A Phase I Trial of a Potent P-Glycoprotein Inhibitor, Zosuquidar Trihydrochloride (LY335979), Administered Intravenously in Combination with Doxorubicin in Patients with Advanced Malignancy

Since the initial description of cancer cells exhibiting multidrug resistance to cytotoxic agents, and the subsequent identification and cloning of the human gene that confers this phenotype, multiple drug resistance 1 (MDR1), much effort has focused on the development of clinically useful compounds capable of modulating this effect (1, 2). Several agents have undergone substantial testing in the clinical trial setting. The results to-date are at best inconclusive and in some cases, disappointing (3). However, few of the first or second-generation MDR-modulating agents have had the requisite potency or selectivity toward the target to adequately assess questions of efficacy in the clinical trial setting.

Zosuquidar trihydrochloride (LY335979) is a third generation modulating agent that was developed specifically as a selective, MDR1/P-glycoprotein (P-gp) inhibitor (4). In vitro concentrations of zosuquidar from 50 to 100 nm are capable of circumventing P-gp-mediated drug resistance in virtually every cell culture system tested (5, 6). In experiments using tumor-bearing mice with syngeneic and human xenografts, zosuquidar was efficacious in restoring drug sensitivity to P-gp expressing implants (5). Moreover, preclinical studies using murine and canine models demonstrated that zosuquidar had no observed effect on the pharmacokinetic profile of coadministered P-gp substrates, including doxorubicin and paclitaxel (5). These data suggest that zosuquidar warrants investigation in clinical trials aimed at reversing drug resistance mediated by P-gp.

We initiated a Phase I trial designed to determine whether pharmacologically effective plasma concentrations of zosuquidar could be safely achieved in cancer patients who received the drug i.v. The study was also designed to evaluate the effects of zosuquidar on doxorubicin pharmacokinetics and toxicity. The results indicate that biologically effective plasma concentrations of zosuquidar are associated with minimal toxicity and without significant alteration of doxorubicin pharmacokinetics.

Patients who were at least 18 years of age and met all of the following criteria were eligible for the study: (a) histological or cytological diagnosis of metastatic or locally advanced cancer (must have failed conventional therapy, have disease considered refractory to standard chemotherapy regimens, or have disease for which no standard chemotherapy was available); (b) prior radiation therapy and chemotherapy completed at least 3 weeks before study enrollment (6 weeks if prior treatment was nitrosourea or mitomycin C); (c) lifetime cumulative anthracycline dose must have not exceeded the doxorubicin equivalent of 300 mg/m²; (d) resting blood pool scan with a cardiac ejection fraction of >45%; (e) performance status of 0 to 2 on the Eastern Cooperative Oncology Group scale; (f) estimated life expectancy of at least 12 weeks; (g) and adequate organ function (granulocytes ≥ 1.5 × 10⁹ cells/liter, platelets ≥ 100 × 10⁹/liter, hemoglobin ≥ 9 g/liter, bilirubin ≤ upper limit of normal, alanine transaminase and aspartate transaminase ≤ 2.5 times normal, serum creatinine ≤ 1.5 mg/dl). Written informed consent was obtained from all patients according to institutional, state, and federal guidelines. This study was approved by the Institutional Review Board at the Indiana University School of Medicine.

Patients were entered in the study in cohorts of three and received treatment during multiple cycles of either 35 days (cycle 1) or 21 days (cycle 2 and subsequent cycles; Table 1). The initial zosuquidar dose was chosen based on studies in the dog; a 2-week i.v. infusion demonstrated that a dose of 10 mg/kg/day produced peak plasma concentrations that exceeded 1000 nm and was without toxicity (7). The starting dose of zosuquidar was 20 mg/m²/day. It was administered as a continuous i.v. infusion via central venous access over 48 h beginning on day 1 of every cycle. A 48-h infusion was chosen to balance the anticipated half-life of zosuquidar in man, the duration of effective P-gp inhibition, and the known pharmacokinetic parameters of doxorubicin. In the first two cohorts, this same dose of zosuquidar was administered and the doxorubicin dose increased. For each subsequent cohort, the zosuquidar dose was escalated by a maximum of 100%. Dose escalation was to cease in any of the following circumstances: (a) the plasma concentration of zosuquidar reached or exceeded a predefined upper limit (defined as 2000 nm or approximately 1000 μg/liter); (b) toxicity of any type (Common Toxicity Criteria version 1 grade 2 or higher except nausea, vomiting, or alopecia) attributed to zosuquidar alone was observed in cycle 1, or (3) unacceptable toxicity of zosuquidar in combination with doxorubicin occurred in cycle 2. There was no intrapatient dose escalation of zosuquidar.

