Model Describing the Relationship Between Pharmacokinetics and Hematologic Toxicity of the Epirubicin-Docetaxel Regimen in Breast Cancer Pat
http://www.100md.com
《临床肿瘤学》
the Division of Pharmacokinetics and Drug Therapy, Department of Pharmaceutical Biosciences, Faculty of Pharmacy, Uppsala University, Sweden
Department of Oncology, Radiology and Clinical Immunology, Uppsala University Hospital, Uppsala, Sweden
Radiumhemmet, Karolinska Institute and Hospital, Stockholm, Sweden
ABSTRACT
PATIENTS AND METHODS: Forty-four patients with advanced disease received EPI and DTX every 3 weeks for up to nine cycles. The initial doses (EPI/DTX) were 75/70 mg/m2. Based on leukocyte (WBC) and platelet counts, the subsequent doses were, stepwise, either escalated (maximum, 120/100 mg/m2) or reduced (minimum, 40/50 mg/m2). Hematologic toxicity was monitored in all patients, whereas pharmacokinetics was studied in 16 patients. A semiphysiological model, including physiological parameters as well as drug-specific parameters, was used to describe the time course of WBC count following treatment.
RESULTS: In the final pharmacokinetic model, interoccasion variability was estimated to be less than interindividual variability in the clearances for both drugs. The sum of the individual EPI and DTX areas under concentration-time curve correlated stronger to WBC survival fraction than did the corresponding sum of doses. A pharmacokinetic-pharmacodynamic (PK-PD) model with additive effects of EPI and DTX could adequately describe the data.
CONCLUSION: The final PK-PD model might provide a tool for calculation of WBC time course, and hence, for prediction of nadir day and duration of leukopenia in breast cancer patients treated with the EPI/DTX regimen.
INTRODUCTION
In the process of investigating the EPI-DTX combination, the pharmacokinetics (PK) of the components have been characterized following combination treatment.8,11,12 An indication of a redistribution phenomenon of EPI when followed immediately by DTX administration was found in a study on 11 patients.12 This phenomenon was not apparent from the data presented in another study in which the same schedule was used; whereas in that study, a different metabolic pattern was found for EPI in the combination, compared with single-drug administration.11 However, no obvious clinically relevant PK interaction between the two parent compounds has been demonstrated.
Knowledge of PK may help in the optimization of dose and schedule selection of polychemotherapy combinations.13 However, optimizing the dosing requires knowledge of the contribution of the component drugs to the antitumoral effect and toxicity. The former is usually not definable from clinical data; therefore, dosing individualization has to be based on dose-limiting toxicity rather than desirable effect. In the case of EPI and DTX, one of the dose-limiting and joint toxicities is hematologic toxicity. The hematologic toxicity from each drug alone has been previously described, and for both drugs, a relationship between exposure in terms of area under the concentration-time curve (AUC) and hematologic toxicity has been shown.14-17 However, no report on the characterization of the relationship between PK and hematologic toxicity for the EPI-DTX combination has been published. Hence, the contribution of each drug to hematologic toxicity in the combination therapy has not yet been quantified.
The aims of the present study were (1) to characterize the PK of both component drugs and (2) to describe the relationship between PK and the dose-limiting hematologic toxicity for the EPI-DTX regimen in breast cancer patients, as this regimen has promising clinical utility.
PATIENTS AND METHODS
Pharmacokinetic Sampling
Blood samples for PK were collected in EDTA-coated Vacutainer tubes (Becton Dickinson, Plymouth, United Kingdom) at five time points within the following sampling windows: immediately before the end of the EPI and DTX infusions, respectively; 30 to 60 minutes after the end of DTX infusion; 5 to 8 hours and 15 to 27 hours after the start of EPI infusion. However, deviations occurred due to practical reasons or patient convenience, but the actual time of sampling was recorded and used in the data analysis. The samples were immediately put in an ice-water bath and were centrifuged at 4°C within 20 minutes to obtain plasma that was stored at –70°C until analysis. Sampling for determination of WBC was scheduled for days 1 (dosing day), 8, 11 or 12, 15, and 22 in association with each treatment occasion. However, for practical reasons, the true sampling times diverged ± 1 to 2 days from those scheduled.
