Abstract
<title>Abstract</title> <p>Cancer cachexia is a systemic metabolic syndrome driving an estimated 20–30% of cancer-related deaths, yet early diagnostic biomarkers and actionable therapeutic targets remain elusive. Leveraging CT radiomics and plasma metabolomics in 586 gastric cancer patients, we develop a high-fidelity classifier (AUC = 0.91) that outperforms conventional clinical indices. To mechanistically anchor these macroscopic signatures, we profiled the visceral adipose tissue (VAT) of seven representative patients using single-cell RNA sequencing. This analysis identified a distinct Macro_IL1B subpopulation that drives local fibro-inflammatory remodeling and adipocyte atrophy via NAMPT-dependent trophic withdrawal. Pharmacological inhibition of PLA2 activity in an in vivo cachexia model attenuates VAT wasting and delays systemic cachexia progression. Our findings reposition VAT as an early effector organ and provide a translational framework for intercepting cachexia at its metabolic-immune origin.</p>