Susceptibility of acute myelogenous leukemia blasts to lysis by lymphokine-activated killer (LAK) cells and its clinical relevance.

To help understanding host-tumor relationships in acute myelogenous leukemia (AML) and better define indications for interleukin 2 (IL-2) therapy in this disease, we studied the relationship between the susceptibility of leukemic cells of 44 AML patients to lysis by autologous (26 cases) and/or allogeneic (41 cases) lymphokine-activated killer (LAK) cells and characteristics of the leukemia. Lymphocytes were activated in the presence of 1000 u/ml recombinant IL-2 for 5 days. Lysis of AML cells was studied by 51Cr release. Average lysis of AML cells by autologous LAK cells was 9 +/- 13% and by allogeneic LAK cells 10 +/- 9% with a significant correlation between lyses by both effectors (p = 0.01). Autologous (p = 0.005) and allogeneic (p = 0.004) lyses were higher in patients with initial infection. Allogeneic lysis was correlated with initial WBC count (p = 0.009), serum lactic-dehydrogenase level (p = 0.05), and expression of CD13 (p = 0.01). Autologous lysis was inversely correlated with expression of CD34 (p = 0.003). Expression of adhesion molecules CD54 (ICAM-1) and CD58 (LFA-3) by the leukemic cells did not correlate with their lysis by LAK cells. Susceptibility of leukemic cells to lysis by LAK cells did not correlate with prognosis of the leukemia.     

Susceptibility of adult acute lymphoblastic leukemia blasts to lysis by lymphokine-activated killer cells

To help understand host-tumor relationships in adult acute lymphoblastic leukemia (ALL) and to better define potential indications for interleukin-2 (IL-2) treatment in this disease, the relationship between the susceptibility of leukemia cells of 22 patients with ALL to lysis by allogeneic lymphokine-activated killer (LAK) cells and characteristics of the leukemia was studied. Lymphocytes were activated in the presence of 1000 U/ml recombinant IL-2 for 5 days. The lysis of ALL cells was studied by the release of 51Cr. The average lysis of ALL cells by control, unactivated lymphocytes was 1.2 +/- 2.4% and by LAK cells 8.9 +/- 8.6%. The susceptibility of leukemic cells to lysis did not correlate with the expression of lymphoid or myeloid differentiation markers or expression of the adhesion molecules CD54 (ICAM-1) and CD58 (LFA-3). Leukemic cells of the FAB I2 subtype were significantly more resistant to lysis than those of the other subtypes (average lysis 1.4 +/- 3.0% versus 12.3 +/- 8.2%, p = 0.003). The susceptibility to lysis did not correlate with the other initial characteristics of the leukemia. The 11 patients in whom 8% or more of leukemic cells were lysed by allogeneic LAK cells survived significantly longer than the 11 patients whose blast cells were less susceptible to lysis (p = 0.04). It is concluded that IL-2 treatment might be of benefit in adult ALL, particularly in non-L2 FAB subtypes and during complete remission to possibly delay relapse and prolong survival.     

Mechanisms of selective killing of neuroblastoma cells by natural killer cells and lymphokine activated killer cells. Potential for residual disease eradication.

Widely disseminated neuroblastoma in children older than infancy remains a very poor prognosis disease. Even the introduction of marrow ablative chemotherapy with autologous rescue has not significantly improved the outlook for these children, presumably because of a failure to eradicate minimal residual disease. One additional approach which may hold promise is the use of immunomodulation with cytokines such as IL2 in the setting of minimal residual disease (MDR), for example after intensive chemotherapy and ABMT. However, considerable variability in the susceptibility of neuroblastoma cells to natural killer (NK) and lymphokine-activated (LAK) killing has been observed, and it is presently unclear how NK and LAK cells recognise neuroblastoma cells. In this paper we examine expression of cell adhesion molecules on neuroblastoma to determine which of these modify interaction with NK and LAK cells. We find that LFA-3 (CD58), the ligand for CD2 is of predominant importance in predicting susceptibility of neuroblastoma to the cytotoxic actions of NK and LAK cells, while expression of ICAM-1 (CD54) may also modify susceptibility. These findings were confirmed by blocking experiments in which co-culture of target cells with ICAM-1 and LFA-3 reduced LAK and NK cytotoxicity. Study of the immunophenotypic features of each patient''s neuroblastoma cells before induction of MRD may be valuable in determining the likely effect of IL2 in predicting disease reactivation.     

