Inhibited apoptosis and drug resistance in acute myeloid leukaemia


Dr Richard J.Jones Johns Hopkins Oncology Center, Room 2-127, 600 North Wolfe Street, Baltimore, MD 21287-8967, U.S.A.


Despite extensive investigation into mechanisms of drug resistance in acute myeloid leukaemia (AML), the aetiology of therapeutic resistance is unclear. We found that five leukaemia cell lines (K562, HL-60, CEM, CEM induced to overexpress bcl-2, and REH) displayed parallel sensitivity to four antileukaemia drugs with different mechanisms of action, with K562 generally being the least sensitive and REH being the most sensitive. The amount of spontaneous apoptosis in the cell lines after serum-free culture paralleled their drug sensitivity: K562 cells displayed the least apoptosis at 24 h (2.50 ± 0.24%) and REH the most (24.47 ± 8.22%). The extent of spontaneous apoptosis of leukaemic blasts from 39 patients with newly diagnosed de novo AML also correlated with the success of the intensive, infusional cytarabine-based induction therapy. There was a median of 19.5% (range 3.6–64%) apoptotic AML cells after 24 h of serum-free culture in patients who entered a complete remission compared with 4.2% (1.8–7.0%) apoptotic AML cells in patients who did not achieve a complete remission (P = 0.0007). Thus, inhibited apoptosis was associated with both in vitro and in vivo pan-resistance to antileukaemic chemotherapy. The cause of inhibited apoptosis in AML is probably a function of interactions among multiple signals that influence apoptosis. Assessment of spontaneous apoptosis may serve as an important prognostic factor for AML.

Even with the introduction of new anti-leukaemic agents and approaches, most patients with acute myeloid leukaemia (AML) are still not cured. Re-emergence of the original leukaemic clone remains the predominant cause of treatment failure. Despite extensive investigation into mechanisms of drug resistance, the cause of the treatment resistance in most leukaemias remains unclear. Clinically, most leukaemias display parallel sensitivity or cross-resistance to most cytotoxic anticancer agents, despite differing mechanisms of action. This suggests that most resistant leukaemias have developed multiple mechanisms of resistance concurrently, or one single mechanism of pan-resistance. However, no classic mechanisms of drug resistance, either alone or in combination, appear to explain pan-resistance to cytotoxic agents in leukaemia. MDR1 expression has been reported to be a poor prognostic marker for AML ( Campos et al, 1992 ), and yet plays no role in cellular resistance to cytarabine (the principal component of most non-transplant regimens for AML) or alkylating agents and irradiation (the principal components of transplant regimens for AML). Other examples of determinants of drug sensitivity which fail to explain resistance in acute leukaemia include the glutathione pathway ( Joncourt et al, 1993 ) and topoisomerase II ( Kaufmann et al, 1994 ).

The efficacy of cytotoxic antineoplastic therapy is determined by a sequential cascade of events including drug delivery, drug–target interaction, the induction of cellular damage, and the response of the cell to the damage ( Chabner & Myers, 1989; Bedi et al, 1995 ). Most classic mechanisms of resistance to specific cytotoxic anticancer agents (e.g. those resulting from MDR1 and the glutathione pathway) result in interference with either drug delivery or individual drug–target interactions. However, the cell's response to damage (death or repair) also influences the activity of cytotoxic anticancer agents ( Bedi et al, 1995 ). The final effect of successful cytotoxic therapy is the induction of apoptosis or programmed cell death, a genetically determined process of cell suicide. There are many well-studied anti-apoptotic signals which lead to enhanced cell survival and are crucial mechanisms of oncogenesis. Moreover, emerging evidence demonstrates that many anti-apoptotic signals, including bcl-2, BCR-ABL and the loss of p53 function, are major causes of resistance to anticancer drugs ( Bedi et al, 1995 ; Steinmann et al, 1991 ; Lowe et al, 1993 ; Clark et al, 1993 ; McGahon et al, 1994 ; Miyashita & Reed, 1993; Strasser et al, 1991 ; Sentman et al, 1991 ; Collins et al, 1992 ). As apoptosis is the final common pathway of death from virtually all cytotoxic agents, signals blocking apoptosis would be expected to produce pan-resistance to cytotoxic agents.

