The efficient stored specimen design for evaluating multiple screening technologies: Application to multicancer detection tests

ObjectiveMulticancer detection (MCD) tests use blood specimens to screen for various cancers with the hope of reducing cancer mortality. Various companies are developing MCD tests, each with a different technology. For evaluating <i>k</i> MCD tests, a traditional design randomizes participants to either a group with no MCD testing or one of <i>k</i> groups each receiving a different immediate (administered at blood draw) MCD test. To reduce sample size, investigators can store some specimens for delayed MCD testing. A standard stored specimen (SSS) design randomizes participants to either a control group with <i>k</i> different delayed MCD tests applied to the same stored specimen or one of <i>k</i> groups each receiving a different immediate MCD test. Our goal was to further increase efficiency via a novel design and analysis.MethodsThe efficient stored specimen (ESS) design randomizes participants to delayed MCD testing, as in the SSS design, or a single group with <i>k</i> different immediate MCD tests applied to the same specimen. We developed a simple method to separately evaluate each MCD test when any positive MCD test yields a work-up.ResultsThe sample size per randomization group is the same for the ESS and SSS designs. For evaluating <i>k</i> MCD tests, the ESS design involves 2 randomization groups while the SSS design involves <i>k</i> + 1 randomization groups of the same size.ConclusionThe ESS design is superior to the SSS design because, without additional assumptions, it substantially reduces sample size when evaluating multiple MCD tests.