In cycle 1 only, doxorubicin was administered 14 days after zosuquidar. For cycle 2 and subsequent cycles, doxorubicin was administered on day 2; which was 24 h after the start of the zosuquidar infusion. The first cohort received 45 mg/m² of doxorubicin administered i.v. over 30 min. Subsequent cohorts received 60 mg/m² of doxorubicin administered i.v. over 30 min. If the target plasma level of zosuquidar was reached without defining maximum tolerated dose, doxorubicin was dose escalated to 75 mg/m² in an additional cohort. This last cohort was to be expanded to a total of 10 patients, provided that dose-limiting toxicity had not been reached.

The maximum tolerated dose was defined as one dose level lower than the dose level at which at least 2 of 6 patients experienced unacceptable toxicity. If unacceptable toxicity occurred in 1 of the first 3 patients at a given dose-level, 3 additional patients were treated at that dose level. If no other patient experienced unacceptable toxicity, dose escalation continued. If unacceptable toxicity occurred in 2 of 6 patients, dose escalation was stopped. Unacceptable toxicity for zosuquidar given in combination with doxorubicin (determined in cycle 2) was defined as Common Toxicity Criteria grade 4 hematological toxicity lasting for 5 days or longer or grade 3 or higher nonhematologic toxicity (excluding alopecia, nausea, or vomiting).

Doxorubicin doses were adjusted for neutrophil and platelet nadirs occurring during the preceding course of therapy, as previously described (8). The doxorubicin dose was also modified if patients experienced Common Toxicity Criteria grade 2 or higher neurological toxicities or hyperbilirubinemia. The doxorubicin administration was discontinued if there was clinical congestive heart failure or if the left ventricular ejection fraction fell, on the gated blood pool scan, to less than 45% or if the total decrease was ≥10% from baseline.

Antitumor response was evaluated, where appropriate, after the third cycle of therapy and after every other subsequent cycle. Responses were quantified by either physical examination or by appropriate imaging studies according to World Health Organization criteria (9).

The data for hematological parameters were analyzed using a linear-mixed-effects model with cycle as a fixed effect and subject as a random effect. This technique accounted for any incomplete data and repeated measures on each patient. The dose level of zosuquidar and the baseline value were included as covariates. Observations corresponding to doxorubicin dose reductions were excluded. The raw data were log-transformed before analysis because the data on this scale better satisfied the normality assumption underlying the analysis (10).

Least square means (LS means) and 95% confidence intervals were generated from the model for each cycle. Ratios of the LS means and their 95% confidence intervals were derived to compare means between cycles. A 95% confidence interval that excludes 1 is equivalent to statistical significance at the 5% level. The percentage decrease in blood counts was obtained using the following formula: (1-ratio LS means) × 100.

Plasma samples were obtained predose and serially up to 96 h after dose to determine concentrations of zosuquidar, doxorubicin, and the metabolite doxorubicinol when the drugs were administered separately (cycle 1) and in combination (cycle 2).

Zosuquidar and doxorubicin and doxorubicinol concentrations were analyzed using validated high-performance liquid chromatography methods which have been previously described (8).

Plasma pharmacokinetic parameters of zosuquidar, doxorubicin, and doxorubicinol were evaluated using noncompartmental methods (WinNonlin Professional version 2.1). Plasma area under the curve (AUC), clearance (CL), volume of distribution at steady state (V_ss), and terminal half-life (t_1/2) were calculated for doxorubicin and plasma AUC and C_max (observed maximum plasma concentration) for doxorubicinol as follows:

where t_n is the last time point where the plasma concentration is above the limit of quantification, C(t_n)′ is the prediction for the concentration at the last quantifiable time point, and λ_z is the calculated terminal rate constant;

with T the infusion time and

where AUMC is area under the moment curve.

