Phase I Trial with BMS-275183, a Novel Oral Taxane with Promising Antitumor Activity

Purpose: BMS-275183 is an orally administered C-4 methyl carbonate analogue of paclitaxel. We did a dose-escalating phase I study to investigate its safety, tolerability, pharmacokinetics, and possible antitumor activity.

Experimental Design: A cycle consisted of four weekly doses of BMS-275183. The starting dose was 5 mg, which was increased by 100% increments (i.e., 5, 10, 20 mg/m², etc.) in each new cohort consisting of one patient. Cohorts were expanded when toxicity was encountered, and 20 patients were treated at the maximum tolerated dose (MTD). Plasma pharmacokinetics were done on days 1 and 15.

Results: A total of 48 patients were enrolled in this trial. Dose-limiting toxicities consisted of neuropathy, fatigue, diarrhea, and neutropenia. First cycle severe neuropathy was reported in four patients treated at 320 (n = 1), 240 (n = 2), and 160 mg/m² (n = 1), whereas eight patients treated at dose levels ranging from 160 to 320 mg/m² experienced grade 2 neuropathy in cycle one. The MTD was 200 mg/m², as 3 of 20 patients experienced grade 3 or 4 toxicity in cycle one [fatigue (n = 2), and neutropenia/diarrhea (n = 1)]. BMS-275183 was rapidly absorbed with a mean plasma half-life of 22 hours. We observed a significant correlation between drug-exposure and toxicity. Tumor responses were observed in 9 of 38 evaluable patients with non–small cell lung cancer, prostate carcinoma, and other tumor types.

Conclusions: BMS-275183 is generally well tolerated on a weekly schedule. The main toxicity is peripheral neuropathy, and the MTD is 200 mg/m². Promising activity was observed in several tumor types, and a phase II trial in non–small cell lung cancer has been initiated.

Paclitaxel is active against a wide spectrum of malignancies, including ovarian, breast, and lung cancer (1). It is a schedule-dependent drug that benefits from prolonged tumor exposure times in which plasma concentrations are above a pharmacologic threshold level (2). Weekly treatment regimens seem to have less side effects than the most commonly given treatments every three weeks (3–5) with comparable or superior activity in lung and breast cancers, respectively (6, 7). Oral administration allows more prolonged or continued schedules, is often more convenient than i.v. administration, and is preferred by the majority of patients (8). Furthermore, oral formulations do not require the vehicle Cremophor EL, which contributes largely to the hypersensitivity reactions caused by i.v. paclitaxel (9). The oral bioavailability of paclitaxel is, however, very poor due to a low passive absorption and active excretion by the P-glycoprotein pump, which is abundantly present in the intestinal brush border (10). Several attempts have been made to increase the bioavailability of paclitaxel by coadministration of P-glycoprotein inhibitors (11–14).

BMS-275183 is a C-4 methyl carbonate analogue of paclitaxel containing modifications to the side chain (15), that has an oral bioavailability of 24% in humans (16). Its mechanism of action is, like paclitaxel, stabilization of microtubules and the antitumor activity of oral BMS-275183 comparable to that of i.v. paclitaxel in in vivo tumor models. Moreover, BMS-275183 was active in vitro against paclitaxel-resistant tumors including those harboring tubulin mutations or overexpressing P-glycoprotein (17). The aims of this dose-escalating phase I study were to investigate the safety, tolerability, pharmacokinetics, and possible antitumor activity of BMS-275183 in patients with advanced solid tumors, and to determine a recommended phase II dose.

Patients. Adult patients with histologically or cytologically confirmed diagnosis of a solid tumor not amenable to standard therapy were eligible for this study. They had to have an Eastern Cooperative Oncology Group performance status of ≤2, a life expectancy of at least 3 months and adequate renal, liver, and bone marrow function, defined as creatinine <1.5 times the upper limits of normal, bilirubin <1.5 times the upper limits of normal, alanine-aminotransferase <2.5 times the upper limits of normal, absolute neutrophil count >1.5 × 10⁹/L and platelets >100 × 10⁹/L. An adequate method of birth control had to be used, and women of child-bearing potential had to have a negative serum or urine pregnancy test. At least 4 weeks had to have elapsed from prior anticancer treatment (including taxanes) and toxicities (except alopecia) had to be recovered to ≤grade 1 (according to the National Cancer Institute Common Toxicity Criteria version 2.0, NCI-CTCv2.0; ref. 18). Exclusion criteria for patients were serious uncontrolled medical disease, active infection, significant pulmonary or cardiovascular disorder, QTc interval >450 ms, sensory or motor neuropathy ≥grade 2, active brain metastasis, inability to swallow capsules, history of gastrointestinal disease (surgery or malabsorption that could affect the absorption of BMS-275183), concomitant use of known inducers or inhibitors of cytochrome P450 isoform CYP 3A4, and any psychiatric or other disorders such as dementia that would impair compliance. Concomitant radiotherapy or systemic anticancer therapy was not allowed. The study was approved by the medical ethics committees of the two participating institutes, and all patients gave written informed consent prior to study entry.

