Construct validity and evidence boundaries of a 17-gene transcriptomic proxy for mitoxyperilysis in colorectal aging and cancer Analysis

Mitoxyperilysis is a recently described lytic cell-death process defined by prolonged mitochondria–plasma-membrane contact, localized oxidative injury, and membrane rupture under combined innate-immune and metabolic stress. We evaluated the construct validity of a 17-gene transcriptomic proxy across perturbation, aging, colorectal cancer (CRC), independent senescence measures, genetics, prognosis, and pharmacogenomics. Public datasets were reanalyzed with culture-, donor-, patient-, pair-, or cell-line-level inference, multiplicity control, cell-composition sensitivity, and cross-platform evaluation. All 17 genes responded to combined stress in GSE235046, but the directions and Torin responses were mixed. In GTEx colon, the proxy–age association attenuated after adjustment (Spearman rho 0.125 to 0.052). In GSE44076, proxy–senescence covariance was inverse in healthy mucosa (rho − 0.324), absent in adjacent mucosa, and positive in tumor (rho 0.354), without a paired proxy increase. Bulk CRC covariance replicated across cohorts (random-effects rho 0.339, 95% CI 0.172–0.486) and persisted after cell-composition adjustment. An exploratory CPTAC-2 reduced-feature check showed protein-layer covariance using 10/17 proxy and 8/12 reference genes (rho 0.292, P = 0.0037). However, none of seven panel-disjoint patient-level tests using two independent senescence algorithms met predefined validation criteria. The panel did not overlap 208 prioritized CRC effector genes; adjusted survival and observed drug-response analyses were null. Thus, this context-sensitive proxy is not a validated measure of mitoxyperilysis, senescence, inherited risk, prognosis, or drug sensitivity. Direct validation requires colorectal-relevant perturbation and rescue linked to spatial, oxidative, and membrane-integrity phenotypes.