Purpose: Paclitaxel is a taxane derivative with a profound antitumor activity against a variety of solid tumors. In a previous clinical study in patients with non-small cell lung cancer (NSCLC) treated with paclitaxel, it was shown that paclitaxel plasma concentrations of 0.1 μmol/liter for ≥15 h were associated with prolonged survival. The purpose of this study was to evaluate the feasibility of Bayesian dose individualization to attain paclitaxel plasma concentrations >0.1 μmol/liter for ≥15 h.

Experimental Design: Patients with stage IIIb-IV NSCLC were treated with paclitaxel and carboplatin once every 3 weeks for a maximum of six courses. During the first course, a standard paclitaxel dose of 175 mg/m2 was administered i.v. in 3 h. In subsequent courses, the paclitaxel dose was individualized based on observed paclitaxel concentrations in plasma during the previous course(s) using a Bayesian algorithm. The paclitaxel dose of a subsequent course was increased to the lowest dose for which the predicted time period during which the paclitaxel plasma concentration exceeds 0.1 μmol/liter was >15 h.

Results: A total of 25 patients have been included in the study (92 evaluable courses). During the first course, the median time period above the threshold concentration was 16.3 h (range, 7.6–31.6 h), and was <15 h for 9 patients (36%). During subsequent individualized courses, the time period above the threshold concentration was <15 h in 23% (5 of 22), 14% (2 of 14), 23% (3 of 13), 11% (1 of 9), and 11% (1 of 9) of the patients in the second, third, fourth, fifth, and sixth course, respectively. Dose increments, ranging from 5 to 65 mg/m2, were performed in 29 of the 67 individualized courses. Patients with increased individualized doses had similar regimen related toxicities compared with those remaining at a dose of 175 mg/m2. Toxicity was reversible and manageable, and was mainly hematological (granulocytopenia CTC grade 3/4 in 80% of the patients). The objective response rate was 20%.

Conclusions: The results indicate that the applied pharmacokinetically guided dosing strategy for paclitaxel is safe and technically feasible. A randomized study is necessary to demonstrate whether dose individualization may result in improved activity and efficacy in patients with NSCLC.

Paclitaxel (Taxol®) is a taxane derivative that acts on microtubular structures by binding directly to tubulin, causing microtubular stabilization that arrests cell division in the G2-M phase of the cell cycle. The drug has a profound antitumor activity against a variety of solid tumors, especially against lung, breast, ovarian, and head and neck tumors (1, 2). Paclitaxel administered as a 3-h infusion has been shown to result in equivalent efficacy as a 24-h infusion scheme but is associated with significantly less myelosuppression (3). The combination of paclitaxel with a platinum compound, cisplatin or carboplatin, is a frequently used combination in the treatment of patients with advanced non-small cell lung cancer (NSCLC; Ref. 4). To offset cisplatin-related toxicity, cisplatin is often substituted by carboplatin (4, 5).

The combination of paclitaxel and carboplatin administered to 55 nonpretreated patients with advanced NSCLC in a Phase I design resulted in a significant positive relationship between exposure to paclitaxel and overall survival. Favorable survival figures were observed in patients with a time period above a chosen threshold paclitaxel concentration of 0.1 μmol/liter >15 h, compared with a time period above the threshold concentration of <15 h (6). Significant relationships between plasma pharmacokinetics and pharmacodynamics of paclitaxel have also been reported in ovarian cancer patients receiving paclitaxel monotherapy whereby myelotoxicity was related to the time period during which paclitaxel concentrations in plasma were >0.05 or 0.1 μmol/liter (7, 8).

Due to substantial interindividual pharmacokinetic variability, individuals may obtain different paclitaxel plasma concentrations over time after fixed doses of paclitaxel [e.g., after administration of 175 mg/m2 in 3 h, a 4–5-fold difference in paclitaxel area under the concentration-time curve (AUC) was reported; Ref. 6]. Pharmacokinetically guided dosing can assist in attaining desired plasma levels by tailoring individual dosages based on estimated individual pharmacokinetic parameters. We developed previously a population pharmacokinetic model for paclitaxel (9). On the basis of this population analysis, Bayesian predictions of the pharmacokinetics of an individual can be made with only a few collected blood samples. The use of pharmacokinetically guided dosing to obtain desired plasma levels has been studied for several anticancer agents, with promising results (10).

The aim of this study was to evaluate the performance and the technical feasibility of pharmacokinetically guided dosing of paclitaxel, administered as a 3-h infusion, in attaining paclitaxel concentrations >0.1 μmol/liter for ≥15 h. Moreover, toxicity and antitumor activity using this approach for the first-line treatment of advanced NSCLC were determined.

Patients.

Patients with histologically proven NSCLC stage IIIB or stage IV were included in the study. Eligibility criteria included a performance status ≤2 on the WHO scale, life expectancy of ≥3 months to allow adequate follow-up of toxicity, adequate hematopoietic (absolute neutrophil count ≥1.5*109/liter and platelet count ≥100*109/liter), hepatic (total bilirubin ≤25 μmol/liter, aspartate amino transferase and alanine amino transferase ≤ 2.5 times the normal upper limit; in case of liver metastasis aspartate amino transferase and alanine amino transferase ≤5 times the normal upper limit) and renal function (serum creatinine ≤140 μmol/liter and creatinine clearance ≥50 ml/min). Exclusion criteria consisted of active bacterial infections, clinical signs of active brain tumor involvement or leptomeningeal disease, known alcoholism, drug addiction, and/or psychotic disorders leading to inadequate follow-up, pregnancy, or breast-feeding. Written informed consent was obtained from all of the patients. The study was approved by the Medical Ethics Committee of the Institute.

Treatment.

