Objectives: To produce an enzyme linked immunosorbent assay (ELISA) to detect antibodies against Epstein Barr virus (EBV) specified nuclear antigen 1 (EBNA 1) and the 18 kD EBV matrix protein, and to determine and optimize its sensitivity and specificity for the diagnosis of nasopharyngeal carcinoma (NPC).
Methods: We used a combination of highly purified glutathione transferase fusion proteins of the 40 kD carboxy domain of EBNA1 and the 18 kD EBV matrix protein for coating ELISA plates. In three separate studies, we tested for IgA antibodies in serum specimens from 28 EBV seronegative donors, 284 EBV seropositive donors and 160 newly diagnosed NPC patients. By comparing the sensitivity and specificity of diagnosis obtained for different cutoff values, we derived several quantitative parameters to evaluate assay performance, establish objective diagnostic criteria which optimize the intrinsic diagnostic capability of the assay and assess the significance of individual test results, respectively. Optimum cutoff optical density (OD) is defined as the cutoff OD where sensitivity of the assay equals its specificity, and resolution of the assay is indicated by the value of sensitivity (or specificity) determined at the optimum cutoff OD. Diagnosis of NPC was achieved by setting a cutoff zone at +/- 20% of this value.
Results: All the EBV seronegative donors tested were not reactive, and most of the EBV seropositive donors were weakly reactive, while the majority of NPC patients were moderately or strongly reactive. While the assay was thus shown to be specific for EBV, there was an overlap in the level of these serum antibodies between few individuals of the two latter groups. It was shown that the assay performed equally well in two separate studies conducted under different testing conditions and using different collections of sera in that assay resolution determined on these occasions were 86% and 87% respectively. Diagnosis of NPC can be achieved at the same expected sensitivity of 89% and 83% determined at the lower and upper limits of the cutoff zones, with the corresponding values of specificity being 78% and 91%. It was further shown in the third study that resolution of the assay can be increased to 90% using an assay produced with a higher concentration of the same antigens, and that diagnosis of NPC can be achieved at a higher sensitivity ranging between 86% and 95% at a corresponding specificity of 93% and 86%.
Conclusions: After optimization and standardization, the ELISA can achieve a sensitivity ranging from 86% to 95%, with corresponding specificities of 93% and 86% respectively for the diagnosis of NPC.