180 pacli y filtro

Nonlinear Pharmacokinetics and Metabolism ofPaclitaxel and Its Pharmacokinetic/Pharmacodynamic

Relationships in Humans

By Luca Gianni, Christine M. Kearns, Antonio Giani, Giuseppe Capri, Lucia Vigan6, Alberta Locatelli,Gianni Bonadonna, and Merrill J. Egorin

Purpose: To characterize and model the dispositionof paclitaxel in humans and define a pharmacodynamicrelationships between paclitaxel disposition and its tox-icity and efficacy.

Patients and Methods: Paclitaxel pharmacokineticswere studied in 55 courses of therapy in 30 patients.Paclitaxel was administered at 135 mg/M 2 or 175 mg/m 2 by either a 3- or a 24-hour infusion schedule to pa-tients with advanced ovarian cancer (n = 15), or at 225mg/m2 by 3-hour infusion to patients with advancedbreast cancer (n = 15). Paclitaxel and 6a-hydroxylpacli-taxel were quantified by high-performance liquid chro-matography (HPLC). Pharmacokinetics were assessedby noncompartmental and model-dependent methods.Pharmacodynamic correlations were evaluated statisti-cally and by regression models.

Results: Paclitaxel disposition is nonlinear in humansan nthe 3-hour schedule, 6a-hydroxylpaclitaxel wasidentified in the plasma of all patients treated. Theplasma disposition of paclitaxel and 6a-hydroxylpacli-taxel was well described by a model that featured multi-

RECENTLY, several studies have evaluated the ad-ministration of paclitaxel by 3-hour infusion. These

trials have generated unexpected clinical observations thatare at variance with those in earlier studies that used 24-hour infusions. The most intriguing finding is that identi-cal doses of paclitaxel were markedly less myelosuppres-sive when delivered by 3-hour than by 24-hour infusion.1

Also, the description and model of the pharmacokineticsof paclitaxel derived from a large number of early clinicalinvestigations 2

7 did not adequately describe the disposi-tion of the drug when administered over 3 hours.

From the Division of Medical Oncology, Laboratory of ClinicalPharmacology, Istituto Nazionale per lo Studio e la Cura dei Tumori,Milano, Italy; and Division of Developmental Therapeutics, Univer-sity of Maryland Cancer Center, Baltimore, MD.

Submitted March 7, 1994; accepted August 31, 1994.Supported in part by grants no. CNR 92.02330.PF39, CNR

93.02309.PF39 from the Italian Research Council, andUIO.CA44691 from the National Cancer Institute, Department ofHealth and Human Services, Bethesda, MD, and by grants fromAssociazione Italiana Ricerca sul Canero, Italy, and Bristol-MyersSquibb, Princeton, NJ.

Address reprint requests to Luca Gianni, MD, Division of MedicalOncology, Laboratory of Clinical Pharmacology, Istituto Nazionaleper lo Studio e la Cura dei Tumori, Via Venezian, 1 20133-Milano,Italy.

© 1995 by American Society of Clinical Oncology.0732 -183X/95/1301-0025$3.00/0

pie nonlinear processes. Neutropenia was not related tothe areas under the curves (AUCs) of paclitaxel or 6a-hydroxylpaclitaxel, or to palitaxel peak concentrations(C.,ax). Neutropenia was related to the duration thatplasma concentrations were : 0.05 /mol/L, a relation-ship that is well described by a sigmoid maximum re-sponse (Ema,) model.

Conclusion: The disposition of paclitaxel in humansis nonlinear. Paclitaxel metabolism to 6a-hydroxylpacli-taxel is likely an important detoxification pathway. My-elosuppression is related to the duration that plasmapaclitaxel concentrations are : 0.05 jtmol/L. Trials ofnew doses and schedules of paclitaxel should take intoaccount its nonlinear disposition to rule out adverse clini-cal consequences, especially if the drug is administeredby short infusion. Our pharmacokinetic model shouldprove to be a powerful tool in predicting paclitaxel dis-position, regardless of dose and schedule, and shouldfacilitate further pharmacodynamic investigations.

J Clin Oncol 13:180-190. C 1995 by American So-ciety of Clinical Oncology.

The comparative safety of short (3-hour) versus longer(24-hour) infusions of paclitaxel with premedication wasrecently addressed by a multicenter European-Canadianstudy.' This study also explored the dose-response rela-tionship of paclitaxel in patients with ovarian cancer whohad relapsed after prior therapy with platinum-containingregimens. In the bifactorial design of the study, patientswere randomized to receive either 135 or 175 mg/m2 ofpaclitaxel by either 3- or 24-hour intravenous (IV) infu-sion. This bifactorial design provided the unique opportu-nity to evaluate the pharmacokinetics of paclitaxel admin-istered at two different dosages and by two differentschedules to a group of homogeneously pretreated pa-tients. Accordingly, we studied the pharmacokinetics ofpaclitaxel in 15 patients accrued to the trial at the IstitutoNazionale Tumori in Milan.

The Istituto Nazionale Tumori was also concurrentlyengaged in a phase II trial of paclitaxel in women withbreast cancer who had relapsed after receiving anthracy-cline-containing chemotherapy regimens.8 In this study,225 mg/m2 of paclitaxel was administered as a 3-hour IVinfusion. We studied the pharmacokinetics of paclitaxelin 15 women enrolled onto this study.

From our analysis of these studies, we now describethe nonlinear disposition of paclitaxel in these patients;describe a metabolite of paclitaxel, ostensibly 6a-hydroxylpaclitaxel, that is measurable in plasma; and de-

Journal of Clinical Oncology, Vol 13, No 1 (January), 1995: pp 180-190180

Downloaded from jco.ascopubs.org by HENK-JAN GUCHELAAR on October 11, 2010 from 186.136.221.124Copyright © 1995 American Society of Clinical Oncology. All rights reserved.

PACLITAXEL PHARMACOKINETICS AND METABOLISM 181

Table 1. Noncompartmental Analysis of Paclitaxel Pharmacokinetics

Dose Infusion No. of No. of Cax AUC Apparent CIT Apparent tl/2 Urinary Excretionmg/m2 (hours) Patients Cycles (pmol/L) (pM-h) (L/h/m

2) (hours) (%)

135 3 4 12 3.3 - 0.4 10.9 ± 1.1 14.8 - 1.4 9.2 - 4.2 3.5 ± 1.4

175 3 3 8 5.9 - 0.9 18.5 3.0 11.4 - 1.8 6.5 - 3.4 2.1 0.5

225 3 15 15 7.6 + 1.9 24.3 - 6.8 11.6 + 3.10 7.4 - 2.0 -

135 24 4 9 0.3 + 0.1 12.4 - 2.2 13.9 ± 3.5 16.2 - 6.8 2.2 + 1.3

175 24 4 11 0.5 + 0.1 16.0 - 4.4 13.9 ± 4.2 14.6 ± 8.9 4.4 - 1.6

Total 30 55

scribe relationships between the pharmacokinetics ofpaclitaxel and the dose-limiting neutropenia associatedwith its use.