Patients were treated with carboplatin (Paraplatin; Bristol-Myers Squibb, Syracuse, NY) and paclitaxel (Taxol®; Bristol Myers Squibb) for a maximum of six courses every 3 weeks. All of the courses consisted of a carboplatin i.v. infusion in 30 min followed by administration of paclitaxel in a 3-h infusion. In each course, the carboplatin dosage was calculated based on the Calvert formula (11) with a target AUC of 6 mg/ml*min:

\[\mathrm{Dose}\ {=}\ \mathrm{target\ AUC}\ {\times}\ (\mathrm{GFR}\ {+}\ 25)\]

where dose is expressed in mg, AUC in mg/ml*min, and the glomerular filtration rate (GFR) in ml/min. The glomerular filtration rate in this formula was approximated by the creatinine clearance (Ccr) as estimated by the Cockcroft-Gault formula: (12) 

\[Ccr\ {=}\ \frac{(140\ {-}\ \mathrm{age})\ {\times}\ \mathrm{weight}\ {\times}\ 1.23}{\mathrm{serum\ creatinine}}\ {+}\ 0.85\ (\mathrm{for\ female})\]

where Ccr is expressed in ml/min, age in years, weight in kg, and serum creatinine in μm. Serum creatinine was determined using the (adjusted) Jaffé method (subtracting 26 μmol from the result of the Jaffé method; Ref. 13).

Standard premedication with dexamethasone (20 mg orally at 12 and 6 h before paclitaxel administration), clemastine (2 mg i.v. 30 min before paclitaxel administration), and cimetidine (300 mg i.v. at 30 min before paclitaxel administration) was administered to prevent potential hypersensitivity reactions. Granisetron (1 mg i.v.) was administered as antiemetic agent. Treatment was discontinued in case of patient refusal, noncompliance of the patient with the protocol, serious adverse events, and progressive disease. Both paclitaxel and carboplatin doses were reduced by 25–50% in case of severe toxicity in a preceding cycle.

Pharmacokinetics and Bioanalysis.

Pharmacokinetic sampling for paclitaxel was performed during all of the courses. During the first course, blood samples were collected in heparinized tubes before start; 1.5 h after the start of the paclitaxel infusion; at the end of the infusion; and 0.5, 1, 2, 4, 6, 9, and 19 h after the end of the infusion, to obtain accurate information on the pharmacokinetics of the respective individual. During subsequent courses of paclitaxel, blood samples were obtained before the start of the paclitaxel infusion, at the end of the infusion, and 9 and 19 h after the end of the infusion. The blood samples were immediately centrifuged at 3000 × g for 5 min. Plasma was transferred to polypropylene tubes and stored at −20°C until analysis.

Paclitaxel was determined in plasma using a validated isocratic high-performance liquid chromatographic method with solid phase extraction as sample pretreatment as described previously (14, 15). The quantitation range of the method was 10–10,000 ng/ml.

During each first carboplatin course pharmacokinetic sampling was performed to confirm achievement of the target carboplatin AUC of 6 mg/ml*min. Samples were collected in heparinized tubes before the start of the carboplatin infusion; at the end of infusion; and at 4, 8, and 23 h after the end of the carboplatin infusion. Plasma was separated by centrifuging the samples at 3000 × g for 5 min. Plasma ultrafiltrate was prepared immediately by transferring 1 ml of plasma in an Amicon micropartition system with a YMT-14 membrane (Millipore Corporation, Bedford, MA) and centrifuging the system at 2500 × g for 20 min. Ultrafiltrate was stored at −20°C until analysis. Carboplatin was quantitated using a validated Zeeman atomic absorption spectrometry method (16). The AUC of carboplatin was determined by noncompartmental analysis using the pharmacokinetic WinNonlin program (Standard Edition Version 3.0, 1999) applying the log-linear trapezoidal method with extrapolation to infinity.

Pharmacokinetically Guided Dosing.

During the first course, paclitaxel was administered in a standard dose of 175 mg/m2. In the second course and subsequent courses the paclitaxel dosage was individualized to achieve a plasma concentration of 0.1 μmol/liter for ≥15 h after the start of the infusion. Individualized doses were calculated using Bayesian analyses applying a previously developed and validated population pharmacokinetic model (9). This population model was developed in our institute based on 100 concentration-time curves from 61 chemonaive NSCLC stage IIIB and IV patients (n = 55) treated with paclitaxel as a 3-h i.v. infusion in combination with carboplatin, and from advanced ovarian cancer patients (n = 6), pretreated with platinum, treated with paclitaxel (24-h i.v. infusion) as single agent. The model was developed using nonlinear mixed effect modeling software (NONMEM, version V 1.1, double precision; Ref. 17) using the first order method and logarithmically transformed concentration-time data. The population model consisted of a three-compartment model with saturable distribution to one peripheral compartment and Michaelis-Menten elimination from the central compartment, and is schematically represented in Fig. 1. Interindividual variability and interoccasion variability were estimated using a proportional error model. Interindividual variability was quantified for the volume of the central compartment, maximal transport rate from the central compartment to peripheral compartment 1, and the maximal elimination rate. Interoccasion variability was quantified for the intercompartmental rate constants (k21, k13, and k31) and maximal elimination rate. Simulation studies showed that interindividual variability in the time above the threshold was considerably larger than interoccasion variability. The residual variance was modeled using an additive error model with the assumption that the population was a mixture of two subpopulations (subpopulations 1 and 2) with a different variance. Body surface area (BSA) was significantly correlated with volume of the central compartment and maximal transport rate from the central compartment to peripheral compartment 1 (P < 0.005). The model-predicted population pharmacokinetic parameters are summarized in Table 1.

