mixture-cure model to extrapolate progression-free and overall survival. Overall survival informed by progression-free survival due to immature overall survival data.

Company: Use mixture-cure model with overall survival informed by PFS; assume continued OS benefit and no waning

ERG: Overall survival extrapolations highly uncertain; applied treatment waning to account for uncertainty and subsequent treatments

Committee: Mixture-cure model is reasonable; company approach more clinically plausible than applying waning to cured population

ICER impact: decreases

survival extrapolation

Uncertainty about the appropriateness of the company's correction to progression-free survival modelling where PFS is capped to match/not exceed OS extrapolation

Company: Corrected the model to cap PFS extrapolation at the point it would meet or exceed OS extrapolation

ERG: Raised concerns that the correction provides counter-intuitive results and the mixture-cure model is inflexible to such changes; unable to adequately scrutinise the correction

Committee: Unclear if the company's correction was appropriate; acknowledged the correction lowered the ICER

survival extrapolation

Company included progression-free survival curve correction after consultation; unclear if appropriate and lowered ICER

Company: Included PFS curve correction

Committee: Removal of the progression-free survival curve correction in updated base case

ICER impact: increases

treatment effect waning

Whether treatment effect for overall survival diminishes over time. Company argues no waning appropriate for curative intent disease; ERG applied waning to account for uncertainty and subsequent treatments.

Company: No waning effect; DLBCL is curable in first line

ERG: Applied waning equal OS in each arm after 60 months to account for uncertainty and subsequent treatments

Committee: No treatment effect waning; company approach more clinically plausible despite uncertainty

ICER impact: decreases

treatment effect waning

Whether treatment effect wanes over time

Committee: No treatment effect waning assumed

treatment sequencing

Inclusion of CAR-T therapies as subsequent treatments

Company: Initially included CAR-T therapies (axicabtagene ciloleucel and tisagenlecleucel) as subsequent treatments; agreed to remove them after technical engagement

ERG: CAR-T therapies are not routinely commissioned; should not be included; when removed, did not redistribute CAR-T use to other treatments (resulting in 97% subsequent treatment use)

Committee: CAR-T therapies should not be included; redistribution of CAR-T therapy use to other subsequent treatments is acceptable and appropriate

utility source

Source of utility values for progressed disease state

Company: Used GOYA trial utility values because of longer follow-up than POLARIX; noted some POLARIX progressed disease patients did not report HRQoL and those who did had better outcomes; timing of GOYA data collection not disclosed

ERG: Agreed with GOYA utility values as similar to TA649; applied age adjustment using UK general population utilities from Ara and Brazier 2010

Committee: GOYA utilities acceptable for decision making but uncertain; would have preferred data collected after second-line treatment; noted toxicity of later line treatments would affect quality of life

utility source

Utility values for progressed disease were uncertain but approach in base case was acceptable

Committee: Base case approach acceptable for decision making

ICER impact: negligible

Evidence gaps

immature overall survivalOverall survival data from POLARIX is immature at 24-month follow-up (HR 0.94, 95% CI 0.65-1.37). Cannot determine if overall survival benefit exists. Future data cuts (August 2022, 2024) unlikely to meaningfully address uncertainty given event rate.
short follow up24-month follow-up insufficient to determine long-term overall survival benefit in mixture-cure model context
immature overall survivalLong-term overall survival estimates highly uncertain
short follow upGOYA utility values collected before second-line treatment started; would expect later line treatment toxicity to impact quality of life

Commercial arrangement

simple discount pas · confidential · critical for recommendation

Special considerations

End of life considered (not met) Cancer Drugs Fund eligible Innovation acknowledged

Cross-references

TA649precedent — Previous NICE appraisal of polatuzumab vedotin; utility values from GOYA used in this appraisal were similar to those in TA649
TA306comparator guidance — NICE TA on pixantrone monotherapy; progressed disease resource data sourced from this TA in company's original submission
TA243comparator guidance — NICE TA on rituximab; ERG used progressed disease resource use data from this TA in original base case
TA559same drug — NICE TA on axicabtagene ciloleucel (CAR-T); currently in Cancer Drugs Fund; committee acknowledged relevance to this appraisal and noted it is being reviewed
TA567same drug — NICE TA on tisagenlecleucel (CAR-T); currently in Cancer Drugs Fund; was initially included by company as subsequent treatment