BENCHMARKING IMPOWER133 AND EXTENDING EVIDENCE TO ECOG PS2 PATIENTS USING A REGISTRY-BASED TARGET TRIAL EMULATION
Author(s)
Erik Lampa, PhD1, Jens Ellingsen, PhD2, Marcus Skribek, PhD3, Anders Berglund, PhD1, Johan Isaksson, PhD4.
1Epistat AB, Uppsala, Sweden, 2Uppsala University Hospital, Uppsala, Sweden, 3Karolinska Institutet, Stockholm, Sweden, 4Uppsala University, Uppsala, Sweden.
1Epistat AB, Uppsala, Sweden, 2Uppsala University Hospital, Uppsala, Sweden, 3Karolinska Institutet, Stockholm, Sweden, 4Uppsala University, Uppsala, Sweden.
OBJECTIVES: Patients with ECOG performance status (PS) 2 are underrepresented in randomized clinical trials (RCTs), limiting evidence on the effectiveness of first-line chemoimmunotherapy in extensive-stage small-cell lung cancer (ES-SCLC). We benchmarked a target trial emulation (TTE) of atezolizumab plus carboplatin-etoposide versus carboplatin-etoposide against IMpower133 and subsequently applied the framework to PS2 patients.
METHODS: Using the Swedish National Lung Cancer Registry, we emulated a target trial among patients with ES-SCLC initiating first-line treatment. In PS0-1 patients, inverse probability of treatment weighting (IPTW) was used to estimate the average treatment effect (ATE) and benchmark overall survival (OS) against IMpower133. Patients receiving chemotherapy alone were identified in 2018-2020 and those receiving chemoimmunotherapy in 2021-2023. Following benchmarking, the framework was applied to PS2 patients using ATT weighting to estimate treatment effectiveness in a population underrepresented in the RCT.
RESULTS: Compared with IMpower133, registry patients in the benchmarking cohort (N=270, 65 receiving chemoimmunotherapy) were older, more often female, and had PS1 more frequently. After IPTW, the estimated OS hazard ratio (HR) was 0.72 (95% CI 0.51-1.01), closely matching the IMpower133 estimate (HR 0.76). Survival estimates at 12 and 18 months were similar to those observed in the trial, and subgroup effects were generally directionally consistent. No evidence of secular survival trends was identified. Among PS2 patients (N=393; 23 receiving chemoimmunotherapy), treatment effects were strongly time-varying. An apparent early survival advantage attenuated during follow-up, with HR estimates converging toward 0.6-0.8 by 6 months and remaining imprecise.
CONCLUSIONS: The TTE successfully reproduced IMpower133 despite differences in patient characteristics and treatment era, supporting the validity of the emulation framework. Extension to PS2 patients showed possible benefit, emphasizing both the value and limitations of registry-based causal inference for informing evidence gaps not addressed by randomized trials.
METHODS: Using the Swedish National Lung Cancer Registry, we emulated a target trial among patients with ES-SCLC initiating first-line treatment. In PS0-1 patients, inverse probability of treatment weighting (IPTW) was used to estimate the average treatment effect (ATE) and benchmark overall survival (OS) against IMpower133. Patients receiving chemotherapy alone were identified in 2018-2020 and those receiving chemoimmunotherapy in 2021-2023. Following benchmarking, the framework was applied to PS2 patients using ATT weighting to estimate treatment effectiveness in a population underrepresented in the RCT.
RESULTS: Compared with IMpower133, registry patients in the benchmarking cohort (N=270, 65 receiving chemoimmunotherapy) were older, more often female, and had PS1 more frequently. After IPTW, the estimated OS hazard ratio (HR) was 0.72 (95% CI 0.51-1.01), closely matching the IMpower133 estimate (HR 0.76). Survival estimates at 12 and 18 months were similar to those observed in the trial, and subgroup effects were generally directionally consistent. No evidence of secular survival trends was identified. Among PS2 patients (N=393; 23 receiving chemoimmunotherapy), treatment effects were strongly time-varying. An apparent early survival advantage attenuated during follow-up, with HR estimates converging toward 0.6-0.8 by 6 months and remaining imprecise.
CONCLUSIONS: The TTE successfully reproduced IMpower133 despite differences in patient characteristics and treatment era, supporting the validity of the emulation framework. Extension to PS2 patients showed possible benefit, emphasizing both the value and limitations of registry-based causal inference for informing evidence gaps not addressed by randomized trials.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
CO117
Topic
Clinical Outcomes, Real World Data & Information Systems
Topic Subcategory
Comparative Effectiveness or Efficacy
Disease
Oncology