Bayesian estimates of the individual pharmacokinetic parameters, based on both the population pharmacokinetic parameter values (Table 1) and the individual paclitaxel concentration-time data from previous course(s), were obtained with NONMEM using the POSTHOC option. The obtained individual pharmacokinetic parameter estimates were used to predict the concentration-time data during the subsequent course after the administration of 175 mg/m2. The time period that the paclitaxel concentration was >0.1 μm was derived from these predicted concentration-time data. If the predicted time period of the paclitaxel plasma concentrations >0.1 μmol/L was <15 h in the subsequent course, the paclitaxel dosage was increased until the predicted time period was ≥15 h (doses were increased in multiples of 5 mg). Due to the presence of interoccasion variability in the population model, it is possible that a patient with a paclitaxel exposure above the target in the first course may receive a dose higher than 175 mg/m2 in the second course based on predictions of the pharmacokinetic parameters in this course. For every course this strategy was repeated.

Evaluation of the performance of the dose individualization strategy was performed by estimating the actual time period that the plasma concentrations exceeded 0.1 μmol/liter in the concerning individualized course based on the drawn plasma samples using, again, Bayesian estimation applying the population model.

Patient Evaluation.

Pretreatment evaluation included a complete medical history and complete physical examination (including neurological examination). Before each course, toxicities and performance status were recorded, and a physical examination was performed. Hematological parameters were determined weekly during all of the courses. Blood chemistry including liver and renal function, liver enzymes, serum electrolytes, total protein, albumin, and glucose levels were checked weekly during the first course. During subsequent courses blood chemistry was checked before each course. The observed toxicity was graded according to the National Cancer Institute Common Toxicity Criteria (CTC), Version 1 (18). Tumor measurement was performed every other cycle. Responses were evaluated according to the RECIST criteria (19).

Relationships between hematological toxicity and the time period that the paclitaxel concentration in plasma is >0.1 μmol/liter, as found in previous studies (7, 8), were evaluated. For this purpose, hematological toxicity was evaluated as a percentage decrease (%D) in granulocyte, leukocyte, or platelet counts using the following equation:

\[\%D\ {=}\ \frac{\mathrm{pretreatment\ value}\ {-}\ \mathrm{value\ of\ the\ nadir}}{\mathrm{pretreatment\ value}}\ {_\ast}\ 100\%\]

The relationships were modeled both linear as well as with the sigmoidal maximum effect equation (7, 8).

Patients.

A total of 25 patients with NSCLC stage IIIB or IV were included in this dose individualization study during a period of 17 months. Patient characteristics and biochemical serum values at baseline are summarized in Table 2. Totally, 9 females and 16 males with a median age of 58 years were included. Two patients were pretreated with surgery (lobectomy and resection of a cerebral metastasis), 7 patients with radiotherapy, and 2 with chemotherapy (both platinum containing chemotherapy).

Nine patients completed all six of the courses of chemotherapy according to the protocol. Patients received less than six courses because of progressive disease (n = 8), deterioration in performance status (n = 3), unexpected disease-related death (n = 2), or change of diagnosis (n = 1). Three patients received dose reductions of 25% of both paclitaxel and carboplatin from course 2, 3, and 4 on, respectively, because of development of granulocytopenia grade 4 complicated with fever (n = 2) or thrombopenia grade 2 (n = 1) after the preceding course (additional dose reduction to 50% of the original dose was performed in 1 of these patients from course 4 on). The time period that the paclitaxel plasma concentrations exceeded 0.1 μmol/liter in these patients in the course before 25% dose reduction were 17.5, 25.6, and 16.9 h, respectively. Two of these patients did complete all six of the courses, whereas 1 went off study after course four due to disease progression.

Five courses in different patients were administered with a week delay because patients had not recovered sufficiently from toxicity at the planned day of their retreatment. Four of these patients had not received a pharmacokinetically guided paclitaxel dose increment whereas 1 had.

Pharmacokinetically Guided Dosing.

A total of 103 courses of paclitaxel and carboplatin were administered, and 92 were pharmacokinetically monitored for paclitaxel. Patients receiving reduced doses due to toxicity were not pharmacokinetically monitored during the courses with reduced doses (a total of 10 courses). In 1 patient the second course was accidentally not monitored. Drug monitoring was performed during the first cycle (n = 25), second cycle (n = 22), third cycle (n = 14), fourth cycle (n = 13), fifth cycle (n = 9), and sixth cycle (n = 9).

The paclitaxel concentrations in plasma as obtained during the first course, after administration of a standard dose of 175 mg/m2, are presented in Fig. 2. The time period that the paclitaxel concentration was >0.1 μmol/liter during the first course ranged from 7.6 to 31.6 h. In 9 of these patients (36%), the time period above the threshold concentration was <15 h, ranging from 7.6 h to 13.4 h (Table 3).

In the second course and subsequent courses, the paclitaxel dosage was individualized based on the observed concentrations in the previous course(s). The dose individualization resulted in the administration of a dose of 175 mg/m2 or higher, to achieve a plasma concentration of 0.1 μmol/liter for ≥15 h after start of the infusion. Dose increments (ranging from 5 to 65 mg/m2; 3–37% adjustment) were performed in 29 of the 67 individualized courses (43%) resulting in total doses of 180–240 mg/m2 (8 × 180 mg/m2, 5 × 185 mg/m2, 3 × 190 mg/m2, 3 × 195 mg/m2, 2 × 200 mg/m2, 1 × 205 mg/m2, 5 × 215 mg/m2, 1 × 230 mg/m2, and 1 × 240 mg/m2). In the remaining 57% of the courses, a dose of 175 mg/m2 was administered. During the second course paclitaxel concentrations were above the threshold concentration for [lt 15 h in 23% of the patients (Table 3). In 8 of the 9 patients in whom the time period above the threshold concentration was <15 h during the first course, it was accomplished to obtain a time period >15 h during the second course. During the third course and subsequent courses, the desired time period above the threshold concentration was not achieved in 7 of 45 courses (16%) ranging from 10.9 h to 13.1 h.