PATIENTS AND METHODS

The pharmacokinetics of paclitaxel were studied in 55 cycles oftherapy administered to 30 patients (Table 1). Fifteen patients hada diagnosis of relapsed or refractory ovarian cancer, and 15 patientshad a diagnosis of relapsed or refractory breast cancer. Both groupsof patients were of comparable age and performance status. Themedian age of all patients was 54 years (range, 45 to 69), and themedian Eastern Cooperative Oncology Group performance statuswas 0 (range, 0 to 2). No patient had evidence of major alterationsof hepatic, renal, or cardiac function at the time of study. All patientswith ovarian cancer had received at least one and a maximum oftwo prior chemotherapy regimens that contained cisplatin and/orcarboplatin. All breast cancer patients had received at least oneand no more than two prior chemotherapy regimens that containeddoxorubicin or an equivalent anthracycline.

Ampules that contained 30 mg of paclitaxel formulated in Cremo-phor EL:ethanol (1:1 vol/vol) were provided by Bristol-MyersSquibb (Wallingford, CT). For patient administration, paclitaxel wasdiluted in 5% dextrose solution to a final concentration 5 0.6 mg/mL. The drug was infused via a Life Care Model 4 volumetricpump (Abbott Laboratories, N Chicago, IL) into a large peripheralor central vein. For women who had undergone mastectomy, pacli-taxel was not infused in the arm on the mastectomy side. Becausea small number of fibers (within acceptable limits of the UnitedStates Pharmacopeia particulate matter test for large-volume paren-teral solutions) have been observed in paclitaxel solutions, in-linefiltration was mandated. All solutions were administered through0.22-pm pore-size cellulose acetate filters (IVEX II; Millipore, Mols-heim, France).

Premedication was uniform for all patients and consisted of thefollowing: (1) prednisone (25 mg orally) 12 hours before infusion,(2) chlorphenamine (10 mg intramuscularly [IM]) 1 hour beforeinfusion, (3) hydrocortisone (250 mg IV) 30 minutes before the startof the paclitaxel infusion, and (4) cimetidine (300 mg IV) 30 minutesbefore the start of the paclitaxel infusion.

Pharmacokinetic Study Design

Evaluation of paclitaxel pharmacokinetics was planned for thefirst three cycles of therapy for ovarian cancer patients and for onlythe first cycle of therapy for breast cancer patients. Because patientsin the ovarian cancer study were randomized by each participatingcenter, patients were evenly distributed among the four arms of thatstudy (Table 1).

In each patient, sufficient plasma was obtained before paclitaxel

administration for preparation of individual standard curves andevaluation of possible interfering peaks in the high-performanceliquid chromatographs (HPLC). Blood samples for analysis of pacli-taxel in patients who received 24-hour infusions were obtained atthe following times: before infusion, 1 hour, 22 hours, 23 hours,and 23 hours 55 minutes during infusion; and 5 minutes, 15 minutes,30 minutes, 1 hour, 2 hours, 3 hours, 6 hours, 12 hours, and 21hours postinfusion. For patients who received paclitaxel by 3-hourinfusion, blood samples were collected at the following times: beforeinfusion; 1 hour, 2 hours, and 2 hours 55 minutes during infusion;and 5 minutes, 15 minutes, 30 minutes, 1 hour, 2 hours, 3 hours, 6hours, 12 hours, and 21 hours postinfusion. Seven patients from thebreast cancer study group had additional samples obtained at 10minutes and 1.5 hours during the infusion. All blood samples weredrawn from a vein in the arm opposite to that used for paclitaxelinfusion. Samples were collected in tubes that contained potassiumedetic acid. Plasma was immediately separated by centrifugation at1,000 x g for 15 minutes at 4°C and stored in polypropylene vialsat -20 0C until analysis.

For patients with ovarian cancer, urine was collected from the startof the infusion through 21 hours postinfusion. Whenever possible,samples of the infusate were collected before and after passagethrough the IVEX II filter so that actual drug delivery could beassessed.

HPLC Determination of Paclitaxel and

6a-Hydroxylpaclitaxel

Paclitaxel (NSC 125973) and cephalomannine were provided bythe Pharmaceutical Resources Branch of the National Cancer Insti-tute, Bethesda, MD. Pure 6a-hydroxylpaclitaxel was a gift of JamesHarris of the Division of Clinical Pharmacology, Food and DrugAdministration, Rockville, MD.9

Paclitaxel in plasma and urine was measured by a previouslydescribed HPLC method 0o with modification as described later. Forsimultaneous quantitation of both paclitaxel and 6a-hydroxylpacli-taxel, the HPLC methodology was further modified as described.

A stock solution of paclitaxel 1.2 mmol/L was prepared in ethanoland stored at -80'C. Standard curves for the quantitation of pacli-taxel were prepared for each patient using that patient's pretreatmentplasma. Standard curves for paclitaxel initially spanned concentra-tions from 0.0625 to 4.0 gmol/L, but were subsequently expandedto encompass concentrations between 0.005 and 14 pmol/L. A 500-gL aliquot of standard or sample was mixed with 10 pL of 0.065-mmol/L cephalomannine internal standard and applied to a 100-mgBond Elut LRC C18 cartridge (Varian, Harbor City, CA) that hadbeen preconditioned with sequential washings of 2 mL of acetonitrileand 2 mL of distilled water. After the plasma sample had been applied,the cartridges were washed with 3 mL of distilled water, and paclitaxeland internal standard were then eluted from the cartridge with 2


GIANNI ET AL

mL of acetonitrile. The eluant was vacuum-dried on a Speed-Vacconcentrator (Savant Instruments, Inc, Farmingdale, NY). The driedresidues were reconstituted with 150 pL of acetonitrile:distilled water(50:50 vol/vol), and 70 pL of the reconstituted solution was injectedinto the HPLC system. The HPLC system used a Superspher CIs 4-pm, 125- x 4-mm column (Hewlett-Packard, Palo Alto, CA) protectedwith a Lichrosphere RP-18 5-,um, 4- x 4-mm guard column (Merck,Darmstadt, Germany). Samples were isocratically eluted with acetoni-trile:distilled water (50:50 vol/vol) at a rate of 1 mL/min, and columneluate was monitored at 230 nm with a Hewlett-Packard model 1050diode array detector. The detector signal was processed with Hewlett-Packard HPLC Chemstation software implemented on a personal com-puter running under Microsoft Windows 3.0 (Microsoft Corp, Red-mond, WA). Concentrations of paclitaxel were calculated by determin-ing the ratio of paclitaxel signal to the internal standard signal in thatsample and comparison of that ratio with a concomitantly performedstandard curve. The retention times for cephalomannine and paclitaxelwere 5.3 and 6.1 minutes, respectively. There were no materials inpretreatment plasma samples that interfered with the peaks of interestin the HPLC. Recovery of paclitaxel was 95% + 1.9% at 0.1 ymol/L, and recovery of cephalomannine was 90% + 3.7% at 1.3 Aimol/L. Standard curves were fitted by a linear equation with proportionalweighting. Paclitaxel standard curves were linear between 0.005 and14 Mmol/L, with an RZ value greater than .99 in all instances. Theintraassay coefficient of variation (CV%) was 10.6 at 0.0625 ysmol/L and 2.8 at 8 pmol/L (n = 6). The interassay CV% values were10.2 and 9.7 at 0.0625 and 8 pmol/L, respectively.