Recent data suggest that signals which block apoptosis are also specifically associated with a poor response to chemotherapy in AML ( Campos et al, 1993 ; Norgaard et al, 1996 ; Banker et al, 1997 ; Maung et al, 1994 ; Porwit-MacDonald et al, 1995 ). However, the importance of this mechanism of drug resistance in AML remains unclear. Any cause of drug resistance will be associated with decreased apoptosis after treatment, since apoptosis is the final mechanism of death from all cytotoxic anticancer agents. The amount of spontaneous apoptosis after serum-free culture is a measure of intrinsic cellular sensitivity to apoptosis that is distinct from apoptosis following cytotoxic agents. Therefore to further study the role anti-apoptotic signals play in drug resistance in AML, the amount of spontaneous apoptosis after serum-free culture and the sensitivities of leukaemia cells to antileukaemic agents (both in vitro and in vivo) were correlated.


Leukaemia cell lines and clonogenic assays

Five human leukaemia cell lines, representative of several forms of acute leukaemia, were analysed: CEM (T-lineage acute lymphoblastic leukaemia), and CEMbcl2 (CEM cells induced to over-express bcl-2) ( Miyashita et al, 1995 ), HL-60 (promyelocytic leukaemia), K562 (CML in blast crisis) and REH (B-lineage acute lymphoblastic leukaemia). The leukaemia cell lines were incubated at a concentration of 250 000 cells/ml in serum-free medium (Minimum Essential Medium Alpha, Gibco BRL, Grand Island, N.Y.) for 24–48 h in a humidified atmosphere at 37°C with 7.5% CO2. For the clonogenic assays to assess drug sensitivities, the cell lines (5 × 106 cells/ml) were incubated with graded concentrations of 4-hydroperoxycyclophosphamide (4HC) for 30 min or daunomycin for 60 min at 37°C. The drug incubations were arrested by the addition of chilled media and the cells were washed twice. Graded concentrations of cytarabine and topotecan were incubated continuously in the culture plates. Cells were placed into 1.32% methyl cellulose (Sigma, St Louis, Mo.), 30% fetal bovine serum, 1% bovine serum albumin, and 10−4 M 2-mercaptoethanol (Sigma, St Louis, Mo.), and the cell culture suspensions (between 200 and 104 cells per ml) were plated into 35 mm petri dishes and incubated in a humidified atmosphere containing 7.5% CO2 at 37°C, as previously described ( Jones et al, 1990 ; Miller et al, 1991 ). Parallel experiments have demonstrated that all five cell lines show linear colony growth over this range of plating densities, with inhibition of growth occurring at cell densities > 104 cells/ml. Quadruplicate cultures were scored with an inverted microscope on days 7–12 with an aggregate of geqslant R: gt-or-equal, slanted 40 cells representing a colony. Each concentration of drug was tested for a minimum of three independent experiments.


After giving informed consent for research use of bone marrow cells, patients with newly-diagnosed de novo AML admitted to the Johns Hopkins Oncology Center between 1993 and 1997 had bone marrow cells removed by posterior iliac crest aspiration. Leukaemia blasts represented a median of 70% (range 50–100%) of the nucleated bone marrow cells by flow cytometry. The mononuclear cells, obtained by Ficoll-Hypaque (Pharmacia Biotech, Sweden) density centrifugation, were essentially devoid of all cells but leukaemia cells. The mononuclear cells were incubated at 500 000 cells/ml in serum-free medium for 24–48 h as described for the cell lines. Patients were treated with intensive, infusional cytarabine, daunomycin and etoposide as previously described ( Gore et al, 1995 , 1994). At the time of consolidation therapy (typically 50–60 d following induction therapy), individual patient responses to induction therapy were determined as either complete response if they had < 5% blasts in their marrow and normalization of their peripheral counts, or no complete response by other evidence of residual tumour. None of the patients considered in to be complete remission had re-emergence of their original cytogenetic abnormality at the time of consolidation.

