EBNA-1 sequences in endemic and sporadic Burkitt's lymphoma.

In their recent manuscript, Habeshaw et al. (4) used mutations in EBNA-1 to identify Epstein-Barr virus strains in immortalized cell lines and Burkitt lymphomas (BL) from different geographic regions. Like us (1, 2), Habeshaw et al. found a near absence of P-ala in BL from any part of the world but frequently identified V-leu and P-thr. However, whereas we concluded that some subtypes present in normal B cells are not found in BL, implying a differential pathogenicity in the context of BL, Habeshaw et al. concluded that within any geographic area, the EBNA-1 subtypes present in BL simply reflect those prevalent in the population. If the data on BL subtypes are similar in these studies, why are the conclusions different? One problem is that Habeshaw et al. studied only three tumors from Europe (including the only P-ala-positive tumor in their series), with most of their data coming from East African tumors. The lack of European tumor data makes it impossible to demonstrate a geographical bias in the distribution of EBNA-1 subtypes in tumors, and the concordance of EBNA-1 subtypes between tumors and subtypes present in nonmalignant B cells cannot be proved, except in the context of East African BL. They will need to study additional tumors from Europe and show that P-ala is prevalent in such tumors to confirm their conclusion. If they were able to demonstrate this, it would contrast with our data, for we have not found P-ala in American BL. A second problem is that Habeshaw et al. do not think it important that they used lymphoblastoid cell lines as opposed to peripheral lymphocytes (PBLs) as controls. It seems probable, from the sum total of the data, that lymphoblastoid lines do not permit identification of the complete virus repertoire in an individual. We and others have observed multiple EBNA-1 subtypes (and multiple subtypes based on other genetic loci) in single individuals as well as in T-cell nasal lymphomas and have proposed that some of these subtypes evolve postinfection (3, 5, 7). We have also observed differences in the relative proportions of the major subtypes associated with BL (P-thr and V-leu) in BL from different regions, as is hinted at by Habeshaw’s data from New Guinea, but neither we nor Habeshaw et al. have observed other subtypes that can be detected in peripheral blood or saliva, namely V-val and V-pro, in BL from any of the combined regions we have examined. In contrast, we and others frequently find V-val in nasopharyngeal carcinomas (2, 3, 6). Habeshaw et al. did not find either V-pro or V-val in the lymphoblastoid lines they examined. This does not surprise us—we have found V-val only in saliva in nonimmunosuppressed individuals, and neither we nor anyone else, to our knowledge, has found V-pro in lymphoblastoid lines, suggesting that it is in some way transformation defective. In patients from Europe and America, we have found V-leu (along with multiple other subtypes) in PBLs from individuals with infectious mononucleosis or with immunosuppression, but we have yet to find V-leu in PBLs from normal individuals. This could be a quantitative phenomenon, but it nonetheless suggests that it is not simply geography (as suggested by Habeshaw et al.) that dictates the distribution of EBNA-1 subtypes in tumors: host factors and the biology of the tumor cells themselves may well play a role and, indeed, may be more important than geography.