Use of External Control Data to Adjust for Treatment Switching in Cost-Effectiveness Analyses: Lisocabtagene Maraleucel (Liso-Cel) Versus Standard of Care (SOC) in Second-Line Relapsed or Refractory Large B-Cell Lymphoma

Author(s)

Chandler C1, Liu FF2, Klijn S2, Elsada A3, Proskorovsky I4
1Evidera PPD, Waltham, MA, USA, 2Bristol Myers Squibb, Princeton, NJ, USA, 3Bristol-Myers Squibb, Denham, Buckinghamshire, UK, 4Evidera PPD, Bethesda, MD, USA

OBJECTIVES: In oncology trials, patients sometimes switch from the control arm to investigational therapy if certain criteria are met and permitted by study design. This may lead to challenges in health technology assessment where the decision problem is to compare treatments with current clinical practice, which requires an estimate of long-term overall survival (OS) in the absence of switching. Several statistical methods are commonly used to correct for switching by using data solely from the trial under consideration. Alternatively, observational-based methods historically used for indirect comparisons could be considered when appropriate external control data not confounded by switching exist. This case study analyzed an external control arm (ECA) versus liso-cel in the absence of switching to estimate long-term OS and compare this approach to conventional methods used in adjusting for crossover.

METHODS: An ECA was constructed to adjust for switching by matching and weighting SOC from CORAL (NCT00137995; n=216) to align with the baseline characteristics of liso-cel in TRANSFORM (NCT03575351; n=92) using inverse probability of treatment weighting (IPTW) and average treatment effect on treated weights. Survival times for the ECA were compared with liso-cel in a Cox regression analysis and extrapolated using mixture cure models for the cost-effectiveness analyses. Counterfactual survival and relative efficacy based on 2 conventional methods, rank-preserving structural failure time model (RPSFTM) and 2-stage estimation (TSE) using the Weibull distribution, were compared with IPTW analysis.

RESULTS: Using IPTW-adjusted ECA, the OS hazard ratio (HR) (95% confidence interval [CI]) for liso-cel versus SOC in the absence of switching was 0.45 (0.29‒0.71). RPSFTM and TSE methods resulted in HRs (95% CI) of 0.28 (0.15‒0.54) and 0.42 (0.25‒0.69), respectively.

CONCLUSIONS: Findings were consistent between IPTW and TSE analyses. Adjusting for switching using ECA may be useful to supplement results from conventional analyses to demonstrate the range of estimates associated with the methodology used.

Conference/Value in Health Info

2023-11, ISPOR Europe 2023, Copenhagen, Denmark

Value in Health, Volume 26, Issue 11, S2 (December 2023)

Code

MSR78

Topic

Methodological & Statistical Research

Disease

Oncology, Personalized & Precision Medicine

Explore Related HEOR by Topic


Your browser is out-of-date

ISPOR recommends that you update your browser for more security, speed and the best experience on ispor.org. Update my browser now

×