Cancer Therapy: Clinical

The Effect of Ketoconazole on the Pharmacokinetics and Pharmacodynamics of Ixabepilone: A First in Class Epothilone B Analogue in Late-Phase Clinical Development Sanjay Goel,1Marvin Cohen,4 S. Nilgu« n C¸ o« mezoglu,4 Lionel Perrin,5 Franc¸ois Andre¤ ,5 DavidJayabalan, 1Lisa Iacono,4 Adriana Comprelli,4 Van T. Ly,4 Donglu Zhang,4 Carrie Xu,4 W. Griffith Humphreys,4 Hayley McDaid,1, 2 Gary Goldberg,1, 3 Susan B. Horwitz,1, 2 andSridhar Mani 1

Abstract Purpose:To determine if ixabepilone is a substrate for cytochrome P450 3A4 (CYP3A4) and if its metabolism by this cytochrome is clinically important, we did a clinical drug interaction study in humans using ketoconazole as an inhibitor of CYP3A4. Experimental Design: Human microsomes were usedto determine the cytochrome P450 enzyme(s) involvedin the metabolism of ixabepilone. Computational docking (CYP3A4) studies were done for epothilone B and ixabepilone. A follow-up clinical study was done in patients with cancer to determine if 400 mg/d ketoconazole (inhibitor of CYP3A4) altered the pharmacokinet- ics, drug-target interactions, and pharmacodynamics of ixabepilone. Results: Molecular modeling and human microsomal studies predicted ixabepilone to be a good substrate for CYP3A4. In patients, ketoconazole coadministration resulted in a maximum ixabepi- lone dose administration to 25 mg/m2 when comparedwith single-agent therapy of 40 mg/m 2. Coadministration of ketoconazole with ixabepilone resulted in a 79% increase in AUC0-1.The relationship of microtubule bundle formation in peripheral blood mononuclear cells to plasma ixabepilone concentration was well described by the Hill equation. Microtubule bundle formation in peripheral bloodmononuclear cells correlatedwith neutropenia. Conclusions: Ixabepilone is a goodCYP3A4 substrate in vitro ; however, in humans, it is likely to be cleared by multiple mechanisms. Furthermore, our results provide evidence that there is a direct relationship between ixabepilone pharmacokinetics, neutrophil counts, and microtubule bundle formation in PBMCs. Strong inhibitors of CYP3A4 should be used cautiously in the context of ixabepilone dosing.

Ixabepilone (BMS-247550; Bristol-Myers Squibb; IXEMPRA) is a Authors’ Affiliations: 1Albert Einstein Cancer Center, 2Department of Molecular semisynthetic analogue of epothilone B that interacts with 3 Pharmacology, and Division of Gynecologic Oncology, Department of Obstetrics microtubules and induces them to form stable bundles in cells andGynecology, Albert Einstein College of Medicine, Bronx, New York; 4Bristol- Myers Squibb Research andDevelopment, Princeton, New Jersey; and 5CEA, (1–3). Epothilones have some similarities to taxanes in targeting iBiTec-URA 2096 du Centre National de la Recherche Scientifique, SB2SM, and stabilizing microtubules but also have important differences; CE-Saclay, Gif sur Yvette, France the class is structurally unrelated to taxanes, and unlike taxanes Received9/10/07; revised1/11/08; accepted1/13/08. and anthracyclines, has low susceptibility to multiple mecha- Grant support: DamonRunyonCancerResearchFoundation(CI:15-02; nisms of drug resistance (reviewed in ref. 1). Promising phase I/II S. Mani), DPI-Centre National de la Recherche Scientifique (contract number ST 00332/2004.PD005; F. Andre¤ ), andBristol-Myers Squibb. studies have led to a completion of a phase III study in breast The costs of publication of this article were defrayed in part by the payment of page cancer and approval by the United States Food and Drug charges. This article must therefore be hereby marked advertisement in accordance Administration for use in the treatment of breast cancer. with 18 U.S.C. Section 1734 solely to indicate this fact. We and others have shown that ixabepilone has moderately Note: Supplementary data for this article are available at Clinical Cancer Research variable interpatient pharmacokinetics due in part to variations Online (http://clincancerres.aacrjournals.org/). M. Cohen, S.N. C¸ o« mezoglu, L. Perrin, andF. Andre¤ contributedequally to this work. in drug metabolism (2, 4, 5). Because related epothilone Current address for D. Jayabalan: Cornell University, Weil Medical College, New analogues serve as cytochrome P450 3A4 (CYP3A4) substrates, Yo r k , N Y 10 0 21. it was hypothesized that ixabepilone would also undergo Requests for reprints: Sridhar Mani, Albert Einstein College of Medicine, Albert metabolism by liver cytochromes (e.g., CYP3A4; ref. 6). Einstein Cancer Center,1300 Morris Park Avenue, Chanin 302D-1,Bronx, NY10461. Phone: 718-430-2871; Fax: 718-904-2830; E-mail: [email protected]. To determine the effect of cytochrome P450s on ixabepilone F 2008 American Association for Cancer Research. biotransformation in vitro, a series of studies with liver doi:10.1158/1078-0432.CCR-07-4151 microsomes were done together with molecular dynamic