In Fig. 3 the concentration-time curves of all six of the courses in a typical patient are shown. This patient received adapted doses during courses two to six, because after the standard dose of 175 mg/m2, the time period that paclitaxel concentrations were above the threshold was only 11.6 h. After dose adaptation during subsequent courses (215 mg/m2) the time period exceeded 15 h. All five of the individualized courses could be administered without delay. In Fig. 4 for all of the individuals and courses the obtained time period that paclitaxel plasma concentrations were >0.1 μmol/liter are depicted. Seven of 13 individualized courses, in which the exposure was too low after dose individualization, were in 2 individuals (with BSA >2 m2).

Monitoring of carboplatin AUCs after all of the first administrations resulted in a median deviation of 18% from the target AUC of 6 mg/ml*min (range, −7% to 73%) indicating that most patients received at least the intended exposure to carboplatin.

Patient Evaluation

Toxicity.

Hematological toxicity consisted mainly of granulocytopenia (range, CTC grade 1–4). Leukocytopenia was mainly CTC grade 1–3, and thrombocytopenia CTC grade 1 and 2 (Table 4). The baseline levels of WBCs, granulocytes, and platelets decreased during later courses resulting in more pronounced leukocytopenia, neutropenia, and thrombocytopenia during those courses. Neutropenic fever was observed in 2 patients with granulocytopenia grade 4 after course one in 1 patient (time paclitaxel concentrations >0.1 μmol/liter = 25.6 h with dose 175 mg/m2) and after course two in the other (time paclitaxel concentrations >0.1 μmol/liter = 17.5 h with dose 175 mg/m2), necessitating hospitalization and administration of i.v. antibiotics as well as dose reduction in subsequent courses. A relationship between hematological toxicity, in terms of percentage decrease in cell counts, and the time period that the paclitaxel concentration in plasma is >0.1 μmol/liter (7, 8) could not be confirmed in the 25 patients we studied when taking into account first-course data as well as all data (data not shown).

Nonhematological toxicity consisted mainly of alopecia CTC grade 1 and 2 after the first course, as well as neurotoxicity, reversible muscular, and joint pain CTC grade 1 and 2 after the first course. Furthermore, in most patients, gastrointestinal complaints CTC grade 1 and 2 were observed after the first course, and were mainly nausea and vomiting. Some patients experienced itching and fatigue CTC grade 1 after the first or after subsequent courses, and one patient developed skin erythema CTC grade 1 after the first course.

Both hematological and nonhematological toxicity appeared to be similar in courses with dose increment compared with courses without dose increment.

Response.

In 1 patient the response was not evaluable. The observed tumor response in 24 patients were a confirmed partial response in 5 patients, stable disease in 4 patients, and progressive disease in 15 patients. The objective response rate was 20%. All 5 of the patients showing a partial response completed six courses of chemotherapy, and all but 1 had exposures above the defined target in courses two to five (the 1 with exposures below the target after individualized doses was a patient with BSA >2 m2 as depicted in Fig. 4). Three patients presenting with metastatic (stage IV) disease died early due to progression, and 22 patients are still alive (at 3–28 months after the start of the first course).

The main purpose of this study was to demonstrate the feasibility and the performance of a Bayesian pharmacokinetically guided dosing strategy of paclitaxel to achieve drug concentrations in plasma >0.1 μmol/liter for ≥15 h. The target exposure was selected based on a study with both paclitaxel and carboplatin resulting in favorable survival figures for patients with paclitaxel exposures above the target (6). No relevant upper limit for the time above the threshold could be defined, and therefore more empirical guidelines for dose reductions in case of toxicity were used. To be able to generate Bayesian predictions for individuals, a population pharmacokinetic model for i.v. administered paclitaxel was applied. This population model, together with observed individual paclitaxel plasma concentrations, was used to predict the time period that the paclitaxel concentration in plasma would exceed 0.1 μmol/liter in the subsequent course for an individual.

The population pharmacokinetic model used in this study was developed and validated in our institute, and based on an earlier published model (20). It describes nonlinear distribution of paclitaxel from the central compartment to the peripheral tissues (Fig. 1), based on measured total plasma paclitaxel concentrations. This nonlinear distribution has been attributed to the presence of Cremophor EL, a soap-like solvent in which the poorly soluble paclitaxel is formulated. Circulating Cremophor EL micelles entrap paclitaxel in the central compartment. This results in a more than proportional increase in plasma paclitaxel levels with increasing doses (and concomitant increasing amount of Cremophor EL), but lower free paclitaxel fraction available for cellular partitioning (21, 22, 23, 24, 25). Both our population model as well as the defined pharmacokinetic-pharmacodynamic relationships of paclitaxel (Refs. 6, 7, 8; on which this study was based) are based on measured total paclitaxel plasma levels, including “free” paclitaxel, and paclitaxel bound to both Cremophor EL and plasma proteins. Because only the fraction of free paclitaxel is available for (tumor) tissue distribution, relationships between the unbound drug concentration and hematological toxicity have also been studied (26, 27). Whether stronger pharmacokinetic-pharmacodynamic relationships can be found when free paclitaxel is monitored should be subject of future studies.