In vitro and in vivo susceptibility of human leukemic cells to lymphokine activated killer activity

The susceptibility of human myeloid and lymphoid leukemic blasts to the lytic action of recombinant interleukin-2 (rIL-2)-generated lymphokine activated killer (LAK) cells was analyzed. With the exception of the K562 cell line, all 9 leukemic cell lines tested were resistant to the natural killer activity of freshly isolated peripheral blood lymphocytes (PBL) from healthy donors but were susceptible to the lytic action of PBL cultured for 3 days in the presence of rIL-2. Of the 32 primary myeloid and lymphoid acute leukemia samples investigated, the great majority were natural killer cell-resistant but were variably sensitive to LAK effectors. Variations in LAK activity were observed according to the donor of PBL, while little or no difference was documented in the capacity to elicit LAK activity of PBL cultured with 100 or 1,000 U of rIL-2/ml. Pretreatment of the leukemic target cells with neuraminidase did not increase substantially their sensitivity to LAK activity. LAK cells generated from the PBL of patients at the onset of the disease or in complete clinicohematological remission lysed Raji cells as efficiently as normal LAK effectors. Finally, LAK cells were capable of abrogating the tumor growth in nude mice of a human leukemic T cell line. These findings demonstrate the susceptibility in vitro and in vivo of human leukemic blasts to the lytic effect of LAK cells and point to a possible clinical exploitment of this new form of adoptive immunotherapy in the management of acute leukemia.     

Relation of natural killer cell line NK-92-mediated cytolysis (NK-92-lysis) with the surface markers of major histocompatibility complex class I antigens, adhesion molecules, and Fas of target cells.

Natural killer (NK) cell line NK-92 has recently been established by Klingemann et al. In this study, we compared the NK-92-mediated cytolysis (NK-92-lysis) with the killing of healthy volunteers' NK cells and lymphokine-activated killer (LAK) cells. The NK-92-lysis was partially different from the NK- and LAK-lysis. 1) The NK-92 could kill most of major histocompatibility complex (MHC) class I antigen-positive tumor cells. 2) The NK cells killed a myeloid leukemia cell line K562, but the NK-92 showed low killer activity against it. 3) The LAK cells could not kill a CD58-deficient cell line OKM-2T, whereas the NK-92 could kill it sufficiently. 4) The NK-92 could not kill CD54-, CD102-deficient cell lines T98G and U373MG; however, the LAK cells could kill them. Blocking tests using specific antibodies revealed the reason for these differences. The K562 expressed relatively low levels of CD54 and CD102. When the K562 was pretreated with anti-CD54 and anti-CD102, the NK-92 could not kill it at all, whereas the NK cells could still kill it, although the killing level decreased. The NK-92 could not kill the anti-CD54- and anti-CD102-treated OKM-2T. The LAK cells could not kill anti-CD58-treated U373MG and T98G. These findings suggest that NK-92-lysis may require the CD54 and CD102 but that NK-lysis does not require them as much, whereas the LAK-lysis may be rather in relation with the CD58. The NK-92 has high killer activity, and may be applicable for clinical use. However, it should be considered that the NK-92 cannnot kill CD54-, CD102-deficient tumor cells.     

Increased susceptibility to lymphokine activated killer (LAK) lysis of relapsing vs. newly diagnosed acute leukemic cells without changes in drug resistance or in the expression of adhesion molecules.

The NK and LAK activity of peripheral blood lymphocytes of leukemic patients as well as the susceptibility of their acute myeloid (AML) and lymphoblastic (ALL) leukemia cells to autologous and allogeneic LAKs were examined. In addition, neoplastic cells at diagnosis and at relapse were compared in the same patients for several features, including in vitro susceptibility to LAKs and to the drugs used in the induction phase, expression of MDR phenotype and of adhesion molecules, and differentiation markers. The NK activity of patients' LAK cells on K562 was significantly lower than that of a group of healthy donors whereas no differences were found in LAK activity as evaluated on Daudi cells. Three of 5 AML and 3 of 4 ALL were significantly more susceptible to autologous and allogeneic LAK lysis when blasts obtained at relapse were compared with leukemic cells of the same patients at diagnosis. This different lysability was not associated with in vitro modified sensitivity to drugs used in induction treatment. Moreover, no elevation in the expression of the multidrug-resistance (MDR)-related P170 glycoprotein was noted in relapsing leukemic cells. Even the expression of adhesion molecules and differentiation markers did not correlate with lysability of leukemic cells. These data demonstrate that relapsing leukemic blasts can be significantly lysed by LAK cells and suggest a rationale for adoptive immunotherapy with IL-2 and LAK cells in the treatment of acute leukemic patients.     