Study design. BMS-275183 (Bristol-Myers-Squibb, Princeton, NJ) was given orally on a continuous weekly schedule. One cycle consisted of 4 weeks of treatment. The starting dose was 5 mg and dose escalation occurred according to a two-stage (accelerated/standard) design (19). In the accelerated dose escalation phase, one patient per cohort was treated and the dose was increased by 100% (i.e., 5, 10, 20 mg/m², etc.) in each next cohort if no toxicity ≥grade 2 was observed during the first cycle. Upon occurrence of any toxicity ≥grade 2, the accelerated phase ended, and the standard dose escalation phase began. In this phase, at least three patients per dose level were enrolled, and the dose was escalated according to a modified Fibonacci scheme. When a dose-limiting toxicity (DLT) was encountered, the cohort was expanded to six patients. Dose escalation was continued until a DLT was observed in two out of two to six patients. The maximum tolerated dose (MTD) was defined as the highest dose at which no more than one out of six patients experienced a DLT. A minimum of 15 patients were to be treated at the MTD, to further establish the safety of a recommended phase II dose.

DLTs were predefined as any of the following drug-related side effects occurring during the first cycle: grade 4 neutropenia ≥5 consecutive days, febrile neutropenia, grade 4 thrombocytopenia or grade 3 with a bleeding episode requiring platelet transfusion, any grade ≥3 nonhematologic toxicity, retreatment delay of >1 week due to drug-related toxicity, QTc interval >500 ms, and any clinically significant arrhythmia within 24 hours following drug administration. Hypersensitivity reactions were not defined as DLTs. Dose reductions by one level were done when a DLT or grade 2 neurotoxicity occurred. During the study, it was noted that several patients treated at the highest dose of 320 mg/m² had to be dose-reduced in the first cycle for grade 2 toxicity that did not qualify for a DLT. We therefore added the following DLT criterion: “dose reduction or omission due to any drug-related toxicity before completion of the first cycle.”

Drug administration. BMS-275183 was provided in 5 and 25 mg capsules solubilized in polyethylene glycol 400/1450 with Gelucire 44/14 as the excipient system at a loading of 4% w/w. The calculated dose was rounded to the nearest 5 mg. Patients ingested the capsules with 150 mL of water for up to 10 minutes. Patients were fasted for at least 8 hours prior to drug administration and for 2 hours post-dose. No prophylactic medication was prescribed. Patients were to receive at least two cycles.

Patient evaluation. Pretreatment evaluation included a complete history and physical examination, urinalysis including pregnancy test, tumor assessment, chest X-ray, ECG, a full blood count, coagulation tests, serum chemistries, and determination of serum tumor markers. All blood tests and toxicity assessment were repeated weekly. Physical examination was done before each cycle. Toxicities were graded according to NCI-CTCv2.0 (18). Patients were considered evaluable for toxicity if they received at least one dose of the study drug.

Because BMS-275183 moderately prolonged the action potential duration in isolated Purkinje fibers,⁴

ECG-monitoring was done for 24 hours (at baseline, 2, 6, and 24 hours) after the first drug administration to monitor potential prolongation of the QTc interval (calculated using the Bazett's formula: QTc = QT / √R − R). ECG monitoring was again done after the second dose if a QTc interval >450 ms was observed.

Response to therapy was assessed every other cycle according to WHO criteria (20). To be evaluable for response, patients had to complete two cycles, unless they had to prematurely discontinue treatment because of rapidly progressive disease.