Flow cytometric analysis of apoptosis

After serum-free culture, the leukaemia cell lines (250 000 cells/ml) and mononuclear bone marrow cells (500 000 cells/ml) from leukaemia patients were fixed in 70% ethanol, permeablized with 0.1% Triton X-100 (Sigma, St Louis, Mo.), treated with 5 μg/ml RNAse (Boehringer Mannheim, Germany) and incubated at 37°C for 15 min prior to staining with 50 μg/ml propidium iodide (Sigma, St Louis, Mo.) for 60 min at 4°C. The fraction of subdiploid cells with oligonucleosomal DNA degradation characteristic of apoptosis was quantified by flow cytometric analysis (FACScan, Becton Dickinson), as previously described ( Darzynkiewicz et al, 1992 ; Bedi et al, 1994 , 1995). Annexin V FITC Apoptosis Detection Kit (Oncogene Research Products, Cambridge, Mass.) was used to quantitate apoptosis based on annexin V binding ( Boersma et al, 1996 ; van Engeland et al, 1996 ; Vermes et al, 1995 ). 5 × 105 cells were used at each time point following serum-free culture.

Statistical evaluations

Each data point on the dose–response curves represented the mean ± standard error of the mean (SEM) of three to six separate experiments. Dose–response curves were established by plotting the logarithm of the surviving fraction of clonogenic leukaemia cells against the incubation drug concentrations. The relationship between the individual cell lines and their responses to the various drugs and serum-free conditions was determined by ranking each from most sensitive to most resistant to the condition (designated as 1–5). The average values of the cell lines' ranks were analysed using an ANOVA test. The association between degree of spontaneous apoptosis and clinical response was tested by the Wilcoxon Rank Sum Test and Fisher's exact test.


We investigated the sensitivity of five leukaemia cell lines to four antileukaemic drugs with different mechanisms of action: cytarabine (an antimetabolite), daunomycin (an intercalating agent/topoisomerase II inhibitor), 4HC (an alkylating agent) and topotecan (a topoisomerase I inhibitor). The leukaemia cell lines displayed parallel sensitivities to the four drugs (Fig 1, Table I). The average rank-order of each leukaemia cell line's sensitivity to the four drugs tested was significantly different (P < 0.0001, ANOVA). The K562 cells ranked as the most resistant cell line to three (4HC, daunomycin, topotecan) of the four drugs tested and the second most resistant cell line (after HL-60) to cytarabine. Similarly, REH cells were the most sensitive line to three of the four drugs and the second most sensitive (after CEM) to topotecan. HL60, CEMbcl-2 and the parental CEM cell line (in order of decreasing resistance to the four drugs) were intermediate in drug sensitivity.

Figure 1.

Fig 1. Dose–response of the five leukaemia cells lines (○, HL60; •, K562; □, BCL2; ▿, CEM; ▴, REH) to cytarabine, topotecan, and 4HC daunomycin. Each data point represents the mean ± SEM of three to six separate experiments. Dose–response curves were established by plotting the logarithm of the surviving fraction of clonogenic leukaemia cells against the incubation drug concentrations.