Alpha-interferon and lymphokine-activated killer cells in hairy cell leukemia

Lymphokine-activated killer (LAK) cell activity generated from peripheral blood was tested in 6 patients with typical hairy cell leukemia, 3 not on treatment with alpha-interferon (alpha-IFN) and 3 receiving therapy. In all cases, substantial killing of the LAK-sensitive target Daudi was observed, but hairy cells, whether or not they had been pretreated with alpha-IFN, were uniformly resistant to LAK lysis. The hairy cells were also resistant to LAK cell killing generated from normal peripheral blood mononuclear cells. alpha-IFN added at various times during LAK generation had little or no effect on LAK activity. It is concluded that LAK cells are not important in mediating the beneficial effects of alpha-IFN in hairy cell leukemia.     

脐血CD_3AK细胞的制备、生物活性测定及临床应用初探

 目的　研究用脐血单个核细胞制备CD3 AK细胞的抗肿瘤作用 ,探讨肿瘤生物治疗近期疗效的免疫指标。方法　分离脐血单个核细胞 ,分别用IL 2和IL 2 +CD3 Ab诱导LAK和CD3AK细胞 ,并测其扩增数量和对K5 6 2细胞的杀伤活性 ;测定肿瘤患者用CD3 AK细胞治疗前后外周血单核细胞(PBMC)的NK杀伤活性。结果　脐血LAK细胞和脐血CD3 AK细胞均于培养后第 11天时扩增倍数最高 ,分别是培养前的 18倍、2 4倍 ,对K5 6 2的杀伤活性分别是培养前的 2 .6倍和 3.2倍 ;肿瘤患者输注CD3 AK细胞一疗程后 ,其PBMC的NK杀伤活性由 6 2 %升高到 82 % ,平均升高 32 %。结论　 (1)脐血单个核细胞是LAK细胞和CD3 AK细胞良好的前体细胞。 (2 )LAK和CD3 AK细胞的数量和杀伤能力在培养的第 11天时达高峰 ,而且CD3 AK细胞数量和杀伤活性明显优于LAK细胞。 (3)CD3 AK细胞输注能明显提高肿瘤患者PBMC的NK活性。 (4 )肿瘤病人PBMC的NK活性测定可望成为肿瘤生物治疗的一个近期疗效参数     

Construction of a minimal HIV-1 variant that selectively replicates in leukemic derived T-cell lines: towards a new virotherapy approach

T-cell acute lymphoblastic leukemia is a high-risk type of blood-cell cancer. We analyzed the possibility of developing virotherapy for T-cell acute lymphoblastic leukemia. Virotherapy is based on the exclusive replication of a virus in leukemic cells, leading to the selective removal of these malignant cells. We constructed a minimized derivative of HIV-1, a complex lentivirus encoding multiple accessory functions that are essential for virus replication in untransformed cells, but dispensable in leukemic T cells. This mini-HIV virus has five deletions (vif, vpR, vpU, nef, and U3) and replicated in the SupT1 cell line, but did not replicate in normal peripheral blood mononuclear cells. The stripped down mini-HIV variant was also able to efficiently remove leukemic cells from a mixed culture with untransformed control cells. In contrast to wild-type HIV-1, we did not observe bystander killing in mixed culture experiments with the mini-HIV variant. Furthermore, viral escape was not detected in long-term cultures. The mini-HIV variant that uses CD4 and CXCR4 for cell entry could potentially be used against CXCR4-expressing malignancies such as T-lymphoblastic leukemia/lymphoma, natural killer leukemia, and some myeloid leukemias.     

