When magic bullets miss their mark

The exquisite specificity of the adaptive immune response underlies much of the promise of cancer immunotherapy. Ironically, however, this same specificity represents an Achilles heel if the antigenic target mutates to no longer be recognizable by the ensuing immune response. In this issue, 2 striking examples of this feared occurrence are reported. Furukawa et al (page 987) examine changes in the expression of HTLV-1 Tax protein in patients with adult T-cell leukemia (ATL). Tax protein plays a pivotal role in immortalization of HTLV-1–infected lymphocytes but, interestingly, is also the source of the dominant MHC class I epitope recognized by HTLV-1–specific cytoxic T lymphocytes (CTLs). Several ATL patients were identified in which a mutation resulted either in a truncated Tax protein (with loss of defined epitopes) or in an amino acid substitution in the dominant HLA-A*02 restricted epitope that was no longer recognized by Tax specific CTLs in HLA-A*02⁺ patients. Comparison of HTLV-1 isolates from ATL patients and their family members indicated that the truncation was present only in patient samples. Given that Tax is required for HTLV-1 transactivation, the truncated (nonfunctional) Tax likely arose in the ATL patients after lymphocyte immortalization, resulting in transformed cells that no longer require Tax and that are invisible to Tax-specific CTLs.

Loss of tumor-associated antigens can have a profound impact on the outcome of therapy, as graphically demonstrated by Gottschalk et al (page 835). This elegant case report describes a patient with post–marrow transplantation lymphoproliferative disease that was unresponsive to adoptive immunotherapy with an Epstein-Barr virus (EBV)–specific T-cell line. The authors systematically identify the likely basis for this treatment failure to be the outgrowth of tumor-harboring EBV with a deletion in EBNA-3B.Whereas the T-cell line used for therapy was technically polyclonal (and was stimulated using lymphoblastoid cells expressing all 9 EBV latency antigens), as typically occurs, it demonstrated highly restricted antigen specificity, in this case almost exclusively recognizing 2 EBNA-3B epitopes, both of which were lost by the tumor. Before treatment, both wild-type and truncated EBV sequences were present in the patient, but after T-cell therapy only the truncated variant could be identified, suggesting selective tumor-cell killing and outgrowth of the antigen loss variant. In this case, as well as in the studies of HTLV-1, it remains unclear what the origin of the mutated viruses might be, and similar isolates have not been reported in normal carriers. Taken together, these reports underscore the inherent risks of taking aim at a heterogeneous tumor target with the laser-like precision of an oligoclonal T-cell response. It is likely that future strategies will need to engineer a more broadly polyvalent response (such as the infusion of independently generated lines targeting multiple antigens) or the identification of antigens that are indespensible for maintaining the transformed phenotype.