Timed Sequential Therapy of Acute Leukemia with Flavopiridol: In Vitro Model for a Phase I Clinical Trial¹

Adult acute leukemias remain a formidable therapeutic challenge. Only ∼70% of adults with newly diagnosed AML³ achieve CR after cytotoxic induction chemotherapy. Although these CRs may be prolonged in 35–40% of younger adults (ages <60 years; Refs. 1, 2, 3), the remainder relapse and die. Certain subgroups, including older adults (1, 4, 5, 6), patients with AML linked to environmental or occupational exposures (including therapy-induced AML; reviewed in Ref. 7) and patients with previous MDS or other antecedent hematological disorders (5, 7) have extremely poor outcomes, with CR rates ≤40%, CR durations <12 months, and cure rates <10–15% (1, 2, 3, 4, 5, 6). The overall outlook for adult ALL is similar (8, 9, 10, 11), with a particularly poor prognosis in Philadelphia chromosome disease (12, 13). Thus, new approaches are needed to improve the outcome for adults with refractory leukemias.

One strategy to augment the cytotoxicity of cycle-dependent antileukemic agents is TST (14, 15). TST is based on the findings that the initial cytoreductive drug induces the remaining malignant cell cohort to enter a proliferative state at a predictable time after drug administration (14, 15, 16). At least hypothetically, this recruited residual tumor would be expected to exhibit an increased sensitivity to drugs administered at the time of predicted proliferation (15, 16). The heightened sensitivity of this proliferative cohort to cycle-dependent drugs, in particular the classical antileukemic nucleoside analogue ara-C, has been demonstrated in in vitro models designed to mimic the in vivo situation (16) and in vivo where the clinical response to ara-C-based TST relates to a linkage between AML cell growth and intracellular ara-C pharmacology (17). Indeed, TST has met with prolongation of disease-free survival in certain groups of adults and children with AML (18, 19, 20, 21, 22).

Nonetheless, whereas TST regimens increase the DFS and cure rate of some patients with AML, there is still a significant proportion of patients for whom TST alone is not curative. These refractory and relapsed leukemias are characterized by dysregulation of proliferation, differentiation, and the balance between cell survival and death, with the balance shifted toward prolonged survival. Such survival likely plays a major role in the resistance to cytotoxic agents exhibited by these refractory leukemias and is likely related to abrogation of programmed cell death (apoptosis) pathways that are activated by key antileukemic drugs.

Activation of apoptosis pathways is a convergence point for many cytotoxic agents, independent of specific drug mechanism. In this regard, the ability to trigger apoptosis independent of cell cycle status might augment overall cytotoxicity and improve clinical results. One such drug is ara-C, a classical cycle-dependent agent with significant activity against both AML and ALL. The drug is a pyrimidine analogue that must be phosphorylated intracellularly to an active triphosphate form (ara-CTP) that, in turn, is incorporated into DNA and leads to a net inhibition of DNA synthesis with eventual cytotoxicity (23). The incorporation of ara-CTP into DNA exerts direct DNA damage and at the same time impedes the enzymatic repair of that damage. These insults culminate in at least two types of cytotoxic activities: (a) S-phase arrest with abrogation of DNA replication and (b) apoptosis (23, 24). In fact, induction of apoptosis in response to DNA strand breakage may at least, in part, explain the cytotoxic activity of ara-C against noncycling cells in addition to the cycling cell population.

Given the pivotal role of apoptosis in drug-induced cytotoxicity, it is important to identify agents that could induce directly and/or overcome resistance to the induction of apoptosis in a clinically meaningful fashion. Flavopiridol (L86–8275), a synthetic flavone derivative that was initially isolated from the stem bark of the Indian tree Dysoxylum binectariferum (25, 26), is a potent growth inhibitor of diverse human tumor cell lines (26), including the induction of apoptosis in hematopoietic cell lines derived from B-cell lymphomas (SUDHL4 and 6), T cell lymphomas (Jurkat, MOLT4), and AML (HL60; Refs. 26, 27, 28, 29). Recent studies in the U937 monoblastic leukemia cell line indicate that flavopiridol induces apoptosis by triggering mitochondrial release of cytochrome c in a fashion that is independent of procaspase-8 cleavage and activation (30).

