CD8(+) T cells in the circulation of patients with head and neck cancer (HNC) were previously shown to be significantly more sensitive to, and preferentially targeted for, apoptosis than CD4(+) T cells (Hoffmann et al., Clin Cancer Res, 8:2553-2562, 2002). To distinguish global from CD8(+) subset-specific apoptosis, we studied Annexin-binding to naïve, memory, and effector subsets of CD8(+) cells by multicolor flow cytometry. Age-related changes in naïve and effector CD8(+) cell subsets were observed in patients and normal controls (NC). The frequencies of naïve (CD28(+)CD45RO(-)) CD8(+) T cells were lower and those of memory (CD28(+)CD45RO(+)) and effector (CD28(-)) CD8(+) T cells significantly higher in the circulation of HNC patients relative to age-matched NC. Among CD8(+) T cells, the CD28(-) effector cell subset contained the highest proportion of Annexin-binding cells, while the naïve CD28(+)CD45RO(-) subset contained the lowest. This suggested a high turnover rate of the CD8(+)CD28(-) effector cell subset in patients with HNC, which was being compensated by a rapid transition of naïve CD8(+) T cells to the effector cell pool. Following tumor resection, the frequency of CD8(+)CD28(-) T cells normalized in the patients, an indication that the presence of tumor had an influence on the size of CD8(+)CD28(-) T-cell pool. Ex vivo, in mixed lymphocyte-tumor cultures (MLTC) with semiallogeneic T cells as responders, CD8(+)CD28(-) T cells could be generated from CD8(+)CD28(+) cells by repeated stimulations with tumor cells. These CD8(+)CD28(-) effector cells lysed the tumor, produced IFN-gamma in response to the tumor, and strongly expressed granzyme B. Thus, the high rate of their apoptosis in the circulation of patients with HNC might be expected to contribute to tumor progression. However, the ex vivo generation of this cell subset was suppressed by strong CD28/B7 ligation or by overexpresson of MHC molecules on tumor cells, suggesting that adequate costimulation is necessary for protection from apoptosis. It appears that interactions of immune and tumor cells might determine the fate of this terminally differentiated effector cell subset.