Exercise-linked serum proteomics reveals a modifiable pre-cancer continuum

Exercise reduces cancer incidence, yet the circulating molecular intermediates linking physical activity to pre-cancer biology remain unknown. Within the Singapore Longitudinal Ageing Studies (n = 6,050; ClinicalTrials.gov NCT03405675), we performed pre-cancer screening based on prospective cancer-free status at baseline serum collection followed by confirmed cancer diagnosis during longitudinal follow-up, with pre-diagnostic samples collected a median 7.76 years before cancer-specific death. Using serum proteomics (>2,400 proteins) across 674 samples - healthy controls (n = 89), pre-cancer individuals (n = 148) and cancer individuals (n = 57) - alongside longitudinal serum samples from two exercise paradigms: long-term unstructured vigorous activity (n = 134) and a short-term supervised structured intervention (n = 56), we identified 52 exercise-responsive hit proteins (HITs) that distinguish healthy from pre-cancer states, map a graded healthy-to-pre-cancer proteomic continuum and shift longitudinally toward healthier profiles following both exercise exposures. Both exercise signatures robustly discriminate pre-cancer from healthy participants, with performance that is predominantly protein-driven and minimally augmented by clinical covariates. A shared three-protein overlap signature retains comparable discrimination, with directional concordance independently confirmed in ~9,800 UK Biobank participants via Olink proteomics. Together, these findings establish a biologically coherent, replicable exercise-linked proteomic signature that maps the pre-cancer state across two independent paradigms and motivates prospective, adherence-monitored interventional studies to determine whether these exercise-induced proteomic shifts are causally protective against cancer.