Glycometabolic, inflammatory and exocrine plasma proteins predict cancer in individuals with cardiovascular disease

Abstract

Cancer development is preceded by systemic immune, metabolic and tissue perturbation. We asked whether a minimal circulating-protein signature anticipates incident cancer in patients with atheromatous cardiovascular disease (ACVD) exposed to tobacco. Discovery used a nested case-control set drawn from the longitudinal FLEMENGHO cohort (n=156; 38 incident lung cancers). Multi-omic profiling returned a two-step classifier: four proteins, smoking status, personal cancer history. Preprocessing constants, classifiers and both thresholds were fixed before transfer to PREVALUNG (n=397; 58 incident cancers, 26 of them lung). There, low O-GlcNAcase (OGA) with low interleukin-6 (IL-6) defined a very-low-risk stratum holding 1 of 58 cancers (sensitivity 98.3%, negative predictive value 98.8%); low chymotrypsin C (CTRC) with high beta microseminoprotein (MSMB) defined a high-risk stratum in which 15 of 28 participants developed cancer. No additional omic feature (metabolomic, immunophenotypic, clonal-haematopoietic or metagenomic) was retained as a reproducible improvement to the classifier under our selection framework. Discrimination was greater for cancers diagnosed more than 12 months after sampling than for near-term cancers (AUC 0.766, 95% CI 0.671-0.857, versus 0.589, 0.485-0.692), arguing against performance being driven principally by occult disease. These estimates are conditional on the cancer frequency of this population. They require prospective confirmation before any imaging schedule is changed.