For zosuquidar, when it was not possible to estimate the terminal half-life, t_1/2α was assessed instead. Only C_endinf, t_1/2α and λ_zwere reported for these subjects. C_endinf was determined directly from the observed concentration-time data. It is defined as the concentration observed at the end of the infusion scheduled to occur at 48 h for zosuquidar and 0.5 h for doxorubicin.

A surrogate assay of P-gp function in patients was used that has been previously described using peripheral blood natural killer lymphocytes that express P-gp (6, 8).

The pharmacokinetic/pharmacodynamic modeling of zosuquidar plasma concentrations versus % inhibition of rhodamine 123 efflux (relative to spike) were carried out using WinNonlin (professional version 2.1).

A total of 41 patients were entered into the study, one of whom died before receiving study-directed therapy and is therefore excluded. Forty patients received zosuquidar and doxorubicin in nine cohorts as defined by the protocol and outlined in Table 1. Patients in cohorts 1 and 2 and cohorts 7 and 8 received the same zosuquidar dose (20 mg/m²/day and 640 mg/m²/day, respectively), in each case this was to allow the doxorubicin dose to be increased in a step-wise manner. Because preliminary pharmacokinetic analysis of zosuquidar suggested that some patients who received 640 mg/m²/day may have reached or exceeded plasma levels of 2000 nm, cohort 9 (the final cohort) received 480 mg/m²/day.

The demographics of the 40 patients are summarized in Table 2. Patients were enrolled with a variety of tumor types, the most common being soft tissue sarcoma (15) and renal cell carcinoma (6). Twenty-one patients had prior chemotherapy, six had prior doxorubicin.

The noncompartmental pharmacokinetic parameters of zosuquidar are shown by cohort in Table 3. Pharmacokinetic parameters could not be calculated for cohorts 1–3 because most values were below the limit of quantification. For subsequent cohorts mean C_max and AUC values were generally proportional to dose with the possible exception of cohorts 7 and 8 (both 640 mg/m²/day) in which there appears to be a greater than proportional increase. However, zosuquidar plasma clearance was shown to be independent of administered dose (Fig. 1).

The arithmetic means and SDs for selected pharmacokinetic parameters of doxorubicin and doxorubicinol both in the presence and absence of zosuquidar are shown in Table 4. Across the study as a whole, in the presence of zosuquidar, there was a 17% decrease in CL and 25% increase in AUC of doxorubicin (60 mg/m2) and a similar change (22% and 15% in CL and in AUC, respectively) with the higher dose of doxorubicin (75 mg/m²; Table 5). However, doxorubicin CL decreased with increasing zosuquidar dose in an exponential manner (Fig. 2, upper panel). Thus, when comparing the doxorubicin CL in cycle 1 to that in cycle 2, deviation from the line of identity was expected for the higher doses where the impact on CL was most marked. This trend was observed when the total dose of zosuquidar administered over the 48 h period was ≥500 mg (Fig. 2, lower panel).

Doxorubicin is metabolized to the primary metabolite doxorubicinol. Similarly, in the presence of zosuquidar there was an increase in AUC and C_max for doxorubicinol (32% and 12% for CL and 26% and 30% for AUC after administration of 60 mg/m² and 75 mg/m², respectively; see Table 5).

Doxorubicin did not have any effect on the pharmacokinetic parameters calculated for zosuquidar (Table 6). In addition, the interindividual variability in both CL and V_ss were similar for cycles 1 and 2.

The effect of zosuquidar on P-gp function was determined using natural killer (CD56⁺) cells collected at predetermined times in an ex vivo assay (11, 12). A direct reversible concentration-effect relationship was observed (Fig. 3). Increasing concentrations of zosuquidar resulted in greater inhibition of P-gp function until maximum inhibition was attained at concentration in excess of 200 μg/liter. Because zosuquidar demonstrated linear pharmacokinetics, this concentration corresponded to an absolute dose of 500 mg. Hence, as the majority of patients received this dose or greater, maximum inhibition of function was readily achieved.

