Tafasitamab with lenalidomide for treating relapsed or refractory diffuse large B-cell lymphoma

Company: Used propensity score matching for RE-MIND2 and matching-adjusted indirect comparisons for published trials; justified based on alignment with published outcomes

ERG: Highlighted that adjusting L-MIND population differently for each comparator may have introduced bias; uncertainty about baseline characteristics varying in RE-MIND2; unclear what type of treatment effect is estimated

Committee: Concluded results were very uncertain due to complexity and potential biases in the indirect comparison methods; preferred to see additional analyses such as partially anchored indirect comparisons

ICER impact: uncertain_direction

Company: Used matching-adjusted indirect comparison to compare against bendamustine and rituximab; did not provide alternative analyses such as partially anchored indirect comparisons

ERG: Raised concerns about the company's indirect comparison methodology and whether it reflected absolute benefits of polatuzumab vedotin with bendamustine and rituximab

Committee: Committee was disappointed that company did not provide additional analyses such as partially anchored indirect comparisons. Found that modelling poorly reflected relative benefit compared with bendamustine and rituximab alone, noting that Sehn et al. reported hazard ratio of 0.42 for overall survival but company modelling showed only small difference in survival.

ICER impact: increases

Committee: Committee expressed concern about indirect comparison approach but noted it could not be resolved through Cancer Drugs Fund evidence collection.

ICER impact: uncertain_direction

Company: Company's base case survival estimate for polatuzumab vedotin with bendamustine and rituximab inconsistent with real-world evidence

ERG: ERG base case more closely aligned with NICE's previous technology appraisal guidance on polatuzumab vedotin, though may overestimate survival compared to real-world experience

Committee: Committee accepted appeal panel's conclusion that real-world survival for polatuzumab vedotin with bendamustine and rituximab should be 10-13 months based on Northend et al. and Sehn et al. studies, but noted company and ERG base cases assumed 29 and 48 months respectively. Committee concluded that ICERs reflecting real-world survival would likely be closer to ERG base case or potentially higher.

ICER impact: increases

Company: Piecewise constant hazard ratio approach (separate hazard ratios up to month 4 and after month 4), using matching-adjusted indirect comparison hazard ratios applied to tafasitamab with lenalidomide survival distributions

ERG: Preferred constant hazard ratio from matching-adjusted indirect comparison, leading to higher survival estimates (3.4 mean life years and 2.2 QALYs) aligned with previous NICE appraisal

Committee: Neither piecewise nor constant hazard ratios appropriate; committee considered the company's extrapolations implausible and preferred modelling approaches that fit underlying hazards and reflect both absolute and relative benefits of polatuzumab vedotin with bendamustine and rituximab

ICER impact: increases

Company: Initially chose generalised gamma distribution

ERG: Preferred log-normal distribution

Committee: Agreed log-normal distribution was appropriate, noting uncertainty due to heavy patient censoring in Kaplan-Meier curve

ICER impact: negligible

Company: Initially chose generalised gamma distribution for progression-free survival of tafasitamab with lenalidomide

ERG: Preferred log-normal distribution for progression-free survival, noting the generalised gamma resulted in a hazard profile inconsistent with clinical expert predictions and overestimated long-term progression-free survival

Committee: Log-normal parametric extrapolation of L-MIND progression-free survival data. Company and ERG agreed that log-normal parametric extrapolation of L-MIND overall survival data was most appropriate. Committee noted inherent uncertainty due to heavy patient censoring.

ICER impact: uncertain_direction

Company: Company's base case model estimated mean undiscounted survival of 29 months for polatuzumab vedotin with bendamustine and rituximab

ERG: ERG's base case model estimated mean undiscounted survival of 48 months, more closely aligned with NICE's technology appraisal guidance on polatuzumab vedotin (over 48 months undiscounted)

Committee: Committee concluded that company's parametric extrapolations for polatuzumab vedotin with bendamustine and rituximab were implausible but noted that ERG estimates may also be overestimated. Clinical expert feedback suggested estimates from real-world practice were lower (Northend et al. 10.2 months median, Sehn et al. 12.4 months median). Preferred modelling approaches that fit underlying hazards and produce outcomes more closely reflecting benefits compared to bendamustine and rituximab alone.

ICER impact: uncertain_direction

Company: Base case survival of 29 months for polatuzumab vedotin arm

ERG: Base case survival of 48 months for polatuzumab vedotin arm

Committee: Committee concluded that ICERs based on survival closer to real-world expectations (10-13 months) would likely be higher than company base case, and likely closer to ERG estimates. Did not present alternative analysis with real-world survival assumption.

ICER impact: increases