Multidrug resistance in haematological malignancies


  • P. Sonneveld

    1. From the University Hospital Rotterdam – Dijkzigt, Department of Hematology, Dr Molewaterplein 40, 3015 GD Rotterdam, The Netherlands
    Search for more papers by this author

Professor P. Sonneveld, MD, University Hospital Rotterdam – Dijkzigt, Department of Hematology, Dr Molewaterplein 40, 3015 GD Rotterdam, Netherlands (fax: +31 010 463 5814; e-mail: sonneveld@haed.azr nL).


Abstract. Sonneveld P (University Hospital Rotterdam – Dijkzigt, The Netherlands). Multidrug resistance in haematological malignancies (Internal Medicine in the 21st Century). J Intern Med 2000; 247: 521–534.

The development of refractory disease in acute myeloid or lymphoblastic leukaemias (AML, ALL) and multiple myeloma (MM) is frequently associated with the expression of one or several multidrug resistance (MDR) genes. MDR1, MRP1 and LRP have been identified as important adverse prognostic factors in AML, T-ALL and MM. Recently, it has become possible to reverse clinical multidrug resistance by blocking P-glycoprotein-mediated drug efflux. The potential relevance of these reversal agents of MDR and potential new approaches to treat refractory disease are discussed.


The treatment in malignant haematologic diseases such as acute myeloid leukaemia (AML), multiple myeloma (MM) and the non-Hodgkin lymphomas (NHL) results in an initial response in the majority of the patients. However, the remission may be of short duration and it is frequently followed by a relapse of resistant tumour cells associated with clinically refractory disease. Goldie and Coldman have formulated that resistant cells may be present at diagnosis or develop by spontaneous mutations. These cells expand by selection during treatment and overgrow the sensitive cells [1]. Recently, the existence of so-called pleiotropic drug resistance (multidrug resistance, MDR) has been identified [23]. In this review the clinical significance of the MDR phenotype and pharmacologic approaches to reverse MDR in AML, NHL and MM will be discussed.

Multidrug resistance

The presence of a diminished response to chemotherapy can be assessed in AML cells in vitro by determination of the clonogenic capacity. In such an assay, purified AML blast cells are exposed to increasing concentrations of cytotoxic agents, and the inhibitory concentration for cell proliferation is analysed. Drug resistant cells will tolerate higher drug concentrations in vitro. More recently, the methyl-thiazol-tetrazolium (MTT) assay has been used to assess the viability of cells during and after exposure to cytotoxic agents. However, these assays do not contribute to our knowledge of mechanisms of drug resistance.

Subpopulations of drug resistant cancer cells are frequently cross-resistant to several structurally unrelated cytotoxic compounds of natural origin, such as anthracyclines (doxorubicin, daunorubicin, idarubicin), vinca alkaloids (vincristine, vinblastine), epipodophyllotoxins (etoposide, teniposide), taxanes (taxol, taxotere) and amsacrine. These drugs have few structural and functional similarities except that they are small, hydrophobic molecules. The drugs may enter the cell by passive diffusion across the cell membrane lipid bilayer. Alkylating agents such as cyclophosphamide, melphalan and chlorambucil, platinum derivates and antimetabolites do not share these characteristics. The phenotypic characteristics of independently derived multidrug resistant cell-lines are remarkably similar, despite their different tissue origin and the drug(s) used for their selection [4 5]. The most striking phenotypic marker of MDR is the overexpression of P-gp, which is located in the plasma membrane of cells. P-gp is encoded by the mdr1 gene, located on chromosome 7. The classical definition of MDR was established, when increased P-gp expression was found in a large range of cell-lines that were selected through exposure to different cytotoxic agents [6]. Since then, several other proteins have been described that may also lead to the development of pleiotropic resistance to cytotoxic agents. This was first demonstrated in multidrug-resistant cell-lines that do not possess P-gp, but express the multidrug-resistance related protein (MRP). The clinical relevance of this and other pathways through which drug resistance may develop, will be discussed.

