Expected late-stage reduction under extended screening in the first multicancer screening trial

Abstract The first multi-cancer screening trial did not meet its primary endpoint of reducing late-stage (stage III/IV) incidence after three screening rounds. We used a previously developed multi-cancer natural history model to assess whether continued annual screening could have produced larger benefit. The model was calibrated to match observed late-stage reductions by screening round and aggregate episode sensitivity for the trial’s 12 prespecified cancers, then used to project late-stage reductions over 10 annual screening rounds. Among the 15 best-calibrated models, early-stage preclinical detectable period ranged from 0.46 to 1.36 years, and per-cancer early-stage sensitivities from 30%-60% of previously reported sensitivities for clinically diagnosed cancers. The calibrated models reproduced the excess late-stage incidence in round 1 and increasing reductions thereafter. With continued screening, late-stage reductions by screening round plateaued at 13%-17% (range across 15 models), while cumulative late-stage reduction reached 9%-13% by round 10.