Blood sampling and pharmacokinetic analysis. Pharmacokinetic monitoring was done on days 1 and 15 of the first cycle. Blood samples of 5 mL were collected in potassium-EDTA vacutainers (Becton Dickinson, Franklin Lakes, NJ) up to 48 hours after drug administration (time points 0, 0.5, 1, 1.5, 2, 3, 4, 5, 6, 8, 24, and 48 hours). The tube was placed on ice for 10 to 30 minutes, centrifuged for 5 minutes at 2,000 × g at 0°C to 4°C, and plasma was separated and stored at −80°C until analysis. Pharmacokinetic profiles were evaluated by noncompartmental analysis using the software package Kinetica version 4.2 (InnaPhase Corporation, Philadelphia, PA). The elimination half-life (T-half) was assessed from the elimination rate constant, estimated by linear regression of the terminal phase of the semilogarithmic concentration versus time curve. The area under the concentration versus time curve from time 0 to the last experimental time point was estimated by the linear-log trapezoidal rule extrapolated to infinity [AUC (INF)].

Plasma concentrations were determined by a validated liquid chromatography/mass spectrometry method. The internal standard used was [¹³C₆] BMS-275183. Oral taxane was extracted by using 4 mL of toluene, followed by shaking for 20 minutes and centrifugation. The organic layer was transferred to a clean tube and evaporated to dryness. The dried extract was reconstituted in 100 μL of a mixture of mobile phases A and B (50:50, v/v). Twenty microliters of the reconstituted sample were directly injected into the liquid chromatography/mass spectrometry system. Chromatographic separation was achieved by gradient elution on a Keystone Hypersil ODS cartridge (2 × 20 mm, 3 μmol/L). Mobile phase A contained 10 mmol/L ammonium acetate (pH 5.5)/methanol (75:25, v/v), and mobile phase B contained 10 mmol/L ammonium acetate/acetonitrile (5:95, v/v). Detection was by negative ion electrospray mass spectrometry on a Micromass Quattro LC. The standard curve, which ranged from 0.1 to 20 ng/mL, was fitted to a 1 / x² weighted quadratic regression model. All plasma samples were analyzed within a total of 35 analytic runs. The limit of quantification was 0.1 ng/mL, and except for the baseline sample, there were no samples below this limit. Quality control samples were measured along with the study samples to assess the accuracy and precision of the assay. The acceptance criteria established for the analysis of oral taxane in plasma specified that the predicted concentrations of at least three-fourths of the standards and two-thirds of the quality control samples be within ±15% of their individual nominal concentration values (±20% for the lowest concentration standard). In addition, at least one quality control sample at each concentration must be within ±15% of its individual nominal concentration values. Values for the between-run precision and the within-run precision for analytic quality control samples were ≤7.6% and 11.7% coefficient of variation, respectively, with deviations from the nominal concentrations of ≤11.2%.

Statistical analysis. Descriptive statistics were used for baseline characteristics, safety assessment, and pharmacokinetic data. Scatter plots versus dose were used to examine the relationship of pharmacokinetic variables to dose. The response rate was calculated for all response-evaluable patients along with the exact 95% confidence interval (CI; ref. 21). Median duration of response was calculated using the Kaplan-Meier method, along with their 95% CI. Quantitative results were analyzed by a Student's t test. P values resulted from two-sided tests, and were considered significant when <0.05.

Patients and treatment. Between August 2000 and September 2004, 48 patients were enrolled and treated in the trial at two participating centers in the Netherlands. Patient characteristics are depicted in Table 1. At study entry, the median performance status was 1, and most patients (28 of 48) had previously received two or more chemotherapy regimens before entering the study. The enrolled patients were diagnosed with a broad variety of tumor types (Table 1), with a relatively large group of non–small cell lung cancer (NSCLC) patients (40%) and colon carcinoma patients (21%).

Patients were enrolled at the following dose levels (Fig. 1): 5 mg, 5, 10, 20, 40, 80, 160, and 320 mg/m². In the original design of the study, further escalation according to a modified Fibonacci scheme was planned because grade 2 and higher side effects occurred at 320 mg/m². However, because of the frequency of toxicity observed at the highest dose level, the dose was not further increased and the following intermediate dose levels were (re)explored: 240 mg/m² (n = 2), 200 mg/m² (n = 21), 160 mg/m² (n = 5), and 120 mg/m² (n = 6; Fig. 1). The mean administered number of cycles was 3.9 (range, 0.25-18 cycles). Two patients were not able to complete the first cycle for reasons unrelated to the study drug and were replaced: one patient on the 5 mg/m² dose level had to discontinue treatment because of rapidly progressive disease, whereas the other patient treated with a dose of 200 mg/m² developed a bowel perforation shortly after the first dose.