Resistance associated with mdr1

There are two different genes encoding for P-glycoprotein in humans: mdr1 and mdr3, both located on chromosome 7. Despite their homology, only Mdr1 is a drug efflux protein related to multidrug resistance. P-gp belongs to the superfamily of ATP-binding cassette (ABC) transporters, a family of ATP-dependent transport proteins. The 170 kDa P-gp consists of two structurally homologous halves, each with six transmembrane domains, one ATP-binding site and the highly conserved ‘Walker A’ and ‘Walker B’ motif ( Fig. 1) [7]. These two halves are probably derived from an internal gene duplication [8]. Several studies have suggested that phosphorylation of P-gp might be essential for drug transport [9]. However, two different groups showed that a mutation of the major phosphorylation sites within P-gp do not affect its transport function [10 11]. The glycosylated sites of P-gp at the cellular outside are probably involved in routing and stability of the protein [12], but serve also as antigens for monoclonal antibodies recognizing P-gp.

Figure 1.

Schematic representation of mdr1 P-glycoprotein.

P-gp has a wide variety of substrates. All its substrates are large hydrophobic and amphipathic molecules, although they have no structural dissimilarity. These molecules are able to intercalate into the membrane and enter the cytosol by passive diffusion.

It is no longer believed that P-gp is a ‘classical’ pump, which binds substrates from the extracellular fluid and then transports these over the membrane. Hydrophobic compounds that are substrates for P-gp do not fully penetrate into the cytoplasm of cells that express P-gp [13]. Interaction of substrate with P-gp has been shown to take place within the membrane [14]. This mechanism of transport is also postulated for a prokaryotic homologue of P-gp with a similar broad substrate specificity in Lactococcus lactis[15].

Although the exact mechanism by which this protein removes hydrophobic drugs from the cell is still not resolved, P-gp may function as a ‘flippase’ within the plasma membrane [16]. It may translocate drugs actively (ATP-dependent) from the cytosolic inner lipid leaflet of the plasma membrane to the outer lipid leaflet. Then these drugs are able to leave the plasma membrane by diffusion. This model is supported by the observation that the mouse mdr2 (homologous to human mdr3), which shows around 75% homology to the multidrug transporting P-gp, may be involved in the secretion of phosphatidylcholine into the bile as observed in a knock-out system [17]. Human mdr3 P-glycoprotein is probably also a transporter for phosphatidylcholine [18] and restores this defect in the mdr2 knock-out mice [19]. In addition, the human mdr1 P-glycoprotein is able to transport phospholipids , as is the human mdr3 P-glycoprotein [20].

Most modulators, such as verapamil and cyclosporin A, reverse P-gp-mediated multidrug resistance by competition between modulator and substrate. The efficient inhibition of P-gp-mediated drug efflux by these modulators may be due to a high passive diffusion over the membrane in contrast to a slow passive diffusion [21]. In this way, these modulators keep P-gp occupied. This mechanism may not account for PSC 833-mediated P-gp reversal, because PSC 833 does not only antagonize the transport of cytostatic agents by P-gp but also binds to cyclophorin [22].

Expression of mdr1 in haematological malignancies

MDR expression is present in several solid and haematological malignancies [23]. Tumours that are derived from tissues that normally stain P-gp-positive may more frequently express the mdr1 gene [2 24]. Tumours derived from the haematological compartment that frequently express P-gp include AML, NHL and acute lymphoblastic leukaemia (ALL) [25–32]. High P-gp expression is frequently observed in patients who were previously treated with natural product cytostatic drugs [2]. This observation suggests that P-gp expression can be induced by a selection process occurring during repeated exposure to these drugs. In addition, anticancer agents may activate P-gp transcription [33 34].

AML is a clonal disease that finds its origin in the transformation of the uncommitted stem cell. This may explain why P-gp is frequently present in blast cells at diagnosis of de novo AML ( Table 1). P-gp expression varies from 19 to 75% of untreated AML cases [35–47] and high-risk myelodysplasia [48]. The quantitative difference between these studies may result from the variety of analytical assays that were used for P-gp analysis. Some studies have also investigated the level of mdr1 expression in relapsed AML. Generally, patients with refractory and/or relapsed AML more frequently express mdr1 more than de novo patients [3639].