Table 1. Table I. Sensitivity of leukaemia cell lines to spontaneous apoptosis and cytotoxic agents. Spontaneous apoptosis after 24 and 48 h in serum-free culture for each of the cell lines as measured by flow cytometric analysis of propidium iodide staining and the rank-order of drug sensitivity of the cell lines as determined from dose–response curves in Fig 1: 5 is the most resistant cell line to each drug tested, 1 is most sensitive (P < 0.001 , ANOVA). 4HC: 4-hydroperoxycyclophosphamide; Ara-C: cytarabine.* Mean ± SEM of a minimum of three separate experiments. Thumbnail image of

The five leukaemia cell lines also differed in their relative amounts of spontaneous apoptosis. Utilizing propidium iodide to identify subdiploid cells with oligonucleosomal DNA fragmentation characteristic of apoptosis ( Darzynkiewicz et al, 1992 ; Bedi et al, 1994 , 1995), the rank-order of their spontaneous apoptosis after 24–48 h in serum-free conditions closely paralleled the drug sensitivity of the cell lines ( Table I, P < 0.0001, ANOVA). K562 cells displayed the least apoptosis (2.59 ± 0.39%) at 48 h, followed in order by HL60 (6.18 ± 1.55%), CEMbcl2 (8.49 ± 2.27%), CEM (17.29 ± 3.53%) and REH (33.67 ± 10.83%). As expected, the CEMbcl2 cell line showed relatively less spontaneous apoptosis, and was consistently ranked as more resistant to all agents tested, than the CEM parental cell line. Recent observations suggest that DNA fragmentation may not always accurately reflect all apoptotic death ( Schulze-Osthoff et al, 1994 ; Nakajima et al, 1995 ); therefore the apoptosis data was confirmed by annexin V binding (Fig 2).

Figure 2.

%, respectively).

To determine if the relationship between spontaneous apoptosis and drug resistance existed clinically, the extent of spontaneous apoptosis of blasts after serum-free culture was compared with the outcome of induction therapy in 39 patients with de novo AML. The patients' clinical characteristics are described in Table II. Of the group, 12 were not evaluable for statistical analysis because of technical difficulties which precluded laboratory analysis of apoptosis or treatment-related death before response could be determined. Patients who entered a complete remission to induction therapy displayed a significantly higher median value of leukaemic blast spontaneous apoptosis at both 24 h (P = 0.0007, Wilcoxon Rank Sum Test) and 48 h (P = 0.001) than patients who did not achieve complete remission ( Table III). Accordingly, all 16 patients with >7% apoptotic AML cells after 24 h of serum-free culture entered a complete remission compared to only three out of 10 patients with leqslant R: less-than-or-eq, slant 7% apoptotic AML cells achieving CR (P = 0.0001, Fisher's exact test). Twenty patients from our small group achieved CR and 12 remain currently in remission with a median disease free survival of 16 months (range 3–55 months). All underwent consolidation therapy with either high-dose cytarabine-based chemotherapy (10) or autologous bone marrow transplantation (two).

Table 2. Table II. Clinical data on AML patients. CR: complete remission; NR: no complete remission; NE: not evaluable; inv: inversion; del: deletion; M4E: M4 with eosinophilia; ND: not done.* Continues in remission.Thumbnail image of
Table 3. Table III. Spontaneous apoptosis in serum-free culture and response to induction therapy. * Median value of the patient samples (range of samples).† Wilcoxon Rank Sum Test.Thumbnail image of


The cause of the treatment resistance in most leukaemias remains a matter of intense investigation. Signals that block apoptosis are now well recognized to be important causes of resistance to anticancer drugs ( Bedi et al, 1995 ; Steinmann et al, 1991 ; Lowe et al, 1993 ; Clark et al, 1993 ; McGahon et al, 1994 ; Miyashita & Reed, 1993; Strasser et al, 1991 ; Sentman et al, 1991 ; Collins et al, 1992 ); however, their role in acute leukaemia drug resistance has not been well studied. We found that enhanced survival of AML cells in serum-free medium, resulting from inhibited apoptosis, is associated with both in vitro and in vivo pan-resistance to cytotoxic anti-leukaemic therapy.