Toxicity.Table 2 summarizes the drug related adverse events. From the dose level of 5 mg to 160 mg/m², no toxicity greater than grade 1 was observed in the first cycle. The dose level of 320 mg/m² was not well tolerated: although only one of six patients experienced a DLT consisting of grade 3 peripheral neuropathy, three of six patients had to be dose-reduced in the first cycle because of grade 2 sensory neuropathy (n = 2) and persisting grade 2 arthralgia (n = 1; Fig. 1). In addition, one patient at this dose level experienced grade 4 hypotension combined with grade 2 diarrhea and grade 3 fever 48 hours after the first drug administration. He recovered within a day after treatment with both antibiotics and corticosteroids. Because no infectious cause could be identified, we classified this episode as a late and atypical hypersensitivity reaction. Rechallenge was uneventful after dose-reduction and premedication with dexamethasone, clemastine, and cimetidine. None of the other patients received premedication, and no other hypersensitivity reactions were observed. Although the report of dose reductions in the first cycle for persisting grade 2 toxicity in as many as three patients did not qualify for a DLT in the then current version of the protocol, this observation still indicated that the 320 mg/m² dose level was not feasible. We therefore amended the protocol to include dose reductions or dose omissions occurring in the first cycle in the DLT definition.

Both patients treated at the 240 mg/m² dose level experienced a DLT consisting of grade 3 peripheral neuropathy. Subsequently, we explored the 200 mg/m² dose level. One DLT, a grade 4 malaise, was observed in the first six patients treated at this dose level. We thus defined this dose as the MTD, and expanded the cohort to 15 patients to assess this dose for phase II testing. In this expanded cohort, one of nine patients experienced a grade 3 diarrhea combined with grade 4 neutropenia in the first cycle. In addition, three of nine patients had to be dose reduced in the first cycle because of grade 2 peripheral neuropathy. As 5 out of 15 patients were not able to receive a full cycle at 200 mg/m², we decided to re-explore the 160 mg/m² dose level, and expanded this cohort to a total of six patients. At this dose level, three of six patients were not able to tolerate a full cycle: one patient experienced neutropenic fever, and two patients were dose reduced because of grade 2 neuropathy. We therefore explored the next intermediate dose level of 120 mg/m², where we did not observe any toxicity greater than grade 1 in the first cycle. As we anticipated that this dose level would be safe but insufficiently active, we decided to further expand the 200 mg/m² cohort to 20 patients. Of the five additionally enrolled patients, one patient experienced a grade 3 fatigue, and one patient was dose reduced because of grade 2 fatigue which lasted >1 week.

Episodes of mild to moderate (grades 1-2) gastrointestinal toxicity were observed in ∼30% of patients, but they did not interfere with the ingestion of BMS-275183 capsules. ECG monitoring revealed a grade 1 prolongation of the QTc interval in 2 out of 41 monitored patients (treated with 160 and 200 mg/m², respectively). In both cases, this normalized after repeated ECGs and did not recur after the second dose, suggesting that BMS-275183 does not have clinically relevant cardiac side effects. Taken together, the main DLT of BMS-275183 was peripheral neuropathy, whereas other DLTs were infrequent and consisted of diarrhea, neutropenia, and fatigue. In general, hematologic toxicity did not occur frequently. The dose level of 200 mg/m² was identified as the MTD.

Peripheral neuropathy: main DLT. Among the 48 treated patients, 31 had a new neuropathy event or worsening neuropathy compared with the baseline severity occurring at any time during their treatment with BMS-275183. These 31 patients belonged to the following dose cohorts: 3 of 6 patients at the 120 mg/m², 3 of 6 patients at the 160 mg/m², 17 of 21 patients at the 200 mg/m², 2 of 2 patients at the 240 mg/m², and 6 of 6 patients at the 320 mg/m² cohort. The observed neuropathy was severe in 6 patients, moderate in 19 patients, and mild in 6 patients. Sensory neuropathy was more common than motor neuropathy, 31 of 48 patients versus 18 of 48 patients, respectively. Symptoms typically consisted of tingling or numbness of the toes and feet, and/or the fingers and hands, with painful neuropathy in some patients. Strength loss indicating motor neuropathy occurred mainly in the lower extremities and interfered with walking in severe cases. Nerve conduction studies done in patients with severe neuropathy symptoms showed moderate to severe axonal polyneuropathy, both sensory and motor.

