Biological Barriers Contributing to Anthracycline Resistance
The wide variability in both clinical and in vitro sensitivity to conventional chemotherapeutics has generated intense interest in identifying the precise cellular mechanisms contributing to drug resistance in acute myeloid leukemia (AML). Characterization of these processes is essential to a biologically sound rationale for the development of more effective treatment strategies and new antineoplastic agents. Correlative studies suggest that some of the recognized mechanisms have prognostic relevance and, therefore, may contribute to clinical drug resistance.
In this article, we will address: (1) possible mechanisms contributing to anthracycline resistance in AML, in particular, overexpression of the plasma membrane transporter, P-glycoprotein (P-gp); (2) P-gp chemomodulation; and (3) the relative cytotoxicities of idarubicin, daunorubicin and mitoxantrone in multidrug-resistant tumor cell lines.
P-GLYCOPROTEIN-MEDIATED MULTIDRUG RESISTANCE
In vitro selection for resistance to a specific chemotherapeutic agent frequently leads to the acquisition of resistance to many structurally and functionally unrelated compounds, a phenomenon referred to as multidrug resistance (MDR). Overexpression of the MDR1 gene or its membrane product, P-gp, is the best characterized cellular mechanism responsible for resistance to the anthracyclines.1 In P-gp-positive cell lines, the cellís ability to accumulate and retain drugs is impaired as a result of active, outward transport across the plasma membrane. Substrates for P-gp include many of the most active natural products used in leukemia therapy, including the anthracyclines, vinca alkaloids, mitoxantrone and, to a lesser degree, the epipodophyllotoxin, etoposide.2
P-gp overexpression has been implicated as an important cellular mechanism possibly contributing to treatment failure in AML patients. Prospective studies in de novo AML have shown that overexpression of the MDR1 gene message is associated with a lower rate of complete remission and shorter remission duration among patients receiving conventional induction and postremission therapy.3,4 Overexpression of P-gp has been linked to a number of adverse prognostic variables, including age, secondary leukemia, cytogenetic pattern and a stem-cell or CD34 surface phenotype.1,4-6 In AML, expression of P-gp is governed in part by the lineage and stage of blast differentiation, in a pattern that reflects its differential regulation in normal hematopoiesis. P-gp expression is strongest in the CD34-positive progenitor cell population with long-term repopulating capacity7; in contrast, functional and surface detection of P-gp is not demonstrable upon terminal differentiation. The known linkage between P-gp and CD34 in AML suggests that P-gp represents a physiologic function conserved from normal hematopoietic counterparts.
To detect the classic MDR phenotype in clinical specimens, numerous methods have been applied, including molecular and immunologic assays, as well as functional assays of cellular capacity to exclude anthracyclines or fluorescent dyes. The technique with greatest clinical relevance remains to be determined, and each assay has its advantages and shortcomings. At a workshop at St. Judeís Medical Center participants met in an attempt to establish standards for MDR1 detection, and a proposed combination of techniques was essential for optimal sensitivity and specificity of P-gp detection. Consensus recommendations from this workshop are forthcoming. Because P-gp is expressed at high levels by NK cells, T-lymphocytes and other normal hematopoietic elements,8 immunodetection remains essential to distinguish between normal and leukemic-cell P-gp expression.
Numerous noncytotoxic compounds serve as substrates for P-gp and block the cellular extrusion of anthracyclines and other antineoplastic agents, including verapamil, quinine, tamoxifen and cyclosporin-A.9-11 On a molar basis, cyclosporine is a more potent inhibitor of P-gp function both in vitro and in animal models and, for this reason, was selected for testing in a phase I/II trial in patients at the Arizona Cancer Center with poor-risk AML. Treatment with high-dose cytarabine was followed by daunorubicin administered concurrently with cyclosporin-A as a continuous intravenous infusion.12 Steady-state blood concentrations of cyclosporin-A that effectively inhibit P-gp in vitro were achieved in the majority of patients. Overall, 26 of 42 patients (62%) achieved a complete hematologic remission or restored chronic phase. More importantly, overexpression of the MDR1 gene message evident prior to treatment was absent in relapsed specimens from patients who achieved a complete remission, suggesting the elimination of MDR1-positive clones. Whether these results can be attributed to P-gp inhibition or the actions of high-dose cytarabine cannot be discerned from this trial.