Table 1.  Clinical significance of multidrug resistance in de novo AML
expression (%)
Patients (n) Prognostic value
for response (CR)
Sato et al. 1990 [35]67 15Yes, P = 0.01
Pirker et al. 1991 [38]71 63Yes, P = 0
Marie et al. 1991 [37]19 35Yes, P = 0.03
Zhou et al. 1992 [39]43 51Yes, P = 0.005
Campos et al. 1992 [41 56]47150Yes, P < 0.00001
Te Boekhorst et al. 1993 [42]58 52Yes, P = 0.0003
Ino et al. 1994 [43]27 52No
Lamy et al. 1994 [44]53 51Yes, P < 0.01
Leith et al. 1995 [45]75171Yes, P = 0.0001
Leith et al. 1995b [51]72193, > 60 yearsYes, P = 0.0001
Paietta et al. 1994 [46]Not mentioned188No
Te Boekhorst et al. 1995 [47]74 38Yes, P = 0.048
Guerci et al. 1995 [49]41 69Yes, P = 0.0001
Van den Heuvel-Eibrink, 1997 [50]44130Yes, P < 0.001

A significant lower probability to achieve a complete remission was observed in patients with mdr1 expression as determined by either RNA assays [3537–39] or P-gp staining [41 42 44 45 49 50]. In contrast, no such correlation was found in two studies using the same assays [43 46]. The lack of agreement between some studies underlines the need for standardizing the assays that are used for mdr1 analysis in AML specimens. Even with highly specific assays it is uncertain if low numbers of mdr1-expressing cells contribute to a poor outcome of treatment. In an attempt to evaluate the value of different numbers of mdr1-positive blast cells, Te Boekhorst showed that even small numbers of these cells (1–5%) represent an increased risk of refractory disease [47]. These data suggest that small numbers of mdr1 cells are relevant for the response to treatment in de novo AML, and that assays should be developed which are capable of detecting mdr1 expression in such small cell fractions.

In one study it was found that mdr1 expression at diagnosis has a negative impact not only on CR rate, but also on remission duration and on survival [38]. More recently, Leith [51 52] demonstrated that elderly AML patients have a higher probability of P-gp expression and enhanced drug efflux. In these patients, P-gp expression was associated with a lower chance of achieving a complete remission and a shorter survival period. These data confirm earlier reports that mdr1 expression is an independent prognostic variable for response in AML and that it is independent of age, cytogenetic profile and proportion of CD34+ AML cells [50]. In acute lymphoblastic leukaemia (ALL), P-gp is observed in 38% of cases. In a multivariate analysis it was shown to be an independent, poor prognostic factor for response and survival in both children and adults [53 54].

Several assay methods of mdr1 expression are available for evaluation of clinical samples; however, they do not necessarily produce comparable results ( Table 2 ). Generally, bulk methods such as mRNA-PCR or Northern blot are not suitable for quantitative differences of mdr1 expression in subpopulations of cells with a certain morphology, and the results may be influenced by contaminating T-cells. With these assays, it is not possible to correlate mdr1 expression with maturation and/or differentiation markers. Therefore, most investigators prefer to determine mdr1 expression at the protein level (P-gp). P-gp-specific antibodies such as C219 and JSB1 have been used in immunocytochemistry, but these assays are not highly reproducible and they are not quantitative. Flow cytometry can be used to determine P-gp expression in viable cells using monoclonal antibodies such as MRK16 and UIC2 which bind to an extracellular epitope. This technique allows the detection of P-gp in different subsets of cells. Such studies have demonstrated that mdr1 is heterogenously expressed in subsets of normal as well as leukaemic blood cells. High mdr1 expression was observed in immature, CD34+ haematopoietic cells, T-cell lymphocytes, CD56+ natural killer cells and macrophages. In leukaemic blast cells, P-gp expression is also frequently associated with the expression of the CD34 or CD7 antigens [41 55]. More recently, coexpression of P-gp and CD34 was demonstrated in AML blast cells [42 45]. In a recent study by Leith et al., discordant expression of mdr1 was observed in CD34+ cells, i.e. P-gp staining and the Rhodamin fluorescent dye retention assay identified AML cases with different properties [45]. This study points to the possibility that mdr1 expression may vary and/or have multiple functional properties in CD34+ AML cells. These data suggest that the expression of mdr1 in immature stem cells is conserved during leukaemic transformation [41 42 44 48 56–58].