Neuropathy usually developed rapidly (Fig. 2), with a median time to onset of 1.2 months (95% CI, 0.3-10.8 months), and was partially reversible. In 14 of 31 patients, the neuropathic symptoms resolved to baseline with a median time to resolution of 8.6 months (95% CI, 5.8-17.5 months). In all patients with severe neuropathy due to BMS-275183, the symptoms decreased to moderate or mild severity after treatment discontinuation. Interestingly, 25 of 34 patients without neuropathic signs at baseline developed neuropathy during BMS-275183 treatment, versus 6 of 14 patients with grade 1 baseline neuropathy, suggesting that baseline grade 1 neuropathy does not predispose to worsening of neuropathy symptoms during BMS-275183 treatment. It is, however, noteworthy that patients with neuropathy greater than grade 1 resulting from prior therapies were excluded from participation in this trial.

Pharmacokinetic analysis. Blood samples for pharmacokinetic analysis were available from 48 patients following the first dose, and from 39 patients following the third dose. Table 3 presents the mean pharmacokinetic variables of BMS-275183 for the first dose. The compound is rapidly absorbed, with a median T_max of 1.0 hours and a mean T-half of 22 hours. In the lower dose ranges, the systemic exposure (AUC) increased with dose level, but this relationship was less evident for doses >160 mg/m² (Table 3). Figure 3 shows the mean concentration versus time profile of BMS-275183 (doses, 120-320 mg/m²). The mean interpatient variability at the 200 mg/m² dose level (n = 21) was 53%. In the 25 patients treated with the same dose on days 1 and 15, the mean intrapatient variability was 21% with both higher (n = 12) and lower (n = 13) exposures on day 15, suggesting that no clinically relevant accumulation or induction of metabolism occurred.

Variables of plasma exposure were more predictive of toxicity than the dose level. Patients in which a DLT or dose reduction after the first dose occurred were treated with doses ranging from 160 to 320 mg/m². Their mean AUC was 6,695 ng/h/mL (n = 16; SD, 2,224 ng/h/mL), compared with a mean AUC of 2,478 ng/h/mL (n = 32; SD, 1,828 ng/h/mL) in patients not experiencing severe toxicity, resulting in a highly significant difference in means test (P < 0.001). This relationship was also observed for the C_max, but with a greater variability (data not shown). We investigated whether the high exposure to the study drug could be predicted by factors influencing the metabolism of the drug. The concomitant use of medications inhibiting or inducing isoform CYP 3A4 of cytochrome P450, the liver enzyme system responsible for the metabolism of BMS-275183, was excluded by the protocol. We hypothesized that the presence of liver metastases could influence the metabolism of BMS-275183. We therefore analyzed the clearance of BMS-275183 by liver involvement. The mean clearance in patients with liver metastases (n = 16) was 94 compared to 133 L/h in patients without liver involvement (n = 32), but this did not reach statistical significance (P = 0.10).

Tumor response. We observed nine partial tumor responses in 38 response-evaluable patients (23.7%; 95% CI, 11.4-40.2%). The tumor types of responding patients included: NSCLC (4 of 13 response-evaluable patients), prostate carcinoma (2 of 2 patients), primitive neuroectodermal tumor (1 of 1 patient), cholangiocarcinoma (1 of 2 patients), and undifferentiated sarcoma (1 of 1 patient). Partial responses were observed after a mean of 3.5 cycles (range, 1-7 cycles) and their median duration was 8.7 months (range, 5.4-40.4 months). Computed tomography scans of a responding NSCLC and cholangiocarcinoma patient are shown in Fig. 4.

As most activity was observed in NSCLC, we conducted further analyses in this tumor type. Intent-to-treat analysis of all 19 enrolled NSCLC patients showed a median progression-free survival of 3.2 months (95% CI, 2.9-7.3 months) and a median overall survival of 10.6 months (95% CI, 8.8-18.1 months). Eleven out of 19 NSCLC patients had received a taxane-containing chemotherapy regimen (docetaxel or paclitaxel) prior to BMS-275183 therapy. Interestingly, two out of five patients with a response to prior taxane therapy responded to BMS-275183 as well. In addition, one patient with a stable disease as best response to prior docetaxel had a partial remission on BMS-275183 treatment. These data suggest that BMS-275183 is active in NSCLC after prior taxane treatment.