The results of this study show that dose individualization results in a higher percentage of patients with a time period that the paclitaxel concentration of 0.1 μmol/liter was achieved for ≥15 h compared with standard dosing of the compound. After administration of the standard dose of 175 mg/m2 (n = 25), 36% of the courses (n = 9) did not result in a time period above the threshold >15 h, whereas after individualizing the doses (n = 67) this was 19% (n = 13; Table 3). From Fig. 4 it becomes clear that 7 of 13 individualized courses in which the exposure was too low after dose individualization are from 2 individuals. In these patients no dose increment was indicated according to the Bayesian predictions of the paclitaxel concentrations in plasma, whereas the time period that the paclitaxel concentration was >0.1 μmol/liter was actually <15 h during each course. It may be noteworthy that both patients had a BSA exceeding 2 m2 (2.1 and 2.25 m2, respectively). In the original patient population used for the development of the population pharmacokinetic model of paclitaxel, in which BSA was included as a covariate, only a few patients had a BSA value >2 m2. Pharmacokinetics of paclitaxel may deviate in patients with high BSA. Because Bayesian fitting initially assumes that each patient is a typical member of the population (defined by the prior values), the applied model cannot predict the pharmacokinetics in patients with BSA >2 m2 adequately. Individualized courses of other patients with concentrations >0.1 μmol/liter <15 h (9%) can be attributed to the interoccasion (course-to-course) variability of the pharmacokinetic parameters (15–39%).

Toxicities encountered during the courses were mainly hematological (primarily granulocytopenia; Table 4) and comparable with those observed in studies with similar chemotherapy regimens (6, 7, 8). Patients with an increased individualized dose appeared to have similar toxic events compared with the patient group receiving unchanged doses. However, because the study protocol did not provide pharmacokinetically guided dose reductions, it was expected that patients receiving an unchanged individualized dose (because paclitaxel concentrations were >0.1 μmol/liter for >15 h already) had a greater risk of toxicity than patients with increased individualized dose (who initially had exposures that were too low, but that after individualization obtained exposures around the target). This is suggested by the observation that delayed administration of a subsequent course was only necessary in 1 patient with an increased individualized dose, whereas 4 patients with an individualized dose of 175 mg/m2 had to receive their doses a week after the intended date.

No significant positive relationship was found between exposure to paclitaxel and hematological toxicity, in contrast with previous studies (7, 8). An explanation for this discrepancy is that pharmacokinetic variation between patients is decreased due to the dosing strategy applied in this study. Therefore, more uniform exposures are reached, and as a result variation in hematological toxicity is decreased. Elucidation of existing relationships may be hampered by the limited range of paclitaxel exposures as well as the coadministration of carboplatin, which itself is myelotoxic.

Table 4 shows that the severity of the bone marrow toxicity, especially granulocytopenia, was more pronounced during later courses. This can be attributed to decrease of baseline levels of WBCs, granulocytes, and platelets during later courses. This cumulative effect has already been observed in an earlier study in patients with NSCLC treated with the combination of paclitaxel and carboplatin (6).

We were able to report the paclitaxel concentrations in plasma during each course and to calculate an individualized paclitaxel dose before the start of the subsequent course. This technical feasibility was mainly due to the short lines between the clinic and laboratory, and a robust analysis method for the quantification of paclitaxel. For the patient, however, the drug monitoring was an extra burden, because the patient needed to be attended to the clinic for a full day and night or return multiple times to the hospital at the indicated sampling times. When patients benefit substantially from the adaptive dosing approach, the inconvenience will be acceptable. From this study no conclusions can be drawn as to whether patients had a prolonged survival due to the dose individualizations. This should be investigated in a significantly powered randomized study comparing standard doses with adaptive pharmacokinetically guided doses.

In conclusion, the results of this study show that Bayesian pharmacokinetically guided dosing of paclitaxel is feasible in achieving a predefined exposure duration above a threshold plasma concentration. With this strategy, patients can be identified who may benefit from higher doses than the standard dose. Furthermore, the dosing strategy appeared technically and logistically feasible. The basis for the execution of future trials should be to test whether our adaptive dosing strategy results in increased response rate, time to progression, or survival.

Grant support: The Dutch Cancer Society (project NKI 2001-2420) and Bristol Myers Squibb BV, Woerden, the Netherlands.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Requests for reprints: M. E. de Jonge, Department of Pharmacy and Pharmacology, The Netherlands Cancer Institute, Slotervaart Hospital, Louwesweg 6, 1066 EC, Amsterdam, the Netherlands. Phone: 31205124657; Fax: 31205124753; E-mail: apmil@slz.nl

Fig. 1.

Graphical representation of the population pharmacokinetic model of i.v. administered paclitaxel.

Fig. 1.

Graphical representation of the population pharmacokinetic model of i.v. administered paclitaxel.

Close modal
Fig. 2.

Plasma concentration-time curves of paclitaxel of all first courses (n = 25). The dashed line represents the threshold paclitaxel plasma concentration of 0.1 μmol/liter.

Fig. 2.

Plasma concentration-time curves of paclitaxel of all first courses (n = 25). The dashed line represents the threshold paclitaxel plasma concentration of 0.1 μmol/liter.

Close modal
Fig. 3.

Typical patient benefiting from pharmacokinetically guided dosing during course 2–6. After a standard dose of 175 mg/m2 (-♦-) paclitaxel in course 1, the time period the paclitaxel concentration was above 0.1 μmol/liter was only 11.6 h in this patient. After individualized paclitaxel doses of 215 mg/m2 (-•-) in courses 2–6, the time period above 0.1 μmol/liter ranged from 15.07 to 16.8 h.

Fig. 3.

Typical patient benefiting from pharmacokinetically guided dosing during course 2–6. After a standard dose of 175 mg/m2 (-♦-) paclitaxel in course 1, the time period the paclitaxel concentration was above 0.1 μmol/liter was only 11.6 h in this patient. After individualized paclitaxel doses of 215 mg/m2 (-•-) in courses 2–6, the time period above 0.1 μmol/liter ranged from 15.07 to 16.8 h.

Close modal
Fig. 4.

