Single-arm Trials with Historical Controls: Study Designs to Avoid Time-related Biases

Area of Science:

• Clinical Trials
• Oncology
• Biostatistics

Background:

• Single-arm trials with external historical controls are increasingly used by regulatory agencies for drug approval, especially for rare diseases or specific indications.
• This trial design is common in oncology for assessing novel therapeutics.
• However, time-related biases can arise when comparing outcomes over time with external controls.

Purpose of the Study:

• To identify and illustrate potential time-related biases in single-arm trials using external historical controls.
• To demonstrate how these biases can affect treatment effectiveness estimates.
• To emphasize the importance of understanding these biases for valid drug approval.

Main Methods:

• Described two potential biases: calendar time differences and choice of time zero (cohort entry).
• Utilized simulated data emulating a blinatumomab trial for relapsed/refractory acute lymphoblastic leukemia (ALL) mortality.
• Compared blinatumomab arm (2010-2014) with historical controls (1990-2013).

Main Results:

• Non-concurrent calendar times led to apparent survival differences, with median survival of 6.1 months (blinatumomab) vs. 3.3 months (controls), which varied significantly over time.
• Selection bias due to defining control cohort entry by the latest salvage treatment altered the hazard ratio from 0.56 to 0.98.
• Adjusting cohort entry to matched salvage treatment improved the comparison accuracy.

Conclusions:

• Single-arm trials with external controls require careful management of time-related biases.
• Understanding and mitigating biases from calendar time and cohort selection is essential for generating valid evidence for regulatory drug approval.
• Accurate assessment of promising treatments in oncology relies on robust trial designs that account for temporal effects.