www.aacrjournals.org2701 Clin Cancer Res 2008;14(9) May 1, 2008 Cancer Therapy: Clinical simulations of CYP3A4 ixabepilone docking within the context human liver microsomes in the presence of NADPH for 15 min in a of data derived from published crystal structures of this enzyme shaking water bath before ixabepilone was added. After substrate (7–9). Finally, a phase Ib study was designed to assess the addition, the samples were incubated for 10 min at 37jC and 0.5 mL effects of ketoconazole (a strong CYP3A4 inhibitor) on of acetonitrile was added to each incubation mixture to stop the reaction. ixabepilone pharmacokinetics and pharmacodynamics (10). Then, the samples were centrifuged for 10 min at 3,000 rpm. The supernatants were then analyzed by LC-MS/MS to determine the concentration of ixabepilone (see Bioanalytical Methods). Immunoinhibition. Monoclonal anti-human CYP antibodies Materials and Methods (anti-1A2, anti-2C8, anti-2C9, anti-2C19, anti-2D6, and anti-3A4) were obtained from Dr. R. Krausz at NIH, Bethesda, MD (12). Amixture In vitro studies containing human liver microsomes, 0.1 mol/L potassium phosphate 14 Enzyme kinetics. Under linear conditions, [ C]ixabepilone buffer [containing 0.5 mmol/L of MgCl2 (pH 7.4)], and monoclonal was incubated in triplicate with pooled human liver microsomes anti-CYP antibody was preincubated for 10 min at 37jC. Then, (BD Biosciences; n = 20; at drug concentrations of 0.1, 0.3, 1, 2.5, 5, 10, NADPH and ixabepilone were added to the incubation mixture. The and 25 Amol/L) and expressed CYP3A4 (at drug concentrations of 0.1, 0.3, final mixtures consisted of 0.5 mg/mL human liver microsomes 1, 2.5, 5, 10, 25, 50, and 100 Amol/L) in the presence of NADPH (preincubated with monoclonal antibody), 0.5 Amol/L ixabepilone, (1 mmol/L). Incubations (0.5 mL) in 0.1 mmol/L phosphate buffer 1 mmol/L NADPH, and 0.1 mol/L phosphate buffer (pH 7.4) that were j containing 0.5 mmol/L of MgCl2 (pH 7.4) were carried out at 37jCfor5 incubated for 10 min at 37 C with shaking. The reaction was stopped or 10 min and stopped with ice-cold acetonitrile (1:1, v/v). The protein by addition of 0.5 mL of acetonitrile. The samples were centrifuged for concentration of human liver microsomes was 0.24 mg/mL 10 min at 3,000 rpm, and aliquots of the supernatant were analyzed by (for 0.1 and 0.3 Amol/L ixabepilone concentrations) or 0.5 mg/mL LC-MS (see Bioanalytical Methods). Positive control incubations were (for 1-25 Amol/L ixabepilone concentrations). The CYP3A4 concentration carried out in the absence of antibody solution as described above. An was 25 pmol/mL (for 0.1 and 0.3 Amol/L ixabepilone concentrations) incubation without NADPH and without antibody served as negative and 50 pmol/mL (for 1-100 Amol/L ixabepilone concentrations). For controls. In addition, another control incubation was carried out in the both human liver microsomes and CYP3A4, the incubation time was same manner described above but the incubation mixture contained 5 min for incubations at 0.1 to 2.5 Amol/L ixabepilone concentrations anti–egg white protein antibody to assess nonspecific reactions. and 10 min for 5 to 100 Amol/L ixabepilone concentrations. After Computational methods for substrate docking. Adetailed description quenching with acetonitrile (0.5 mL), the samples were centrifuged for of the molecular dynamic simulations done and metabolism site 5 min at 3,000 rpm. The supernatants were then analyzed by liquid predictions are presented in Supplementary Information. chromatography tandem mass spectrometry (LC-MS/MS) to determine the concentration of ixabepilone (see Bioanalytical Methods). Clinical ketoconazole interaction study

Km and Vmax values for ixabepilone metabolism in human liver Patient eligibility. Advanced cancer patients unresponsive to stan- microsomes and CYP3A4 were determined by fitting the rates of parent dard therapeutic interventions were enrolled in this open label phase I drug disappearance in the presence of ixabepilone to the Michaelis- study (10). Additional criteria included an Eastern Cooperative

Menten equation [V =(VmaxS)/(Km+S)] using a nonlinear regression Oncology Group performance score of 0 to 1 and no more than three analysis in Sigmaplot (version 8). prior chemotherapeutic regimens. The remaining criteria were identical Metabolite identification. Metabolites were identified by LC-MS/MS to previously published reports (1, 2, 4, 13–23). after analyzing samples from a [14C]ixabepilone (20 Amol/L) incuba- Study design, treatment administration, and dose escalation. This tion with pooled human liver microsomes. The results from these was a study administering ketoconazole, a nanomolar inhibitor of analyses showed that ixabepilone was mainly metabolized to oxidative CYP3A4, with ixabepilone. In cycle 1 (3 wk), ixabepilone was metabolites. In addition to parent drug (P), the identified metabolites coadministered in escalating doses per treatment cohort with a fixed included P+16, P+14, and P-2 metabolites (11). dose of ketoconazole. In cycle 2, ixabepilone was administered as a cDNA-expressed CYP enzymes. Human cDNA-expressed enzymes fixed dose (40 mg/m2) without administering ketoconazole. (BD Biosciences; CYP1A2, CYP2A6, CYP2C8, CYP2C9, CYP2C19, During cycle 1, patients received a 3-h i.v. infusion of ixabepilone on CYP2D6, and CYP3A4) were incubated with ixabepilone. The day 1 at a starting dose of 10 mg/m2. Ketoconazole (400 mg/d) was incubation mixtures, in duplicate, consisted of 100 pmol/mL of CYP given orally with a meal on day -1 (24 h before the infusion of enzymes, 0.5 Amol/L ixabepilone, 1 mmol/L NADPH, and 0.1 mol/L ixabepilone), on day 1 (2 h before the infusion of ixabepilone), and on phosphate buffer containing 0.5 mmol/L MgCl2 (pH 7.4). The final days 2 to 5 (six doses). The dose of ixabepilone in cycle 1 was to be volume of the incubation mixtures was 0.5 mL. After 10 min of increased to 20, 30, and 40 mg/m2 for subsequent cohorts of subjects incubation at 37jC in a shaking water bath, 0.5 mL of ice-cold based on safety evaluations. However, the dose escalation increment acetonitrile was added to each aliquot to stop the reaction. The samples could be <10 mg/m2 based on observed events in the preceding dose were centrifuged for 5 min at 3,000 rpm and the supernatants were level and the pharmacokinetic and pharmacodynamic data obtained at analyzed by LC-MS/MS to determine the concentration of ixabepilone prior dose levels. Based on a standard minimum enrollment design of (see Bioanalytical Methods). three subjects at each dose level, any dose level could be expanded to at Chemical inhibition. CYP inhibitors or antibodies were incubated, in least six patients based on the observation of one or more dose-limiting triplicate, with human liver microsomes and ixabepilone (0.5 Amol/L). toxicities (DLT). The maximum tolerated dose was exceeded when Human liver microsomes were incubated with 0.5 Amol/L of ixabepilone defined by the dose at which at least 33% of six patients had DLT(s). in the presence of the following CYP inhibitors: furafylline (CYP1A2, One dose level below this, the maximum tolerated dose was fully 10 Amol/L), tranylcypromine (CYP2A6, 2 Amol/L), sulfaphenazole explored in terms of toxicities observed in at least 6 patients. For each (CYP2C9, 10 Amol/L), benzylnirvanol (CYP2C19, 1 Amol/L), quinidine dose level at the maximum tolerated dose or below, additional subjects (CYP2D6, 1 Amol/L), ketoconazole (CYP3A4, 1 Amol/L), and montelu- could be enrolled to obtain additional safety data. kast (CYP2C8, 3 Amol/L). Each 0.5 mL incubation mixture contained During cycle 2 (3 wk), based on cycle 1 toxicities, patients received a 2 430 AL of 0.1 mol/L phosphate buffer [containing 0.5 mmol/L of MgCl2 3-h i.v. infusion with a maximum of 40 mg/m of ixabepilone on day 1. (pH 7.4)], 50 AL of 10 mmol/L NADPH solution, 2.5 AL of selective The toxicity-based dose reductions and DLT criteria for ixabepilone chemical inhibitor solution, 5 AL of 0.05 mmol/L ixabepilone solution, were followed as published in prior studies (2, 13). Similarly, and 12.5 AL of human liver microsomes (20 mg/mL). The time- prophylaxis regimens for ixabepilone were strictly adhered to as dependent inhibitor of CYP1A2, furafylline, was preincubated with previously published (2, 13).

