EBV genome-guided transcriptomic re-annotation reveals molecular subtypes of nasopharyngeal carcinoma informing prognosis and treatment.

Non-keratinizing nasopharyngeal carcinoma (NPC) is closely related to Epstein-Barr virus (EBV) infection. Patients with NPC often exhibit diverse treatment responses due to tumor heterogeneity. Thus, identifying molecular subgroups based on EBV involvement holds promise for refining personalized treatment strategies and improving treatment outcomes in NPC patients. 193 treatment-naïve NPC specimens with comprehensive clinical and pathological data were procured from Fujian Cancer Hospital. RNA sequencing was employed to acquire the gene expression profiles, followed by the re-annotation of 100 EBV-associated genes leveraging the EBV sequence. Molecular subtypes were conducted via consensus clustering, with an external NPC cohort serving as a validation dataset. Scissor method was applied to identify survival-associated cell subpopulations from single-cell data, following comprehensive bioinformatic analyses. Three molecular subtypes of NPC-CoriLyt, Cneg, and CEB1-were identified, each with specific clinical profiles. The CEB1 subtype is distinguished by its heightened metabolic activity and immunosuppressive environment. A hub-gene-based risk model for these subtypes strongly predicted disease-free survival, with replicated results in the validated cohort. The model's predictive accuracy was high, with areas under the curve for 1, 3, and 5-year survival rates at 0.79, 0.86, and 0.88, respectively. M2-type macrophages exhibit a high-risk score profile and play a critical role in EBV infection, with prominent activation of the TNF-II and TGF-β signaling pathways. This study introduced a new EBV-related transcriptomics-based classification system for NPC that showed great promise in predicting patient survival outcomes.