Drug Assays
EPI plasma concentration was determined as previously described,19 with minor modifications: The column used was a Zorbax SB-CN (Agilent Industries, Hgersten, Sweden), and the mobile phase consisted of formiate buffer and acetonitrile (ratio, 675:325). Between-run and within-run coefficients of variation, measured at five concentrations (6.172 to 1,111 ng/mL), were less than 13%. Accuracy was 92% to 104% within the same concentration range. DTX plasma concentrations were analyzed according to a method originally developed for analysis of paclitaxel in plasma.20 Between-run and within-run coefficients of variation, measured at five concentrations (8.1 to 7,800 ng/mL), were less than 12%. Accuracy was 87% to 103% within the same concentration range.
Data Analysis
PK and PK-pharmacodynamic (PD) modeling was carried out using mixed-effects modeling in the NONMEM program, Version V and VI beta, using the FOCE INTER/CENTER methods.21 The model building process was guided by graphical evaluation within the program Xpose, Version 3.0,22 as well as by the objective function value, which is a goodness-of-fit statistic produced by NONMEM. For hierarchical models, a drop in the objective function value by more than 3.84, 6.63, and 10.83 denotes an improved fit at P < .05, .01, and .001, respectively, for a one-parameter difference. Whereas the population model-predicted drug concentrations and WBC are based on the estimated typical values of the model parameters, the individual model predictions are based on the individual parameter estimates, obtained in NONMEM as empirical Bayes (POSTHOC) estimates.
The modeling was performed in steps. First, the PK models were developed using the ADVAN 11 subroutine. In the PK-PD analysis that followed, the individual PK parameter estimates from the final PK models were fixed, and the predicted individual drug concentration-time profiles were used as input functions in the pharmacodynamic model. PK in the individuals from whom PK information was missing was assumed to be the same as in the typical individual in the studied population. For this part of the analysis, the ADVAN 6 was used.
Since it is well known that EPI and DTX show three-compartment PK,19,23 only such models were considered in the PK part of the analysis. During the development of the DTX PK model, variability terms including interoccasion variability were tried sequentially on all PK parameters and where kept in the model only when significant on the P < .01 level. EPI data did not support a three-compartment model. Therefore, EPI data were analyzed using the PRIOR routine within NONMEM.24 Parameter estimates from a previous PK analysis of data from a similar population of 79 breast cancer patients (unpublished data) were used as frequentist priors, where a penalty is added to the objective function on deviation from these priors. The structural and statistical parts of the previously developed model were kept when characterizing the EPI data. The estimate used as prior for total-body clearance (CL) and the intercompartmental clearances (Q2 and Q3) were (with coefficient of variation [CV%] in parentheses) 71.7 (2.7) L/h, 70.6 (3.9) L/h, and 17.8 (10) L/h, respectively; whereas the estimates for central volume of distribution (V1) and the peripheral volumes of distribution (V2 and V3) were 13.1 (7.5) L, 776 (3.7) L, and 14.6 (7.4) L, respectively. The estimates used for interindividual variability in CL, interoccasion variability in V1, and residual error were 15% (39), 73% (16), and 25.7% (7.7), respectively.
Model Describing Hematologic Toxicity
The model that was used in the present PK-PD modeling procedure was a recently developed semiphysiological model shown to be able to describe the WBC concentration-time profile following single-agent chemotherapy,15 and that is now used to describe the effect of two drugs in combination.
The model consists of five compartments, of which one represents proliferating cells. That compartment is linked to a compartment representing circulating leukocytes via a maturation chain that is composed of three sequential compartments (Fig 1). The prediction at any time (t) after dose in the fifth compartment, WBC(t), is hence fit to the WBC observations obtained from the patients.