Impaired Cytotoxic Activity of Interleukin 2 (IL 2)-cultured Large Granular Lymphocytes (LGL) in Childhood Acute Lymphoblastic Leukemia. Quantitative Measurement of IL 2 Receptor Expression of IL 2-cultured LGL

Large granular lymphocytes (LGL) of the natural killer (NK) cell lineage were highly purified and their interleukin 2 (IL 2) receptors (IL 2R) were investigated in childhood acute lymphoblastic leukemia (ALL). Not only NK activity but also 3-day recombinant IL 2-cultured lymphokine activated killer (LAK) activity were decreased in ALL, despite normal percentages of LGL, CD16 (Leul1)⁺ and CD56 (NKH1)⁺ cells. The cytotoxic activity of IL 2-cultured LGL, but not IL 2-cultured T cells, was significantly decreased in ALL, indicating a selective defect of LGL among the killer cells. IL 2R (CD25) numbers/cell of 3-day IL 2-cultured LGL were not decreased in ALL by flow cytometric analysis. Scatchard plot analysis demonstrated that high affinity IL 2R numbers/cell of 3-day IL 2-cultured LGL were normal but the binding affinity level of the receptors in these cells in ALL was one-third of that in the similarly-cultured control cells, suggesting inadequate IL 2-IL 2R interaction is responsible, at least in part, for their reduced cytotoxic activity.     

The relationship between multi-drug resistance and resistance to natural-killer-cell and lymphokine-activated killer-cell lysis in human leukemic cell lines.

Results concerning a possible link between susceptibility to natural-cell-mediated immune cytolysis and the multi-drug resistance (MDR) phenotype are conflicting. We evaluated in human acute lymphocytic leukemia the relationship between acquired drug resistance and susceptibility to cytolysis mediated by endogenous, interferon-activated, and interleukin-2-activated natural cytotoxic cells. Eight human leukemia drug-resistant/sensitive cell line pairs were evaluated; drug-resistant sub-lines included those selected for primary resistance to adriamycin, etoposide, teniposide, vincristine, and vinblastine. A majority of P-glycoprotein-positive MDR sub-lines displayed slight but statistically significant resistance to endogenous and/or interferon-activated natural-killer(NK)-cell-mediated lysis, as compared with the drug-sensitive parental type. P-glycoprotein-negative sub-lines displayed variable NK susceptibility; within this group, the variants selected for primary etoposide resistance were more susceptible to NK cytolysis than parental cells. Results of cold-target-inhibition experiments suggest that altered NK susceptibility does not arise solely from modulation of NK target recognition and adherence structures. IL2-activated killer (LAK) cells lysed both drug-sensitive and drug-resistant lines. Two MDR lines selected for primary etoposide resistance displayed enhanced LAK susceptibility. In contrast, the 2 variants selected for resistance to adriamycin exhibited partial resistance to LAK-mediated killing, which could be overcome at high effector-to-target ratios. Our results support the development of interleukin-2/LAK immunotherapy for the treatment of leukemias with acquired drug resistance.     

Lymphokine-activated killer cell plus recombinant interleukin-2 therapy of erythroleukemia in mice

The antileukemic effects of lymphokine-activated killer (LAK) cells plus recombinant interleukin-2 (rIL-2) therapy were assessed in mice with Friend virus (FV)-induced erythroleukemia. LAK cells were generated by incubating normal spleen cells for 72 hr in the presence of rIL-2 (1000 units/ml). At the time of injection, the LAK cells were cytotoxic in vitro against FV-infected fibroblasts and NK-sensitive and -resistant tumor targets but not normal controls. To determine in vivo activity, fully leukemic mice (spleen weight greater than 0.75 g) were injected with either PBS or LAK cells (10(8) cells/mouse IV at 14 and 17 days post virus) and rIL-2 (10,000 units/mouse IP every 8 hr on days 14 through 18 post virus). More than 70% of the progressively leukemic mice experienced permanent leukemia regressions (disease-free for greater than 100 days) following LAK cell plus rIL-2 therapy. Regressions were characterized by return of spleen and liver weights to normal and elimination of virus-infected erythroid (CFU-E) and macrophage (CFU-C) progenitor cells from spleen and marrow. Leukemic animals treated with either LAK cells alone or IL-2 alone experienced only transient leukemia regressions. These results demonstrate that LAK cell plus rIL-2 treatment can induce permanent regressions in progressively leukemic mice and provide a responsive and manipulable model system to elucidate the mechanisms involved in this form of immunotherapy.     

Gene transfer of CD154 and IL12 cDNA induces an anti-leukemic immunity in a murine model of acute leukemia.

