Although concurrently randomized controls remain the gold standard for evaluating investigational therapies, the use of historical controls represents more efficient and economical options. Because historical controls derived from a single clinical trial have the biases of that trial, it may be preferable to synthesize controls from multiple trials. Researchers utilized a proprietary clinical trial database to develop a synthetic control arm for a particular phase 1/2 single-arm trial in acute myeloid leukemia (AML).
The synthetic control arm was built using data from 7 relapsed/refractory AML trials completed in the past 5 years. The 7 trials had similar eligibility criteria to the particular phase 1/2 trial under study. Patients who had baseline covariates matching the patients in the trial were selected for the synthetic control arm.
The utility of this approach was addressed by establishing early end points as predictors of long-term clinical outcomes. The primary outcomes were complete remission (CR; with or without hematologic recovery) and overall survival (OS). The synthetic control arm included 340 patients.
The researchers concluded that the example synthetic control arm identified well-defined patients for whom a CR or CR without hematologic recovery is associated with longer OS. Synthetic control arms provide improved context for the interpretation of uncontrolled experimental results, allowing for quantitative estimation of treatment effect size and more reliable decision-making in the early phases of clinical trials. Synthetic control arms may represent an efficient and informative adaptive approach to clinical trial design in some circumstances.
Berry DA, et al. ASCO Abstract 7021.