Table 2.  Analytical methods for mdr1 expression
In situ hybridizationMdr1 gene not amplified
PCRNot quantitative
RT-PCR, RNase protectionSemi-quantification of mRNA
Few cells needed, cytology possible
Low sensitivity, high background
MRK16 or U1C1 staining by flow cytometrySpecific for subsets of cells
Rhodamin retention
Analysis of functional efflux and
effect of P-gp modulating agents

In addition to Rhodamin 123, the P-gp-mediated efflux of cytostatic agents such as daunorubicin, doxorubicin or vincristine can be determined in cell suspensions. Using such a functional assay, the effect of P-gp inhibition by a drug-resistance modulating agent can also be evaluated [59–61]. Together, these assays may be used to analyse the MDR profile of AML cells obtained from individual patients. Only a few studies have attempted to correlate the expression of mdr1 with in vitro drug sensitivity [37 62–64]. Beyond the considerable technical difficulties of these clonogenic assays, interpretation of the results is hampered by the fact that other mechanisms of drug resistance may also be involved.

In pretreated non-Hodgkin’s lymphoma, the reported incidence of P-glycoprotein staining varied from 2 to 49% in untreated patients and 64% in pretreated patients, whilst with mRNA analysis these figures were 22–50% and 30–60%, respectively [25–32]. As summarized by Yuen & Sikic [32], it is presently unclear whether MDR expression has a significant impact on the response to therapy in lymphomas. Most studies suggest, however, that P-gp-positive patients may have a poor prognosis compared to negative patients. A high frequency of P-gp expression is also observed in relapse multiple myeloma [57 58 65–67]. In untreated myeloma, MDR does not seem to have an important impact on the outcome of treatment [66], whilst in VAD refractory myeloma, mdr1 expression is almost unvariably observed [57 65]. In these patients, the frequency and intensity of P-gp expression is correlated with the prior exposure to doxorubicin and vincristine, respectively [60].

Other mechanisms of pleiotropic drug resistance in AML


Typical MDR has now been recognized as an important cause of in vitro resistance to many antileukaemic drugs such as anthracyclines, epipodophyllotoxins and amsacrine. However, in spite of the fact that mdr1 confers clinical resistance in AML, other mechanisms of resistance seem to be involved as well (Table 3). The MRP (multidrug-resistance associated protein) gene is a member of the superfamily of membrane drug transporters and is located on chromosome 16. Like P-gp, it confers resistance to anthracyclines. MRP expression has been reported in a variety of untreated and refractory haematologic malignancies, including AML and chronic leukaemias [68–71]. The frequency of MRP expression in untreated AML to a level surpassing that of normal blood leucocytes is ≈ 50% [68]. The expression in relapsed AML patients is higher [70]. In cell lines and in AML specimens, coexpression of MRP or MRP1/MRP2 subclasses with P-gp is frequently observed [68 71 72]. Although the clinical relevance of MRP expression alone in AML is unclear [73 74], combined expression has a strong negative impact on response and survival [75]. An Australian group reported that deletion of the mrp gene in AML patients with the 46,inv(16) karyotype was associated with a favourable effect on disease-free survival and overall survival [76]. There are at present no possibilities of circumventing MRP resistance through reversal agents. Genistein is the single effective reversal agent of MRP, but it cannot be used in patients. Recently, a vault transporter protein was identified in doxorubicin resistant cell lines and designated LRP [77]. Expression has been observed in blast cells of AML patients and seemed increased in patients who responded poorly to anthracyclines [73 78 79].