Mechanistically, flavopiridol inhibits the activities of multiple serine-threonine CDKs by at least two mechanisms: (a) noncell cycle-dependent binding to the ATP-binding site in the CDK molecules (31); and (b) cell cycle-dependent interference with CDK phosphorylation (32, 33, 34). In turn, the blockade of CDK function leads to cell cycle arrest at both the G₁-S and G₂-M checkpoints. The cell cycle arrest induced by flavopiridol relates not only to inhibition of CDKs 1,2,4,6, and 7 but also to direct transcriptional repression of cyclin D1 with depletion of cyclin D1 levels. CDK inhibition occurs at flavopiridol concentrations of 40–200 nm, whereas cyclin D1 depletion requires slightly higher drug doses (100–300 nm; Ref. 35). Additionally, low concentrations of flavopiridol (50–100 nm) decrease the production of VEGF in response to a hypoxic stimulus, at least in part by decreasing the stability of VEGF mRNA (36). VEGF is a growth and survival factor for diverse tumor cell types, including certain acute leukemias (37, 38), and it is conceivable that flavopiridol might exert some part of its growth inhibition through this mechanism.

Recent studies demonstrate that flavopiridol’s cytotoxic effects also relate to the drug’s ability to interrupt transcription by multiple mechanisms (reviewed in Ref. 39). One such mechanism relates to flavopiridol’s ability to bind to and inactivate the CDK9/cyclin T1 (a noncycling cyclin that binds the antiretroviral Tat protein) complex known as P-TEFb (39, 40). P-TEFb facilitates transcription elongation by phosphorylating the COOH-terminal domain of the RNA polymerase II complex (41). By targeting P-TEFb, flavopiridol impedes transcriptional elongation and thereby exerts a net blockade of RNA polymerase II action (41, 42). Another mechanism by which flavopiridol may inhibit transcription is by binding to double-stranded DNA (43). This binding, in turn, may disrupt the interaction between DNA and transcription factors, including the binding of the transcription factor STAT3 to DNA. Interestingly, D cyclins are downstream targets of STAT3 (35), and it is reasonable to speculate that the inhibition of cyclin D expression by flavopiridol results from the drug interfering with STAT3-mediated transcription. Mcl-1 is an antiapoptotic protein of the Bcl-2 family whose expression is related to chemoresistance in AML, at least in the setting of leukemia relapse (44). Like D cyclins, Mcl-1 is a transcriptional target of STAT3 (45), and its expression is down-regulated with concomitant induction of apoptosis in myeloma cell lines and in fresh CD138+ cells from myeloma patients exposed to flavopiridol in vitro (46). Moreover, disruption of STAT3 function and resultant down-regulation of Mcl-1 has been observed in AML blasts obtained from patients receiving flavopiridol as part of a National Cancer Institute- and Institutional Review Board-approved clinical trial (47, 48).

Thus, transcriptional inhibition via multiple pathways is likely to be a central mechanistic theme by which flavopiridol exerts its antitumor activities (39). This hypothesis is substantiated by data generated in the human lung cancer cell line A549, suggesting that flavopiridol is active against noncycling as well as cycling cells and that such activity is cytotoxic as well as cytostatic (27). Furthermore, flavopiridol-induced cytotoxicity is followed by recruitment into cycle and synchronization of residual A549 cells surviving a 24-h drug exposure (49). An increase in the proportion of tumor cells entering S phase is observed 48–72-h postflavopiridol initiation and persists for a minimum of 3–4 days thereafter (after which point cultures reached confluence and were not evaluable for ongoing growth). Treatment with ara-C at the time when flavopiridol-treated residual tumor cells are actively proliferating leads to synergistic growth inhibition and cytotoxicity in vitro (49). This outcome mimics the recruitment of residual leukemic cells observed after initial cytoreductive therapy, thus simulating the principles of TST.