In this dose-escalating study, we found that BMS-275183 is a potent taxane analogue that could be safely administered orally without the use of prophylactic medication for hypersensitivity reactions. The main DLT consisted of peripheral neuropathy, whereas other grades 3 and 4 side effects comprised fatigue, diarrhea, and neutropenia, and were infrequent. These findings are remarkable, because hematologic toxicity and not neuropathy was dose-limiting for both paclitaxel and docetaxel (reviewed in refs. 1, 22). Severe paclitaxel-induced neuropathy was mainly observed with high doses of 250 mg/m² or greater (23, 24), whereas clinically significant neuropathy is not a common side effect of docetaxel treatment (22). The clinical neuropathy symptoms of patients treated with BMS-275183 were partially reversible and comparable to those induced by paclitaxel, although we observed a relatively high incidence of motor neuropathy. The distal and symmetrical neurologic deficits suggest that taxanes cause a “dying back” axonopathy resulting from disruption of axoplasmic transport from the drug's effect on microtubule assembly (25). Several attempts have been made to reduce the neurotoxic effects of paclitaxel by neuroprotective drugs (26, 27), but the clinical effects have been disappointing thus far (26, 28).

We identified 200 mg/m² as the MTD. The definition of this dose was not unequivocal: (re)exploration of several intermediate dose levels was necessary to identify a safe dose. This was due to the design of the dose escalation schedule with only one patient per dose level, combined with a relatively high pharmacologic interpatient variability (53%) and strict definitions of safety which prompted dose reduction upon occurrence of grade 2 peripheral neuropathy. Retrospectively, a conventional design with three to six patients per cohort might have resulted in a faster and more straightforward conduct of the trial. The dose of 200 mg/m² was generally well tolerated, although some patients had a relatively high exposure to the drug and experienced severe side effects. This relationship between drug exposure and severity of adverse events has also been reported for other taxanes (29). A recently initiated phase II trial in NSCLC therefore randomizes between weekly administration of 120 and 200 mg/m², to compare the activity and tolerability of these two different dose levels of BMS-275183.

The interpatient variability of 53% that we observed for BMS-275183 at 200 mg/m² is approximately twice as high as for i.v. paclitaxel (historical data; ref. 30). This is not unprecedented for an oral drug, and is most likely caused by variations in the absorption of the drug combined with individual differences in metabolism due to CYP3A4 polymorphism. These two factors probably also explain the apparent nonlinear relationship between the dose of BMS-275183 and AUC or C_max. The small patient number in many cohorts strengthens this impression, and future studies will allow more firm conclusions on the linearity of the pharmacokinetic variables of BMS-275183. Recently, several studies have been conducted to predict docetaxel and irinotecan pharmacokinetics by use of cytochrome P450 CYP3A4 phenotyping probes in order to minimize toxicity and to maximize efficacy (31–34). Although such individualized docetaxel dosing has significantly decreased its pharmacokinetic variability compared with body surface area–based dosing (34), the lack of an easily administered, low-cost, and widely available test for CYP3A4 activity still limits its applicability (35).

For paclitaxel, it is known that short infusion times (1-3 hours) allow higher doses to be administered but induces neuropathy as a common side effect, whereas prolonged infusion times of 24 hours or longer trigger less neuropathy, but give a higher incidence of myelosuppression (36). As neutropenia was not an important DLT in this trial with weekly dosing of BMS-275183, we have initiated a phase I study investigating twice weekly administration in order to spread the systemic exposure over a longer period of time and hopefully minimize the neurotoxicity of BMS-275183. Preliminary data suggests that the incidence of neuropathy is indeed lower in a twice weekly administration regimen (37), indicating that the dosing schedule of BMS-275183 may be optimized.

In summary, BMS-275183 is a potent oral taxane analogue that is generally well tolerated at the MTD of 200 mg/m² weekly. Its safety profile differs from other taxanes: the principal side effect is neuropathy rather than myelosuppression, and premedication for hypersensitivity reactions is not needed. The observed response rate of 24% in the heavily pretreated patient group of this phase I trial indicates that BMS-275183 is a potent new taxane analogue. In addition, our results suggest that BMS-275183 has significant activity in NSCLC and the observed responses in other tumor types warrant further investigation. A phase II study in NSCLC as well as phase I trials to investigate other treatment schedules are currently ongoing.