The rate at which an average cell moves (matures) from one compartment (stage) to another is governed by the first-order maturation rate constants (kMTT). The product of the proliferation rate constant (kprol) and the baseline value of the amount of proliferative cells needs (from mass balance considerations) to equal the product of the elimination rate constant for circulating leukocytes (kcirc) and the baseline WBC. Without loss of generality, kpro1 was set to equal kMTT, resulting in a baseline value of the amount of proliferative cells equaling the baseline leukocyte count.15 The circulating WBC elimination rate constant (kcirc) was, as a default, fixed to be the same value as kMTT. This is a procedure that has been used before to handle the lack of information in this parameter from this type of clinical data.15 The lack of information about kcirc stems from the fact that the rate-limiting step for the temporal aspects of the leucopenia profile is the maturation time, not the half-life of leukocytes in circulation. In addition to the default procedure, kcirc was fixed to a value corresponding to a half-life of 6 hours, a value previously reported in the literature.25 Since the present maturation chain consists of three compartments, the average time it takes for a cell to mature and appear in the systemic circulation (ie, the mean transit time through the chain [MTT]) is equal to 4/kMTT.
Also included in the model is a feedback mechanism that imitates the endogenous colony stimulation factor feedback system. The system responds to a decrease in blood-cell concentration, and it is modeled as a function of the baseline WBC (WBCbase) and WBC(t) according to (WBCbase/WBC[t]){gamma}. The drug-related bone marrow toxicity is a function of the drug plasma concentration (CEPI and CDTX, respectively) and for each drug, a specific slope parameter (SEPI and SDTX) that linearly relates the drug effect in the proliferating compartment to concentration of drug in plasma. Thus, the changes in the amount of proliferative cells (Prol, compartment 1) can be expressed by the following equation:
SDTX and the uncertainty in that estimate, obtained in a previous analysis on data from 601 patients who received DTX as single therapy (7.91 L/mg [SE, 0.49 L/mg]), was used as a prior in the model.15
No data were available from patients receiving EPI as single therapy in the present study. Therefore, to assess whether the assumption of additivity of EPI and DTX effects was appropriate, a model which as well as the effects of EPI and DTX alone also included an effect component from the product of the two drug concentrations was compared with the final model. In addition, the estimated SEPI was evaluated versus previously published data.17 The WBC-time profile after the doses given in that study was simulated, and the model-predicted decrease in WBC 12 days after administered dose was compared with the observations obtained 10 to 14 days after given dose. Doses used in the simulation were hence 40, 60, 90, and 135 mg/m2.
Investigated Correlations
The correlations between the PK parameters for the two drugs were assessed. In addition, relationships between different exposure measurements and nadir WBC, WBC absolute decrease at nadir, as well as survival fraction of WBC, were evaluated. Exposure measurements evaluated were the sum and product of mg- and body surface area–based doses as well as the sum and product of the AUCs. Statistical significance (P < .05) was tested using Pearson’s product moment correlation coefficient.
RESULTS
One individual showed unexplainable diverging DTX PK compared with the remaining patients during the third and last occasion (Fig 2D). The data from that patient at that occasion were therefore excluded from the data set and analyzed separately. The DTX clearance was estimated to 6.3 L/h compared with the estimates from the first and second occasion (26.9 and 24.1 L/h, respectively). The EPI PK parameters did not differ, and the EPI clearance for that individual was estimated to 74 L/h, which is near the population parameter estimate of 69.6 L/h.
Hematologic Toxicity
A total of 805 WBC observations at 187 occasions from 42 individuals were used in the PK-PD modeling (Fig 3A). Twenty-one of the patients received G-CSF support at one or several occasions; one patient was treated with G-CSF during all dosing occasions. Data from those occasions were not included in the analysis and are not shown.
The leukocyte count at baseline (ie, observed WBC at the start of therapy during the first occasion, was on average (CV and range within parentheses) 6.8 x 109/L (27%; 3.9 to 12 x 109/L). WBC nadir (ie, the lowest observed count between dosing occasions, occurred on average at 10 days (23%; 7 to 14 days) after treatment and averaged 1.2 x 109/L (47%; 0.3 to 3.3 x 109/L).
Exposure-Toxicity Relationships
In all correlation analyses, the survival fraction (SF) of WBC at nadir, defined as the ratio between the measured WBC at nadir and measured WBC before the start of the first course of therapy, showed the highest correlation with exposure measures. Correction for fraction unbound drug, 0.2326,27 and 0.0328,29 for EPI and DTX, respectively, did not improve the fits.