For samples in which both paclitaxel and 6a-hydroxylpaclitaxelwere quantitated, the analytic procedure was identical to that de-scribed, except that the mobile phase was adjusted to acetonitrile/distilled water (48:52 vol/vol). Separate standard curves for pacli-taxel and 6a-hydroxylpaclitaxel were prepared for each patient'spretreatment plasma level. Paclitaxel standard curves were generatedas described earlier. Standard curves for 6ac-hydroxylpaclitaxelranged between 0.005 and 4.0 uimol/L. Again, concentrations ofpaclitaxel and 6a-hydroxylpaclitaxel in samples were determined bycalculating the ratio of the paclitaxel or 6a-hydroxylpaclitaxel signalto that of the internal standard and then comparing that ratio to theappropriate standard curve. The retention times of 6a-hydroxylpacli-taxel, cephalomannine, and paclitaxel were 4.8, 6.6, and 7.6 minutes,respectively. Recovery of 6a-hydroxylpaclitaxel was 104% + 2.6%at 1.0 ftmol/L. The 6a-hydroxylpaclitaxel standard curve was linearbetween 0.0125 to 3.5 Mmol/L, with an intraassay CV% of 2.5 at 1ymol/L. The interassay CV% values for 6a-hydroxylpaclitaxel were14.8 and 11.5 at 0.025 and 2 ymol/L, respectively. Recovery andassay CV% values for cephalomannine and paclitaxel were essen-tially unchanged from those stated.

A slightly modified procedure was used to determine the concen-tration of paclitaxel in urine and in the infusate. Briefly, sampleswere diluted 1:10 (vol:vol) with mobile phase. External standardcurves that encompassed paclitaxel concentrations between 0.5 and8.0 ymol/L for urine, and between 8.0 and 64.0 pmol/L for infusatemeasurement, were prepared in mobile phase without internal stan-dard, and were consistently linear with an R2 value greater than .98.Identical volumes (70 pL for urine and 10 yL for infusate) of samplesand standards were injected directly into the HPLC system.

Pharmacokinetic and Pharmacodynamic Analysis

The pharmacokinetics of paclitaxel and 6a-hydroxylpaclitaxelwere evaluated by both noncompartmental and model-dependentmethods. For the noncompartmental analysis, total-body clearance(ClB), steady-state volume of distribution (Vdss), area under the

concentration versus time curve from time zero to infinity (AUC),and terminal half-life (t1/2) were estimated using the LaGrange func-tion as implemented by the computer program LAGRAN." Theextrapolated areas accounted for 3.5% + 2.8% for 3-hour infusionsand for 9.3% _ 8.1% of the total AUC for 24-hour infusions. Theapparent clearance was calculated using the following formula: CITB= (Dose/AUC).

For compartmental analysis, pharmacokinetic models were fit toindividual patient's concentration-versus-time data using the IDmodule in the ADAPT II pharmacokinetic software package.' 2 Mod-els were designed to fit paclitaxel and 6a-hydroxylpaclitaxel concen-tration-time data simultaneously. Data on both parent compound andmetabolite were available from the 15 patients with breast cancer.In the 15 ovarian cancer patients, for whom data were available forpaclitaxel only, parameters concerned with metabolism of the parentcompound to 6a-hydroxylpaclitaxel were held constant at the meanvalues calculated from breast cancer patients.

Initially, all data were fit by weighted least-squares regression.Because the total number of concentration-time observations perindividual was relatively small compared with the number of param-eters estimated, refinement of the parameter estimates was soughtthrough an iterative two-stage approach."3 The mean values andvariances for each parameter were calculated and entered as the priorinformation for subsequent Bayesian estimation. For individuals inwhom information was available from multiple courses, pharmacoki-netic estimations from each course were weighted by the reciprocalof the total number of courses studied, so that all patients wereequally represented in the final calculations. This iterative two-stageapproach was repeated, in each cycle updating the Bayesian priors,until the mean estimates of all parameters differed by less than10% from the previous mean estimate, which was our arbitrarilypredefined stopping point.

The goodness of fit of constructed models was assessed by mini-mization of the sums of squares, examination of the residuals forlack of heteroscedasticity,' 4 and dose independence of the estimatedparameters. As the complexity of the model increased, previousassignments of nonlinear processes and additional compartmentswere reassessed as to their necessity.

Relationships were sought between a variety of paclitaxel pharma-cokinetic parameters and the dose-limiting neutropenia associatedwith paclitaxel administration. Pharmacokinetic parameters exploredincluded plasma peak concentration (C,,,x), AUC, mean residencetime, and durations spent above several arbitrarily selected plasmaconcentrations. Because plasma paclitaxel concentrations were mea-sured at intervals too widely spaced to allow precise definition ofduration at or above given thresholds, the time spent at or abovevarious plasma paclitaxel concentrations was determined from con-centration-versus-time profiles generated for each patient by usingthat patient's unique pharmacokinetic parameter estimates and theSIM module in ADAPT II.12 These simulations were designed toreport plasma paclitaxel concentrations in 30-minute incrementsfrom time 0 to 50 or 60 hours postinfusion. Neutropenia was evalu-ated in two ways. Within individual patients, myelosuppression wasdescribed as the continuous variable, percentage reduction in abso-lute neutrophil count (ANC). The pretherapy ANC (ANCpre) wasdetermined on the day of therapy or on the day before therapy.Following treatment with paclitaxel, patient ANCs were obtainedon a weekly or twice-per-week basis. The nadir of patients whowere sampled twice per week was not different from the one thatwould have been measured with a weekly complete blood cell sam-pling. The lowest measured ANC value was used in the equation asthe ANCadir,. Relative neutropenia was calculated as follows: ANC

182


PACLITAXEL PHARMACOKINETICS AND METABOLISM

% change = ([ANCp, - ANCnadi]/[ANCp]) x 100. To avoid poten-tially confounding information due to cumulative toxicity, neutro-penia data from only course 1 of paclitaxel therapy in each patientwas used.