Clin Cancer Res 2008;14(9) May 1, 2008 2702 www.aacrjournals.org Inhibition of Ketoconazole Metabolism

Fig. 1. A, concentration-dependent metabolism of ixabepilone in human liver microsomes. B, concentration-dependent metabolism of ixabepilone in human cDNA expressing CYP3A4 enzymes (see Materials andMethods, In vitro Studies, for details).

Pretreatment and posttreatment investigations. Standard physical, The mobile phase, composed of 10 mmol/L ammonium formate and objective antitumor response, and laboratory assessments were carried 0.1% formic acid in water (A) and acetonitrile (B), was delivered at a out as previously described (2, 13). In addition to standard protocol flow rate of 0.3 mL/min. The gradient was as follows: 95% A(0.2 min), tests, the [14C]erythromycin breath test was administered before to 5% Ain 2 min, 5% A(1 min), to 95% Ain 0.1 min. The high- protocol treatment to assess for hepatic CYP3Aactivity (Metsol). performance liquid chromatograph was interfaced to a Micromass Toxicity assessment. Because toxicity was the major end point of this Quattro Ultima LC-MS/MS tandem mass spectrometer (Micromass) clinical study, clinical and laboratory assessments were done weekly. In equipped with an electrospray interface operating in the positive certain patients, especially those with abnormal laboratory variables, ionization mode. Detection of each respective analyte was achieved repeat laboratory assessments were done twice within the same week through selected reaction monitoring. The transitions monitored were and continued as clinically indicated. as follows: m/z 507 to m/z 320 for ixabepilone and m/z 494 to m/z 306 Pharmacokinetic and pharmacodynamic samples. Blood samples for the internal standard. (4 mL per sample) were collected from an indwelling catheter or by Standard curves. Standard curves and quality control samples direct venipuncture using tubes containing K3EDTAas the anticoagu- defining the dynamic range of the bioanalytical method were prepared lant. Every reasonable attempt was made to obtain a dedicated in blank microsomes processed in the same manner as the test samples. peripheral catheter on the opposite arm or side of body from that The analysis of ixabepilone was conducted against an eight-point used for drug infusion. For the purposes of assessing ixabepilone blood standard curve ranging from 4 to 10,000 nmol/L. The standard curve concentrations, the collection times were predose, at 90 and 180 min was fitted with a linear regression weighted by reciprocal concentration from start of dosing, then at 0.25, 0.5, 1.0, 3.0, 5.0, 22, 45, 69, 93, and squared (1/x2). 117 h from the end of infusion. Within 1 h of collection, for each blood Statistical analysis. To assess the effect of ketoconazole on the sample, plasma was collected by centrifugation (10 min at 1,000 Â g pharmacokinetics of ixabepilone, two-way ANOVA was done on dose- at 4jC) and stored at -20jC. Peripheral blood mononuclear cells were normalized log(Cmax) and log(AUC0-1) for all patients who had values isolated at various time points (pre-dose and 3, 6, 24, and 48 h) after for cycles 1 and 2. The factors in the analysis were patient and cycle. the start of drug infusion. Because a 40 mg/m2 maximum tolerated dose from cycle 1 was not

Bioanalytical methods. Human EDTAplasma samples were ana- achieved, Cmax and AUC0-1 from cycle 1 were dose normalized to a lyzed for ixabepilone using LC-MS/MS method as previously validated 40 mg/m2 dose. Point estimates and 90% confidence intervals for and described for clinical use (2, 13). The standard curves were well means and differences between means on the log scale were fitted by a 1/x-weighted quadratic equation over the concentration exponentially factored to obtain estimates for geometric means and range of 2 to 500 ng/mL. The between-run variability and within-run ratios of geometric means on the original scale. The ratio of population variability for the analytical quality controls of ixabepilone were no geometric means of Cmax and AUC0-1 for ixabepilone given in more than 8.94% coefficient of variation, 16.62% coefficient of combination with ketoconazole (cycle 1) to ixabepilone given alone variation, respectively, with deviations from the nominal concentration (cycle 2), along with respective 90% confidence intervals, was reported. of no more than F2.89%. Samples from in vitro studies were analyzed as summarized below. Results Sample preparation. Aliquots (50 AL) of the biological matrix from in vitro studies were treated with 150 AL of acetonitrile containing In vitro studies. The consumption of ixabepilone by human A 0.5 mol/L of the internal standard BMS-212188, followed by vortex liver microsomes and human cDNA–expressed CYP3A4 can be mixing for 2 min. The supernatant was then separated from the characterized by monophasic Michaelis-Menten kinetics. The precipitated proteins after a 10-min centrifugation at 3,900 rpm and Km values for the oxidative metabolism of ixabepilone were transferred to autosampler vials or a 96-well plate. The injection volume A for LC-MS/MS analysis was 10 AL. 8.2 and 4.3 mol/L in human liver microsomes and CYP3A4, Instrumentation. Ahigh-performance liquid chromatography sys- respectively. Vmax values were 1,399 nmol/mg protein/min tem consisted of two Shimadzu LC-10ADvp pumps with a SCL-10Avp (f13 nmol/pmol CYP3A4 in human liver microsome/min) System Controller (Shimadzu) was used. The analytical column was an and 17.9 nmol/pmol CYP/min for human liver microsomes Atlantis dC18, 3 Am, 2.1 mm  50 mm (Waters Corporation) column. and CYP3A4, respectively (Fig. 1A and B).

www.aacrjournals.org2703 Clin Cancer Res 2008;14(9) May 1, 2008 Cancer Therapy: Clinical

Table 1. Oxidative metabolism of ixabepilone in human liver microsomes in the presence of CYP inhibitors (n =3)

Treatment/inhibitor Enzyme inhibited Ixabepilone remaining, nmol/L* (%c) Ixabepilone metabolized (%) % Inhibitionb Positive control — 88.6(35.5) 64.5 — x Heat killed — 249.5 (100 )0— x No NADPH — 234.7 (100 )0— Furafylline CYP1A2 117.4 (47.1) 52.9 17.9 Tranylcypromine CYP2A6111.9 (44.9) 55.1 14.5 Montelukast CYP2C8 135.6(54.4) 45.6 29.2 Sulfaphenazole CYP2C9 110.1 (44.1) 55.9 13.4 Benzylnirvanol CYP2C19 96.3 (38.6) 61.4 4.8 Quinidine CYP2D6100.7 (40.4) 59.67.5 Ketoconazole CYP3A4 234 (93.8) 6.2 90.4

*Average of triplicates. c% Remaining was calculated as follows: nmol of ixabepilone in the incubation mixture with inhibitor/nmol of ixabepilone in the negative control (heat killed) Â 100 (nanomole values were determined from LC/MS analysis). b% Inhibition = (% ixabepilone remaining - % ixabepilone in positive control) / (% ixabepilone in negative control - % ixabepilone in positive control). x The amount of ixabepilone in the negative control samples was designated as 100% as the peak area ratio of the parent drug to internal standard in the incubations was designated as 100%.