No significant correlation between the sum of EPI and DTX doses and SF was found when considering either all dosing occasions or the first occasion only.
Relationships between different exposure measurements and SF were evaluated separately for the dosing occasions from which PK was available. For those occasions, regardless of considering all doses or the first dose only, no significant relationship was found between SF and any dose measurement, whereas a significant correlation was obtained when instead the sum of the AUCs was considered as a predictor for SF.
The PK/PD Model for Hematologic Toxicity
The parameter estimates in the final model describing development of hematological toxicity after administration of EPI in combination with DTX are presented in Table 3. Only small changes in parameter estimates were observed when only individuals with pharmacokinetic information (n = 16; relative change in population mean parameters < 8%) were included in the analysis, or when kcirc was fixed to 0.12 hours–1 (relative change in population mean parameters < 22%). Predictions from the final model versus observations are apparent in Figure 3B–C. The model predicts a WBC of 1.1 * 109/L at nadir, occurring 9.5 days after treatment, after the combination of the median doses administered in the study (ie, 160 mg and 140 mg of EPI and DTX, respectively). The hematologic toxicity caused by each milligram of DTX is, according to the final model, predicted to be less pronounced than that caused by EPI, considering not only the differences in steepness of the estimated slopes but also the differences in pharmacokinetics. In Figure 4 the model predicted WBC-time profiles after administrations of the actual median dose of each drug, either alone or in combination, are shown.
DISCUSSION
WBC count at nadir, and other toxicities, have previously been suggested as biologic markers for the adjuvant chemotherapy efficacy in breast cancer patients.35-37 Accordingly, the dosing strategy in the present patient cohort was to treat the patients with individualized doses that resulted in equivalent hematological toxicity in terms of a nadir WBC of ≤ 1.0 * 109/L. This was carried out by stepwise dose escalation/reduction, based on the hematologic toxicity. Using this principle, it was previously demonstrated that individually tailored and dose escalated FEC (fluorouracil, epirubicin, cyclophosphamide) resulted in statistically significantly fewer breast cancer relapses compared with standard FEC followed by marrow supported high-dose therapy.33 However, the potential disadvantage with the stepwise strategy used in the present study is that patients showing good tolerability will not receive their individual optimal dose until the fourth dosing occasion (ie, 9 weeks after start of therapy). This delayed delivery of a potentially optimal dose may facilitate the development of chemotherapy resistant tumor cell clones.
If highest dose possible were the aim of the therapy, the ideal would be to rapidly determine the MTD for an individual patient, if not possible already before the first dosing occasion, at least during the first course of therapy. The variability in the pharmacokinetic parameter clearance was for both drugs in the present study found to be less within than between individuals. Hence, therapeutic drug monitoring (TDM) is a possible method to use, provided that a target concentration or exposure profile has been defined. TDM has been successfully carried out in anticancer therapy.13 However, the fact that anticancer regimens often include drugs with overlapping toxicity and that models that accounts for the contribution of more than one drug to the pharmacodynamic profile has been lacking, have precluded the use of TDM for most drug combinations. Development and implementation of PK/PD models, as the model proposed in the present study, might increase the chance of successfully performing TDM. On the other hand, measuring WBC only and obtaining individualized PK/PD parameter estimates may be sufficient for predicting the outcome of future courses under joint escalation/reduction of DTX and EPI. However, no evaluation of various prospective strategies for dose individualization was performed on the present data and any possible benefit of TDM using this model needs to be established in future studies.