A sigmoid maximum response (EMAX) model was used to describethe relationship between relative neutropenia and duration at orabove a number of defined threshold plasma concentrations of pacli-taxel (0.1, 0.05, and 0.03 pmol/L). A modified form of the Hillequation,1 5

16 as follows was used: E = EMIN ([EMAx - DH]/[DH%+ DH]). Again, the ID module of ADAPT II was used to fit themodel to the data. In this equation, E represents effect, defined asthe percentage reduction in ANC. EMIN is the minimum reductionpossible, which was fixed at a value of 0. The maximum response,or EMAX, was fixed at value of 100, which represented a theoreticmaximum 100% reduction from baseline ANC. D is the durationof time at or above evaluated threshold plasma concentrations ofpaclitaxel, and D50, is the duration of time predicted to result in50% of the maximum response. The Hill constant, H, is a term thatdescribes the sigmoidicity of the curve. Model fits were evaluatedfor goodness of fit by minimization of sums of the squared residualsand by reduction of the estimated CV% for fitted parameters. Thebest fit was that to the data generated with the 0.05-gLmol/L threshold.

In addition, occurrence of grade 3 or 4 neutropenia, assessed as adiscontinuous variable, was evaluated with respect to dose, schedule,peak concentration, AUC, and duration of time at or above specifiedthreshold concentrations. Significance of these relationships was as-sessed by construction of contingency tables with subsequent X2analysis.

RESULTS

Because preparation of paclitaxel for infusion is oftenassociated with fiber formation, in-line filtration was man-datory for all patients described in this study. To rule outthe possibility that filtration caused a significant loss ofthe planned dose, samples of infusate distal and proximalto the filter were analyzed in 32 cycles of therapy adminis-tered at 135 or 175 mg/m 2 by either 3- or 24-hour infusion.At all doses and schedules, the mean postfilter concentra-tions of paclitaxel were 97% ± 10% of the respectiveprefilter concentration. This was within the range of vari-ance of the assay procedure.

Identification of 6a-Hydroxylpaclitaxel

Samples from patients who had received paclitaxel onthe multicenter ovarian cancer study were analyzed bythe initial HPLC procedure described. Chromatographsfrom all patients on this study treated with 3-hour IVinfusions contained a peak with a retention time of ap-proximately 3.5 minutes. This peak reached maximumheight immediately after the end of the infusion and de-clined rapidly thereafter. The assay procedure was subse-quently modified, as described, so that baseline resolutionof this peak could be achieved. This modified HPLCprocedure was used to quantify both paclitaxel and thepresumed metabolite, in plasma, from all patients in thebreast cancer study. The retention times of the suspectedplasma metabolite and of a pure standard of 6a-hydroxyl-

183

paclitaxel were identical (4.8 minutes). Furthermore, theultraviolet spectra of 6a-hydroxylpaclitaxel and the sus-pected metabolite collected during the chromatographicprocedure had 99.5% concordance in the region from 210to 300 nm. In the absence of sufficient plasma for defini-tive mass spectrographic analysis and nuclear magneticresonance spectroscopy, these observations were taken aspresumptive evidence of identity between the unknownpeak in plasma and 6a-hydroxylpaclitaxel. Although asuspected metabolite peak was present in the HPLCs ofall patients who received paclitaxel by 3-hour infusion,insufficient pretreatment plasma and posttreatment sam-

ples remained to reanalyze for 6a-hydroxylpaclitaxel thesamples from ovarian cancer patients treated at paclitaxeldoses of 135 and 175 mg/m2. Therefore, as stated earlier,quantitation of 6a-hydroxylpaclitaxel in plasma was lim-ited to patients with breast cancer who received 3-hourinfusions of paclitaxel at 225 mg/m2.

Pharmacokinetics of Paclitaxel and

6a-Hydroxylpaclitaxel

Mean pharmacokinetic parameters for paclitaxel, ascalculated by noncompartmental methods, are listed inTable 1. It was planned that the patients enrolled ontothe ovarian cancer study would have pharmacokineticstudies performed for cycles 1 through 3 of therapy. All15 patients were studied during cycle 1. Data from cycles1 and 2 were available in 14 patients, and data fromcycles 1, 2, and 3 were available in 11 patients. In thesepatients, the ratios of cycle 2 AUC to cycle 1 AUC andcycle 3 AUC to cycle 1 AUC were 114% + 33% (mean+ SD) and 129% ± 36%, respectively. For patients whoreceived paclitaxel by 3-hour infusion, examination ofthe relationship between dose administered and both themeasured plasma paclitaxel CMAX and the AUC indicateda nonlinear relationship (Fig IB; Table 1). For example,3-hour paclitaxel infusions at 135 mg/m 2 resulted in amean CMAX of 3.3 pm and a mean AUC of 10.4 pmol/L-h, while 3-hour infusions at 175 mg/m 2 resulted in amean CMAX of 5.9 ymol/L and a mean AUC of 18.0 /mol/L h. Thus, a 30% increase in dose resulted in an 80%increase in CMAX and a 75% increase in AUC (Table 1).

For patients who received 24-hour infusions, the rela-tionships between dose and the mean end of infusionpaclitaxel concentrations (CMAx) and calculated AUCswere also disproportionate. However, these nonlinear re-lationships were less pronounced than those observedwith 3-hour infusion schedules. Representative patientplasma paclitaxel concentration-versus-time profiles fromboth 3- and 24-hour infusions appear in Fig 2.

In patients who received 3-hour paclitaxel infusions,


GIANNI ET AL

4A

:L

X

I-

:3

W

J1

W2

o.4

a-

4uJa.

,).

ul

Ao

0

o0

i0P

O

<

IL

o

<

B0o

o

o

DOSE (mg/m2) DOSE (mg/m 2)

Fig 1. Relationships between paclitaxel dose and (A) measured plasma paclitaxel C..., or (B) paclitaxel AUC in patients who received 3-hour paclitaxel infusions. Lines indicate linear relationships expected if either C..x or AUC increased proportionally with dose.

plasma paclitaxel concentrations decreased rapidly imme-diately after cessation of the infusion. This was followedby a more prolonged terminal phase. The concentration-time course of the metabolite followed the same generalpattern as the parent, but metabolite concentrations inplasma were always well below corresponding concentra-tions of parent compound (Fig 3). The plasma CMAX Of6a-hydroxylpaclitaxel ranged from 0.48 to 3.00 pmol/Land occurred between 5 and 15 minutes after completionof the 3-hour paclitaxel infusion (Table 2). Noncompart-mental pharmacokinetic parameters for 6a-hydroxyl-paclitaxel are listed in Table 2. In that measured concen-trations of 6a-hydroxylpaclitaxel were always muchlower than corresponding parent drug concentrations, therelative AUC of paclitaxel was 4.5 to 18 times greaterthan that of the metabolite.