In sum, flavopiridol exhibits some novel mechanisms of action that may not be cross-resistant with the cytotoxic antileukemic agents currently in use and might potentiate the net antileukemic activity of such agents. On the basis of the above-noted studies, we designed an in vitro model of TST to test the hypotheses that: 1) flavopiridol can induce apoptosis and cytotoxicity in fresh human leukemia bone marrow cells; and 2) flavopiridol can recruit the surviving leukemic cells into a proliferative state and thus “prime” such cells for the S-phase-related cytotoxicity of ara-C. Our in vitro model was developed with the in vivo situation in mind, using clinically achievable doses of flavopiridol (26, 50) and ara-C (17, 51) to facilitate translation of our findings to the clinical setting. The results, in turn, serve as a springboard for designing a clinical trial of flavopiridol-based TST in refractory leukemias.

Bone marrow cells were obtained from 20 adults with relapsed and refractory acute leukemia (17 AML and 3 ALL) by routine needle aspiration. Six patients had newly diagnosed AML with poor risk features (3 treatment-related AML or MDS/AML and 3 MDS/AML with complex cytogenetics). Five patients had relapsed leukemia (3 AML and 2 ALL) and 9 had refractory disease (8 AML and 1 ALL). All 14 patients with relapsed or refractory leukemia had received ara-C before the study. No patient received flavopiridol before in vitro study. All patients provided informed consent for marrow aspiration and in vitro studies according to respective University of Maryland, Baltimore Institutional Review Board guidelines before marrow aspiration. Patient and disease characteristics are presented in Table 1.

HL-60 cells were cultured at an initial cell concentration of 5 × 10⁵ cells/ml in the presence of 10% (v/v) heat-inactivated FCS (Life Technologies, Inc., Grand Island, NY) and RPMI 1640 containing l-glutamine (2 mm), penicillin (50 IU/ml), and streptomycin (50 μg/ml). Leukemic blast populations from 20 patients were enriched from heparinized marrow aspirates by Ficoll Hypaque gradient sedimentation. The percentage of blasts in each marrow population was ≥50% before sedimentation and increased to ≥80% after sedimentation. Blast-enriched populations were cultured at a final concentration of ∼5 × 10⁵ cells/ml (range, 2.8–6.7 × 10⁵⁾ with 15% autologous serum (v/v) and maintained at 37°C in a humidified atmosphere containing 5% CO₂. Cell counts in control cultures, as measured by Coulter counting, did not change significantly during the first 2 days of culture (mean Δ day 2/day 0 = 0.95; range, 0.6–2.1). Cell counts in day 5 control cultures exhibited modest count decreases relative to day 2 (mean Δ day 5/day 2 = 0.8; range, 0.3–1.3).

Flavopiridol was kindly provided by the Developmental Therapeutics Program, Division of Cancer Treatment and Diagnosis, National Cancer Institute and Aventis Pharmaceuticals. Stock solutions of flavopiridol (1,000-fold concentrated) were prepared in DMSO and stored at −20°C until use.

For each of the 20 marrow populations, blast cell cultures were initiated on day 0 in the presence of flavopiridol 250 nm (F₂₅₀) for 24 h (Day 1), removed from flavopiridol for 24 h (day 2), and then cultured in ara-C 1 μm (A₁) for an additional 72 h (F₂₅₀A₁, Fig. 1). Parallel controls consisted of cultured cells that: (a) did not receive either flavopiridiol or ara-C (F₀A₀); (b) received flavopiridol day 0 but did not receive ara-C on day 2 (F₂₅₀A₀); and (c) did not receive flavopiridol on day 0 but received ara-C on Day 2 (F₀A₁).

Apoptosis and cell cycle phase distribution were estimated on days 1 and 2 (when sufficient sample permitted), as indicated in Fig. 1, from cell nuclei stained with (50 μg/ml) in hypotonic citrate buffer (52). Apoptosis and cell cycle phase distribution were estimated from the flow cytometric data using the Modfit LT 3.0 flow cytometry modeling program (Verity Software House, Topsham, ME).

The number of viable cells in culture was determined at the end of the 72 h ara-C (or control) exposure on day 5 (see Fig. 1) by double cytofluorescence assay, which involves simultaneous staining of cells with FDA and PI under isotonic conditions, as described previously (53).

