Phase I Clinical Trial of BMS-247550, A Derivative of Epothilone B, Using Accelerated Titration 2B Design

Among the many new cytotoxic agents introduced over the last two decades, the taxanes have shown significant antitumor activity in many tumor types. Despite the impressive clinical benefits observed, the overall antitumor activity of taxanes still is limited, both due to the lack of activity in certain tumor types and development of resistance. The development of drugs that can improve upon the activity and circumvent resistance of taxanes is ongoing.

Epothilones A and B belong to a new class of nontaxane tubulin polymerization agents obtained by fermentation of the myxobacteria Sorangium cellulosum (1). These agents show impressive in vitro antitumor activity including activity in taxane-resistant cell lines (2). Despite this impressive in vitro activity, studies with these agents in in vivo models have revealed only modest activity. This has been attributed to their metabolic instability, unfavorable pharmacokinetics, and narrow therapeutic window (3).

BMS-247550 is a semisynthetic derivative of epothilone B developed to overcome the metabolic instability and the narrow therapeutic window of the natural product. This analogue is metabolically more stable partly due to the substitution of the lactone with a lactam, which reduces the product's metabolism by carboxylestrase. The mechanism of cytotoxicity of BMS-247550 is related to the stabilization of the microtubules, as observed with taxanes, resulting in mitotic arrest. In vitro and in vivo data has shown impressive antitumor activity with BMS-247550 in many cancer types and cell lines (4). BMS-247550 has also shown activity in tumor models that are taxane resistant, either through multidrug resistance or through tubulin mutations. The activity shown in MDR cell lines suggests that BMS-247550 is not a substrate of the p-glycoprotein (5, 6).

In animal toxicology studies, BMS-247550 was well tolerated, with the main toxicities being gastrointestinal, peripheral neuropathy, and myelotoxicity. The severely toxic dose 10% (STD₁₀) in rats was 74 mg/m² and a dose of 10 mg/m² was well tolerated in rats and dogs; therefore, a dose of 7.4 mg/m² (10% of the STD₁₀ in rats) was considered an appropriate starting dose for phase I studies. Preclinical data also suggested that the efficacy of the drug would be better with less frequent dosing schedules (4). Based on these impressive preclinical efficacy and toxicity studies, we conducted a phase I trial, using the accelerated titration “2B” design (7), of BMS-247550 given every 3 weeks over 1 hour. This schedule was also chosen because it complimented the schedules being evaluated in other studies, daily ×5 days every 3 weeks and weekly schedule.

The objectives of this study were to determine the maximum tolerated dose (MTD) of BMS-247550 and the recommended phase II dose for patients with advanced solid tumors when given once every 21 days, to determine the toxicity profile of this drug, to study the pharmacokinetics of BMS-247550 and correlate the plasma levels with the toxicity.

Patients with histologically or cytologically confirmed nonhematologic cancer refractory to conventional therapy or for which there was no known effective therapy were eligible. Other eligibility criteria included age ≥18 years, Eastern Cooperative Oncology Group performance status of ≤2, and no chemotherapy or radiation therapy for ≥4 weeks before initiation of therapy (6 weeks for mitomycin C or nitrosourea). Prior taxane therapy was allowed. Patients had to have adequate bone marrow function [absolute neutrophil count (ANC) ≥ 2,000/mm³, platelets >100,000/mm³], adequate hepatic function (total bilirubin ≤ 1.5 mg/dl and aspartate aminotransferase and alanine aminotransferase ≤2.5 times the institutional upper limit of normal), and adequate renal function (creatinine ≤1.5 mg/d). All patients with child-bearing potential were required to use effective birth control and a negative pregnancy test 72 hours before start of study medication in premenopausal females. Patients were excluded for preexisting neuropathy (Common Toxicity Criteria, CTC grade >1), documented hypersensitivity reaction to prior paclitaxel or other therapy containing Cremophor-EL, history of brain metastases, a serious uncontrolled medical disorder or active infection, HIV-positive status, and pregnant or lactating females. The study was approved by the Institutional Review Board of Wayne State University, and all patients provided a signed informed consent as per institutional guidelines.