Randomized studies are now in progress in the Southwest Oncology Group (SWOG) to determine the contribution of cyclosporine to this regimen in patients with poor-risk AML. Common toxicities of cyclosporine chemomodulation include nausea and vomiting, hypomagnesemia, prolonged myelosuppression and transient hyperbilirubinemia, the latter of which appears to be the result of direct inhibition of bilirubin transport proteins. Not surprisingly, the Arizona Cancer Center study found elevations in serum bilirubin levels to be associated with altered daunorubicin pharmacokinetics, making interpretation of randomized trials problematic.12
Improved response with a P-gp modulator might then be attributed to increased systemic drug exposure or MDR modulation. Other variables possibly influencing response to this type of treatment approach include the prevalence of P-gp-positive cases in each treatment arm, the ability to achieve and sustain effective chemomodulator blood levels and the presence of non- MDR mechanisms affecting anthracycline resistance. These factors will be assessed in the SWOG trial to ensure a valid comparison of response according to drug-resistance phenotype and relative drug exposure.
The lack of specificity of currently available P-gp modulators emphasizes the need for more selective and more potent chemosensitizers. Several such compounds, including the cyclosporin-D analogue, PSC 833 and the carboxamide derivative, GF120918, have shown promise in preclinical investigations.13,14 On a molar basis, both compounds exhibit approximately five- to tenfold greater in vitro potency than cyclosporin-A. Preliminary experience with PSC 833 in phase I trials has not shown significant alterations in serum bilirubin. PSC 833 is now under investigation in multi-institutional studies in the US and Europe as a modulator of daunorubicin resistance in AML.
NON-P-GLYCOPROTEIN-MEDIATED MULTIDRUG RESISTANCE
P-gp is only one of several biochemical mechanisms now recognized to mediate the MDR phenotype in tumor cell lines. To date, at least two additional, non-P-gp MDR mechanisms associated with decreased drug accumulation have been identified (Table 1). These include the MDR- associated protein (MRP) described by Cole et al15 and the MDR phenotype identified by the monoclonal antibody LRP56.16 Although outward drug transport may be the principal mechanism limiting nuclear drug exposure in these MDR variants, alterations in intracellular drug sequestration also contribute to cellular resistance.17,18 Cell lines displaying these drug-resistance phenotypes are cross-resistant to a broad range of natural products, yet are weakly sensitive to modulation by conventional chemosensitizers, such as verapamil and cyclosporin-A.
The MRP gene encodes a 190-kD glycoprotein localized to the plasma membrane as well as endomembrane structures.19 Highly specific monoclonal antibodies for the MRP gene product have become available only recently and, for this reason, studies in human leukemia to date have been limited to molecular assays of gene message. Preliminary investigations indicate that the MRP gene is constitutively expressed at low levels in peripheral blood and bone marrow mononuclear cells of diverse lineage.20 Unlike P-gp, lineage-dependent differences in levels of gene message have not been discerned. Among the hematologic malignancies, only chronic lymphocytic leukemia consistently shows MRP overexpression.20,21 In contrast, MRP overexpression appears to be uncommon in AML, acute lymphoblastic leukemia and chronic myeloid leukemia. Immunodetection studies are necessary to exclude possible post-transcriptional modifications affecting expression of the MRP gene product. The monoclonal antibody LRP56 identifies a 110-kD cytoplasmic protein expressed in a P-gp-negative MDR lung-cancer cell line and, thus, termed lung-resistance protein (LRP).16 Similar to P-gp, LRP is overexpressed in a number of normal tissues, including peripheral blood lymphocytes, bronchial epithelium, renal tubules and intestinal epithelium. The LRP56 antibody produces a coarsely granular pattern of cytoplasmic staining, suggesting close association with endomembrane structures.