As a prelude to our studies with fresh human marrow leukemia populations, we examined the effects of flavopiridol on the cells from the human-derived AML cell line HL-60, using the flow cytometric FDA/PI assay (53). Fig. 2 demonstrates that graded doses of flavopiridol were capable of altering cell growth and survival patterns of HL-60 cells in a dose-dependent fashion with an LC₅₀ of ∼500–600 nm for HL-60 cells after 72 h of exposure. These data substantiate the original findings of Bible and Kaufmann (27) using another method for measuring cytotoxicity and validate the use of FDA/PI (53) for our studies in fresh human leukemia marrow cells. Moreover, apoptosis could be detected within the first 24 h with flavopiridol doses as low as 250 nm. On the basis of these findings, we selected the dose of 250 nm for additional studies in fresh acute leukemia cell populations to allow us to detect cytotoxic effects of flavopiridol while at the same time we were able to observe the behavior of cells that would survive the flavopiridol exposure. Additionally, this dose corresponds to drug levels that are readily achievable in vivo (35).

As depicted in Table 2 and Fig. 3, flavopiridol induced a ≥20% (1.2-fold) increase in apoptosis in 14 (70%) of the 20 human leukemia samples within the first 24 h of culture as determined by hypotonic PI nuclear staining. Mean increase in the presence of flavopiridol relative to drug-free culture was 4.3-fold (median, 1.7; range, 0.6–27). Flavopiridol caused a ≥1.5-fold increase in apoptosis relative to control populations in 11 (55%) of patients, with ≥3-fold increase detected in 6 (30%). Only 5 (25%) leukemic populations failed to exhibit any increase in apoptosis during initial flavopiridol exposure. There was no clear relationship between the response to flavopiridol and prior exposure to ara-C and/or known ara-C resistance. This apoptosis led to eventual cytotoxicity on day 5, where mean survival for flavopiridol-exposed cells was 56 ± 15% (median, 52%; range, 5–124%) of that detected in nonexposed control cells, with 12 (60%) of the 20 marrow populations exhibiting cell survival <40% of control (discussed in TST Model section in conjunction with Table 3 and Fig. 4, below).

Cell cycle determinations on Day 2 before ara-C exposure demonstrated an increase in S phase for cells initially cultured in flavopiridol followed by 24 h in flavopiridol-free medium in 8 (53%) of the 15 leukemia populations analyzed for this parameter (Fig. 5,a). Removal from flavopiridol resulted in a mean 1.7-fold increase in the percentage of cells in S phase on day 2 (median, 1.2; range, 0.2–7.0), with 7 (47%) of the marrow blast populations evincing increases of ≥1.5-fold on day 2 after flavopiridol removal (Table 4). Increases in S phase were detected in 6 (86%) of 7 AML marrow populations, with a mean 2.8-fold increase for all AMLs (median, 2.1; range, 0.2–7.0) but only 1 of 5 MDS/AML (mean, 0.9; range, 0.4–1.2) and 1 of 3 ALL (mean, 0.7; range, 0.5–1.5) populations (6 of 7 versus 2 of 8, χ² = 5.53; P = 0.019). In contrast, increases in S phase were not detected in the majority of parallel drug-free control cultures (mean, 0.9; median, 0.8; range, <0.01–3.2-fold). The proportion of cells in S phase decreased by at least 20% (0.8-fold) in eight marrows, with six marrows exhibiting decreases of 40–>95% (0.6–<0.05-fold), and only four marrows demonstrating increases in S-phase percentage (1.2–3.2-fold). Interestingly, these four marrows were from AML patients in contrast to no such increases in any of the MDS/AML or ALL marrows (4 of 7 versus 0 of 8, χ² = 6.23; P = 0.013). Nonetheless, the effect of flavopiridol on recruitment of cells into S phase exceeded that seen in control cultures in 5 of 7 AML marrow populations. In addition, Fig. 5 b demonstrates that while exposure to flavopiridol does not increase (and may actually decrease) the percentage of cells in S phase on day 1 relative to drug-free control cultures, there is a trend (P = 0.067) toward increases in the S-phase fraction in the flavopiridol-exposed cultures on day 2 just before adding ara-C.