Pretreatment evaluation. A complete history, physical examination, and laboratory studies including a complete blood count with differential and platelet count, prothrombin time, partial thromboplastin time, fibrinogen, bilirubin, alkaline phosphatase, electrolytes, magnesium, calcium, phosphorous, creatinine, alanine aminotransferase, aspartate aminotransferase, lactate dehydrogenase, and glucose were done within 2 weeks before treatment. Chest X-ray, EKG, and radiographic studies for tumor measurements were required within 4 weeks before treatment.

Drug preparation and administration. BMS-247550 was produced by Bristol-Myers Squibb Pharmaceutical Research Institute and supplied by National Cancer Institute/Division of Cancer Treatment and Diagnosis as a three-vial system (two vials of diluent and one of BMS-247550) for the 20-mg vial potency. The diluent consisted of ethanol plus polyoxyethylated castor oil (Cremophor-EL) mixture (1:1 by volume). Before constitution of the drug, the vehicle was placed at room temperature for 1 hour. Eleven milliliters of the vehicle were injected into the BMS-247550 vial, and after gently mixing the contents, a concentration of 2 mg/mL was achieved. This solution was further diluted with Lactated Ringer's solution in a non-PVC container for a final concentration of 0.1 to 0.6 mg/mL.

The starting dose was 7.4 mg/m² (10% of the rat STD₁₀), repeated every 21 days infused i.v. over 1 hour. All patients received prophylaxis for hypersensitivity reactions, related to Cremphor, consisting of dexamethasone 20 mg orally at 12 and 6 hours before therapy and diphenhydramine 50 mg i.v. and ranitidine 50 mg i.v. 30 minutes before therapy. Heart rate and blood pressure were monitored every 15 minutes during the infusion.

Study design. This phase I study was designed according to the accelerated titration “2B” design described by Simon et al. (7). This design has an initial accelerated phase followed by a standard escalation phase. During the accelerated phase, initial cohorts contained one patient until the first instance of first-course CTC grade ≥2 toxicity (except for nausea, vomiting, fatigue, anorexia, anemia, and alopecia). With the occurrence of grade ≥2 toxicity, two additional patients were to be treated at that dose. If these two patients did not develop CTC grade ≥2 toxicity, the accelerated phase of the study with one patient cohorts continued. During the accelerated phase, all patients were observed for at least 3 weeks before dose escalation. The dose escalations in this phase were in increments of 40%. Toxicity in patients for this study was graded according to CTC version 2.0.

With the second occurrence of first-course CTC grade ≥2 or the first instance of first-course dose-limiting toxicity (DLT), the accelerated phase was terminated and a standard escalation phase with 33% dose increases was initiated with three patient cohorts.

DLT was defined as grade ≥3 nonhematologic toxicity (except for nausea, vomiting, and alopecia), grade ≥3 nausea/vomiting despite adequate antiemetic therapy, grade 4 neutropenia lasting ≥5 days or febrile neutropenia (i.e., one fever > 38.5°C with ANC < 1,000 cells per mm³), platelet count ≤ 25,000 cells per mm³ or thrombocytopenia associated with a bleeding episode requiring platelet transfusion, or treatment delay of >2 weeks due to delayed recovery from drug-related toxicity. If a patient experienced a first-course DLT, a total of six patients were to be treated at that dose level. If two or more patients developed first-course DLT, that dose was to be considered the maximum given dose. The dose just below the maximum given dose was the MTD, provided less than two of six treated patients experienced first-course DLT.

Continuation of therapy, dose adjustment, and response assessment. Patients underwent weekly assessment of blood counts, serum chemistry, and toxicity assessment. Tumor measurements were conducted after every two cycles of BMS-247550. Patients were required to have an ANC of ≥1,500/mm³, platelets of ≥100,000/mm³, adequate performance status, and hepatic and renal function as per the eligibility criteria to receive further therapy with BMS-247550. Additionally, all toxicities related to BMS-247550 should have recovered to baseline or CTC grade ≤1 (except alopecia). If a longer than 3-week delay was required for recovery from toxicity, the patient was removed from study. Patients with DLT could be retreated at the next lower dose if the toxicity improved to grade ≤1 with a ≤ 2-week treatment delay. Intrapatient dose escalation was permitted if the patient had developed CTC grade <2 toxicity in the prior cycle and if at least one patient had completed one cycle at the subsequent dose level. Patients were required to receive at least two courses to be considered eligible for response assessment. Response Evaluation Criteria in Solid Tumors were used for response assessment.