In contrast to the nonlinearity observed in plasma, doseand schedule did not have measurable effect on the rela-tively small amount of paclitaxel excreted in the urineduring the first 24 hours after initiation of the infusion.Urinary elimination accounted for only 2% to 4% of thetotal dose administered (Table 1).

We sought to define a pharmacokinetic model thatwould explain the observed nonlinearity in paclitaxel dis-position. Initially, the data were fit to a two-compartmentmodel with a single, nonlinear Michaelis-Menten processthat described distribution from the central to the periph-eral compartment. In this model, both return from theperipheral to central compartment and central elimination

were described by first-order rate constants." From thenoncompartmental analysis, it appeared likely that centralelimination also included a nonlinear component, as AUCincreased disproportionately to administered dose (Table1; Fig IB). Central elimination of paclitaxel and the ap-

4

-jI

a.

0.0

TIME (h)

Fig 2. Plasma paclitaxel concentration-v-time profiles of represen-tative patients who received the drug at various indicated doses andinfusion durations. Symbols represent actual measured plasma pacli-taxel concentrations and lines represent model fit curves.

184



pearance of the metabolite in plasma were both best de-scribed by Michaelis-Menten processes. The final modeland the mean + SD for all model-derived processes andvolumes appear in Fig 4. Comparison of the parametersfor central elimination (oVm, oKm) and appearance of 6a-hydroxylpaclitaxel in plasma (mVm, mKm) shows that thecentral elimination pathway is, by far, the predominantroute of elimination. The pathway that describes the ap-pearance of 6a-hydroxylpaclitaxel accounts for only me-tabolite identified in plasma, which is likely only a frac-tion of the total 6a-hydroxylpaclitaxel formed. Because6a-hydroxylpaclitaxel pharmacokinetics could not beevaluated independently of parent compound, our esti-mates for volume and elimination of the metabolite aresimply descriptive of the observed data. However, thismodel does describe well the measured plasma concentra-tions of both paclitaxel and 6a-hydroxylpaclitaxel (whenavailable) at all doses and schedules studied.

Pharmacokinetic/Pharmacodynamic Relationships ofPaclitaxel

Neutropenia was the most common and relevant toxic-ity that could be investigated in our patients. We wereable to evaluate neutropenia in 29 of 30 patients whoreceived the first course of paclitaxel. Table 3 lists theabsolute, relative, and World Health Organization-graded neutropenia observed with respect to dose andschedule of paclitaxel. Within an individual schedule, ei-

zi2

z0z0

I

0.1

0TIME (h)

Fig 3. Plasma concentrations of paclitaxel (0-) and 6a-hydroxyl-paclitaxel (0) in plasma of a patient treated with 225 mg/m 2 ofpaclitaxel as a 3-hour infusion. Symbols represent actual measuredplasma paclitaxel concentrations and lines represent model fit curves.

185

Table 2. Summary of Noncompartmental Pharmacokinetic Parametersof 6-a-Hydroxypaclitaxel After Administration of

225 mg/m2 by 3-Hour Infusion

Paclitoxel-to-6a-C•x TMA AUC Hydroxypoclitaxel

Variable (pmol/L) (hours) (pmol/L-h) AUC Ratio

Mean + SD 1.27 - 0.62 3.19 - 0.09 3.22 - 1.7 9.28 + 4.47Minimum 0.48 3.08 0.94 4.47Maximum 3.00 3.25 7.15 18.69CV% 49 2.7 53 48

Abbreviation: TMx, time to CAx.

ther 3 hours or 24 hours, there was an increase in neutro-penia, both in terms of incidence and severity, as doseand AUC increased. However, the most striking observa-tion was that for any dose, the 24-hour infusion schedulewas associated with significantly more neutropenia thanthe identical dose administered as a 3-hour infusion (Ta-ble 3). As a result, total plasma exposure to paclitaxel,as measured by AUC, and plasma CMAX of paclitaxelwere not predictive of toxicity. In an attempt to explainthe relationships between neutropenia, schedule, anddose, we reasoned that neutropenia could be related tothe time that plasma paclitaxel concentrations were ator above a threshold concentration. This duration wouldtherefore be a function of dose, infusion schedule, andthe disposition of paclitaxel in any individual patient.

We initially investigated a threshold value of 0.1 gmol/L, as this has been reported by some investigators to bethe lowest cytotoxic concentration of paclitaxel that wasclinically relevant.5',"8 9 However, other investigatorshave observed that paclitaxel concentrations as low as0.05 pmol/L cause microtubular abnormalities and cyto-toxicity in vitro.20 Moreover, in a study of patients whoreceived prolonged infusions of paclitaxel, in whichplasma concentrations remained less than 0.1 pmol/L,significant toxicity and grade 4 neutropenia were ob-served.21 It was evident, therefore, that a threshold pacli-taxel concentration of 0.1 pmol/L would not be an appro-priate predictor of neutropenia. Thus, we evaluatedthreshold concentrations less than 0.1 pmol/L. For indi-vidual patients, the duration that plasma paclitaxel con-centrations remained > 0.05 pmol/L is depicted in Fig5; only data from course 1 of therapy are included. Asexpected, the mean duration of time that plasma paclitaxelconcentrations were > 0.05 Mmol/L increased with in-creasing dose and length of infusion. However, there wasconsiderable overlap among the doses and schedules, pre-sumably reflective of interpatient pharmacokinetic vari-ability.

We found that by using a threshold paclitaxel concen-tration of 0.05 pmol/L, our sigmoid EMAX model describedwell the observed relative neutropenia versus duration ator above the threshold (Fig 6). For this model, the EMAX


GIANNI ET AL

infusion

k4l Vm/Km

kk4l4 1,

k14 k21

oVm mVm

Km mK

3I

eVm

eKm

Vm = 17.7 ± 7.2 PM/h

Km = 0.23 + 0.12 jCM

k21 = 1.4 + 0.2 h-1

k1 4 = 2.6 ± 1.3 h-1

k4 1 = 0.60 + 0.30 h-1

oVm = 29.9 + 1.5 /pM/h

oKm = 7.0 + 1.6 pM

Vi = 3.8 + 0.3 L/m2

mVm = 1.61 + 0.01 ipM/h

mKm = 60.4 ± 16.8 pM

eVm = 1.9 + 0.7 pM/h

eKm = 0.43 ± 0.21 pM

V 3 = 0.37 ± 0.01 L/m2

Fig 4. Structure of the pharmacokinetic model developed for paclitaxel and 6a-hydroxylpaclitaxel. 1, Paclitaxel central compartment; 2,first paclitaxel peripheral compartment; 3, 6a-hydroxylpaclitaxel compartment; 4, second peripheral compartment of paclitaxel. Solid linesrefer to the parent compound, dotted lines to the metabolite. V, and K. are Michaelis-Menten constant estimates for the saturable distributionof paclitaxel from 1 to 2. k21 is the first-order constant for return from 2 to 1. k14 and k4 1 are the first-order rate constants for movement outof 1 to 4 and return from 4 to 1. mV. and mK, are Michaelis-Menten constants for formation of 6a-hydroxylpaclitaxel in plasma, and eV.and eK, are Michaelis-Menten constants for its elimination from plasma. V1 is the estimated volume of the central compartment of paclitaxel.V3 is the estimated volume of the compartment of the metabolite. Mean + SD values for each parameter in the model are indicated.

was fixed at 100% and the EMIN at 0%; the D50% wasestimated to be 17.4 hours and the Hill constant was 2.3.Of particular importance is that duration at or above thethreshold, irrespective of the actual dose or schedule ad-ministered, was the defining parameter predictive of neu-tropenia.