The major metabolic pathways have yet to be identified, but metabolism of ixabepilone was inhibited by f90%. The preliminary analysis shows that beyond its two degradants inhibitors of the other CYP enzymes (furafylline for CYP1A2, (a diol and an oxazine derivative), several P+16, P+14, and P-2 tranylcypromine for CYP2A6, montelukast for CYP2C8, sulfa- metabolites have been characterized (11). Using cDNA- phenazole for CYP2C9, benzylnirvanol for CYP2C19, and expressed CYP enzymes, ixabepilone was completely metabo- quinidine for CYP2D6) did not significantly inhibit the oxidative lized by CYP3A4 and remained almost unchanged (f90%) by metabolism of ixabepilone (inhibition in the range of 5-29%). other CYP enzymes (1A2, 2A6, 2C8, 2C9, 2C19, and 2D6). Consumption of ixabepilone in human liver microsomes When ixabepilone was incubated with human liver micro- was almost entirely inhibited by anti-CYP3A4 monoclonal somes at a clinically relevant concentration (clinical Cmax = antibodies (83%; Table 2). No significant inhibition was 0.6 Amol/L following a single 70 mg of i.v. administration; observed in the presence of other CYP inhibitory antibodies ref. 24) of 0.5 Amol/L, 64.5% of ixabepilone was consumed (anti-1A2, anti-2C8, anti-2C9, anti-2C19, and anti-2D6). under these conditions (Table 1). When ixabepilone was Computational methods for substrate docking. Detailed incubated with human liver microsomes in the presence of results are described in Supplementary Information. The ketoconazole (1 Amol/L), an inhibitor of CYP3A4, the oxidative dynamic docking of epothilone B to CYP3A4 active site shows

Table 2. Oxidative metabolism of ixabepilone in human liver microsomes in the presence of monoclonal antibodies (n =3)

Treatment Ixabepilone remaining, nmol/L* (%c) Ixabepilone metabolized (%) % Inhibitionb Positive control 110.7 (47.2) 52.8 — x No NADPH 234.7 (100 )0— Control-with antibody against hen egg whitek 93.8 (40.0) 60.0 NI{ With anti-CYP3A4 213.1 (90.8) 9.2 82.6 With anti-CYP1A2 87.1 (37.1) 62.9 NI{ With anti-CYP2C8 96.0 (40.9) 59.1 NI{ With anti-CYP2C9 93.6(39.9) 60.1 NI { With anti-CYP2C19 88.0 (37.5) 62.5 NI{ With anti-CYP2D699.0 (42.2) 57.8 NI {

Abbreviation: NI, no inhibition. * Average of triplicates. c% Remaining was calculated as follows: nmol ixabepilone in the incubation mixture with inhibitor / nmol ixabepilone in the negative control (heat killed) Â 100 (nanomole values were determined from LC/MS analysis). b% Inhibition = (% ixabepilone remaining - % ixabepilone in positive control) / (% ixabepilone in negative control - % ixabepilone in positive control). x The amount of ixabepilone in the negative control samples was designated as 100% as the peak area ratio of the parent drug to internal standard in the incubations was designated as 100%. k Control incubation contained a monoclonal antibody against hen egg white lysozyme to assess for nonspecific reactions. { % Ixabepilone remaining was less than or equal to the % ixabepilone in the positive control incubation sample.

Clin Cancer Res 2008;14(9) May 1, 2008 2704 www.aacrjournals.org Inhibition of Ketoconazole Metabolism

Table 3. Summary statistics for ixabepilone pharmacokinetic variables

Cycle Ixabepilone (mg/m2) Pharmacokinetic variables

C max (ng/mL) AUC0-1 (ng h/mL) T 1/2 (h) MRT (h) VSS (L) Clearance (L/h) 110(n = 4) 54.3 (14) 996.0 (22) 48.3 (26.3) 70.3 (31.3) 1,227.9 (382.8) 18.5 (4.9) 20 (n = 12) 97.1 (44) 1,694.1 (76) 66.0 (29.7) 81.6 (40.7) 1,689.3 (719.3) 25.7 (13.2) 25 (n = 7) 200.3 (61) 2,740.8 (52) 55.0 (11.4) 66.2 (13.7) 1,194.3 (670.0) 18.8 (12.8) 30 (n = 4) 246.9 (49) 3,144.6 (56) 93.9 (27.4) 100.7 (38.2) 1,920.7 (881.3) 20.5 (11.9) 240(n = 20) 206.1 (38) 1,848.8 (46) 53.2 (25.7) 51.9 (23.9) 2,171.3 (1,329.2) 48.9 (18.3)

NOTE: The Cmax and AUC0-1 values are geometric means (% coefficient of variation) and the remaining variables are means (FSD). The sample size (n) represents number of patients assessed.

that only one position is possible with high interaction energy genotype. Nine and three patients had one or two CYP3A5 (-81.0 kcal; Supplementary Fig. S2A). In contrast, for ixabepi- alleles that coded for active enzyme, respectively, based on a lone, two distinct final positions are possible (-90.0 and lack of either the *3 or *6 variant. Seven patients were -96.3 kcal; Supplementary Fig. S2B and S2C, respectively). The homozygous wild-type for CYP2D6 (see Supplementary Infor- latter positioning (Supplementary Fig. S2C) is characterized by mation for Methods). two hydrogen bond networks and the higher interaction All patients received ketoconazole administered by study energies are consistent with higher number of specific personnel on-site and after administration of ketoconazole; a interactions between substrate and the protein. mouth check was done to ensure that the subjects had swallowed the dose. Atotal of two patients (7%) missed at Clinical ketoconazole interaction study least 1 day of ketoconazole dosing in cycle 1. In addition, treatment compliance was monitored by drug accountability, Demographics and treatment compliance. Twenty-nine medical record, and case report forms. In terms of concomitant patients were enrolled into this study, of which 2 (7%) were medicines, there were no xenobiotics used that were known to never treated. The median age of enrollment was 62 years interfere with CYP3A4- or CYP3A5-directed metabolism. (range, 33-84 years) and 18 patients (67%) were female. Pharmacokinetics and interaction effect analysis. The expo- Overall, the Eastern Cooperative Oncology Group performance sure and clearance of ixabepilone was affected by coadmin- scores were either 0 (n =1)or1(n = 26). Twenty-four patients istration of ketoconazole (Tables 3 and 4). Specifically, (89%) had extensive (metastatic) disease and 26 (96%) had ixabepilone clearance decreased in 19 of 22 patients when undergone chemotherapy previously. The number of chemo- coadministered with ketoconazole. As shown in Table 4, therapy regimens used were two and three in 10 (37%) and 13 coadministration of ketoconazole with ixabepilone resulted in (48%) patients, respectively. The most common diagnosis was an increase of f7% in ixabepilone Cmax. However, the 90% gynecologic malignancy (carcinoma of ovaries, cervix, endo- confidence interval indicates that this effect is not statistically metrium) followed by prostatic adenocarcinoma. Liver metas- different from the null hypothesis (i.e., no difference in tases were observed in 2 patients (f7%) and the median effect). Coadministration of ketoconazole with ixabepilone (range) pretreatment bilirubin (mg/dL), alanine aminotrans- resulted in a 79% increase in AUC0-1. The 90% confidence ferase (IU/L), and albumin (g/dL) levels were 22 (14-43), 19 interval indicates that this effect is statistically different from (8-60), and 3.8 (2.8-4.3), respectively. The main reason for the null hypothesis. The mean plasma ixabepilone concentra- discontinuation of treatment was disease progression/relapse tion versus time for all patients by dose is shown in (12 patients, 44%). Supplementary Fig. S4. The erythromycin breath test, a measure of hepatic CYP3A The clearance of ixabepilone did not correlate with CYP3A4 activity, was administered to subjects before treatment for activity as measured by the erythromycin breath test (Supple- exploratory evaluation as a marker of the rate of clearance of mentary Fig. S5). Furthermore, the mean clearance of eight ixabepilone. The percent of [14C]ixabepilone metabolized per subjects with at least one CYP3A5 allele that coded for active hour ranged from 0.96 to 4.72, a 4.9-fold range. Genomic DNA enzyme did not differ from that of 10 subjects without active from 24 subjects was analyzed for the CYP3A5 and CYP2D6 enzyme (data not shown).

