Multi-Omic and Spatial Profiling Defines Outcome-Associated Tumor States in Head and Neck Adenoid Cystic Carcinoma

Abstract

Abstract Head and neck adenoid cystic carcinoma (HNACC) shows highly variable clinical behavior that histopathological features only partly predict, and prognostic biomarkers for aggressive disease at diagnosis remain limited. Here, we show by integrating bulk, single-cell, and spatial transcriptomes from 25 HNACC tumors that poor survival is marked by profound loss of myoepithelial tumor cells and expansion of luminal tumor states. Both outcome groups share a myoepithelial-like origin, but aggressive tumors converge on an actively dividing, MYC/E2F-driven luminal state concentrated in discrete spatial niches. Aggressiveness therefore reflects where tumors end along a shared developmental path rather than a distinct lineage or altered immune infiltration. A four-gene classifier capturing this state predicted survival in an independent cohort, outperforming larger signatures and flagging aggressive subgroups in two others. These findings define a lineage proliferation axis linked to HNACC outcome, providing a basis for prospective risk stratification and proliferative-directed therapy.