Pharmacokinetic studies. Blood samples were collected during the first two cycles to assess the pharmacokinetics of BMS-247550, using liquid chromatography tandem mass spectrometry methodology. Pharmacokinetic samples were collected at pretreatment, 30 minutes, 1 (end of infusion), 1.15, 1.30, 1.45, 2.00, 3.00, 4.00, 6.00, 8.00, 24.00, 48.00, and 72.00 hours. Within 1 hour of collection, plasma was separated from blood by centrifugation at 1,000 × g for 15 minutes at 0°C to 5°C. Plasma samples were stored at −20°C until analysis. Concentrations of BMS-247550 were measured by a validated assay. After addition of internal standard (BMS-212188) to 0.2 mL of each study sample, calibration standard, and quality control sample, the samples were precipitated with acetone. The supernatant was further extracted with 1-chlorobutane. The organic layer was removed and evaporated to dryness. The residue was reconstituted and injected into a YMC ODS-AQ column (4.6 × 50 mm; YMC, Wilmington, NC). Chromatographic separation was achieved isocratically at a flow rate of 0.3 mL/min with detection by electrospray tandem mass spectrometry using a Quattro LC mass spectrometer (Micromass, Beverly, NJ). The mobile phase contained acetonitrile/0.01 mol/L ammonium acetate (pH 5.0, 65:35). The retention times were 2.9 minutes for BMS-247550 and 4.2 minutes for the internal standard. The standard curve, which ranged from 2 to 500 ng/mL, was fitted to a 1/x weighed quadratic regression model. The within-run and between-run coefficient of variation of quality control samples for BMS-247550 were within 15% and 10%, respectively. Deviations of the predicted concentrations from nominal values for quality control samples were within 9%. Different pharmacokinetic variables of BMS-247550 were computed employing a noncompartmental method using WinNonlin 4.0.1 software.

Statistical methods. Nonparametric statistical methods of analysis were used throughout. The relationship of BMS-247550 dose level to drug clearance (and other continuous variables) was assessed using the Jonckheere-Terpstra k-sample rank sum test for ordered alternatives (8). That test is more sensitive (i.e., powerful) for specific patterns of association (e.g. trends) than a general test for any departure from equality. For this test, a one-sided significance level was used, because we had a directional hypothesis. We hypothesized that as dose level increased, then clearance and ANC would decrease and half-life would increase. Because the sample sizes were quite small, these analyses were conducted using exact inference methods based on permutation testing (9). The relationship between dose level and clinically relevant neutropenia (ANC <1.0, yes/no) was examined via Fisher's exact test. Spearman's rank correlation was used to assess the relationship between the percent change in ANC from pretreatment to nadir level, and the length of time that the BMS-247550 plasma concentration was >15 ng/mL.

Eighteen patients (10 men and 8 women) were registered onto the study. The median age of patients was 54.5 years (43-80 years). Six patients had melanoma, four patients had colon cancer, and two patients had sarcoma with one each of other tumor types. One patient with ovarian cancer was never treated with the drug due to significant clinical deterioration soon after registration. A total of 45 (range, 1-9) cycles of BMS-247550 were given to 17 patients (Table 1).

Toxicity and response. The first cycle toxicities are presented in Table 2 and the overall toxicities are presented in Table 3. Seven dose levels, ranging from 7.4 to 56 mg/m², were evaluated. One patient cohorts occurred with the first four dose levels. Due to grade 3 thrombocytopenia that did not satisfy the definition of DLT during the first cycle in one patient at the 28.6 mg/m² (fifth) dose level, this cohort was expanded and a total of four patients were treated at this dose. Because no additional patients developed grade ≥2 toxicity, the accelerated phase was continued.

At the 56 mg/m² dose, two of three patients experienced first-course dose-limiting myelotoxicity, terminating the accelerated escalation phase. One patient had prolonged (>5 days) grade 4 neutropenia and the other patient had neutropenic sepsis. Neutrophil counts in both patients recovered to baseline. Both patients refused further therapy with the drug. The third patient treated at 56 mg/m² received a total of three cycles at this dose and discontinued further therapy due to grade 3 fatigue and grade 2 myalgias.

