TA904 · STA

Pembrolizumab with lenvatinib for previously treated advanced or recurrent endometrial cancer

Recommended with restrictionsTechnology Appraisal Committee AMay 2023

Pembrolizumab plus lenvatinib is recommended only if the companies provide them according to the commercial arrangements

Source documents

Interventions

pembrolizumab (Keytruda)
monoclonal antibody · humanised monoclonal antibody acting on the programmed cell death protein-1 immune checkpoint receptor pathway · intravenous
lenvatinib (Lenvima)
· tyrosine kinase inhibitor
pembrolizumab plus lenvatinib (pembrolizumab with lenvatinib)
· pembrolizumab is a PD-1 agent and lenvatinib is a VEGFR/FGFR inhibitor that work synergistically
pembrolizumab with lenvatinib
· PD-1 inhibitor combined with tyrosine kinase inhibitor

Conditions

advanced or recurrent endometrial canceroncology · metastatic
previously treated advanced or recurrent endometrial canceroncology · metastatic

Comparators

NameType Established Committee preferred
paclitaxelactive drugYes
doxorubicinactive drugYes
paclitaxel or doxorubicin monotherapyactive drugYes

Clinical trials

TrialDesignPhasePivotal
KEYNOTE-775RCTphase IIIYes
KEYNOTE-146single_arm1b/2
ECHOobservational
Heffernan (2022)observational

Economic model

partitioned survival (company)
Time horizon: 40 years
Cycle length: 1 week
markov
Time horizon: 40-year
Cycle length: not stated

ICER

£30,000–£50,000 (pembrolizumab with lenvatinib versus paclitaxel or doxorubicin monotherapy) · moderate uncertainty
£30,000–£50,000 (pembrolizumab with lenvatinib vs paclitaxel or doxorubicin monotherapy) · moderate uncertainty

Methodological decisions (10)

crossover adjustment

Adjustment for treatment switching - proportion of control arm patients receiving pembrolizumab plus lenvatinib or other PD-1/PD-L1 or VEGF/VEGFR inhibitors after paclitaxel or doxorubicin monotherapy

Company: Applied two-stage estimation (TSE) method without recensoring; TSE method considered least biased

ERG: Noted TSE method assumes same treatment effect for all treatments after switching, which may not be appropriate given variety of treatments; true effect likely lies between adjusted and unadjusted values

Committee: Used EAG scenario of all patients waning at 5-7 years; noted adjusted result is less biased than unadjusted but the unadjusted is more conservative; true result likely between adjusted and unadjusted

ICER impact: decreases

crossover adjustment

Treatment switching adjustment in comparison between EAG scenarios

ERG: Provided scenarios both adjusted and unadjusted for treatment switching

Committee: Most plausible ICERs between EAG's scenario adjusted for treatment switching and unadjusted for treatment switching

ICER impact: uncertain_direction

model structure

Partitioned survival model with 3 health states and 24-month pembrolizumab stopping rule

Company: Proposed model with 24-month stopping rule matching KEYNOTE-775

ERG: Considered the model structure reasonable

Committee: Model structure generally appropriate

ICER impact: negligible

population generalisability

Average age of population modelled

Company: Used age from KEYNOTE-775 trial

ERG: Clinical experts felt average age in KEYNOTE-775 was slightly lower than likely in UK clinical practice

Committee: Mean age of 67 years, between trial estimate and EAG estimate, consistent with real-world studies ECHO

ICER impact: negligible

survival extrapolation

Choice of parametric curve for extrapolating overall and progression-free survival beyond trial follow-up

Company: Originally used standard parametric and 2-piece parametric curves; later revised to one-knot spline model using odds scale

ERG: Recommended more sophisticated flexible models (cubic splines) rather than standard parametric curves with arbitrary breakpoints

Committee: One-knot spline model was an appropriate choice for extrapolation of both overall and progression-free survival in both arms, given that EAG scenarios testing alternatives had minimal impact on ICER

ICER impact: uncertain_direction

treatment effect duration

Whether treatment effect continues indefinitely or wanes over time after pembrolizumab stops at 2 years

Company: No treatment effect waning for the duration of the 40-year model time horizon; cites evidence from KEYNOTE-775 showing sustained benefit and long-term data from melanoma and other immunotherapy trials

ERG: Some evidence to support some duration of effect after stopping pembrolizumab but not sufficient to conclude no waning; preferred EAG scenario of waning from years 3-5

Committee: Waning from years 5-7 after starting treatment applied to all patients, based on evidence that treatment effect is likely durable but must assume some waning; committee concluded true result likely between adjusted and unadjusted values

ICER impact: increases

treatment effect waning

Application of waning from years 5 to 7 after starting treatment (3 to 5 years after treatment with pembrolizumab stops)

Company: Did not incorporate waning in base case

Committee: Apply waning from years 5 to 7 after starting treatment

ICER impact: increases

treatment sequencing

Adjustment for treatment switching: proportion of comparator arm receiving subsequent PD-1/PD-L1 or VEGF/VEGFR inhibitor therapies not available in NHS practice

Company: Applied two-stage estimation (TSE) method to adjust for treatment switching and incorporated adjusted data in base case

ERG: Noted that committee did not request treatment switching adjustment in preferred base case; company had independently chosen to include it

Committee: Not explicitly stated; committee noted the adjustment gives most optimistic estimate of treatment benefits

ICER impact: increases

utility source

Approach to deriving health-state utilities - time-to-death versus progression status

Company: Time-to-death approach with 6 TTD categories allows finer gradations and captures utilities across full spectrum of disease; more comprehensive for immunotherapy trials with limited post-progression utility assessments

ERG: Progression status approach is more consistent with model structure; company approach divorces health-related quality of life from disease status

Committee: EAG approach of deriving utilities using progression status is more appropriate; committee noted increased granularity from company's 6 TTD categories increases uncertainty; maintained that progression status approach is more appropriate

ICER impact: uncertain_direction

utility source

Using progression status to derive utilities

Company: Did not incorporate this approach in base case

Committee: Using progression status to derive utilities

ICER impact: uncertain_direction

Evidence gaps

no direct comparisonTrial population slightly younger than UK clinical practice; generalisability to older, higher weight patients noted but assessed as having minimal impact
immature overall survivalUncertainty about how long treatment effect lasts after people stop taking pembrolizumab at 2 years
short follow upTrial had median 14.7 months follow-up at final data cut; limited follow-up noted in subgroup analyses by MMR status
short follow upLimited long-term follow-up data from KEYNOTE-775; median follow-up 34.7 months for KEYNOTE-146 with considerable censoring at later timepoints
surrogate not validatedUncertainty regarding progression-free survival as surrogate endpoint and its relationship to overall survival
otherUncertainty in treatment waning assumptions; insufficient evidence to definitively determine appropriate time point when waning should begin

Commercial arrangement

simple discount pas · confidential · critical for recommendation

Special considerations

End of life criteria met Innovation acknowledged Equality issues raised

Cross-references

TA779comparator guidance — Dostarlimab recently appraised but could not be considered as comparator because recommended for use in Cancer Drugs Fund
TA779same condition — Dostarlimab appraisal for dMMR endometrial cancer; committee noted that TA779 used time-to-death utility approach with disease progression as covariate
comparator guidance — Previous NICE appraisals for renal cell carcinoma appraisal where pembrolizumab plus lenvatinib was assessed; committee noted that conclusions from that appraisal should not be applied to this indication