Preliminary investigations performed at the Arizona Cancer Center suggest that the LRP phenotype may adversely influence treatment outcome in AML.22 LRP overexpression detected by immunocytochemical techniques was evident in 35% of patients with de novo AML, 48% with secondary AML and 38% of relapsed specimens, but in only one of ten cases of blast phase chronic myeloid leukemia. Remission rate was significantly lower (P = .003) in patients with LRP overexpression (36%), compared to LRP-negative cases (68%). This difference was the result of a higher incidence of resistant failures in patients with LRP overexpression. When compared with P-gp, both markers have prognostic significance; however, only LRP had independent prognostic relevance in a logistic regression model.22 LRP overexpression was associated with adverse prognostic variables, including advanced age, P-gp and CD7 surface phenotype, but not CD34.
The prognostic relevance of the LRP phenotype in other malignancies is now under investigation. In addition, LRP overexpression in AML will be evaluated further in the SWOG trial to determine its effect on outcome in patients randomly assigned to treatment with cyclosporin-A.
RELATIVE CYTOTOXICITIES OF IDARUBICIN, DAUNORUBICIN AND MITOXANTRONE
Until MDR modulators have demonstrated benefit in well-controlled, randomized trials, achieving optimal results with currently available agents remains a prerequisite. Idarubicin, the dimethoxy-derivative of daunorubicin, has shown superior remitting activity in standard-risk de novo AML in randomized trials.23,24 At equimolar concentrations, idarubicin is more potent than daunorubicin in both sensitive and MDR leukemia cell lines.25 Whether idarubicin has a clinical advantage when administered at dosages that approximate 25% of those used for daunorubicin remains in question.
Relative cytotoxicities of idarubicin, daunorubicin and mitoxantrone in P-gp-positive leukemia and myeloma cell lines are listed in Table 2.26 Idarubicin shows greater relative cytotoxicity in P-gp-positive cell lines, as reflected by its lower resistance index, perhaps related to its inherent lipophilicity and more rapid cellular penetration. Cytotoxicity is enhanced by verapamil, confirming that the two anthracyclines and mitoxantrone are substrates for P-gp; however, a clear benefit with idarubicin is not discernible at concentrations simulating in vivo drug exposure, raising concerns that none of the available anthracyclines or anthracenediones offers a substantial advantage in MDR leukemias. The principal alcohol metabolites of idarubicin and daunorubicin have a longer half-life than the parent compounds, but show approximately 10- to 20-fold less activity in P-gp-positive cell lines. None of the agents investigated exhibits a cytotoxicity advantage in P-gp-negative MDR cell lines.26
CONCLUSIONS AND RECOMMENDATIONS
An understanding of the biological mechanisms contributing to treatment failure with current regimens is necessary to develop more effective therapeutic approaches for acute leukemia. It is now clear that P-gp is only one of several possible cellular mechanisms leading to anthracycline resistance in AML. Treatment strategies that address isolated mechanisms of resistance merit investigation, but may provide only minor clinical benefit. Nevertheless, better understanding of these mechanisms will advance the development of new antineoplastic agents with alternate mechanisms of action. Until then, participation in cooperative group trials is encouraged to discern the potential benefit of MDR-directed therapies. Optimization of treatment with currently available agents by dose-escalation of less cardiotoxic anthracyclines/anthracenediones should be investigated. In addition, high-dose cytarabine, a non-P-gp substrate, may offer a rational alternative in MDR-positive cohorts.
ALAN F. LIST, MD, is an Associate Professor of Medicine, Bone Marrow Transplant Program and the Arizona Cancer Center, University of Arizona College of Medicine, Tucson, Arizona.