The time period that paclitaxel plasma concentrations were >0.1 μmol/liter in patients during course 1 after administration of the standard dose (175 mg/m2) and after the administration of the individualized doses (175 mg/m2 or higher) during subsequent courses. -□- and -×- are patients with body surface area >2 m2.

Fig. 4.

The time period that paclitaxel plasma concentrations were >0.1 μmol/liter in patients during course 1 after administration of the standard dose (175 mg/m2) and after the administration of the individualized doses (175 mg/m2 or higher) during subsequent courses. -□- and -×- are patients with body surface area >2 m2.

Close modal
Table 1

Population pharmacokinetic parameter estimates of i.v.-administered paclitaxel

EstimateRSEa (in %)
Pharmacokinetic parameter   
 V (L) 12.1 5.4 
  Influence of BSA (θ12)b 1.16 33 
 Tmax (μmol/h) 216 5.6 
  Influence of BSA (θ13)c 1.05 49 
 Tm (μmol/liter) 2.59 10 
 k21 (h−10.916 7.0 
 k13 (h−11.63 8.7 
 k31 (h−10.0722 7.9 
 Vmax (μmol/h) 47.0 10 
 Km (μmol/liter) 0.656 15 
 Population 1–2 0.380 27 
Interindividual variability   
 IIV-V (%) 20 31 
 IIV-Tmax (%) 26 27 
 IIV-Vmax (%) 22 33 
Interoccasion variability   
 IOV-k21 (%) 15 48 
 IOV-k13 (%) 26 25 
 IOV-k31 (%) 39 30 
 IOV-Vmax (%) 18 32 
Residual error   
 Additive error 1 (μmol/liter) 0.202 11 
 Additive error 2 (μmol/liter) 0.105 7.6 
EstimateRSEa (in %)
Pharmacokinetic parameter   
 V (L) 12.1 5.4 
  Influence of BSA (θ12)b 1.16 33 
 Tmax (μmol/h) 216 5.6 
  Influence of BSA (θ13)c 1.05 49 
 Tm (μmol/liter) 2.59 10 
 k21 (h−10.916 7.0 
 k13 (h−11.63 8.7 
 k31 (h−10.0722 7.9 
 Vmax (μmol/h) 47.0 10 
 Km (μmol/liter) 0.656 15 
 Population 1–2 0.380 27 
Interindividual variability   
 IIV-V (%) 20 31 
 IIV-Tmax (%) 26 27 
 IIV-Vmax (%) 22 33 
Interoccasion variability   
 IOV-k21 (%) 15 48 
 IOV-k13 (%) 26 25 
 IOV-k31 (%) 39 30 
 IOV-Vmax (%) 18 32 
Residual error   
 Additive error 1 (μmol/liter) 0.202 11 
 Additive error 2 (μmol/liter) 0.105 7.6 
a

RSE, relative standard error; additive error 1, residual variability of subpopulation 1; additive error 2, residual variability of subpopulation 2; BSA, body surface area; IIV, interindividual variability; IOV, interoccasion variability; k21, rate constant from the first peripheral compartment to the central compartment; k13, rate constant from the central to the second peripheral compartment; k31, rate constant from the second peripheral to the central compartment; Km, plasma concentration at half Vmax; Population 1–2, fraction of the patient population (n = 61) belonging to subpopulation 1; Tmax, maximal transport rate from the central to the first peripheral compartment; Tm, plasma concentration at half Tmax; V, volume of the central compartment; Vmax, maximal elimination rate.

b

V = θ1 * (BSA/1.81)θ12.

c

Tmax = θ2 * (BSA/1.81)θ13 where BSA is expressed in m2, θ1 and θ2 represent the population value of V and Tmax, respectively, of a patient with a (median) BSA or 1.81 m2, θ12 and θ13 are exponential factors.

Table 2

Baseline characteristics of the patients

NumberMedian value (range)
Patients  25  
 Female   
 Male  16  
Age   58 (40–72) 
BSA (m2)a   1.9 (1.55–2.25) 
WHO Performance status    
 0   
 1  17  
 2   
Tumor stage    
 IIIB   
 IV  20  
Histologic type    
 Squamous cell carcinoma   
 Adenocarcinoma  11  
 Large cell carcinoma   
Pretreatment    
 None  16  
 Prior chemotherapy   
 Radiotherapy   
 Surgical therapy   
Serum biochemical parameters Normal range   
 Creatinine (μmol/liter) (40–105)  77 (46–124) 
 Total bilirubin (μmol/liter) (<16)  6 (3–15) 
 Albumin (g/liter) (35–50)  40 (31–45) 
 Alkaline phosphatase (unit/liter) (40–120)  104 (49–364) 
 Aspartate amino transferase (unit/liter) (<40)  21 (12–54) 
 Alanine amino transferase (unit/liter) (<45)  20 (7–198) 
 Gamma glutamyl transpeptidase (unit/liter) (<50)  47 (9–252) 
 Lactate dehydrogenase (unit/liter) (<450)  374 (121–859) 
NumberMedian value (range)
Patients  25  
 Female   
 Male  16  
Age   58 (40–72) 
BSA (m2)a   1.9 (1.55–2.25) 
WHO Performance status    
 0   
 1  17  
 2   
Tumor stage    
 IIIB   
 IV  20  
Histologic type    
 Squamous cell carcinoma   
 Adenocarcinoma  11  
 Large cell carcinoma   
Pretreatment    
 None  16  
 Prior chemotherapy   
 Radiotherapy   
 Surgical therapy   
Serum biochemical parameters Normal range   
 Creatinine (μmol/liter) (40–105)  77 (46–124) 
 Total bilirubin (μmol/liter) (<16)  6 (3–15) 
 Albumin (g/liter) (35–50)  40 (31–45) 
 Alkaline phosphatase (unit/liter) (40–120)  104 (49–364) 
 Aspartate amino transferase (unit/liter) (<40)  21 (12–54) 
 Alanine amino transferase (unit/liter) (<45)  20 (7–198) 
 Gamma glutamyl transpeptidase (unit/liter) (<50)  47 (9–252) 
 Lactate dehydrogenase (unit/liter) (<450)  374 (121–859) 
a

BSA, body surface area.