Table 4. Statistics for ixabepilone dose normalized Cmax and AUC0-1

Phamacokinetic variable (n = 22) Adjusted geometricmeans Ratio of geometricmeans Treatment Adjusted Ratio Point estimate geometricmeans (90% confidence limit)

C max (ng/mL) Cycle 1 231.49 Cycle 1 vs cycle 2 1.067 (0.931, 1.223) Cycle 2 216.91

AUC0-1 (ng h/mL) Cycle 1 3,457.91 Cycle 1 vs cycle 2 1.791 (1.547, 2.074) Cycle 2 1,930.43

www.aacrjournals.org2705 Clin Cancer Res 2008;14(9) May 1, 2008 Cancer Therapy: Clinical

Table 5. Percent of PBMCs with microtubule bundles observed in cycle 1 and 2 end-of-infusion

Cycle 1 (with ketoconazole), mean % (FSD) Cycle 2 (without ketoconazole), mean % (FSD) 10 mg/m2 (n = 4) 22.5 (F9.4) 20 mg/m2 (n = 8) 36.6 (F9.5)* 25 mg/m2 (n = 6) 38.7 (F6.2) 30 mg/m2 (n = 4) 56.5 (F4.1) 40 mg/m2 (n = 18) 40.7 (F8.6)*

*P = 0.29.

Drug target effect analysis. The formation of microtubule to that observed in baseline values from our previous study bundles caused by ixabepilone in PBMCs is a plasma (n = 49) and for baseline values in cycle 2 in this study, concentration–dependent effect. For cycle 1 (ixabepilone and suggesting that ketoconazole itself had no influence on tubulin ketoconazole) and cycle 2 (ixabepilone alone), the percent of bundle formation in PBMCs (13). In addition, ixabepilone- PBMCs with microtubule bundles was greatest just before the mediated microtubule bundle formation is not inhibited by end of infusion at 3 hours, remained above baseline at 6 and ketoconazole in HepG2 cells (Supplementary Fig. S1). 24 hours, and returned to approximately baseline at 48 hours Based on these findings, variable estimates using all the data (Supplementary Fig. S6). The percent of PBMCs with microtu- from cycles 1 and 2 together showed that the best model bule bundles also increased in the presence of ketoconazole remained the Hill equation with E0 fixed at zero. The variable 2 and drug (Table 5). At 25 mg/m (n = 6) of ixabepilone, the % estimates were similar to that described above—Emax (%) was (SE) bundle formation was 38.7 (F6.2) compared with that at 46.4%, EC50 (ng/mL) was 35.2, and sigmoidicity factor was 40 mg/m2 (n = 17) of ixabepilone in the absence of 1.18 (figure not shown). These data suggest that ketoconazole ketoconazole (40.7 F 8.6%). The relationship of microtubule increases ixabepilone-mediated microtubule bundle formation bundle formation to plasma ixabepilone concentration after in PBMCs by increasing the concentration of ixabepilone. ixabepilone alone was well described by the Hill equation with Clinical adverse events. Table 6 lists the cycle 1 DLT events effect at zero concentration (E0) fixed to zero. The fitted following administration of ixabepilone and ketoconazole. variable values for the model are presented in Fig. 2A. The Hill During the administration of ketoconazole alone, there were no equation with E0 fixed at zero was the best model compared grade 2 or greater toxic events observed. Most patients with the Hill equation with E0 as a model variable; the Emax complained of grade 1 abdominal discomfort (bloating/nausea) model with E0 fixed to zero; the Emax model with E0 as a model for the duration of this treatment period. The initial dose level variable; the linear model or the log-linear model, based on the of ixabepilone administered for cycle 1 was 10 mg/m2 and objective function, the Akaike criteria, and the Schwartz criteria. subsequent cohorts of 20 and 30 mg/m2 were opened. Based on The plot relationship of microtubule bundle formation to observed DLTs and available pharmacokinetic and phar- plasma ixabepilone concentration (for patients dosed at macodynamic data, the 30 mg/m2 cohort was closed and the 40 mg/m2 only) is shown in Fig. 2A. The percent baseline 20 mg/m2 cohort was expanded to 6 patients. On further tubulin bundles in cycle 1 was 0.8%, which was nearly identical safety data obtained at 20 mg/m2, an intermediate dose level

Fig. 2. A, relationship of percent PBMCs with microtubule bundles and ixabepilone plasma concentration. B, relationship between percent PBMCs with microtubule bundles andnadirabsolute neutrophil count ( ANC).The line describes a negative linear fit equation with r2 = 0.32 andcurvedlines represent 95% confidence intervals.