The most common toxicities associated with zosuquidar alone were reversible neurological symptoms. The most common was tremor (eight patients) occurring only at doses of ≥480 mg/m²/day. An associated symptom of dizziness was reported for five patients and again mainly at higher doses. None of these episodes were severe (Common Toxicity Criteria grade 1 or 2 only). Symptoms typically became apparent on the second day of dosing and all resolved within a few hours after the conclusion of the infusion.

Overall there were 23 episodes of grade 3/4 neutropenia in 16 patients and 2 of grade 3/4 thrombocytopenia (both in the same patient) in cycles 1 and 2. Eight occurred after doxorubicin alone resulting in a dose reduction in four patients, and 17 occurred in cycle 2 leading to 13 dose reductions. All patients with grade 3/4 neutropenia in cycle 1 experienced the same toxicity in cycle 2. A total of six patients required hospitalization for neutropenic fever, three of the patients after cycle 1, and three patients after cycle 2. Three of the episodes occurred in patients who had received a doxorubicin dose of 75 mg/m². The overall incidence of neutropenic fever was 4.9% (6 of 122 cycles of doxorubicin or doxorubicin/zosuquidar). All patients received antibiotics, and all recovered without sequelae.

The analysis of the hematological toxicity and assessment of the impact of zosuquidar is complex because doxorubicin was given repeatedly and, in some cases, the dose of both agents was increased between cohorts. The nadir blood counts both in the presence and absence of zosuquidar are shown in Table 7 and Fig. 4. Overall, it was estimated that, on average, WBC decreased by 32% (95% CI, 21–42%) between cycle 1 and 2; absolute neutrophil count by 61% (95% CI, 42–74%) and platelets by 23% (95% CI, 8–35%) as derived from the ratio of LS means. However, as described above, the decrease in doxorubicin CL was most marked when the total dose of zosuquidar was ≥500 mg corresponding to cohort 5 (group B). In this group there is an average decrease in absolute neutrophil count of 84% (95% CI, 72–91%) compared with 61% (95% CI, 42–74%) overall and only 4% (95% CI, −68% to 45%) when the dose of zosuquidar was <500 mg (group A). Similar trends are shown for WBC and platelet count. In group A, there is no difference between cycle 1 and 2; thus the differences seen in the patients overall is heavily influenced by group B. This latter group makes up the majority of the patient population that received the highest doses of doxorubicin. The impact of the higher dose of doxorubicin in group B is shown by the LS means for cycle 1 where doxorubicin was given alone.

As per protocol specification, six patients were discontinued because of decreased cardiac function as assessed by gated blood-pool imaging studies (Table 8). Patients experienced ejection fraction decreases of 10% to 18% from baseline. Their baseline levels ranged from 60% to 74%, and their levels at discontinuation ranged from 42% to 62%. Lifetime exposures for doxorubicin were 120–360 mg/m². There were no instances of congestive heart failure or other symptomatic cardiac disease. Of note, five of the six patients had ejection fractions above 50% at the time of discontinuation from the study.

Formal evaluation of efficacy was not the intent of this Phase I study; however, 32 patients were evaluated for efficacy. Two patients had a partial response to therapy, one as assessed by bi-dimensional tumor measurements, the other by a serum tumor marker, Prostate-specific antigen decreased by >50%. The latter patient remained on trial for 12 cycles. An additional 12 patients had stable disease as their best response, with six of the patients having stable disease documented for at least 5 cycles.

The patient with a measurable partial response was a male with metastatic breast cancer. He received 480 mg/m²/day zosuquidar for six cycles before being discontinued having achieved maximal clinical response. His doxorubicin dose was initially 75 mg/m², but it was decreased twice because of leukopenia. Before entering the study, this patient had received prior therapy with docetaxel and estramustine.

Effective pharmacological inhibition of P-gp function in the clinical setting has previously been achieved at the expense of the delivered dose of the coadministered cytotoxic agent, mostly as a consequence of pharmacokinetic interactions (6, 13, 14). It has been difficult to evaluate the contribution of pharmacodynamic effects as this has been masked by the larger effect of the pharmacokinetic interactions. Ideally, if pharmacokinetic interactions could be minimized, full-dose antitumor therapy could be administered with an MDR modulator and thereby isolate and more adequately assess the effect of P-gp inhibition on clinical outcomes (15). The development of highly selective, potent inhibitors of P-gp, such as zosuquidar, may help achieve this goal.

