Cyclosporine inhibition of P-glycoprotein in chronic myeloid leukemia blast phase

Blast transformation of chronic myeloid leukemia (CML-BP), the terminal phase of the disease, remains an important therapeutic challenge. Attempts to induce remission or to restore chronic phase with chemotherapy regimens active in acute myeloid leukemia (AML) are generally ineffective, yielding median overall survival (OS) of 2 to 4 months.1,2 Resistance to cytotoxic therapy in CML-BP extends to high-dose chemoradiotherapy conditioning with allogeneic stem cell rescue, offering a potential for extended survival rate that rarely exceeds 10%.1 The bcr/abl fusion gene product is singularly responsible for the hematologic manifestations of chronic phase CML3 and contributes to cellular resistance to cytotoxic therapy by the up-regulation of antiapoptotic molecules and other survival signals.4-8 In CML-BP, however, chemotherapy resistance is multifactorial and cannot be ascribed solely to bcr/abl. Although the bcr/abl-specific tyrosine kinase inhibitor, imatinib mesylate (Gleevec; Novartis, Basel, Switzerland), has shown promising activity in patients with CML-BP, responses in CML-BP are short-lived because of the acquisition of alternative mechanisms of resistance that contribute to treatment failure.9 10 

P-glycoprotein (Pgp) is a highly conserved plasma membrane glycoprotein that functions as an ATP-dependent multidrug exporter with broad specificity for natural product-derived antineoplastics.11-13 In adult AML, the overexpression of Pgp is associated with reduced cellular accumulation and in vitro resistance to anthracyclines that can be overcome by concurrent exposure to competitive Pgp antagonists such as cyclosporin A (CsA).14-16 Indeed in AML, Pgp overexpression is associated with a lower rate of complete remission and with inferior rates of disease-free and overall survival in patients receiving conventional anthracycline-containing induction and consolidation regimens.17-20 Southwest Oncology Group trial SWOG-9126 showed that adding CsA to a chemotherapy regimen containing infusional daunorubicin (DNR) significantly reduces induction resistance in patients with high-risk AML and prolongs the duration of remission and survival, implicating Pgp as an important cellular mechanism of resistance in myeloid leukemia.21 Pgp expression is demonstrable in myeloblasts in as many as 70% of patients with CML-BP.22-24 To determine whether the inhibition of Pgp export function improves survival in CML-BP, the SWOG conducted a randomized trial in adults (18-70 years old) with CML in previously untreated BP (SWOG-9032) comparing treatment with the same induction and consolidation regimen containing infusional DNR and high-dose cytarabine (HDAC) used in SWOG-9126, with or without the addition of CsA. Patients were required to have serum bilirubin levels lower than 1.5 times the institutional upper limit of normal, adequate cardiac function, and performance status of 0 to 2 by SWOG criteria.

Among 82 patients with previously untreated CML-BP registered between September 1992 and January 1998, 5 were ineligible because of prior treatment intended to restore chronic phase or to control leukocytosis 7 days or less before registration (3 patients), performance status 3 (1 patient), or unwillingness to accept randomization (1 patient). All 73 eligible patients with centrally reviewed pretreatment cytogenetics had the Philadelphia chromosome (Ph). Randomization between the HDAC/DNR (40 patients) and HDAC/DNR+CsA (37 patients) arms was stratified by age (younger than 55 vs 55 or older) and blast lineage (myeloid vs lymphoid vs undifferentiated). Blast expression of Pgp, multidrug resistance protein-1 (MRP1), the major vault protein lung resistance protein (LRP), and Pgp dye efflux capacity were assessed by flow cytometry as previously described.20 21 

Characteristics of the eligible patients were generally similar in the CsA and control arms, except that CsA patients had higher leukocyte and peripheral blast counts (Tables1-4). Expression of Pgp and LRP were slightly more common in the CsA arm, but CsA-inhibitable DiOC₂ efflux and expression of MRP1 were not. Among 75 eligible patients evaluable for induction toxicity, 8 (21%; 95% confidence interval [CI], 10%, 37%) of 38 controls died of induction toxicity compared with 7 (19%; 95% CI, 8%, 35%) of 37 CsA patients (one-tailed test, P = .48). Grade 4 hyperbilirubinemia was more frequent with CsA (P < .0001; 14 [38%] HDAC/DNR+CsA vs 1 [3%] HDAC/DNR).