Clin Cancer Res 2008;14(9) May 1, 2008 2706 www.aacrjournals.org Inhibition of Ketoconazole Metabolism

Table 6. Cycle 1 DLTs

Cycle 1 dose No. patients No. DLTs DLT event 10 mg/m2 40 20 mg/m2 12 2 G3 fatigue and G4 neutropenia; G3 mucositis 25 mg/m2 7 2 G4 neutropenia and G3 diarrhea; febrile neutropenia 30 g/m2 4 2 G3 mucositis; G3 fatigue, febrile neutropenia of 25 mg/m2 was opened. Because 2 of 4 (50%) patients at describing a relationship between both variables and gives 30 mg/m2 had DLTs, and 2 of 7 (29%) at 25 mg/m2 had DLTs, variable estimates (e.g., slope at absolute neutrophil count the latter dose was defined as the maximum tolerated dose by values <2,000) that are similar to those obtained using the protocol criteria. The overall treatment-related adverse event exponential decay model (Fig. 2B). Upon analyzing the data by profile of ixabepilone administered concomitantly with keto- neutropenic grade, there is a significant increase in maximum conazole in cycle 1 was similar to that of ixabepilone percent of PBMCs with microtubule bundles in patients with administered alone beyond cycle 1. The most common grade 2 or greater neutropenia compared with those with grade treatment-related adverse events in cycle 1 across all dose 1 or less toxicity (P = 0.002) and similarly for those with grade cohorts were fatigue (n = 19 patients, 70%) and nausea 0 to 2 neutropenia compared with those experiencing grade 3 (n = 13, 48%). The most common treatment-related events to 4 neutropenia (includes febrile neutropenia) [mean (SE), 58 beyond cycle 1 were fatigue (n = 17 patients, 74%) and nausea (14) versus 38 (10), respectively, P < 0.02] (figure not shown). (n = 12, 52%). Unexpected side effects occurred in two patients, Antitumor response. Twenty-three patients (85%) had at both of whom developed acute renal failure while receiving the least one target tumor lesion for assessment. Four patients study medications (see Supplementary Information for details). (15%) had no measurable or evaluable target tumor lesions. Peripheral neuropathy is a dose-limiting side effect after Investigator assessment of tumor response indicated that chronic dosing with ixabepilone. The incidence of treatment- 1 patient (4%) had a complete response (ovarian adenocarci- related sensory peripheral neuropathy was consistent with noma with abdominal metastases) based on serial positron previously published studies (reviewed in ref. 1). Peripheral emission tomography scans; 12 (44%) had stable disease; 10 neuropathy was primarily sensory and grade 1 (n = 6, 22%) or (37%) had progressive disease; and response status was grade 2 (n = 7, 26%). Overall, only 2 patients (7%) reported undetermined in 4 (15%) patients. grade 3 neuropathy and there were no drug discontinuations due to peripheral neuropathy. Laboratory adverse events. Overall, the most frequent side Discussion effect in cycle 1 was hematologic and included neutropenia The human microsomal studies clearly show that CYP3A4 (n = 5; 19%), anemia (n = 4; 15%), thrombocytopenia plays a major role in the oxidative metabolism of ixabepilone at (n = 4, 15%), and febrile neutropenia (n = 2; 7%), whereas concentrations of 0.1 to 25 Amol/L in native human liver the most common hematologic abnormality beyond cycle 1 microsomes. The molecular modeling studies further validated was neutropenia (n = 7; 30%), thrombocytopenia (n = 2; 9%), these observations and showed the utility of various theoretical and anemia (n = 1; 4%). Grade 4 neutropenia was reported in 3 models in predicting CYP3A4 biotransformation. The clinical patients (11%) in cycle 1 and in one patient (4%) beyond cycle study validates the in vitro findings that inhibition of CYP3A4 1. Grade 3 neutropenia were reported in 2 patients (9%) by ketoconazole can significantly increase the exposure to the beyond cycle 1. Grade 3 thrombocytopenia and grade 3 febrile parent compound, ixabepilone, suggesting that inhibition of its neutropenia was reported in 2 patients (7%) each in cycle 1. metabolism directly alters its presence in blood. However, Grade 3 anemia was reported in 1 patient (4%) beyond cycle 1. in vitro experiments cannot predict the extent of drug Similarly, liver function tests (alkaline phosphatase, aspartate metabolism, as this is highly dependent on fraction metabo- aminotransferase, alanine aminotransferase, and total biliru- lized (f ) by a given pathway. Unfortunately, due to bin) during the ixabepilone administration phase remained m degradation, the human Absorption, Distribution, Metabolism grade 0 or 1 in severity. There was only one grade 2 alkaline and Excretion study of ixabepilone did not allow an accurate phosphatase elevation and grade 3 increase in creatinine measurement of f .6 There might be other mechanisms for reported in one patient (4%). There was no grade 4 serum m ixabepilone clearance, such as chemical degradation, which abnormality reported during this study and there was no dose- may not be accounted for in in vitro studies. neutropenia relationship. Ixabepilone-associated neutropenia The modeling studies predicted a fairly tight docking into the was not dependent on age, baseline absolute neutrophil count CYP3A4 catalytic domain and based on docking alone it could value, Eastern Cooperative Oncology Group performance score, be predicted that ixabepilone would likely be a substrate of or prior chemotherapy (two-way ANOVA, P > 0.2). CYP3A4 (compared with epothilone B). Because CYP3A4 is Microtubule bundles in PBMCs as a function of toxicity. As known to have homotropic and heterotropic substrate inter- published previously, we showed a relationship between actions, the docking with ixabepilone suggested that with this percent PBMCs with microtubule bundles at the end of compound, such interactions are not likely to occur because the infusion and absolute neutrophil count nadir (r2 = 0.44, one substrates are large enough to occupy the major part of the active phase exponential decay or r2 = 0.32, linear relationship). Although the mono-exponential decay model gives a slightly 2 better R value, the linear model is the simplest model 6 S.N. C¸ o« mezoglu, unpublishedobservations.