Table 3

Time-period of paclitaxel concentrations in plasma >0.1 μmol/liter during each course

Course numberNumber of patientsMedian time-period of concentrations >0.1 μmol/liter (h) (range)Percentage of patients with concentrations >0.1 μmol/liter for <15 h (%)
25a 16.3 (7.6–31.6) 36 
22 18.0 (13.6–24.8) 23 
14a 16.3 (11.2–21.3) 14 
13a 16.3 (12.4–17.9) 23 
9b 16.0 (12.2–19.5) 11 
9b 16.1 (10.9–19.6) 11 
Course numberNumber of patientsMedian time-period of concentrations >0.1 μmol/liter (h) (range)Percentage of patients with concentrations >0.1 μmol/liter for <15 h (%)
25a 16.3 (7.6–31.6) 36 
22 18.0 (13.6–24.8) 23 
14a 16.3 (11.2–21.3) 14 
13a 16.3 (12.4–17.9) 23 
9b 16.0 (12.2–19.5) 11 
9b 16.1 (10.9–19.6) 11 
a

Including two patients with body surface area >2 m2.

b

Including one patient with body surface area >2 m2.

Table 4

Hematological toxicity during all administered courses (n = 103)

ToxicityCourse #1 (n = 25)Course #2 (n = 24)a,bCourse #3 (n = 16)cCourse #4 (n = 16)cCourse #5 (n = 11)dCourse #6 (n = 11)d
Leukocytes       
 Grade 1 6 (24%) 7 (29%) 6 (38%) 2 (13%) 0 (0%) 2 (18%) 
 Grade 2 5 (20%) 3 (13%) 6 (38%) 6 (38%) 5 (45%) 3 (27%) 
 Grade 3 4 (16%) 6 (25%) 3 (19%) 5 (31%) 4 (36%) 4 (36%) 
 Grade 4 1 (4%) 1 (4%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 
Granulocytes       
 Grade 1 4 (16%) 2 (8%) 1 (6%) 1 (6%) 1 (9%) 0 (0%) 
 Grade 2 5 (20%) 4 (17%) 3 (19%) 2 (13%) 1 (9%) 3 (27%) 
 Grade 3 6 (24%) 6 (25%) 7 (44%) 7 (44%) 3 (27%) 5 (45%) 
 Grade 4 3 (12%) 6 (25%) 4 (25%) 4 (25%) 6 (55%) 3 (27%) 
Platelets       
 Grade 1 5 (20%) 6 (25%) 6 (38%) 8 (50%) 4 (36%) 5 (45%) 
 Grade 2 0 (0%) 2 (8%) 1 (6%) 2 (13%) 2 (18%) 4 (36%) 
 Grade 3 0 (0%) 1 (4%) 0 (0%) 1 (6%) 1 (9%) 0 (0%) 
 Grade 4 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 
ToxicityCourse #1 (n = 25)Course #2 (n = 24)a,bCourse #3 (n = 16)cCourse #4 (n = 16)cCourse #5 (n = 11)dCourse #6 (n = 11)d
Leukocytes       
 Grade 1 6 (24%) 7 (29%) 6 (38%) 2 (13%) 0 (0%) 2 (18%) 
 Grade 2 5 (20%) 3 (13%) 6 (38%) 6 (38%) 5 (45%) 3 (27%) 
 Grade 3 4 (16%) 6 (25%) 3 (19%) 5 (31%) 4 (36%) 4 (36%) 
 Grade 4 1 (4%) 1 (4%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 
Granulocytes       
 Grade 1 4 (16%) 2 (8%) 1 (6%) 1 (6%) 1 (9%) 0 (0%) 
 Grade 2 5 (20%) 4 (17%) 3 (19%) 2 (13%) 1 (9%) 3 (27%) 
 Grade 3 6 (24%) 6 (25%) 7 (44%) 7 (44%) 3 (27%) 5 (45%) 
 Grade 4 3 (12%) 6 (25%) 4 (25%) 4 (25%) 6 (55%) 3 (27%) 
Platelets       
 Grade 1 5 (20%) 6 (25%) 6 (38%) 8 (50%) 4 (36%) 5 (45%) 
 Grade 2 0 (0%) 2 (8%) 1 (6%) 2 (13%) 2 (18%) 4 (36%) 
 Grade 3 0 (0%) 1 (4%) 0 (0%) 1 (6%) 1 (9%) 0 (0%) 
 Grade 4 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 
a

Toxicity was only evaluable in 23 of the 24 administered second courses since 1 patient died 2 days after the administration of the second course.

b

One course was administered at a reduced dose.

c

Three courses were administered at reduced doses.

d

Two courses were administered at reduced doses.

We thank Rianne Van Maanen and Ciska Koopman-Kroon for analyses of the paclitaxel samples.