The impact of flavopiridol pretreatment on the cytotoxic effects of ara-C on the individual leukemic marrow blast populations was determined by the addition of ara-C (1 μm final concentration) on day 2, 24 h after removal of flavopiridol 250 nm, with measurements conducted on day 5 after 72 h exposure to ara-C (F₂₅₀A₁). The 24-h rest period in vitro was intended to mimic the design of TST, where drug administration is timed to coincide with the expected occurrence of cell regrowth after initial cytoreduction (14, 15, 16). Furthermore, the dose of ara-C used in vitro is one that is readily achievable in vivo (51) and has been used in previous in vitro studies of TST-based ara-C pharmacology (17).

As depicted in Table 3 and Fig. 4, F₂₅₀A₁ cultures exhibited significant increases in overall cytotoxicity, as measured by FDA/PI, relative to all other culture conditions. Mean survival for the 20 marrow samples given F₂₅₀A₁ cultures was 35.6% (range, 2.4–96%) of cells receiving no drug, which was significantly less than the survival seen with flavopiridol alone (mean, 56.1%; range, 4.6–123.6; P = 0.0003) or ara-C alone (mean, 65.2%; range, 19–106%; P < 0.00001) when analyzed by paired t test. When used as single agents, flavopiridol and ara-C induced equivalent decreases in cell survival. Cytotoxicity with the flavopiridol/ara-C combination was at least 20% greater than the cytotoxicity detected in flavopiridol alone in 13 cases (65%), ara-C alone in 15 cases (75%), or both single drug cultures in 9 cases (45%). Antagonistic effects of F₂₅₀A₁ were uncommon, occurring in only one blast population each relative to F₂₅₀A₀ and F₀A₁ (Table 3).

Our studies demonstrate that flavopiridol exerts direct cytotoxic effects against a significant proportion of primary human acute leukemia marrow blast populations in vitro. This cytotoxicity occurs at a modest dose of drug, namely 250 nm, which is readily achievable in vivo (26). In response to flavopiridol, more than half (11 of 17) of the AML populations and all 3 of the ALL populations evinced significant increases in apoptosis during the first 24 h of culture, with dramatic increases (≥3-fold) in more than one-third of the leukemias. We chose to measure apoptosis per se only on day 1 of culture because such measurements at later time points could be confounded by death and clearance of the previously apoptotic cell fraction. Interestingly, most of these leukemia populations are at least as sensitive to flavopiridol 250 nm as they are to ara-C 1 μm. This finding may not be so surprising, however, given the refractory nature of these patients and the fact that most of them have received ara-C and have either relapsed despite therapy or failed to respond altogether. The notion that the patients’ blasts are indeed ara-C resistant is substantiated by comparison with HL-60 cells, where the LC₅₀ for ara-C is ∼200 nm, one-fifth of the dose that killed ∼50% of our fresh leukemia blast populations using the same FDA/PI assay (53).

From this perspective, 70% of the marrow cell populations examined in this study were obtained from patients with relapsed and refractory leukemias who had received ara-C previously as part of antileukemic induction and/or consolidation regimens and thus, by virtue of having recurrent or persistent leukemia, would be expected to be resistant to the cytotoxic effects of ara-C. The notion that these populations are indeed ara-C resistant is substantiated by finding that ara-C had only a modest impact on overall cell survival, decreasing survival to 65% (median, 69%; range, 19–106%) of that seen in control cultures (Fig. 4, Table 3). This relative refractoriness to ara-C cytotoxicity was evident for the 6 AML populations that were previously untreated and therefore “ara-C-naïve” (mean, 67%; range, 21–106% of control) as well as for those who had received ara-C previously. Three of the previously untreated patients (patients 2, 14, and 18) had therapy-related AML associated with adverse cytogenetics and 3 (patients 8, 12, and 17) had MDS/AML (2 with complex cytogenetic features; see Table 1). The apparent in vitro refractoriness to ara-C-induced cell death in these previously untreated patients parallels the known primary clinical refractoriness of this AML subgroup to classical induction chemotherapy approaches, all of which rely heavily on moderate to high doses of ara-C (1, 2, 3, 4, 14, 18, 19, 20, 21, 22).