IL12 is an essential cytokine for the generation of T helper 1 response, natural killer (NK) cells and cytotoxic T lymphocyte (CTL) stimulation. CD154 triggers CD40 on antigen-presenting cells, thus inducing antigen presentation to the immune system and production of IL12. As IL12 and CD154 share several pathways mediating immune response, we investigated in an aggressive murine model of acute leukemia the relative antileukemic efficiency of IL12, CD154 and IL12 + CD154 gene transfer. Live leukemic cells transduced by IL12, CD154, and IL12 + CD154 showed reduced leukemogenicity but CD154 protective effect was reduced when 10(6) leukemic cells were injected. Vaccines with lethally irradiated IL12-transduced cells were able to cure mice previously injected with 10(4) leukemic cells and adoptive transfer of IL12-induced antileukemic immunity protected recipient mice. NK cytotoxicity was enhanced in mice vaccinated with leukemic cells transduced by IL12, CD154, and CD154 + IL12. IL12 transduced cells induced IFN-gamma mRNA in CD4(+) and CD8(+) T cells isolated from the spleen of vaccinated animals, however, in vivo depletion experiments showed that IL12 vaccine effect was CD4(+) but not CD8(+) T cell dependent. We conclude that IL12 gene is a more potent candidate than CD154 for gene therapy of acute leukemia.     

Cytokine-induced killer cells: NK-like T cells with cytotolytic specificity against leukemia

Cytokine-induced killer (CIK) cells are a unique population of cytotoxic T lymphocytes (CTL) with the characteristic CD3+CD56+ phenotype. These cells have demonstrated higher proliferative and cytolytic activities in comparison to the reported CD3-CD56+ lymphokine activated killer (LAK) cells that are essentially activated natural killer (NK) cells. CIK cells are non-MHC-restricted in target cell recognition and killing. We have shown the feasibility of generating CIK cells from a series of marrow samples of patients with acute myeloid leukemia (AML) collected at diagnosis. At maturity, the CIK cells exhibit potent cytotoxicity against autologous AML targets as well as allogeneic myeloid leukemia cells, regardless of the HLA types of these targets. This observed cytotoxicity is not entirely due to NK cells as prior pre-absorption of the NK cells cytolytic activities does not abolish the subsequent cytotolytic activities against leukemic targets. It has also been reported by others that CIK cells are cytolytic against chronic myeloid leukemia (CML) cells, both in vitro and in the SCID mouse tumor model. In a mouse transplant model across MHC barrier, the CIK cells generated from the donor do not induce graft vs. host disease as observed for unfractionated donor splenocytes. In comparison to untreated control mice, the infusion of CIK cells results in the prolonged survival of murine leukemia-bearing mice. CIK cells also express CD94, part of the NK receptor comprising of CD94-NKG2 heterodimer. However, only low level of the killer immunoglobulin-like receptors are expressed by the CIK cells. In addition, as reported for the classical CTL, CIK cells could interact with dendritic cells (DC) to result in the enhancement of cytotolytic activities against tumor cells. The characteristic biological properties of the CIK cells would, therefore, enable them to be exploited for anti-leukemic therapy.     

Increased cytotoxicity against B-chronic lymphocytic leukemia by cellular manipulations: potentials for therapeutic use

B-cell chronic lymphocytic leukemia (CLL) is characterized by profound immune dysfunction and a marked resistance to apoptosis. Understanding the cellular biology of immune effector cells from CLL patients as well as leukemic target cells is essential to developing immune mediated therapeutic strategies for CLL. In this study, an immortal CLL cell line called WSU-CLL has been used to study the characteristics of B-cell CLL as a tumor target for natural killer (NK), activated natural killer, and lymphokine activated killer (LAK) cells. The WSU-CLL cells were significantly less (p<0.001) susceptible to NK cell mediated cytotoxicity compared to K562, a standard tumor target cell line. In vitro activation of effector cells with either short term, low dose IL-2 or long term, high dose IL-2 significantly increased the susceptibility of CLL cells for cell mediated killing. The addition of CD1a+/CD3-/CD4+/CD80+/CD83+ dendritic cells derived from human umbilical cord blood increased the cytotoxicity of LAK cells against WSU-CLL. There is an increased expression of Bcl-2 and decreased expression of Fas on WSU-CLL cells as determined by RT-PCR techniques indicating possible roles for these genes in exerting resistance to immune cell mediated lysis. When Bcl-2 expression was downregulated in WSU-CLL cells using gene specific antisense oligonucleotides, the susceptibility of WSU-CLL cells to the cytotoxicity of chemotherapeutic agent Fludarabine was increased. Thus, our results suggest that in vitro activation with cytokines, addition of accessory cell populations such as dendritic cells and/or manipulation of key gene expression i.e. down regulation of Bcl-2 might be potential strategies to increase the antitumor cytotoxicity against CLL cells.     