1
Huizing MT, Sewberath Misser VH, Pieters RC, et al Taxanes: a new class of antitumor agents.
Cancer Investig
,
13
:
381
-404,  
1995
.
2
Rowinsky EK, Donehower RC. Paclitaxel (Taxol).
N Engl J Med
,
332
:
1004
-14,  
1995
.
3
Eisenhauer EA, Ten Bokkel Huinink WW, Swenerton KD, et al European-Canadian randomized trial of paclitaxel in relapsed ovarian cancer: high-dose versus low-dose and long versus short infusion.
J Clin Oncol
,
12
:
2654
-66,  
1994
.
4
Novello S, Le Chevalier T. European perspectives on paclitaxel/platinum-based therapy for advanced non-small cell lung cancer.
Semin Oncol
,
28
:
3
-9,  
2001
.
5
Schiller JH, Harrington D, Belani CP, et al Comparison of four chemotherapy regimens for advanced non-small-cell lung cancer.
N Engl J Med
,
346
:
92
-8,  
2002
.
6
Huizing MT, Giaccone G, Van Warmerdam LJC, et al Pharmacokinetics of paclitaxel and carboplatin in a dose-escalating and sequencing study in patients with non-small-cell lung cancer.
J Clin Oncol
,
15
:
317
-29,  
1997
.
7
Huizing MT, Keung ACF, Rosing H, et al Pharmacokinetics of paclitaxel and metabolites in a randomised comparative study in platinum-pretreated ovarian cancer patients.
J Clin Oncol
,
11
:
2127
-35,  
1993
.
8
Gianni L, Kearns CM, Gianni A, et al Nonlinear pharmacokinetics and metabolism of paclitaxel and its pharmacokinetic/pharmacodynamic relationships in humans.
J Clin Oncol
,
13
:
180
-90,  
1995
.
9
Van Den Bongard HJGD. Development and validation of a population pharmacokinetic model of paclitaxel in cancer patients.
Clinical pharmacokinetics of anticancer agents. Focus on E7070, methotrexate and paclitaxel
, Thesis University Utrecht Utrecht, the Netherlands ISBN 90-6734-045-6. 
2002
.
10
De Jonge ME, Huitema ADR, Rodenhuis S, and Beijnen JH. Individualized cancer chemotherapy: strategies and performance of prospective studies on therapeutic drug monitoring with dose adaptation (Review). Clin Pharmacokinet. In press 2004.
11
Calvert AH, Newell DR, Gumbrell LA, et al Carboplatin dosage: prospective evaluation of a simple formula based on renal function.
J Clin Oncol
,
7
:
1748
-56,  
1989
.
12
Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine.
Nephron
,
16
:
31
-41,  
1976
.
13
Leger F, Seronie-Vivien S, Makdessi J, et al Impact of the biochemical assay for serum creatinine measurement on the individual carboplatin dosing: a prospective study.
Eur J Cancer
,
38
:
52
-6,  
2002
.
14
Willey TA, Bekos EJ, Gaver RC, et al High-performance liquid chromatographic procedure for the quantitative determination of paclitaxel (Taxol) in human plasma.
J Chromatogr
,
621
:
231
-8,  
1993
.
15
Huizing MT, Sparreboom A, Rosing H, Van Tellingen O, Pinedo HM, Beijnen JH. Quantification of paclitaxel metabolites in human plasma by high-performance liquid chromatography.
J Chromatogr B
,
674
:
261
-8,  
1995
.
16
Van Warmerdam LJC, Van Tellingen O, Maes RAA, Beijnen JH. Validated method for the determination of carboplatin in biological fluids by Zeeman atomic absorption spectrometry.
Fresenius J Anal Chem
,
351
:
1820
-4,  
1995
.
17
Beal SL, Sheiner LB. .
NONMEM Users Guides, NONMEM Project Group
, University of California at San Francisco San Francisco  
1998
.
18
.
Guidelines for Reporting of Adverse Drug Reactions
, National Cancer Institute, Division of Cancer Treatment Bethesda, Maryland  
1988
.
19
Therasse P, Arbuck SG, Eisenhauer EA, et al New guidelines to evaluate the response to treatment in solid tumors.
J Natl Cancer Inst (Bethesda)
,
92
:
205
-16,  
2000
.
20
Karlsson MO, Molnar V, Freijs A, Nygren P, Bergh J, Larsson R. Pharmacokinetic models for the saturable distribution of paclitaxel.
Drug Metab Dispos
,
37
:
266
-70,  
1999
.
21
Van Tellingen O, Huizing MT, Nannan Panday VR, Schellens JHM, Nooijen WJ, Beijnen JH. Cremophor EL causes (pseudo-) non-linear pharmacokinetics of paclitaxel in patients.
Br J Cancer
,
81
:
330
-5,  
1999
.
22
Sparreboom A, Van Tellingen O, Nooijen WJ, Beijnen JH. Nonlinear pharmacokinetics of paclitaxel in mice results from the pharmaceutical vehicle Cremophor EL.
Cancer Res
,
56
:
2112
-5,  
1996
.
23
Sparreboom A, Van Zuylen L, Brouwer E, et al Cremophor EL-mediated alteration of paclitaxel distribution in human blood: clinical pharmacokinetic implications.
Cancer Res
,
59
:
1454
-7,  
1999
.
24
Van Zuylen L, Karlsson MO, Verweij J, et al Pharmacokinetic modeling of paclitaxel encapsulation in Cremophor EL micelles.
Cancer Chemother Pharmacol
,
47
:
309
-18,  
2001
.
25
Van Zuylen L, Gianni L, Verweij J, et al Inter-relationships of paclitaxel disposition, infusion duration and Cremophor EL kinetics in cancer patients.
Anti-Cancer Drugs
,
11
:
331
-7,  
2000
.
26
Henningsson A, Karlsson MO, Viganò L, Gianni L, Verweij J, Sparreboom A. Mechanism-based pharmacokinetic model for paclitaxel.
J Clin Oncol
,
19
:
4065
-73,  
2001
.
27
Henningsson A, Sparreboom A, Sandström M, et al Population pharmacokinetic modeling of unbound and total plasma concentrations of paclitaxel in cancer patients.
Eur J Cancer
,
39
:
1105
-14,  
2003
.