For completeness, we evaluated threshold paclitaxelconcentrations both greater than and less than 0.05 ymol/L. A threshold concentration of 0.1 bpmol/L produced astep function, in which profound neutropenia would bemanifest as an all-or-none event. This was inconsistentwith our clinical observations. As the threshold was de-creased to 0.03 pbmol/L, the entire curve was shifted tothe right, with no improvement in our measures of good-ness of fit.

With respect to incidence of grade 3 or 4 toxicity, therewas no correlation between measured plasma paclitaxel

C,,,, AUC, or dose, and occurrence of grade 3 or 4

neutropenia. The duration of exposure to paclitaxel con-centrations ý- 0.05 ymol/L was significantly related tothe incidence of grade 3 or 4 neutropenia: 12 of 16 pa-tients with exposure durations more than 24 hours experi-enced grade 3 or 4 neutropenia, as opposed to only twoof 13 patients with durations < 24 hours (P < .005, X2).Administration schedule, per se, was not an independentpredictor of toxicity. Although six of seven patients whoreceived 24-hour infusions experienced grade 3 or 4 neu-tropenia versus eight of 22 patients who received 3-hourinfusions, the patients who received 24-hour infusionswere actually a subset of all patients who experienced >0.05 ymol/L paclitaxel exposures for more than 24 hours.

DISCUSSION

The data we have generated in the present study com-plement and expand previous knowledge of the clinicalpharmacology of paclitaxel and have important practical

Table 3. Paclitaxel-Associated Neutropenia by Dose and Schedule

Time Dose AUC Mean ANC at Nadir ANCNADIR Cycles With Grade 3-4(hours) (mg/m2) (pmol/L h - SD) (c/pL ± SEM) (% + SEM) ANC (%)

3 135 10.9 1.1 1,841 ± 335* 49- 9 253 175 18.9 - 3.0 1,374+ 331 71 ± 6 373 225 24.3 - 6.8 1,075 + 207 77 - 6 50

24 135 12.4 - 2.2 846 ± 110* 80 ± 8 7824 175 16.0 - 4.4 634 - 79* 80 - 4 89

*Statistically significant difference (P -5 .05, two-tailed t test) between samples.

186



40 -

- 40-

o 35-

_j 30-

25-

4 20-

15-

0. 10-LU

S5-

U_

co

8

0

0 C

o0

.04

0

00

135/3 175/3 225/3 135/24 175/24

DOSE (mg/m 2)/INFUSION SCHEDULE (h)

Fig 5. Values for duration spent at a plasma paclitaxel concentra-tion - 0.05 pmol/L with respect to various doses and schedules ofadministration. Circles represent individual patients. Bars depictmean values for each group.

implications for its optimal clinical use. The discoveryand identification of 6a-hydroxylpaclitaxel as the majorpaclitaxel metabolite in human plasma was an importantfinding. Biliary excretion of paclitaxel and hydroxylatedmetabolites can account for approximately 25% of theadministered dose.22 Recently, it was shown that 6a-hy-droxylation is the prevalent biotransformation of pacli-taxel in isolated human liver microsomes, where it iscatalyzed by cytochrome P450-3A9' 23 24 or -2C accordingto other investigators.2 5 Since the metabolite is approxi-mately 30 times less toxic than paclitaxel, 9 our data indi-cate that 6a-hydroxylation could be an important detoxi-fication pathway. As a consequence, induction orinhibition of paclitaxel 6a-hydroxylation by other drugsmetabolized by cyt P450 3A 23,24 or -2C 25 will eventuallyrequire paclitaxel dose adjustments to avoid under- orover-dosage. In our patients, we observed that the totalamount of 6a-hydroxylpaclitaxel in plasma accounted foronly approximately one fifth to one twentieth of the totalpaclitaxel dose. Also, the metabolite was only detectablein plasma during times of relatively high concomitantparent compound concentrations, like those produced bythe shorter 3-hour infusion schedule. From the in vitrodata on microsomal metabolism, one would expect a largefraction of paclitaxel to be metabolized to 6a-hydroxyl-paclitaxel. In an effort to reconcile this apparent discrep-ancy, we recalled the observations by one of us (M.J.Egorin, unpublished observations) that 6a-hydroxylpacli-

A•

100-

90-

80-

o 70-z

z 60-uJ -

o-

O 50-LU

o 40-o

* 30-

20-

10-

n-

1

S

1300

o

o

A

A 175mg/m2 -3h

o 225 mg/m2

- 3h

* 135mg/m2

-24h

A 175 mg/m 2-24h

U 5 10 15 20 25 30 35 40 45 50

TIME PLASMA [PACLITAXEL] a 0.05 pM (h)

Fig 6. Pharmacokinetic/pharmacodynamic relationship betweenduration spent at a plasma paclitaxel concentration > 0.05 pmol/Land percentage reduction in ANC in the first course of therapy. Sym-bols represent individuals treated at different doses and schedules(see Legend). Curve depicts the sigmoid Em_, model fit to the data.The broken portion of the curve represents that region for which datawere not available.

-

• . , ' . ...

187

taxel accumulates in plasma of patients with biliary ob-struction, and that large amounts of hydroxylated pacli-taxel metabolites,22 and especially 6a-hydroxylpaclitaxel,can be measured in bile. We therefore speculated that themajority of 6a-hydroxylpaclitaxel formed is eliminateddirectly into the bile, and that metabolite measurable inplasma represents overflow at times of relative saturationor blockade of biliary elimination. With respect to ourpharmacokinetic model, metabolic conversion to 6a-hy-droxylpaclitaxel is embedded within both the centralelimination (oVm, oKm) and metabolite in plasma (mVm,mKm) parameters. In many cases, the metabolite inplasma component was relatively insignificant, with re-spect to the central elimination pathway. However, inpatients with liver abnormalities by neoplastic or nonneo-plastic disease that caused intrahepatic or extrahepaticcholestasis, this may not be the case.