A total of six patients were accrued to the next lower dose level of 40 mg/m². Only one patient developed DLT (grade 4 neutropenia during the first cycle); therefore, this dose was declared the MTD. Of the four patients who started at 40 or 56 mg/m² and received at least three cycles, all four developed neutropenia with subsequent cycles and three patients required dose reductions. No grade 3 and/or 4 myelosuppression was observed below 28.6 mg/m². None of the patients on the trial received growth factors to support neutrophil counts.

Among the nonhematologic toxicities, fatigue was a major adverse effect at dose levels of ≥28.6 mg/m² (Table 3). Toxicities not listed in Table 3 included diarrhea (three patients, one grade 4), hand foot syndrome (two patients), rash (three patients), and mucositis (one patient). Grade 1 to 2 neuropathy was observed at the 28.6 and 40 mg/m² dose levels. Of the five patients who received more than two cycles of BMS-247550 at ≥28.6 mg/m² dose, three developed evidence of grade 1 and/or 2 neuropathy.

Three patients (one lung, one melanoma, and one sarcoma) had tumor shrinkage; however, none met the partial response criteria. All these patients received a dose of ≥28.6 mg/m². Of these three patients, one patient (lung) received five cycles and one patient (sarcoma) received nine cycles of therapy. Both patients discontinued therapy due to neuropathy. Four patients (two colon, one melanoma, and one renal) had stable disease as their best response.

Pharmacokinetics. The time course of mean plasma concentrations of BMS- 247550 following 1-hour infusion at different doses in cycle 1 is shown in Fig. 1. The mean values of various pharmacokinetic variables are summarized in Table 4. Pharmacokinetic data of one patient in cycle 1 treated at 40 mg/m² were excluded from analysis as the plasma concentration-time profile was erratic due to blockade of infusion line and nonadherence to the infusion schedule. In the majority of patients, the plasma concentrations exhibited a triexponential decline following infusion. However, in some patients, the plasma concentrations fell in a biexponential fashion. A linear relationship was observed between dose and mean peak plasma concentration (r² = 0.92) or AUC (r² = 0.88).

For the 16 patients in whom the pharmacokinetic data were analyzed, a modest trend toward slightly increasing mean drug clearance with increasing dose level was observed (data not shown). However, this was not statistically significant (P = 0.1567 via the exact Jonckheere-Terpstra test, one sided). Half-life of BMS-247550 by dose level showed a modest trend toward increasing mean drug half-life with increasing dose level, but this was not statistically significant (P = 0.5000 via the exact Jonckheere-Terpstra test, one sided; data not shown). Thus, neither drug clearance nor elimination half-life was dose dependent. Only four patients were treated at the same dose in cycles 1 and 2: one patient at 28.6 mg/m², one patient at 40.0 mg/m², and two patients at 56.0 mg/m². These data are too sparse to yield meaningful results regarding differences in pharmacokinetic variables (e.g., C_max) between cycles 1 and 2.

To better understand the occurrence of neutropenia, associations of cycle 1 dose and plasma levels of BMS-247550 with neutropenia were evaluated. Mean (or median) % change in ANC at nadir increased in a strongly negative fashion (i.e., neutropenia increased) with increasing dose (P < 0.0001 via the exact Jonckheere-Terpstra test, one sided; Table 5). The occurrence rates of clinically relevant neutropenia (ANC <1.0, yes/no) at dose levels of 28.6, 40, and 56 mg/m² were one of four, 25%; one of six, 17%; and three of three, 100%, respectively. This relationship was not quite monotonic and was not statistically significant (P = 0.1867 by Fisher's exact test).

Neutropenia within 7 to 10 days followed by an increase in the neutrophil count during recovery of blood counts was noted in the majority of patients treated at or above 28.6 mg/m². Interestingly, it seems that there may be a relationship between the duration of exposure to plasma levels of BMS-247550 above 15 ng/mL and the occurrence of neutropenia. This relationship is shown as a scatterplot in Fig. 2. The overall association between the percent change from pretreatment ANC to the nadir ANC during treatment and for the length of time that the BMS-247550 plasma concentration was >15 ng/mL (each as a continuous variable) is negative (Spearman's rank correlation coefficient = −0.59) and statistically significantly (P = 0.0122; Table 6).

