Abstract 4660: Epstein-Barr virus transcriptomic signatures of tumor subpopulations predict relapse in nasopharyngeal carcinoma

Abstract Nasopharyngeal carcinoma (NPC), originating from nasopharyngeal epithelium and frequently metastasizes to cervical lymph nodes and distal organs, is endemic in Southeast Asia and highly associated with Epstein-Barr virus (EBV). Radiation therapy with or without concurrent chemo-radiation is the most common treatment, but 30-50% of treated patients still relapse. Plasma EBV DNA test is the gold standard in screening and monitoring relapse. However, 20-50% NPC relapse are plasma EBV DNA-negative. In situ hybridization (ISH) of EBV-encoded small RNAs (EBER) of formalin-fixed paraffin-embedded nasopharyngeal biopsy can identify EBV-positive tumors regardless of plasma EBV DNA, but remains unable to predict relapse at initial diagnosis. Recently, single-cell RNA sequencing (scRNA-seq) technology has emerged as a long-sought solution to low resolution in bulk sequencing. We launched the largest and most comprehensive scRNA-seq NPC study. 74 patients were recruited. Matched biopsies from normal adjacent tumor, primary tumor and lymph node were scRNA-seq profiled. 1298414 cells had been successfully profiled. Our study spanned 5 years and 5 patients relapsed. By unsupervised clustering, cell type classification by canonical markers, and copy number variation, differential abundance and survival analyses, we identified a number of relapse-predicting neoplastic and non-neoplastic subpopulations. For the first time, we also attempted to unfold EBV transcriptomic signature and its association with relapse at single-cell resolution using Vulture developed by our team. Our results confirmed EBV transcripts were highly enriched in neoplastic cells. Surprisingly, EBV transcripts could also be detected in both non-neoplastic epithelial and tumor micro-environment cells especially T cells and B cells. Neoplastic subclones carrying mutations predictive of relapse revealed by our modified Mutect2 pipeline and other relapse-predicting tumor subpopulations were also found to possess higher EBV positivity. Cox regression analysis identified both cell and tissue type-specific EBV transcriptomic signatures implicated in initial attachment and entry during infection, viral gene regulation, immune evasion and cellular transformation that predicted relapse. Receiver operating characteristic analysis suggested EBV transcriptomic signatures outperformed both nasopharyngeal EBER ISH and plasma EBV DNA in relapse prediction at initial diagnosis. To conclude, we successfully discovered previously overlooked relapse-predicting EBV transcriptomic signatures with cell and tissue type specificity that may be implicated in (1) tumorigenic transformation and (2) progression evolution. EBV transcriptomic signatures outperforming current strategies in relapse prediction at initial diagnosis may serve as new diagnostics and therapeutic targets for relapse prevention. Citation Format: Eymon Ka Chun Wu, Jun-Yi Chen, Ken Hung On Yu, Nelson Su, Esther Hyunji Won, Hazel Yap, Hin Kwok, Raymond King-Yin Tsang, Joseph Chun-Kit Chung, Anthony Wing-Ip Lo, Stephanie Nga Sze Wong, Jack Kuipers, Jeffery Yan Ho Lau, Horace Cheuk-Wai Choi, Ka-On Lam, Sum-Yin Chan, To-Wai Leung, Xin Yuan Guan, Anne Wing Mui Lee, Chi Ming Che, Joshua Wing Kei Ho, Victor Ho Fun Lee. Epstein-Barr virus transcriptomic signatures of tumor subpopulations predict relapse in nasopharyngeal carcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 4660.