Another important point of this investigation is thedemonstration that in-line filtration in paclitaxel infusionsystems does not sequester the drug. Thus, the actualdelivered dose is equivalent to the intended dose, andthe different clearances of paclitaxel observed at variousdoses and schedules were not an artifact of the deliverysystem. For the purposes of the present study, the clarifi-cation of this practical point ruled out that the administra-

11 q

I

f


GIANNI ET AL

tion system could be a source of the observed nonlinearityof paclitaxel disposition.

In our study, the disposition of paclitaxel was clearlynonlinear and complex. When analyzing paclitaxel con-centration-versus-time data from patients treated with 3-hour infusions, the measured CMAX and the calculatedAUCs were not proportional to the doses administered.The same trends were evident when the data from the24-h schedule were evaluated, although the nonlinearitywas less evident at the relatively low plasma concentra-tions of paclitaxel associated with the longer infusion.The data were consistent with a saturable eliminationsystem according to which low plasma concentrationswould appear to be cleared at a relatively faster rate thanhigh concentrations. Thus, when concentrations declined,particularly below the saturation point of the eliminationsystem, clearance would approach a linear or proportionalrelationship with respect to concentration. Therefore, dif-ferences in clearance rates would be more evident at rela-tively high concentrations as opposed to low concentra-tions. This may explain why previous investigators failedto identify the nonlinear disposition of paclitaxel at lowdoses infused over 1 or 6 hours.2'6 Still, as other investiga-tors have shown, if the pharmacokinetics of paclitaxelare assessed over a sufficiently wide range of doses, thedeviation from linearity of the drug plasma kinetics isevident even during prolonged infusions.26

The nonlinear disposition of paclitaxel may have con-siderable ramifications with respect to the clinical use ofthe drug. Increases or decreases in dose, with no concur-rent adjustment of time of infusion, will result in nonpro-portionally higher or lower CMAX values and total plasmaexposures or AUCs. Longer or shorter infusion schedulescould also result in nonproportionally lower or higherpredicted plasma concentrations and total AUCs for thesame delivered dose of paclitaxel. The demonstration ofnonlinearity suggests the opportunity that new doses andschedules of paclitaxel should be tried with suitable phar-macokinetic studies that would help the interpretation ofpossible unexpected clinical effects.

The pharmacokinetic model presented here accuratelydescribes the plasma concentration-time profiles of pacli-taxel in patients at all doses and schedules studied, andcould be used to predict paclitaxel disposition at as yetuntested doses and schedules. Until now, simpler linearbiexponential2-6' 27 or triexponential models7 have beenused to describe paclitaxel disposition. At first, thesemodels appear to fit postinfusion plasma disappearance ofpaclitaxel, even in the case of 3-hour infusions.7 However,they consistently underestimate plasma concentrationsduring the infusion and fail to yield dose-independentparameter estimates. Thus, the complexity of the pre-

sented model is necessary for the accurate estimation ofcomplete paclitaxel and 6a-hydroxylpaclitaxel concentra-tion-time profiles.

The mean parameter estimates provided for the modelare the result of the combined estimates from 55 coursesof paclitaxel administered to 30 patients. While we recog-nize that this is a relatively small sample size from whichto estimate population characteristics, these estimateshave thus far proven durable.

Our development of the pharmacokinetic model pre-sented for paclitaxel disposition has allowed us to definea relationship between paclitaxel pharmacokinetics andneutropenia. Neutropenia can be related to the time atwhich plasma paclitaxel concentrations are - 0.05 pmol/L. Total dose and AUC did not correspond with the inci-dence or severity of neutropenia. The relationship be-tween time > 0.05 Ipmol/L and neutropenia is well de-scribed by a sigmoid EMAX model.

Other investigators have suggested that a paclitaxelthreshold of 0.1 jimol/L was informative with respect toneutropenia.7' However, a study of 96-hour infusions ofpaclitaxel in 34 patients at doses of 120, 140, and 160 mg/m 2 resulted in mean steady-state concentrations of 0.05,0.07, and 0.08 pmol/L, respectively. 21 Grade 4 neutropeniawas observed in 14 of these patients. A threshold concentra-tion of 0.1 ymol/L would have predicted no toxicity in thesepatients. Furthermore, examination of the data presented tosupport the 0.1-jimol/L threshold shows that the relationshipbetween percentage reduction in ANC and time at or abovethe threshold is a step function.7 This implies that profoundneutropenia is an all-or-none phenomenon, which is incon-sistent with clinical observations.

Our chosen threshold of 0.05 Aimol/L paclitaxel correlateswell with the neutropenia observed in our patients, and wasselected between three different and arbitrary values (0.1,0.05, and 0.03 /mol/L) on the basis of measures of goodnessof fit to the pharmacodynamic model. However, additionaldata concerning neutropenia from relatively short durationsat > 0.05 pmol/L is necessary to confirm the lower end ofour model. More information on the lower portion of thecurve is also important, because it may change the thresholdvalue. We are currently engaged in such an effort.

In addition to neutropenia, hypersensitivity reactionsand peripheral neurosensory symptoms have been de-scribed as the most common dose-limiting toxicities asso-ciated with paclitaxel administration. 1,3-5,28-30 The defini-tion of pharmacodynamic relationships for these and otherpaclitaxel toxicities is of great interest. In our study, wecould not assess whether a pharmacodynamic relationshipexisted between paclitaxel pharmacokinetics and periph-eral neurosensory symptoms. Half of the patients we stud-ied had been previously treated with platinum-based che-

188


PACUTAXEL PHARMACOKINETICS AND METABOLISM

motherapy regimens and had preexisting neurotoxicity.In these patients, the overall incidence of discomfort frompaclitaxel was much greater than in the 15 patients withbreast cancer who had never received neurotoxic drugs.As far as hypersensitivity is concerned, the clinical find-ings of the European-Canadian trial, which contributedsome of the patients investigated in the present study,ruled out that the infusion schedule is related to the inci-dence and severity of hypersensitivity reactions in pa-tients who received adequate premedication.1

The characterization of the therapeutic range of con-centrations of a drug requires not only the definition ofpharmacodynamic relationships to predict or produce anacceptable level of toxicity, but also the definition ofpharmacodynamic relationships to produce the maximumlikelihood of response. The observation that the probabil-ity of response in previously treated patients with ovariancancer was not related to the duration of infusion seemsto indicate that the biochemical/biologic target(s) of

189

toxicity of paclitaxel is different for bone marrow pro-genitors of neutrophils and for ovarian cancer cells.1

Thus, in the case of paclitaxel, toxic and therapeuticpharmacodynamic relationships will most likely bebased on different pharmacokinetic parameters. More-over, relationships between pharmacokinetics and ther-apeutic response may well vary depending on tumortype. Therefore, the relationship between paclitaxelpharmacokinetics and antineoplastic response will needto be defined for each tumor type against which pacli-taxel demonstrates activity. The pharmacokineticmodel presented here should prove to be a powerfultool in investigating each of these issues and relation-ships. Furthermore, should a range of plasma paclitaxelconcentrations be associated with optimal therapeuticindex, in light of demonstrated interpatient variability,the model will provide the basis for using adaptivecontrol with feedback dosing strategies to achieve thoseconcentrations in individual patients.