Epothilones have generated widespread interest because they inhibit depolymerization of tubulin similar to the taxanes and show activity against taxane-resistant tumors. Animal tumor models have shown that the log cell kill of BMS-247550, a semisynthetic derivative of epothilone B, is equal or superior to paclitaxel in both paclitaxel-sensitive and -resistant tumors (4).

In this phase I study of BMS-247550, the DLT observed was grade 4 neutropenia at 56 mg/m², and 40 mg/m² was the MTD. Only three patients received ≥3 cycles of ≥40 mg/m² dose, and two of these patients required dose reductions. Thus, our preliminary data suggests that in pretreated patients, myelosuppression may limit the number of cycles that can be delivered of BMS-247550 at 40 mg/m² every 3 weeks. Significant gastrointestinal toxicities, including nausea/emesis, were not observed at the dosages evaluated.

Patients developed neuropathic symptoms at the higher dosages (≥28 mg/m²) usually after two or three cycles. Only five patients received the drug for more than two cycles at dosages of ≥28 mg/m². Three of these five patients developed some evidence of neuropathy suggesting that this adverse effect may limit prolonged drug administration. Preclinical data suggested that BMS-247550, unlike paclitaxel, did not induce the production of proinflammatory cytokines and thus may not induce myalgias. Nonetheless, mild, non–dose-limiting myalgias and arthralgias were observed.

Neuropathy has emerged as an important adverse effect in the initial trials of this agent. In our study, the incidence of neuropathy was less than observed in other studies. Two other phase I studies of BMS-247550 evaluated the same schedule and in both these studies a greater incidence of neuropathy was observed (10, 11). In the study by Mani et al., a greater number of cycles (62 of 85 cycles) were given at a dose of ≥40 mg/m² (11). In addition, of the 16 patients (64% of all patients) who developed neuropathic symptoms, 15 patients had received agents known to be associated with neuropathy, previously. In our study, only seven patients (39%) had received therapy known to cause neuropathic symptoms. These differences could explain the relative low incidence of neuropathy observed in our study.

Abraham et al. evaluated a schedule of BMS-247550 given daily ×5 every 3 weeks (12). These investigators also observed a lower rate of neuropathy compared with other studies. This could be attributed to a lower peak concentration because the dose of each cycle was given over 5 days compared with only 1 day. Development of neuropathic symptoms has been correlated with peak concentrations with other drugs such as paclitaxel (13). To reduce the peak concentrations of the drug and thus effect on the development of neuropathy, a decision was made to prolong the duration of infusion of BMS-247550 to 3 hours. However, in a recent phase II study evaluating BMS-247550 in patients with colon cancer given once every 3 weeks with a 3-hour infusion, the drug had to be discontinued in 20% of patients due to neuropathy (14). Thus, further studies with different schedules need to be done to improve the ability to deliver this drug over multiple courses.

Limited evidence of antitumor activity was observed in this phase I trial. One (non–small cell lung cancer) of the three patients who had tumor shrinkage had received prior paclitaxel and was progression free for almost 8 months after BMS-247550. One patient with colon cancer maintained stable disease for 6 months. Phase II studies of this agent have been reported in many tumor types and the drug has shown activity in patients with prior taxane therapy and in tumor types that generally do not respond to taxanes (15–17). Further studies are required to truly define the efficacy of this drug in both taxane-nave and taxane-pretreated patients.

Pharmacokinetic variables of BMS-247550 were computed using a noncompartmental method as the plasma concentrations of the drug exhibited triexponential decline in some patients and biexponential decline in others. A dose-related increase in the systemic exposure (C_max and AUC) of BMS-247550 in the dose range of 7.4 to 56 mg/m² observed in this phase I study indicates that the drug exhibits linear kinetics as was observed in the preclinical pharmacokinetic studies in rats and dogs. The drug is extensively distributed following intravascular administration (V_ss = 376-1,898 L/m²). Although the elimination half-lives ranged from 22 to 72 hours in different subjects in the dose range of 7.4 to 56 mg/m², no statistically significant difference was found among the mean t_1/2 or mean clearance values in that dose range.