We tested a sequence of drug administration in vitro that would allow us to detect cytotoxic activity of flavopiridol on acute leukemia cell populations, namely, the use of flavopiridol as an initial single drug, with the intent of extrapolating the resultant data to the design of a clinical trial that would likewise permit direct assessment of flavopiridol’s antileukemic activity in vivo (47, 48). The sequence tested here, specifically flavopiridol followed by ara-C, differs from studies where administration of the cdk inhibitor flavopiridol (49) or UCN-01 (53, 54) appears to exert its greatest antitumor effects when administered after chemotherapeutic agents. These studies were conducted in solid tumor (49) and leukemic (54, 55) cell lines, and the agents were administered in close temporal proximity (i.e., 6–24 h of initial exposure, followed immediately by the second drug in sequence). In contrast, our studies differ in two potentially important ways. First, we studied fresh leukemia populations, which are inherently heterogeneous in terms of cell growth kinetics and pharmacological determinants. It seems likely that this heterogeneity may account for the relatively modest effects of flavopiridol alone and in timed sequence with ara-C seen in these fresh human leukemias relative to the quantitatively more striking effects seen in studies using cell lines. Second, we designed our sequential approach to allow time for those cells surviving the initial flavopiridol exposure to re-enter and traverse the cell cycle and thereby acquire growth kinetic sensitivity to ara-C (16, 17, 49). However, because we did not examine alternate drug schedules, our data do not preclude the possibility that another sequence of flavopiridol and ara-C might be equally effective in terms of net antileukemic cytotoxicity. Thus, alternate sequences may be worth future exploration in vitro, particularly if an in vivo TST-based clinical trial using the sequence of flavopiridol followed by ara-C does not yield rewarding clinical results in patients with refractory leukemias.

Nonetheless, the sequential administration of flavopiridol followed by ara-C was able to augment net antileukemic cytotoxicity in a substantial majority of these poor-risk leukemias, including all three ALLs (patients 10, 11, and 20). This observation is interesting because we did not detect consistent recruitment into S phase after flavopiridol exposure in these ALL marrows. It is possible that the enhanced ara-C sensitivity of the ALL marrows is, in part, independent of cell growth kinetic changes. Flavopiridol induces apoptosis in diverse lymphoproliferative malignancy cell lines of either B or T origin (28, 29) and in primary malignant cells from chronic lymphocytic leukemia (45, 55) and multiple myeloma (46) patients. Provocatively, the apoptosis may be independent of changes in Bcl-2 and/or functional p53 expression (29, 45, 56). Most recently, a new Burkitt’s lymphoma cell line (GA-10) established from peripheral blood cells from a chemorefractory patient and characterized by absence of EBV and presence of biallelic p53 mutations has been found to exhibit striking sensitivity to flavopiridol, with induction of apoptosis at doses as low as 100 nm (57). Again, the apoptosis seen in GA-10 is associated with activation of caspase-3 but does not rely on p53 function (57). Taken together, the in vitro findings in malignant lymphoid cell lines, primary chronic B-cell malignancies (chronic lymphocytic leukemia and myeloma), and our small experience with primary ALL samples suggest that flavopiridol might be effective for lymphoid malignancies, including ALLs and should be tested clinically in those settings.

In conclusion, our studies demonstrate that flavopiridol can induce apoptotic cell death in leukemic marrow blast cell populations obtained from patients with acute leukemias that would be expected to be resistant to ara-C and other chemotherapeutic agents. Furthermore, in an in vitro model designed to mimic in vivo TST, flavopiridol-treated blast cultures exhibit increased sensitivity to the proapoptotic and cytotoxic effects of ara-C. Although we cannot exclude the possibility that flavopiridol and ara-C are acting in an independent fashion, we observed an apparent advantage of sequential FP₂₅₀A₁ over either agent given singly for newly diagnosed and previously treated refractory leukemias and for both myeloid and lymphoid blast populations. The enhancement of ara-C-based antileukemic activity by flavopiridiol supports the design of a clinical trial of TST in refractory and relapsed acute leukemias where flavopiridol is given for the dual purpose of initial cytoreduction and subsequent priming of the remaining leukemic cell cohort for the purpose of enhancing the response of that cohort to cytotoxic chemotherapeutic agents.

We thank Mary Wasti for secretarial expertise and Dr. Scott Kaufmann for helpful insights.