Antileukemia activity of a natural killer cell line against human leukemias.

We describe here the in vitro and in vivo antileukemia activity of a recently described natural killer (NK) cell line (NK-92), which has features of human activated NK cells. The cytotoxic activity of rhIL2-dependent cultured NK-92 cells against primary patient-derived leukemic target cells [12 acute myelogenous leukemias (AMLs), 7 T acute lymphoblastic leukemias (T-ALLs), 14 B-lineage-ALLs, and 13 chronic myelogenous leukemias (CMLs)], human leukemic cell lines (K562, KG1, HL60, Raji, NALM6, TALL-104, CEM-S, and CEM-T) and normal bone marrow cells was measured in 51Cr-release assay (CRA). The patient-derived leukemias could be subdivided into three groups based on their sensitivity to NK-92 cells: insensitive (< or =19% lysis), sensitive (20-49% lysis), and highly sensitive (> or =50% lysis) at an E:T ratio of 9:1. Of 46 patient-derived samples, 24 (52.2%) were sensitive or highly sensitive to NK-92-mediated in vitro cytotoxicity (6 of 12 AMLs, 7 of 7 T-ALLs, 5 of 14 B-lineage-ALLs, and 6 of 13 CMLs). NK-92 cells were highly cytotoxic against all of the eight leukemic cell lines tested in a standard 4-h CRA. Normal human bone marrow hematopoietic cells derived from 18 normal donors were insensitive to NK-92-mediated cytolysis. In comparison with human lymphokine-activated killer cells, normal NK cells, and T cells, NK-92 cells displayed more powerful antileukemia activity against a patient-derived T-ALL as well as K562 and HL60 cells, both in in vitro CRA and in a xenografted human leukemia SCID mouse model. The NK-92 cells did not induce the development of leukemia in SCID mice after i.v., i.p., or s.c. inoculation. In adoptive transfer experiments, SCID mice receiving i.p. inoculations of human leukemias derived from a T-ALL (TA27) and an AML (MA26) that were highly sensitive to the cytolysis of NK-92 cells in vitro, as well as a pre-B-ALL (BA31) that was insensitive to the in vitro cytolysis of NK-92 cells, were treated by administration of NK-92 cells with or without rhIL2 (2 x 10(7) NK-92 cells i.p.; one dose or five doses). Survival times of SCID mice bearing the sensitive TA27 and MA26 leukemias were significantly prolonged by adoptive cell therapy with NK-92 cells. Some of the animals who received five doses of NK-92 cells with or without rhIL2 administration were still alive without any signs of leukemia development 6 months after leukemia inoculation. In contrast, survival of mice bearing the insensitive BA31 leukemia were not affected by this treatment. This in vitro and in vivo antileukemia effect of NK-92 cells suggests that cytotoxic NK cells of this type may have potential as effectors of leukemia control.