The superior correlation between the sum of the respective drug’s AUC and SF compared with the between sum of doses and SF in the patients from which PK was available, indicates that part of variability in hematologic toxicity is explained by variability in PK. This result is in accordance with previous studies in which EPI and DTX were administered as single drugs.14,16,17
Estimation of the effect of each drug was not possible without using data from another study as prior information in the modeling process. The fractional decreases in WBC 12 days after dosing of EPI alone were, by the model, predicted to be 0.3, 0.42, 0.56, and 0.72, after the doses 68, 102, 153, and 230 mg, respectively, which could be compared with those reported previously (ie, 0.45, 0.52, 0.60, and 0.70, respectively),17 for comparative exposures. The estimated value for docetaxel slope was 7.74 L/mg, which is close to the value of 7.91 L/mg which has been previously shown to adequately describe the leukopenia profile in 601 patients.15 In addition, the model in which multiplicative effects between the drug concentrations were allowed did not result in an improved fit. Thus, the developed model, with additive effects between epirubicin and docetaxel, can describe not only combination treatment, but also monotherapy with epirubicin and with docetaxel. However, for the combination treatment, only a limited dose range was investigated and caution in extrapolation of the results to combinations of the two drugs at other doses should be exercised.
The lower variability within than between individuals in clearance of the studied drugs found in the present study and, in addition, the significant correlation between the sum of the respective drug’s AUC and fractional decrease of leukocyte count at nadir, suggests that individual clearance estimates of the component drugs in the studied regimen may be used to individualize therapy. By in addition estimating the SEPI and SDTX in an individual patient, the chance of defining the MTD early in drug therapy may increase.
The presently proposed model describes the whole time-course of WBC after combination therapy consisting of the two drugs EPI and DTX. The delay in the observed effect is well captured by the model and furthermore, the estimated time at which nadir occurs is in agreement with the observations. Hence, the model might offer the benefit of being predictive of the duration of leukopenia and thus it might be used to calculate the doses of the drugs in the combination to administer.
Authors’ Disclosures of Potential Conflicts of Interest
Acknowledgment
We are grateful to the patients who kindly participated in the study. We also thank Britt Jansson, Ingrid Fallenius, Jessica Nilsson, Carina Andersson, Rita Grnberg, Tina Fornbrandt, Birgitta Ohlander, Gunvor Svensson, and Clementine Molin for technical assistance.
NOTES
Supported by the Swedish Cancer Society.
Authors’ disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
1. Airoldi M, Cattel L, Pedani F, et al: Clinical and pharmacokinetic data of a docetaxel-epirubicin combination in metastatic breast cancer. Breast Cancer Res Treat 70:185-195, 2001
2. Wenzel C, Locker GJ, Schmidinger M, et al: Combined analysis of two phase II trials in patients with primary and advanced breast cancer with epidoxorubicin and docetaxel+granulocyte colony stimulating factor. Anticancer Drugs 13:67-74, 2002
3. de Matteis A, Nuzzo F, D’Aiuto G, et al: Docetaxel plus epidoxorubicin as neoadjuvant treatment in patients with large operable or locally advanced carcinoma of the breast: A single-center, phase II study. Cancer 94:895-901, 2002
4. Venturini M, Michelotti A, Papaldo P, et al: Identification of the highest dose of docetaxel associable with active doses of epirubicin: Results from a dose-finding study in advanced breast cancer patients. Ann Oncol 12:1097-1106, 2001
5. Viens P, Roche H, Kerbrat P, et al: Epirubicin-docetaxel combination in first-line chemotherapy for patients with metastatic breast cancer: Final results of a dose-finding and efficacy study. Am J Clin Oncol 24:328-335, 2001
6. Milla-Santos A, Milla L, Rallo L, et al: High-dose epirubicin plus docetaxel at standard dose with lenograstim support as first-line therapy in advanced breast cancer. Am J Clin Oncol 24:138-142, 2001