Neutropenia within 7 to 10 days followed by an increase in the neutrophil count during recovery of blood counts was noted in the majority of patients treated at or above 28.6 mg/m². This observation was similar to that observed in preclinical toxicity studies in rats (18). Our data suggest that there may be a threshold plasma level of 15 ng/mL for the occurrence of neutropenia. There seems a correlation between length of time that BMS-247550 plasma concentration was >15 ng/mL and the percent change in neutrophil count. The small numbers of patients available for this analysis limits interpretation. Nonetheless, this observation may deserve further evaluation in future trials to optimize the delivery of this drug.

Earlier in vitro studies have revealed that the IC₅₀ values for BMS-247550 in 18 of 21 tumor cell lines tested were in the range of 1.4 to 6 nmol/L (3 ng/mL) for a 72-hour exposure. However, the IC₅₀ values for HCT116/VM 46 (colon cancer cell line with MDR) and MIP (colon cancer cell line) were 24.5 and 24.8 nmol/L, respectively. Furthermore, the minimum effective concentration of BMS-247550 against s.c. Pat-7 human ovarian carcinoma in mice was in the range of 30 to 90 nmol/L (15.2-45.6 ng/mL) for a total infusion duration of 10 hours (4). Thus, it seems that a chronic steady-state concentration of 15 ng/mL might be therapeutically effective while reducing neutropenia.

As stated earlier, BMS-247550 has also been evaluated in other schedules. This includes both 1-hour weekly administrations of the drug, daily ×3 every 3 weeks, and daily ×5 every 3 weeks (12, 19, 20). These schedule modifications may alter the drug's toxicity and dosing profile and it seems that the incidence of neuropathy is less with the daily ×3 and daily ×5 every 3-week schedule. An oral formulation of BMS-247550 has been evaluated in preclinical models and was found to be equally effective (4). The bioavailability with the oral formulation was 31% in mice for pH 8.0 buffered solution and 27% in rats following intraduodenal administration of unbuffered solution. Due to the labile nature of BMS-247550 at low pH, an enteric-coated oral formulation given daily to maintain steady-state plasma levels around 15 ng/mL over several weeks may minimize toxicity and maximize efficacy of the drug.

This study used the novel accelerated titration “2B” design as described by Simon et al. (7). The design is characterized by one-patient cohorts in the initial stages of the study as well as intrapatient dose escalation. Fifteen of 17 patients treated with BMS-247550 received at least one course at ≥28.6 mg/m² dose. Thus, most of the patients received the drug at or close to the MTD and therefore improving the possibility of benefit from the drug. With a modified Fibonnaci design, a lesser proportion of patients accrued to study would have received the drug at dosages close to the MTD.

The duration of enrollment for this study was 20 months. Due to the accelerated design, during the initial portion of the study, only one patient was considered for enrollment at any given time. Some patients considered for enrollment were never registered due to various reasons including ineligibility, decline in performance status, and patient refusal. Thus, consideration of only one patient at a time may have prolonged the enrollment period in this study. This suggests that this accelerated design may not necessarily shorten the study duration.

Three other phase I studies of BMS-247550 also used an accelerated titration design (10–12). In all these three studies, the initial dose escalation was 100% of the previous dose. In two of the three studies, at least one patient died of a drug-related adverse effect, whereas none of our patients died of drug-related toxicities. In general, more patients in these studies received a dose closer to the MTD than in our study. However, we feel that a 40% escalation in the initial stages of the trial may be a safer approach than 100% escalation, especially with a drug that has a relatively narrow therapeutic index. It is unclear whether the number of patients required to complete the study were reduced with the use of this accelerated design compared with a modified Fibonacci design. The authors are not aware of a phase I study evaluating BMS-247550 given every 3 weeks, using the traditional design.

In conclusion, the MTD of BMS-247550 is 40 mg/m² given as a 1-hour infusion every 3 weeks with neutropenia being the DLT. Limited neuropathy was observed though very few patients received prolonged therapy with the drug at the recommended phase II dose of 40 mg/m². BMS-247550 shows linear pharmacokinetics. It is possible that development of neutropenia may be related to sustained drug levels above a certain threshold. However, further investigation is required to confirm this observation. Ongoing phase II studies will provide information on the potential utility of this agent in the treatment of tumors responsive or refractory to existing taxanes.