Reversal of multidrug resistance in patients

Since the recognition of MDR as an independent mechanism of drug resistance, attempts have been made to reverse P-glycoprotein expression or function in vitro by adding anti-mdr1-oligonucleotides [80] or protein kinase C inhibitors like staurosporine, which down-regulate mdr1 expression [81]. An alternative pharmacological approach to modulating P-gp function is based on inhibition of its binding to cytostatic drugs. Competitive binding experiments using [3H]azidopine or [3H]cytostatic drugs showed that MDR-reversing agents compete with certain cytostatic drugs for binding to P-gp [82]. Many noncytotoxic agents can indeed restore the intracellular drug accumulation of cytostatic drugs by blocking P-gp efflux functional sites. Such agents include calcium channel blockers, calmodulin inhibitors, immunosuppressive agents, quinolones, indole alkaloids, detergents, steroids and antiestrogens [83–90]. Several of these reversal agents share common chemical features, such as a planar aromatic domain and two amino groups, one of which has a cationic charge at physiological pH, and they all are highly lipophilic. A combination of different modulators, such as verapamil and cyclosporin A, may result in a synergistic effect [86]. This observation suggests that the exact mechanism of drug reversal may not be identical for different reversing agents. In vitro reversal of drug resistance in fresh AML specimens has also been investigated. Verapamil, cyclosporin A, PSC 833 and other reversing agents increase the intracellular retention of daunorubicin in AML blast cells which express P-gp, but not in drug-sensitive or P-gp-negative AML cells [42 47 63 71 91–94]. This pharmacological effect is associated with increased cytotoxicity of anthracyclines in in vitro clonogenic assays or in the MTT test [95–99]. A different in vitro dye accumulation with and without a reversing agent is now regarded as the key test for functional P-gp expression.

So far, few clinical trials have been performed with effective MDR modulators. Several phase II/III studies in solid tumours combined verapamil, quinidine or trifluoroperazine with doxorubicin and epirubicin, respectively. In these studies no serum levels needed to modulate MDR could be achieved [28 100–102]. Other combinations included verapamil plus vinblastine or etoposide [103], high-dose verapamil or R-verapamil plus chemotherapy, diltiazem plus vincristine, tamoxifen and vinblastine and nifedipine plus etoposide [82 102–110]. These trials have shown that the concept is feasible, although generally the clinical effect in refractory solid tumour patients was limited. However, for optimal modulation of drug resistance, it would be necessary to achieve a steady state or trough , nonprotein-bound plasma concentration, which is one or two times higher than the concentration needed in vitro to circumvent MDR. For many of these modulators, this concentration exceeds that of other pharmacological interactions. In addition, the variability of resorption, protein binding and pharmacokinetics leads to unpredictable plasma levels and frequently to unacceptable toxicity. At present, cyclosporin A and the cyclosporin D analogue PSC 833 (valspodar) [111] are the most promising compounds for clinical drug modulation, since these agents can be administered at sufficient doses to achieve effective serum levels and can also be combined with cytotoxic agents without a concomittant steep increase of toxicity (Table 4).

Administration of cyclosporin A and PSC 833 may result in increased toxicity of the anticancer drugs for two reasons. First, P-gp may be blocked in normal tissues by the modulator. CD34+ haemopoietic progenitor cells are potentially harmed by a combined regimen of a modulator plus myelotoxic drugs, because these cells express P-gp. Severe myelosuppression in these patients may result from inhibition of P-gp in CD34+ stem cells and their progeny [112]. Secondly, these modulators appear to alter the pharmacokinetics of cytostatic drugs through modulation of ABC transporter proteins in the biliary canaliculi and the renal tubuli, thereby blocking biliary and renal drug elimination. Such an effect was first observed in mice and in patients treated with verapamil and doxorubicin [113]. Because of the cross-over design in that study one could demonstrate that the peak levels, the elimination half-life and the volume of distribution of doxorubicin were higher in the presence of verapamil at a plasma concentration of 5 μm. Increased toxicity was also observed in several clinical studies of verapamil with VAD (vincristine/doxorubicin/dexamethason) [28 62], bipredil plus vinblastine [114], cyclosporin A plus daunorubicin and high-dose cytarabine [115], cyclosporin A with vincristine, doxorubicin, and dexamethasone [116] and cyclosporin A plus etoposide [117]. These studies showed that cyclosporin A at effective blood levels leads to an approximately two-fold increase of the plasma retention time of etoposide, daunorubicin and doxorubicin. As a consequence, in trials attempting to modulate resistant tumour cells, the dose of drugs should be reduced by 25–50% in order to attain equitoxic treatment.