7. Mavroudis D, Alexopoulos A, Ziras N, et al: Front-line treatment of advanced breast cancer with docetaxel and epirubicin: A multicenter phase II study. Ann Oncol 11:1249-1254, 2000
8. Pagani O, Sessa C, Nole F, et al: Epidoxorubicin and docetaxel as first-line chemotherapy in patients with advanced breast cancer: A multicentric phase I-II study. Ann Oncol 11:985-991, 2000
9. Kouroussis C, Xydakis E, Potamianou A, et al: Front-line treatment of metastatic breast cancer with docetaxel and epirubicin: A multicenter dose-escalation study. The Greek Breast Cancer Cooperative Group (GBCCG). Ann Oncol 10:547-552, 1999
10. Pagani O, Sessa C, Martinelli G, et al: Dose-finding study of epidoxorubicin and docetaxel as first-line chemotherapy in patients with advanced breast cancer. Ann Oncol 10:539-545, 1999
11. Esposito M, Venturini M, Vannozzi MO, et al: Comparative effects of paclitaxel and docetaxel on the metabolism and pharmacokinetics of epirubicin in breast cancer patients. J Clin Oncol 17:1132, 1999
12. Ceruti M, Tagini V, Recalenda V, et al: Docetaxel in combination with epirubicin in metastatic breast cancer: Pharmacokinetic interactions. Farmaco 54:733-739, 1999
13. Evans WE, Relling MV, Rodman JH, et al: Conventional compared with individualized chemotherapy for childhood acute lymphoblastic leukemia. N Engl J Med 338:499-505, 1998
14. Bissett D, Setanoians A, Cassidy J, et al: Phase I and pharmacokinetic study of taxotere (RP 56976) administered as a 24-hour infusion. Cancer Res 53:523-527, 1993
15. Friberg LE, Henningsson A, Maas H, et al: Model of chemotherapy-induced myelosuppression with parameter consistency across drugs. J Clin Oncol 20:4713-4721, 2002
16. Extra JM, Rousseau F, Bruno R, et al: Phase I and pharmacokinetic study of Taxotere (RP 56976; NSC 628503) given as a short intravenous infusion. Cancer Res 53:1037-1042, 1993
17. Jakobsen P, Bastholt L, Dalmark M, et al: A randomized study of epirubicin at four different dose levels in advanced breast cancer: Feasibility of myelotoxicity prediction through single blood-sample measurement. Cancer Chemother Pharmacol 28:465-469, 1991
18. Cockroft D, Gault M: Prediction of creatinine clearance from serum creatinine. Nephron 16:31-41, 1976
19. Sandstrom M, Freijs A, Larsson R, et al: Lack of relationship between systemic exposure for the component drug of the fluorouracil, epirubicin, and 4-hydroxycyclophosphamide regimen in breast cancer patients. J Clin Oncol 14:1581-1588, 1996
20. 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 621:231-238, 1993
21. Beal S, Sheiner L: NONMEM Users Guides. San Fransicso, CA, University of California, 1999
22. Jonsson EN, Karlsson MO: Xpose: An S-PLUS based population pharmacokinetic/pharmacodynamic model building aid for NONMEM. Comput Methods Programs Biomed 58:51-64, 1999
23. Bruno R, Vivler N, Vergniol JC, et al: A population pharmacokinetic model for docetaxel (Taxotere): Model building and validation. J Pharmacokinet Biopharm 24:153-172, 1996
24. Gisleskog PO, Karlsson MO, Beal SL: Use of prior information to stabilize a population data analysis. J Pharmacokinetics Pharmacodynamics 29:473-505, 2002
25. Cartwright GE, Athens JW, Wintrobe MM: The kinetics of granulopoiesis in normal man. Blood 24:780-803, 1964
26. Chassany O, Urien S, Claudepierre P, et al: Comparative serum protein binding of anthracycline derivatives. Cancer Chemother Pharmacol 38:571-573, 1996
27. Robert J: Clinical pharmacokinetics of epirubicin. Clin Pharmacokinet 26:428-438, 1994
28. Urien S, Barre J, Morin C, et al: Docetaxel serum protein binding with high affinity to alpha 1-acid glycoprotein. Invest New Drugs 14:147-151, 1996
29. Loos WJ, Baker SD, Verweij J, et al: Clinical pharmacokinetics of unbound docetaxel: Role of polysorbate 80 and serum proteins. Clin Pharmacol Ther 74:364-371, 2003
30. Henderson IC, Hayes DF, Gelman R: Dose-response in the treatment of breast cancer: A critical review. J Clin Oncol 6:1501-1515, 1988
31. Colleoni M, Rocca A, Sandri MT, et al: Low-dose oral methotrexate and cyclophosphamide in metastatic breast cancer: Antitumor activity and correlation with vascular endothelial growth factor levels. Ann Oncol 13:73-80, 2002