There are a variety of anti-apoptotic signals that have been associated with the malignant transformation of AML, including bcl-2 ( Campos et al, 1993 ; Banker et al, 1997 ; Maung et al, 1994 ; Porwit-MacDonald et al, 1995 ), bcr-abl ( Tien et al, 1992 ), loss of p53 ( Prokocimer & Rotter, 1994) and autonomous production of growth factors ( Griffin & Lowenberg, 1986). Moreover, these signals are generally associated with a poor prognosis. Although all these anti-apoptotic signals probably participate in drug resistance in AML, individually these events are relatively uncommon in newly-diagnosed de novo AML. The intrinsic susceptibility of AML cells to apoptosis is probably a function of interactions among multiple signals that either induce or inhibit apoptosis. Thus, measurement of spontaneous apoptosis may be more predictive for prognosis than the assessment of individual signals that influence apotosis (e.g. bcl-2 or p53), because it evaluates the cumulative effect of all signals that influence apoptosis. In fact, the five cell lines we studied varied widely in their drug sensitivities and sensitivity to spontaneous apoptosis, even though they all have lost wild-type p53 expression ( Sugimoto et al, 1992 ), a strong anti-apoptotic signal.

The degree of apoptosis in a cell population could be influenced by cell proliferation as well as cell survival. Autonomous proliferation of AML cells in serum-free culture has been found to predict a poor response to induction therapy ( Lowenberg et al, 1993 ). However, the degree of apoptosis in the AML cells in our study was strictly a function of cell survival; little or no cell proliferation occurred over the 48 h in serum-free conditions, as total numbers (viable and non-viable) remained steady. The results of spontaneous apoptosis in the clinical samples could also be confounded by the presence of normal cells. However, the assayed samples (mononuclear cells recovered by density-centrifugation) from all patients were essentially entirely composed of leukaemic blasts.

Although this is a small series, it is a prospective analysis and the prognostic significance of the degree of spontaneous apoptosis with response to induction therapy was strong. In addition to correctly predicting the successful induction of six of seven patients with good-risk AML (inversion of chromosome 16, translocation of chromosomes 8 and 21, translocation of chromosomes 15 and 17), the degree of spontaneous apoptosis also correctly predicted the successful induction of all the three patients with bad-risk features (abnormalities of chromosome 5, translocation of chromosomes 6 and 9). The extent of spontaneous apoptosis after serum-free culture also successfully predicted the response of the 10 AML patients with normal or unknown chromosomes ( Table II). Another small study similarly demonstrated a positive association between AML blast survival in vitro and lack of response to induction therapy ( Norgaard et al, 1996 ). However, whether assessment of spontaneous apoptosis will have independent prognostic significance will require multivariate analysis of a larger prospective study.

Identifying leukaemias resistant to spontaneous apoptosis by flow cytometry was inexpensive and the results were available in less than 48 h, so this assay could potentially be used to guide initial therapy. However, since apoptosis is the final common mechanism of death induced by virtually all cytotoxic agents, leukaemias that are drug resistant as a result of anti-apoptotic signals are likely to be relatively unresponsive to all cytotoxic therapy. Accordingly, the most resistant cell lines (K562 and HL60) were relatively resistant to all four cytotoxic agents, despite different mechanisms of action. There are, however, a number of approaches being studied that directly overcome or circumvent blocks in the apoptotic pathway and may prove effective in treating leukaemias that are pan-resistant as a result of anti-apoptotic signals. Cellular immunotherapy may be one such approach; it appears to induce apoptosis possibly by activating a late event in the apoptotic pathway ( Fuchs et al, 1995 ; Vaux et al, 1992 ; Roger et al, 1996 ; Shi et al, 1994 ; Chen et al, 1995 ).


We thank Marie C. Moineau for her helpful assistance with manuscript preparation. This work was supported in part by NIH grant CA15396 and American Cancer Society grant DHP-177. R.J.J. is a Leukemia Society of America Scholar.