The first case report of clinical reversal of drug resistance in haematologic patients was published in 1988. A myeloma patient refractory to VAD (vincristine, doxorubicin, dexamethasone) was treated with VAD plus verapamil, which resulted in a response of some duration [118]. Based on this experience, a larger group of patients with multiple myeloma or non-Hodgkin’s lymphoma was treated with high-dose verapamil with the purpose of achieving effective plasma concentrations of verapamil [28 64]. The high plasma concentrations of verapamil required for P-gp inhibition led to cardiac arrhythmias in the majority of patients. In a subsequent phase I/II study of verapamil combined with VAD, cardiac monitoring was needed and most patients had EKG irregularities. It should be noted, however, that approximately half of the patients achieved a response [62].

In an attempt to avoid the cardiovascular side-effects of calcium antagonists, an NCI phase I trial with dexverapamil was performed in patients with refractory lymphoma or sarcoma [119], which demonstrated the feasibility of achieving effective and safe plasma concentrations. In a subsequent cross-over trial of dexverapamil combined with EPOCH (etoposide, doxorubicin, vincristine, cyclophsphamide and prednisone), several responses were noted in 64 analysed patients [120]. Notably, half of mdr1-positive patients responded, compared with one-eighth of patients with no or weak expression. In these extensively pretreated patients, other acquired or intrinsic mechanisms of drug resistance may, of course, also be of influence.

The feasibility of cyclosporin A as an MDR reversal agent, combined with VAD, was evaluated in 21 myeloma patients with advanced disease, who had progressed after or while receiving VAD [116]. In this heavily pretreated group of patients, 58% of patients with mdr1-positive plasma cells responded, compared with 33% of mdr1-negative patients. The steady state serum concentration of cyclosporin A which was achieved with continuous infusion of 7.5 mg kg–1 day–1 in these patients was, however, suboptimal at 1000–1100 ng mL–1. The toxicity was mild, due to the fact that the pharmacokinetics of doxorubicin were not significantly different from historical controls. More recently we have performed a co-operative trial in 22 myeloma patients who were refractory to primary treatment with alkylating agents or VAD, using oral PSC 833 combined with VAD by continuous infusion. In this dose-escalating schedule of PSC 833, given at the highest dose level of 15 mg kg–1 day–1, peak plasma levels were > 2500 ng mL–1 and trough levels were > 1000 ng mL–1, respectively. A twofold increase of the plasma area under the curve (AUC) of doxorubicin and of its hydroxyl metabolite was observed in the majority of the patients. For this reason, the doses of doxorubicin and vincristine had to be reduced. The dose-limiting toxicities were intestinal neuropathy and marrow hypoplasia, which seemed to be associated with an increased organ exposure to vincristine and doxorubicin, respectively. Interestingly, a number of clinical responses were noted, which were associated with a proportional reduction of plasma cells expressing the MDR phenotype [121].

