Pembrolizumab with platinum- and fluoropyrimidine-based chemotherapy for untreated advanced oesophageal and gastro-oesophageal junction cancer

Company: Assumed equivalent efficacy between dual and triple regimens; used dual therapy regimen in model

ERG: Stated that all relevant scope comparators, including triple therapy should be considered

Committee: Dual chemotherapy regimen is appropriate comparator; clinical experts confirmed triple therapy use is declining

ICER impact: negligible

Company: Base case used cisplatin and fluorouracil as per KEYNOTE-590; presented scenario with blended comparator arm based on UK market share

ERG: Provided scenario analysis with oxaliplatin and capecitabine, assumed equal efficacy to cisplatin/fluorouracil

Committee: ERG's scenario using oxaliplatin and capecitabine is most reflective of current clinical practice, but comparable efficacy between combinations means choice has little effect on cost-effectiveness

ICER impact: negligible

Company: Included PD-L1 testing costs as additional cost to current care

ERG: Agreed with including PD-L1 testing costs

Committee: Introducing PD-L1 testing unlikely to add significant burden to NHS; appropriate to include testing costs

ICER impact: negligible

Company: Dual regimen of cisplatin and fluorouracil from KEYNOTE-590 appropriate for decision making; both company and ERG scenarios had negligible effect on cost-effectiveness

ERG: Oxaliplatin and capecitabine more reflective of current clinical practice; provided scenario analysis with this regimen

Committee: ERG's oxaliplatin and capecitabine scenario most reflective of clinical practice, but concluded that cisplatin and fluorouracil from company base case was appropriate for decision making due to comparable efficacy and negligible effect on cost-effectiveness estimate

ICER impact: negligible

Company: Used Kaplan-Meier data with log-logistic extrapolation from 40 weeks; clinical evidence supports sustained treatment benefit without treatment waning

ERG: Preferred log-logistic piecewise model plus treatment waning (5-7 years) to address uncertainty about lifetime treatment effect; stated all four scenarios plausible

Committee: All 4 scenarios presented are plausible and resulted in range of cost-effectiveness estimates; company and ERG scenarios not widely different

ICER impact: uncertain_direction

Company: Log-logistic extrapolation from 40 weeks appropriate; clinical evidence indicates sustained treatment benefit with pembrolizumab; treatment waning conservative; generalised gamma has poor statistical fit

ERG: Log-logistic piecewise model plus treatment waning (5-7 years) preferred as it produces survival estimates in middle of range and addresses uncertainty around lifetime treatment effect

Committee: All 4 scenarios plausible; acknowledged long-term uncertainty in overall survival; scenarios preferred by company and ERG not widely different

ICER impact: increases

Company: Clinical evidence indicates sustained treatment benefit with pembrolizumab; treatment waning assumption is conservative

ERG: Included treatment waning effect between 5-7 years as more plausible assumption addressing uncertainty

Committee: Clinical experts agreed small proportion could be cured or in long-term remission; unclear if treatment waning occurs but reasonable assumption

ICER impact: uncertain_direction

Company: Updated base case to include costs of nivolumab based on KEYNOTE-590 CPS 10+ subgroup who received anti-PD-1/PD-L1 treatment

ERG: Company approach most suitable as it captures outcomes without assumptions about efficacy; however, some KEYNOTE-590 patients received anti-PD-1/PD-L1 after pembrolizumab (not UK practice) and proportion receiving subsequent treatment likely lower than in UK practice; provided alternative scenario but with limitations regarding uncertainty on nivolumab's overall survival impact

Committee: Both company and ERG approaches had limitations; appropriate to include costs of nivolumab as subsequent treatment

ICER impact: negligible

Company: Used time-to-death approach

ERG: Not fully specified in this chunk

Committee: Progression-based utilities preferred because values are more plausible

ICER impact: uncertain_direction

Company: Time-to-death approach more appropriate as it captures more health states important for short life expectancy condition; capped utility values for group >1 year from death to general population values; EQ-5D collected once from progressed disease patients

ERG: Time-to-death approach reasonable but produced implausibly high utility values compared to general population, especially >1 year from death; progression-based approach values more plausible; not able to fully evaluate interaction approach

Committee: Progression-based utilities preferred because values were more plausible; more common in cancer appraisals; ERG unable to fully critique interaction approach

ICER impact: increases