REFERENCES

1. Eisenhauer EA, ten Bokkel Huinink WW, Swenerton KD, etal: European-Canadian randomized trial of paclitaxel in relapsedovarian cancer: High-versus low-dose and long versus short infusion.J Clin Oncol 12:2654-2666, 1994

2. Grem JL, Tutsch KD, Simon KJ, et al: Phase I study of Taxoladministered as a short iv infusion daily for 5 days. Cancer TreatRep 71:1179-1184, 1987

3. Wiernik PH, Schwartz EL, Einzig A, et al: Phase I trial of Taxolgiven as a 24-hour infusion every 21 days: Responses observed inmetastatic melanoma. J Clin Oncol 5:1232-1239, 1987

4. Brown T, Havlin K, Weiss G, et al: A phase I trial of Taxolgiven by a 6-hour intravenous infusion. J Clin Oncol 9:1261-1267,1991

5. Wiernik PH, Schwartz EL, Strauman JJ, et al: Phase I clinicaland pharmacokinetic study of Taxol. Cancer Res 47:2486-2493,1987

6. Longnecker SM, Donehower RC, Cates AE, et al: High-perfor-mance liquid chromatography assay for Taxol in human plasma andurine and pharmacokinetics in a phase I trial. Cancer Treat Rep71:53-59, 1987

7. Huizing MT, Keung AC, Rosing H, et al: Pharmacokineticsof paclitaxel and metabolites in a randomized comparative study inplatinum-pretreated ovarian cancer patients. J Clin Oncol 11:2127-2135, 1993

8. Gianni L, Capri G, Munzone E, et al: Paclitaxel efficacy inpatients with advanced breast cancer resistant to anthracyclines.Semin Oncol 21:29-33, 1994 (suppl 8)

9. Harris JW, Katki A, Anderson LW, et al: Isolation, structuraldetermination, and biological activity of 6a-hydroxytaxol, the princi-ple human metabolite of Taxol. J Med Chem 37:706-709, 1994

10. Jamis-Dow CA, Klecker RW, Sarosy G, et al: Steady-stateplasma concentrations and effects of Taxol at a 250 mg/m2 dose incombination with granulocyte-colony stimulating factor in patientswith ovarian cancer. Cancer Chemother Pharmacol 33:48-52, 1993

11. Rocci ML, Jusko WJ: LAGRAN program for area and mo-ments in pharmacokinetic analysis. Comput Prog Biomed 16:293-216, 1983

12. D'Argenio DZ, Schumitzky A: A program package for simu-

lation and parameter estimation in pharmacokinetic systems. ComputProg Biomed 9:115-134, 1979

13. Steimer JL, Mallet A, Golmard JL, et al: Alternative ap-proaches to estimation of population pharmacokinetic parameters:Comparison with the nonlinear mixed-effect model. Drug MetabRev 15:265-292, 1984

14. Remington RD, Shork MA: Statistics with Applications tothe Biological and Health Sciences. (ed 2). Englewood Cliffs, NJ,Prentice-Hall, 1985, p 145

15. Wagner JG: Kinetics of pharmacologic response. I. Proposedrelationships between response and drug concentration in the intactanimal and man. J Theoret Biol 20:173-201, 1968

16. DeLean A, Munson PJ: Simultaneous analysis of families ofsigmoidal curves: application to bioassay, radioligand assay, andphysiological dose-response curves. Am J Physiol 235:E97-E102,1978

17. Kearns C, Gianni L, Giani A, et al: Non-linear pharmacoki-netics of taxol in humans. Proc Am Soc Clin Oncol 12:135, 1993(abstr)

18. Rowinsky EK, Donehower RC, Jones RJ, et al: Microtubulechanges and cytotoxicity in leukemic cell lines treated with taxol.Cancer Res 48:4093-4100, 1988

19. Rowinsky EK, Cazenave LA, Donehower RC: Taxol: A novelinvestigational antineoplastic agent. J Natl Cancer Inst 82:1247-1259, 1990

20. Schiff PB, Fant J, Horowitz SB: Promotion of microtubuleassembly in vitro by taxol. Nature 277:665-667, 1979

21. Wilson WH, Berg S, Kang YK, et al: Phase 11II study oftaxol 96-hour infusion in refractory lymphoma and breast cancer:Pharmacodynamics and analysis of multi-drug resistance (mdr-1).Proc Am Soc Clin Oncol 12:134, 1993 (abstr 335)

22. Monsarrat B, Alvinerie P, Wright M, et al: Hepatic metabo-lism and biliary clearance of Taxol in rats and human. J Natl CancerInst Monogr 15:39-46, 1993

23. Kumar GN, Walle UK, Walle T: Cytochrome P450 3A-medi-ated human liver microsomal Taxol 6a-hydroxylation. J PharmacolExp Ther 268:1160-1164, 1994

24. Walle T: 6ca-hydroxytaxol: Isolation and identification of the


GIANNI ET AL

major metabolite of Taxol in human liver microsomes. Drug MetabDisp 22:177-179, 1994

25. Cresteil T, Monsarrat B, Alvinerie P, et al: Taxol metabolismby human liver microsomes: Identification of cytochrome P450 iso-zymes involved in its biotransformation. Cancer Res 54:386-392,1994

26. Sonnichsen D, Hurwitz CA, Pratt CB, et al: Saturable pharma-cokinetics and paclitaxel pharmacodynamics in children with solidtumors. J Clin Oncol 12:532-538, 1994

27. Rowinsky EK, Burke PJ, Karp JE, et al: Phase I and pharma-

codynamic study of Taxol in refractory acute leukemias. Cancer Res49:4640-4647, 1989

28. Sarosy G, Kohn E, Stone DA, et al: Phase I study of Taxoland granulocyte colony-stimulating factor in patients with refractoryovarian cancer. J Clin Oncol 10:1165-1170, 1992

29. Kohn EC, Sarosy G, Bicher A, et al: Dose-intense Taxol:High response rate in patients with platinum-resistant recurrent ovar-ian cancer. J Natl Cancer Inst 86:18-24, 1994

30. Holmes FA, Walters RS, Theriault RL, et al: Phase II trialof Taxol, an active drug in the treatment of metastatic breast cancer.J Natl Cancer Inst 83:1797-1805, 1991

190


Documents

180 pacli y filtro