Susceptibility of multidrug-resistant human leukemia cell lines to human interleukin 2-activated killer cells.

Considering the possibility to overcome drug resistance by other treatment strategies than chemotherapy we investigated the susceptibility of three independently selected multidrug-resistant sublines of the T-lymphoblastoid leukemic cell line CCRF-CEM to lymphokine-activated killer (LAK) cells. We found that two of the multidrug-resistant sublines were significantly less susceptible targets to LAK cells. A third one, however, was as susceptible as the parental CCRF-CEM cell line. Moreover, a multidrug-resistant subline that reverted to an almost drug-sensitive phenotype was observed to be also revertant for resistance against LAK cells. We found an inverse relationship between the expression of the mdr1 gene (P-glycoprotein) and the susceptibility to LAK cells. Verapamil, a calcium channel blocker, while increasing the drug sensitivity of a multidrug-resistant subline, did not induce a reversal of the suppression of LAK susceptibility. The possibility of enhanced resistance to LAK cells of multidrug-resistant cells should be taken into account when one is looking for therapy strategies to overcome multidrug resistance.     

Expression of cell adhesion molecules in human melanoma cell lines and their role in cytotoxicity mediated by tumor-infiltrating lymphocytes.

The role of cell adhesion molecules (CAM) LFA1, ICAM-1, LFA3, VLA1, VLA4, CD29, CD44, and CD56 in tumor-infiltrating lymphocyte (TIL) and natural killer cell (NK)-mediated killing of target cells was studied. Melanoma cell lines and autologous TIL were derived from seven patients with metastatic melanoma, and cytotoxicity assays were done in the presence and absence of monoclonal antibodies (MoAb) to CAM expressed on melanoma cells or TIL. The melanoma cell lines analyzed were all positive for CD29 and LFA3 expression, negative for LFA1 expression, but showed variable expression of ICAM-1, VLA1, VLA4, CD44, and CD56. The effects of anti-CAM antibodies on TIL-mediated melanoma killing fell into three categories: (1) consistent inhibition of TIL-mediated killing was observed when melanoma cells were pretreated with anti-ICAM1 and anti-LFA-3 MoAb or when TIL were pretreated with anti-LFA1; (2) no effect was observed when melanoma cells were pretreated with anti-CD56; or (3) a discreet, but significant, inhibition was observed when target cells were pretreated with anti-CD29, anti-VLA1, anti-VLA4, and anti-CD44. Cytotoxicity was significantly enhanced by pretreatment of target cells with gamma-interferon (gamma-IFN), although gamma-IFN did not augment surface expression of the CAM studied. The NK-mediated killing of K562 cells was blocked by anti-LFA1, anti-CD18, and anti-ICAM, and partially inhibited by anti-CD44 MoAb. Together, these results suggest that several accessory CAM may play a role in regulating cellular cytotoxicity. Because cytotoxicity generally correlated with the level of expression of CAM in melanoma cells, weak CAM surface expression may provide a means for melanomas to escape immune surveillance.     

Lymphokine-activated killer cells can kill target cells not only by early killing but also by late killing,and the late killing level against autotumor cell line

Generally, lymphokine-activated killer (LAK) cells kill target cells early after the LAK cells adhere to them. In this study, we describe that LAK cells can also kill at a later time, such as 24-96 h. LAK cells were generated from a cancer patient and healthy volunteers. As target cells, the patient's autotumor cell line H41 was used. When LAK cells were added to the target cells in a culture well, the LAK cells killed the target cells by cell-cell adhesion within 1-4 h (early killing), but not all cells were killed. The LAK cells were then removed. However, the remaining cells ultimately died 24-96 h later (late killing). The late killing was different from the early killing because numerous granules and vacuoles appeared in the cytoplasm. The late killing was not induced by adding supernatant of the LAK cell culture, suggesting that LAK-target cell contact may be necessary for the killing. The cell injury was inhibited by 3-methyladenine (lysosome inhibitor). It suggests that the vacuoles may be caused by activated lysosome. The patient's LAK cells induced late killing at high levels. There was a high percentage of CD8(+)CD16(+) cells in the peripheral blood lymphocytes (PBL). This subset induced late killing more effectively than the CD8(-)CD16(+) subset. Killing was more conspicuous against H41 than against allogeneic cell line T98G. This type of killing is noteworthy for understanding of killing mechanism of LAK cells.