Increasing power and robustness in screening trials by testing stored specimens in the control arm

Area of Science:

• Biostatistics
• Cancer Screening Trials
• Clinical Trial Design

Background:

• Screening trials need large sample sizes and long durations to show mortality benefits.
• The Intended Effect (IE) design boosts statistical power by testing stored control-arm specimens.
• The US National Cancer Institute is evaluating the IE design's feasibility in a pilot study.

Purpose of the Study:

• To relax key assumptions of the IE design for improved feasibility and robustness.
• To investigate cost-efficiency through stratified sampling of control-arm specimens.
• To assess the impact of "unintended effects" and non-compliance on IE design performance.

Main Methods:

• Incorporated inverse-probability sampling weights to reduce costs by testing a subset of specimens.
• Developed methods to correct for potential loss-of-signal in stored specimens.
• Addressed non-compliance with control-arm specimen collection and examined sensitivity to unintended screening effects.

Main Results:

• Testing a 50% sample of control-arm specimens maintained nearly all IE design power while halving specimen testing.
• The IE design generally maintained power gains over standard analysis, even with unintended effects.
• Corrections mitigated bias and power loss associated with specimen degradation and non-compliance.

Conclusions:

• The IE design can be optimized for cost-efficiency and robustness against signal loss.
• Unintended screening effects typically do not negate the power advantage of the IE design.
• Corrections effectively address bias and power loss from non-compliance, though practical application is needed.