Abstract
PURPOSE: Antibodies to select Epstein-Barr virus proteins can diagnose early-stage nasopharyngeal carcinoma (NPC). We have previously shown that IgA against Epstein-Barr virus nuclear antigen 1 (EBNA1) can predict incident NPC in high- and intermediate-risk cohorts 4 years before diagnosis. Here, we tested EBNA1 variants, with mutants, to define the sequence requirements for an NPC risk assay. EXPERIMENTAL DESIGN: Mammalian-expressed constructs were developed to represent EBNA1 variants 487V and 487A, which can differ by ≥15 amino acids in the N- and C-termini. Denatured lysates were evaluated by a refined IgA and IgG immunoblot assay in a case-control study using prediagnostic NPC sera from two independent cohorts in Singapore and Shanghai, the People's Republic of China. RESULTS: At 95% sensitivity, 487V yielded a 94.9% specificity compared with 86.1% for 487A. EBNA1 deleted for the conserved glycine-alanine repeats (GAr) reduced false positives by 22.8%. NPC sera reacted more strongly to the C-terminus than healthy controls, but the C-terminal construct (a.a. 390-641) showed lower specificity (84.8%) than the EBNA1 GAr-deleted construct (92.4%) at 95% sensitivity. CONCLUSIONS: Although EBNA1 IgA was present in healthy sera, most epitopes localized to the immunodominant GAr. We conclude that a refined EBNA1 antigen deleted for the GAr, but with residues consistently detected in Southeast Asian NPC tumors, is optimized for risk prediction with an extended sojourn time of 7.5 years. Furthermore, distinct EBNA1 serologic profiles enhanced the utility of the EBNA1 IgA assay for risk stratification. This illustrates the importance of serologically relevant EBNA1 sequences for NPC risk prediction and early detection.