www.aacrjournals.org2707 Clin Cancer Res 2008;14(9) May 1, 2008 Cancer Therapy: Clinical site (volume of epothilone B has been estimated to be 709 A˚ 3). model of neutrophil dynamics, in which the inhibition of However, a cooperative effect involving an external allosteric site neutrophil progenitor production was related to the plasma cannot be excluded (7, 25). Furthermore, competitive effects are concentration of ixabepilone by an inhibitory hyperbolic expected with ketoconazole as validated for other compounds. function. Together, this suggests that there is a direct relationship Finally, the interaction energies show that ixabepilone is likely between ixabepilone pharmacokinetics, neutrophil counts, and to be a better substrate for CYP3A4 than epothilone B. The value microtubule bundles in PBMCs. Hence, a count of microtubule of this novel docking strategy is that other clinically effective bundles in PBMCs serve as a surrogate for ixabepilone toxicity. epothilone derivatives can be modeled in a similar manner. Third, comedications (e.g., dexamethasone) routinely used in The observation that ixabepilone is a good substrate cancer patients can serve as potent activators of the human for CYP3A4 catalysis in human liver microsomes alone pregnane X receptor. Pregnane X receptor regulates CYP3A4 at cannot predict in vivo clearance (25–29). There are examples the level of transcription, which could alter drug metabolism (e.g., sorafenib) where the in vitro studies predict that certain by inducing its expression (32). Ketoconazole can attenuate enzyme(s) metabolize drug(s) with high affinity, yet this effect transcriptional regulation of CYP3A4 across the patient popu- is not observed clinically (30). Degradation of ixabepilone and lation (32–34). Furthermore, pregnane X receptor can also the metabolites of degradants might also play a role in in vivo regulate tumor drug metabolism and resistance and ketocona- clearance. This underscores the importance of translating all zole may inhibit this process also (35–41). Together, coadmin- observations seen in vitro to humans. istration of ketoconazole may be advantageous in protecting The clinical study with ketoconazole and ixabepilone clearly against a pregnane X receptor–mediated increased clearance of indicate that inhibition of CYP3A4 by ketoconazole signifi- ixabepilone that may indeed lead to subtherapeutic drug levels cantly increases the exposure of ixabepilone in the blood and and decreased antitumor effects. Supporting these findings are alters drug-target effects. These effects are reflected by increased the observation that ketoconazole can augment cytotoxicity of toxicity when both drugs are coadministered. The implications drugs (e.g., nocodazole) in cancer cells (40, 42–44). of these findings are broad. First, although ketoconazole is one The clinical study was designed as both a dose escalation of the most potent inhibitors of CYP3A4, other clinically safety and tolerability assessment and as an evaluation of important inhibitors of CYP3A4 that are commonly used in the pharmacokinetics of ixabepilone when administered with cancer patients would be predicted to have significant drug a strong CYP3A4 inhibitor such as ketoconazole. A dose of interactions with ixabepilone (8). Second, inhibition of 25 mg/m2 of ixabepilone with ketoconazole was tolerated by CYP3A4 by ketoconazole increases ixabepilone exposure in five of seven patients, whereas 40 mg/m2 of ixabepilone is the the blood, which directly correlates with increased formation of recommended dose as a single agent. This result is consistent microtubule bundles in PBMCs. The latter is inversely with the observed 79% increase in exposure. Based on these correlated with nadir absolute neutrophil counts. These results results, a reduced dose of 20 to 25 mg/m2 of ixabepilone might validate our previous findings in patients undergoing treatment be safely administered to a patient who must have continued with single-agent ixabepilone (2, 4). The implications are that therapy with a strong inhibitor of CYP3A4. by using our technically straightforward assay (approximately <8 hours performance time) to visually quantitate tubulin Disclosure of Potential Conflicts of Interest bundles, one can predict both neutropenia and blood levels of ixabepilone. In our study, there was no clear relationship No potential conflicts of interest were disclosed. between blood concentrations of ixabepilone and neutrophil counts. However, based on a model describing exposure- Acknowledgments response analyses of ixabepilone-induced neutropenia in We thank Drs. Merrill Egorin, University of Pittsburgh, PA, andAmit Roy, Bristol- patients with breast cancer (31), the absolute neutrophil Myers Squibb, for helpful andinsightful discussions; andDr. Mimi Y. Kim, from the count-time profile data were adequately described by the Department of Epidemiology and Biostatistics of Albert Einstein College of Medi- semimechanistic, nonlinear, mixed effects (‘‘population’’) cine, for statistical support andadvise on statistical methodsusedin the manuscript.

References 1. Mani S, Macapinlac M, Jr., Goel S, et al. The clinical tumors andlymphomas. J Clin Oncol 2007;25: 10. Goel S, Goldberg G, Iacono L, et al. Effect of keto- development of new mitotic inhibitors that stabilize 10 82 ^ 8. conazole on the pharmacokinetics andpharmacody- the microtubule. Anticancer Drugs 2004;15:553 ^ 8. 6. Amrita VK, Ming C, Francis YL,Yueping Z, Punit HM. namics of ixabepilone. Proc Am Soc Clin Oncol 2. Mani S, McDaidH, Hamilton A, et al. Phase I clinical Preclinical pharmacokinetics andoral bioavailability of 2006;24:435. andpharmacokinetic studyof BMS-247550, a novel BMS-310705, a novel epothilone B analog. Cancer 11. Comezoglu SN, Ly VT, Everett D, et al. Metabolism derivative of epothilone B, in solid tumors. Clin Cancer Chemother Pharmacol 2005;V56:145. of 14C-ixabepilone in mouse, rat, dog, monkey and Res 2004;10:1289 ^ 98. 7. Williams PA, Cosme J, Vinkovic DM, et al. Crystal human liver microsomes. Drug Metabol Rev 2006; 3. McDaidHM,ManiS,ShenHJ,MuggiaF,Sonnichsen structures of human cytochrome P450 3A4 boundto 38:173. D, Horwitz SB.Validation of the pharmacodynamics metyrapone andprogesterone. Science 2004;305: 12. Gelboin HV, Krausz K. Monoclonal antibodies and of BMS-247550, an analogue of epothilone B, during 683^6. multifunctional cytochrome P450: drug metabolism a phase I clinical study. Clin Cancer Res 2002;8: 8. Marechal JD,YuJ, Brown S, et al. In silico and in vitro as paradigm. J Clin Pharmacol 2006;46:353 ^ 72. 2035^43. screening for inhibition of cytochrome P450 CYP3A4 13 . Mani S, McDaidHM, Grossman A, et al. Peripheral 4. Zhuang SH, Agrawal M, Edgerly M, et al. A Phase I by comedications commonly used by patients with bloodmononuclear andtumor cell pharmacodynam- clinical trial of ixabepilone (BMS-247550), an epothi- cancer. Drug Metab Dispos 2006;34:534 ^ 8. ics of the novel epothilone B analogue, ixabepilone. lone B analog, administered intravenously on a daily 9. Zhou D, Afzelius L, Grimm SW, Andersson TB, Ann Oncol 2007;18:190 ^ 5. schedule for 3 days. Cancer 2005;103:1932^ 8. Zauhar RJ, Zamora I. Comparison of methods for 14. Denduluri N, Lee JJ,Walshe J, et al. Phase II trial of 5. Aghajanian C, Burris HA III, Jones S, et al. Phase I the prediction of the metabolic sites for CYP3A4- ixabepilone, an epothilone B analog, given daily for study of the novel epothilone analog ixabepilone mediated metabolic reactions. Drug Metab Dispos three days every three weeks, in metastatic breast (BMS-247550) in patients with advanced solid 2006;34:976^83. cancer. Invest New Drugs 2007;25:3421 ^ 7.