This Phase I study demonstrated that zosuquidar is well tolerated, exhibiting minimal toxicity, at concentrations in excess of those required to maximally inhibit P-gp function. The only toxicity attributed to i.v. zosuquidar as a single agent is reversible grade 2 neurotoxicity, primarily cerebellar-related tremor. Whether this toxicity occurred as a result of P-gp inhibition at the blood-brain barrier could not be assessed directly in this trial (16). However, when zosuquidar was administered orally, a more frequent and severe (grade 3/4) cerebellar toxicity was evident at lower plasma concentrations where maximal P-gp inhibition may not have been achieved (8). These data suggest a mechanism other than P-gp inhibition may contribute to the observed neurotoxicity.

Analysis of the pharmacokinetic data demonstrated that the primary effect of zosuquidar on doxorubicin appeared to be a decrease in clearance. This most likely represented inhibition of P-gp in bile canaliculi impeding biliary excretion. This was further supported because there was no decrease in the AUC of doxorubicinol, as would be expected if decreased metabolism had accounted for the change in doxorubicin clearance. In fact, AUC was found to increase in the presence of zosuquidar, which again was consistent with the inhibition of canalicular P-gp.

Trials with other MDR modulators, such as valspodar, when combined with doxorubicin have resulted in larger increases in AUC and C_max than have been demonstrated with zosuquidar (13). This effect may be attributed, in part, to the specificity of zosuquidar for P-gp compared with other ATP-binding cassette transporters. Because doxorubicin is transported by both P-gp and multidrug resistance protein 2, the impact of a less selective modulator may be significant. Additionally, the impact on doxorubicinol AUC is influenced markedly by the period of P-gp inhibition. In the studies with valspodar, the period of P-gp inhibition after doxorubicin infusion was substantially longer (approximately 96 h) than in this study.

The pharmacokinetic data also revealed an unexpected decrease in the V_ss of doxorubicin in the presence of zosuquidar. The mechanisms responsible for this effect are unknown. Recent publications have shown decreases in V_ss of P-gp substrates such as talinol and docetaxel by other P-gp inhibitors (Verapamil and R101933, respectively), suggesting the effect may be directly related to P-gp inhibition (17, 18).

The clearance of zosuquidar was independent of BSA, hence dosing using this paradigm is not beneficial. In this study, an absolute dose of ≥500 mg administered over 48 h was associated with both a greater impact on doxorubicin clearance and maximal P-gp inhibition in NK cells. This dose is predicted to give peak plasma concentrations ≥200 μg/liter, the threshold for maximum P-gp inhibition.

Despite the modest changes in the pharmacokinetics of the doxorubicin, there were no clinical sequelae observed in this Phase I trial sufficient to define a dose-limiting toxicity, at the maximal delivered doses of zosuquidar (480–640 mg/m²) and doxorubicin (75 mg/m²/day). However, there was a statistically significant decrease in absolute neutrophil count in the presence of zosuquidar. This decrease was more evident at doses resulting in maximal P-gp inhibition. Because there is a minor decrease in the elimination of doxorubicin and importantly doxorubicinol, which is pharmacologically active, these cannot be ascribed solely to a pharmacodynamic effect on bone marrow stem cells.

Another important effect that potentially would limit the ability to administer a full course of treatment is cardiotoxicity (19). From this study there is no evidence to suggest that cardiac function is adversely affected from the combination although the sample size is small. Further analysis of this potential toxicity would necessitate longer treatment durations in a larger patient number in a randomized setting.

In summary, zosuquidar can be administered safely at doses compatible with maximal inhibition of P-gp function. Some pharmacokinetic interaction was observed that may be explained by inhibition of P-gp in the bile canaliculi reducing elimination. This resulted in an exacerbation of hematological toxicity as determined from nadir blood counts. The combination warrants further exploration in Phase II trials.