Twelve controls (30%; 95% CI, 17%, 47%) achieved complete response (CR; 7 patients) or restored chronic phase (CP; 5 patients) compared with 3 CsA patients (8%; 95% CI, 2%, 22%; 2 CRs, 1 CP). Twenty-one controls (53%; 95% CI, 36%, 68%) had resistant disease compared with 25 (68%; 95% CI, 50%, 82%) CsA patients. All responses (CR or restored CP) were achieved with one course of induction therapy. Twelve of 15 responding patients (2 CsA, 10 controls) had cytogenetic follow-up, and 3 of these (2 with CR and 1 with restored CP, all controls) achieved cytogenetic CR.

Estimated median OS time was 5 months (95% CI, 4-7 months) in the control arm versus 3 months (95% CI, 2-5 months) in the CsA arm (one-tailed test, P = .77). Because the CsA-treated cohort had higher leukocyte and peripheral blast counts and higher Pgp expression, additional analyses were performed to assess whether such differences confounded the treatment comparison. In unstratified proportional hazards (PH) regression models, survival decreased with increasing WBC (2-tailed test, P = .074) or absolute blast count (P = .19), but it varied little in relation to level of Pgp expression (P = .47) or efflux (P = .14). However, in multivariate analysis, the absence of benefit of CsA was not explained by confounding effects of these factors (results not shown).

To determine whether pretreatment multidrug resistance phenotype has prognostic relevance in CML-BP, we examined the relation between blast expression of Pgp and other drug resistance proteins to treatment outcome in univariate analysis. With both treatment arms combined, response to treatment (CR, CR/CP, or RD) was significantly associated with Pgp expression and function (Table5), whereas neither MRP1 nor LRP expression adversely influenced induction outcome. As in the analysis of OS, the lack of benefit of CsA could not be explained by confounding effects of drug resistance protein expression or efflux.

This controlled trial was designed so that the survival comparison between treatment arms would have 91% statistical power if the true CsA:control mortality hazards ratio was 0.5 (one-sided test at critical level, 5%), and this level of power was achieved. Therefore, the results of this study provided compelling evidence that the addition of CsA to an HDAC/DNR-containing induction and consolidation regimen offers no survival benefit of practical importance over chemotherapy alone in patients with CML-BP. We cannot exclude the possibility that failure to achieve blood concentrations of CsA sufficient for Pgp blockade impacted the results of this trial. However, the analogous regimen studied in patients with high-risk AML (SWOG-9126) consistently yielded CsA blood concentrations exceeding 1500 ng/mL and significantly improved survival in Pgp-positive patients.21 Our data indicate that Pgp expression and function represent adverse prognostic variables impacting induction outcome in CML-BP, analogous to the data reported for AML. The negative results of this study do not exclude the potential benefit of more potent, so-called second-generation Pgp antagonists in CML-BP, but they raise the notion that Pgp expression reflects an alternative antigenic marker of a primitive blast phenotype that, in this disease, represents only one of many redundant cell defense mechanisms or survival signals. Indeed, recently characterized large genomic deletions on the derivative chromosome 9 adjacent to the t(9;22) breakpoint identify a subgroup of CML patients with particularly poor prognosis and inferior survival, independent of clinical prognostic variables.25 Strategies to inhibit Pgp function may merit further study in CML-BP when combined with the bcr-abl tyrosine kinase inhibitor, imatinib mesylate, and conventional cytotoxics.26 Imatinib mesylate interrupts the constitutive kinase-induced antiapoptotic signals to yield additive cytotoxicity or synergy with many antineoplastics.27Indeed, Pgp has been implicated as an important cellular mechanism contributing to in vitro resistance to imatinib mesylate. The farnesyl transferase inhibitor SCH66336, a downstream inhibitor of ras-mediated signaling by BCR-ABL, directly inhibits Pgp transport function with a potency comparable to that of CsA.28-30Inhibition of Pgp might accentuate cytotoxicity by augmenting imatinib mesylate and antineoplastic cellular accumulation and, as a consequence, potentially limit the molecular adaptation toBCR-ABL recognized with extended imatinib mesylate treatment.31