Codon Usage Bias Variation and Evolutionary Signatures of Epstein-Barr Virus in Distinct Epithelial Cancers

EBV genomic variation has been shown to contribute to the development of certain EBV-associated cancers. While EBV genomic variation has been extensively studied at the nucleotide level, it remains unclear how synonymous codon usage contributes to viral adaptation across epithelial cancer contexts. Here, we analyzed 1148 EBV genomes with annotated tumor origins to investigate genome-wide genetic differentiation and codon usage patterns of 13 core genes across NPC- and GC-associated viruses and EBV types. SNP-based analyses revealed partial genetic separation between NPC-EBV and GC-EBV, characterized by both rare GC-associated risk variants and common protective haplotypes. Codon usage patterns, however, showed strong gene-specific structure: EBNA2 and EBNA3 clustered primarily by EBV type, whereas EBNA1 and LMP2A were more sensitive to tumor background. Codon bias analyses suggested heterogeneous contributions of mutational pressure and natural selection across genes and lineages, whereas lytic BALF genes displayed highly conserved codon usage despite cancer-associated variants. Collectively, this study demonstrates that codon usage patterns of specific EBV genes are associated with tumor background and are jointly shaped by gene function and viral lineage structure.