CD8⁺ T cells mediate GVL against CML-like leukemia. (A) Survival curve for B10.D2→Balb/c mixed allogeneic chimeras with CML that were untreated (Control) or treated with imatinib alone (100 mg/kg/d) or in combination with weekly DLI (total of 4 infusions beginning at day 14 after BMT) of unfractionated (Total splenocytes, solid line), CD4-depleted, or CD8-depleted splenocytes. Imatinib treatment was stopped at day 77 (vertical dotted line). The difference in survival between recipients of CD4-depleted DLI and recipients treated with imatinib only or with CD8-depleted DLI was significant (P = .0329 and P = .0059, respectively, by Mantel-Cox tests), whereas there was no significant difference in survival of recipients treated with total splenocyte DLI versus CD4-depleted DLI, or of recipients treated with CD8-depleted DLI versus imatinib only. (B) FACS analysis of leukemia burden (percentage of GFP⁺ leukocytes, X-axis) and allogeneic chimerism (β2-microglobulin b allele, Y-axis) at day 49 (2 weeks after the last DLI) in 5 representative recipients of DLI with total (top row), CD4-depleted (middle row), or CD8-depleted (bottom row) splenocytes. Note the eradication of GFP⁺ cells and full allogeneic chimerism in most recipients of CD4-depleted splenocytes, but persistent leukemia in most recipients treated with CD8-depleted splenocytes. (C) Survival curve for an independent transplantation cohort of B10.D2→Balb/c allogeneic chimeras with CML-like leukemia that were treated with imatinib and DLI consisting of CD4- or CD8-depleted splenocytes or purified CD8⁺ splenocytes, with imatinib treatment stopped at day 42 after transplantation (vertical dotted line). The survival of recipients treated with either CD4-depleted or CD8⁺ DLI was significantly improved compared with the CD8-depleted arm (P = .0209 and .0278, respectively, by Mantel-Cox tests), whereas there was no significant difference in survival of recipients of CD8-depeleted DLI versus control (P = .4259).