Clin Cancer Res 2008;14(9) May 1, 2008 2708 www.aacrjournals.org Inhibition of Ketoconazole Metabolism

15. WhiteheadRP, McCoy S, Rivkin SE, et al. A Phase II mass balance study of the novel anticancer agent ixa- androstane receptor and the pregnane X receptor. trial of epothilone B analogue BMS-247550 (NSC bepilone using accelerator mass spectrometry. Invest Mol Pharmacol 2006;70:329^39. #710428) ixabepilone, in patients with advanced New Drugs 2007;25:327 ^ 34. 35. Mensah-Osman EJ,Thomas DG,Tabb MM, et al. Ex- pancreas cancer: a Southwest Oncology Group study. 25. Isin EM, Guengerich FP. Kinetics andthermody- pression levels andactivation of a PXR variant are di- Invest New Drugs 2006;24:515^ 20. namics of ligandbinding by cytochrome P450 3A4. rectly relatedto drug resistance in osteosarcoma cell 16. Ajani JA, Safran H, Bokemeyer C, et al. A multi- J Biol Chem 2006;281:9127 ^ 36. lines. Cancer 2007;109:957 ^ 65. center phase II study of BMS-247550 (Ixabepilone) 26. Cresteil T, Monsarrat B, Dubois J, Sonnier M, 36. Masuyama H, Suwaki N,Tateishi Y, Nakatsukasa H, by two schedules in patients with metastatic gastric Alvinerie P, Gueritte F. Regioselective metabolism SegawaT, HiramatsuY.The pregnane X receptor regu- adenocarcinoma previously treated with a taxane. of taxoids by human CYP3A4 and 2C8: structure- lates gene expression in a ligand- and promoter- Invest New Drugs 2006;24:441 ^ 6. activity relationship. Drug Metab Dispos 2002;30: selective fashion. Mol Endocrinol 2005;19:1170 ^ 80. 17. HussainM,TangenCM,LaraPN,Jr.,etal.Ixabepi- 438^45. 37. Mani S, Huang H, Sundarababu S, et al. Activation of lone (epothilone B analogue BMS-247550) is active 27. Bun SS, CiccoliniJ, Bun H, Aubert C, CatalinJ. Drug the steroidandxenobiotic receptor (human pregnane in chemotherapy-naive patients with hormone- interactions of paclitaxel metabolism in human liver X receptor) by nontaxane microtubule-stabilizing refractory prostate cancer: a Southwest Oncology microsomes. J Chemother 2003;15:266 ^ 74. agents. Clin Cancer Res 2005;11:6359 ^ 69. Group trial S0111. J Clin Oncol 2005;23:8724 ^ 9. 28. Jamis-Dow CA, Klecker RW, Katki AG, Collins JM. 38.Takami N, Sakamoto H,YamamotoT. Steroidandxe- 18. Gadgeel SM,Wozniak A, Boinpally RR, et al. Phase I Metabolism of taxol by human andrat liver in vitro :a nobiotic receptor (SXR) is a key system for the acqui- clinical trial of BMS-247550, a derivative of epothi- screen for drug interactions and interspecies differen- sition of cisplatin resistance in endometrial cancer lone B, using acceleratedtit ration 2B design. Clin ces. Cancer Chemother Pharmacol 1995;36:107 ^ 14. cells. J Int MedRes 2003;31:59 ^ 68. Cancer Res 2005;11:6233 ^9. 29. Jamis-Dow CA, Pearl ML, Watkins PB, Blake DS, 39. Giuliano KA. High-content profiling of drug-drug 19. Low JA, Wedam SB, Lee JJ, et al. Phase II clinical Klecker RW, Collins JM. Predicting drug interactions interactions: cellular targets involvedin the modulation trial of ixabepilone (BMS-247550), an epothilone B in vivo from experiments in vitro.Humanstudieswith of microtubule drug action by the antifungal ketoco- analog, in metastatic andlocally advancedbreast paclitaxel andketoconazole. Am J Clin Oncol 1997; nazole. J Biomol Screen 2003;8:125^ 35. cancer.JClinOncol2005;23:2726^34. 20:592 ^ 9. 40.WangYJ,Yu CF, Chen LC, et al. Ketoconazole poten- 20. Galsky MD, Small EJ, Oh WK, et al. Multi-institu- 30. Lathia C, Lettieri J, Cihon F, Gallentine M, Radtke tiates terfenadine-induced apoptosis in human Hep tional randomized phase II trial of the epothilone B an- M, Sundaresan P. Lack of effect of ketoconazole- G2 cells through inhibition of cytochrome p450 3A4 alog ixabepilone (BMS-247550) with or without mediated CYP3A inhibition on sorafenib clinical phar- activity. J Cell Biochem 2002;87:147 ^ 59. estramustine phosphate in patients with progressive macokinetics. Cancer Chemother Pharmacol 2006; 41. Martinez C, Garcia-Martin E, Pizarro RM, Garcia- castrate metastatic prostate cancer. J Clin Oncol 57:685 ^ 92. Gamito FJ, Agundez JA. Expression of paclitaxel- 2005;23:1439^46. 31. Zhu H RA, Cohen M, Pfister M. Exposure-response inactivating CYP3A activity in human colorectal 21. Smaletz O, Galsky M, Scher HI, et al. Pilot study of analysis for ixabepilone associatedneutropenia. Clin cancer: implications for drug therapy. Br J Cancer epothilone B analog (BMS-247550) andestramus- Pharmacol Ther 2007;81:S121 ^ 8. 2002;87:681 ^ 6. tine phosphate in patients with progressive metastatic 32. Huang H, Wang H, Sinz M, et al. Inhibition of drug 42. Chen RJ, Lee WS, Liang YC, et al. Ketoconazole prostate cancer following castration. Ann Oncol metabolism by blocking the activation of nuclear induces G0/G1arrest in human colorectal andhepato- 2003;14:1518^24. receptors by ketoconazole. Oncogene 2007;26: cellular carcinoma cell lines. Toxicol Appl Pharmacol 22. Abraham J, Agrawal M, Bakke S, et al. Phase I 258 ^ 68. 2000;169:132^ 41. trial andpharmacokinetic studyof BMS-247550, 33. Takeshita A,Taguchi M, Koibuchi N, Ozawa Y. Puta- 43. Wang YJ, Jeng JH, Chen RJ, et al. Ketoconazole an epothilone B analog, administered intravenously tive role of the orphan nuclear receptor SXR (steroid potentiates the antitumor effects of nocodazole: on a daily schedule for five days. J Clin Oncol andxenobiotic receptor) in the mechanism of in vivo therapy for human tumor xenografts in nude 2003;21:1866^73. CYP3A4 inhibition by xenobiotics. J Biol Chem mice. Mol Carcinog 2002;34:199 ^ 210. 23. Lee FY, Borzilleri R, FairchildCR, et al. BMS- 2002;277:32453 ^ 8. 44. HoYS,TsaiPW,YuCF,LiuHL,ChenRJ,LinJK. 247550: a novel epothilone analog with a mode of 34. Duret C, Daujat-Chavanieu M, Pascussi JM, et al. Ketoconazole-induced apoptosis through P53- action similar to paclitaxel but possessing superior Ketoconazole andmiconazole are antagonists of the dependent pathway in human colorectal and hepato- antitumor efficacy. Clin Cancer Res 2001;7:1429^ 37. human glucocorticoidreceptor: consequences on cellular carcinoma cell lines. J Cell Biochem 2002;87: 24. Beumer JH, Garner RC, Cohen MB, et al. Human the expression andfunction of the constitutive 147 ^ 5 9.

www.aacrjournals.org2709 Clin Cancer Res 2008;14(9) May 1, 2008