This study will be followed by phase II and phase III studies in Europe and the USA in relapse patients, in order to explore the potential benefit of PSC 833 in refractory myeloma. The validity of MDR reversal should however, be investigated in randomized studies in patients who failed conventional therapy or who have a high frequency of mdr1 expression. Two such studies have been performed in multiple myeloma. Dalton et al. investigated the effect of verapamil added to VAD in a phase III trial. No effect was observed using verapamil at a suboptimal dose [122]. In a recent co-operative study of the EORTC and HOVON study groups, the effect of cyclosporin A was studied in myeloma patients who were refractory to alkylating agents. This phase II study showed an advantage of VAD/cyclosporin over VAD with respect to progression-free survival [123]. These and other studies were perfomed with unreduced dosages of vincristine, doxorubicin and dexamethasone (VAD) together with the reversal agent. Such an approach may, however, make it impossible to answer the question of whether P-gp inhibition in the tumour cells rather than a pharmacokinetic interaction is responsible for the observed effect. Treatment with cyclosporin A or PSC 833 reduces the biliary clearance of most P-gp transported drugs through an inhibition of ABC transporters such as Mdr3 and c-MOAT. This effect results in a delayed excretion and a prolonged plasma retention time of the cytostatic agent and thereby into an increased plasma area under the curve (AUC). Thus, an unmitigated dose of chemotherapy combined with a reversal agent results in a dose escalation of the involved cytostatic agents. Therefore, future phase III trials with MDR reversing agents have to take into account that the doses of VAD and other P-glycoprotein transported drugs may have to be reduced to equitoxic levels if one wants to properly evaluate the role of MDR reversal. Several trials have been performed in AML (Tables 4 and 5). Approximately 50% of untreated AML patients have P-gp expression in the blast cells, whilst at relapse most patients have even higher levels of P-gp expression. In addition, most antileukaemic agents except cytarabine belong to the natural product class of cytostatic drugs. These conditions make AML a suitable model for MDR reversal. The first attempt to treat a refractory AML patient with daunorubicine and cytarabine in combination with Cyclosporin A was performed in 1990. This patient was refractory to standard induction treatment and he subsequently achieved a short-lasting remission [65]. Several phase I and II trials were initiated in AML patients who were either refractory to primary treatment or had relapsed after a previous response. In one study, 20 refractory or relapse patients were treated with mitoxantrone and etoposide, to which cyclosporin A was added. The dose of the cytostatic agents was not reduced. Consequently, the toxicity of this regimen was considerable and primarily related to severe marrow hypoplasia. A number of objective responses, including complete remissions, were noted; however, the toxicity of this regimen was regarded as unacceptable [124]. In another study, the same approach was used in 42 patients with refractory or relapse AML or a blast crisis of CML. However, here daunorubicine combined with high-dose cytarabine and cyclosporin A were administered in a dose-escalation design. The dose of cyclosporin A was increased until plasma levels were obtained that were sufficient to inhibit P-gp of the patients own blast cells in vitro. The dose of daunorubicine was not attenuated. The toxicity of this regimen correlated with the dose of cyclosporin A. It consisted primarily of myelosuppression, nausea and hyperbilirubinaemia. The plasma levels of daunorubicin were increased, compared with controls [125]. Because of potential side-effects of cyclosporin A, such as nephrotoxicity and immunosuppression, PSC 833 was developed for clinical reversal of drug resistance. In a study from the Arizona Cancer Center, daunorubicine and cytarabine were combined with escalating dosages of PSC 833. In this study, daunorubicine was reduced from 45 mg m–2 to 34 mg m–2 daily. The results suggest that PSC 833 may effectively modulate P-gp-mediated multidrug resistance. It was observed that LRP-positive cases had a lower probability of achieving a complete response compared to LRP-negative cases, regardless of P-gp expression. This study suggests that, in addition to MDR, LRP expression is an adverse biological feature that is not affected by P-gp antagonists. Comparable results were obtained in another study from the USA [126]. Another MDR reversal agent that has been studied in AML is quinine [127]. In a phase III study, the slightly better response in patients using quinine was associated with higher toxicity, leading to an overall similar clinical benefit in patients treated with and without quinine. More recently, the South-Western Oncology Group (SWOG) performed a randomized study in relapsing or high-risk patients using standard induction treatment, compared with an attenuated dose of daunorubicin with cytarabine and cyclosporin A. Although the response rates were not different in both arms, the overall survival and the progression-free survival was significantly better in the patients receiving cyclosporin A (A. List, personal communication). Two abstracts were published of studies in which PSC 833 was administered together with standard induction regimens during frontline treatment in elderly AML patients [128 129]. These studies were designed as a dose finding study for the cytostatic agents combined with a standard dose of PSC 833. The results indicate that this approach is feasible even in elderly patients. The Dutch and British study groups have just completed a prospective randomized phase III trial to evaluate the clinical benefit of PSC 833 used in frontline treatment in this group of high-risk AML patients.


Multidrug resistance represents a form of pleiotropic drug resistance that has an adverse prognostic value in AML, refractory multiple myeloma and non-Hodgkin’s lymphoma. It may affect the outcome of current chemotherapy protocols. Therefore, mdr1 expression should be assessed in prospective randomized studies in these tumours. In addition, reversal of drug resistance should be attempted at an early stage of disease, before other relevant mechanisms of resistance may emerge. Reversal agents that only inhibit P-gp in tumour cells and do not influence the pharmacokinetics of cytostatic agents should be developed.

Received 27 October 1999; revision received 10 February 2000; accepted 9 December 1999.