Exploring the validity of a synthetic control arm (SCA) for augmentation or replacement of a randomized control in difficult-to-study indications: A case study in relapsed or refractory multiple myeloma (R/R MM).

Abstract

e20521 Background: The randomized clinical trial (RCT) is the gold standard in drug development. However, for indications where patients have a strong preference for the investigational product, such as many oncology and rare diseases, the use of a SCA may improve drug development and reduce patient burden. SCA is an external control constructed from patient-level data from previous clinical trials to match the baseline characteristics of the patients in an investigational group and can augment a single-arm trial or a RCT compromised by control arm early withdrawal or noncompliance in order to estimate treatment effects. This research explores whether the treatment effect (difference between arms) based on an SCA can mimic the treatment effect from a RCT. Tipping point analyses were explored to assess the impact of unobserved confounders on the SCA-based demonstration of efficacy. Methods: This case study is based on patient-level data from previous clinical trials in R/R MM. The SCA patients satisfied key eligibility criteria of the target RCT and were further selected using propensity score methods to balance the baseline characteristics in the SCA with the target randomized treatment group (TRT) from the original RCT. Sensitivity analyses utilizing methods proposed by Lin (1998) illustrate the robustness of the treatment effect to unobserved covariate(s). Results: Comparable balance was achieved in observed baseline characteristics between SCA and the matched patients from TRT. The treatment effect utilizing SCA is similar to the original RCT. The Kaplan Meier curve of overall survival for the SCA overlaps with that of the randomized control and the quantified differences between SCA and the matched patients from TRT are very similar to the original RCT. Tipping point analyses show changes in HRs under representative sets of assumptions regarding the unobserved confounder (results not shown). Conclusions: This case study demonstrates an SCA built from previous clinical trials, can be well-balanced at baseline with TRT and can provide similar treatment effect estimates as a RCT. Tipping point analyses can elucidate whether treatment effects are reliable despite a reasonable degree of confounding expected in a clinical setting. This suggests, in some settings, SCA can be used to augment or replace a randomized control in future trials without loss of understanding of the treatment effect. [Table: see text]