Virus-immune signatures reveal distinct clinical phenotypes and predict prognosis in patients with Kaposi’s sarcoma

BACGROUND: Kaposi’s sarcoma (KS) is a Kaposi sarcoma-associated herpesvirus (KSHV)-driven multicentric malignancy characterized by substantial clinical heterogeneity. Current etiological classification systems inadequately capture the underlying biological determinants, thereby limiting prognostic precision and personalized therapeutic strategies. We aimed to establish biology-driven phenotypic classifications of KS to evaluate their clinical relevance and prognostic utility. METHODS: Using an unsupervised clustering approach, patients were stratified into four biologically distinct subtypes based on KSHV viral load, immune cells, and inflammatory cytokines. We examined the associations between these subtypes, disease progression, and survival outcomes and developed a visualized risk prediction model by integrating key determinants. RESULTS: Cluster analysis identified four distinct biological phenotypes: Cluster 1 (“low viral load-immune preserved,” n = 47); Cluster 2 (“high viral load-immune depleted,” n = 45); Cluster 3 (“intermediate viral load-hyperinflammatory,” n = 27); and Cluster 4 (“high viral load-immune dysregulated,” n = 23). Notably, acquired immunodeficiency syndrome-associated KS (AIDS-KS) was significantly enriched in Cluster 4 (65.22%), which was characterized by elevated KSHV viral burden and CD4+ T-cell depletion. The classification stratified prognosis (p = 0.002). Longitudinal monitoring of KSHV viral load dynamics revealed three distinct viral clearance patterns: rapid clearance (10.00%), persistently elevated KSHV viral load (32.50%), and gradual clearance (57.50%). Our constructed risk prediction model accurately predicted the probability of disease progression within two years (AUC: 87.8%). CONCLUSIONS: This study established a biology-driven KS classification system grounded in an integrated virus-immune landscape, revealing critical heterogeneity. The associated prognostic model provides a practical tool for individualized risk assessment and precise KS management.