32. Fossati R, Confalonieri C, Torri V, et al: Cytotoxic and hormonal treatment for metastatic breast cancer: A systematic review of published randomized trials involving 31,510 women. J Clin Oncol 16:3439-3460, 1998
33. Bergh J, Wiklund T, Erikstein B, et al: Tailored fluorouracil, epirubicin, and cyclophosphamide compared with marrow-supported high-dose chemotherapy as adjuvant treatment for high-risk breast cancer: A randomised trial—Scandinavian Breast Group 9401 study. Lancet 356:1384-1391, 2000
34. Benefit of a high-dose epirubicin regimen in adjuvant chemotherapy for node-positive breast cancer patients with poor prognostic factors: 5-year follow-up results of French Adjuvant Study Group 05 randomized trial. J Clin Oncol 19:602-611, 2001
35. Colleoni M, Price K, Castiglione-Gertsch M, et al: Dose-response effect of adjuvant cyclophosphamide, methotrexate, 5-fluorouracil (CMF) in node-positive breast cancer: International Breast Cancer Study Group. Eur J Cancer 34:1693-1700, 1998
36. Saarto T, Blomqvist C, Rissanen P, et al: Haematological toxicity: A marker of adjuvant chemotherapy efficacy in stage II and III breast cancer. Br J Cancer 75:301-305, 1997
37. Poikonen P, Saarto T, Lundin J, et al: Leucocyte nadir as a marker for chemotherapy efficacy in node-positive breast cancer treated with adjuvant CMF. Br J Cancer 80:1763-1766, 1999(M. Sandstrm, H. Lindman, )
Department of Oncology, Radiology and Clinical Immunology, Uppsala University Hospital, Uppsala, Sweden
Radiumhemmet, Karolinska Institute and Hospital, Stockholm, Sweden
ABSTRACT
PATIENTS AND METHODS: Forty-four patients with advanced disease received EPI and DTX every 3 weeks for up to nine cycles. The initial doses (EPI/DTX) were 75/70 mg/m2. Based on leukocyte (WBC) and platelet counts, the subsequent doses were, stepwise, either escalated (maximum, 120/100 mg/m2) or reduced (minimum, 40/50 mg/m2). Hematologic toxicity was monitored in all patients, whereas pharmacokinetics was studied in 16 patients. A semiphysiological model, including physiological parameters as well as drug-specific parameters, was used to describe the time course of WBC count following treatment.
RESULTS: In the final pharmacokinetic model, interoccasion variability was estimated to be less than interindividual variability in the clearances for both drugs. The sum of the individual EPI and DTX areas under concentration-time curve correlated stronger to WBC survival fraction than did the corresponding sum of doses. A pharmacokinetic-pharmacodynamic (PK-PD) model with additive effects of EPI and DTX could adequately describe the data.
CONCLUSION: The final PK-PD model might provide a tool for calculation of WBC time course, and hence, for prediction of nadir day and duration of leukopenia in breast cancer patients treated with the EPI/DTX regimen.
INTRODUCTION
In the process of investigating the EPI-DTX combination, the pharmacokinetics (PK) of the components have been characterized following combination treatment.8,11,12 An indication of a redistribution phenomenon of EPI when followed immediately by DTX administration was found in a study on 11 patients.12 This phenomenon was not apparent from the data presented in another study in which the same schedule was used; whereas in that study, a different metabolic pattern was found for EPI in the combination, compared with single-drug administration.11 However, no obvious clinically relevant PK interaction between the two parent compounds has been demonstrated.
Knowledge of PK may help in the optimization of dose and schedule selection of polychemotherapy combinations.13 However, optimizing the dosing requires knowledge of the contribution of the component drugs to the antitumoral effect and toxicity. The former is usually not definable from clinical data; therefore, dosing individualization has to be based on dose-limiting toxicity rather than desirable effect. In the case of EPI and DTX, one of the dose-limiting and joint toxicities is hematologic toxicity. The hematologic toxicity from each drug alone has been previously described, and for both drugs, a relationship between exposure in terms of area under the concentration-time curve (AUC) and hematologic toxicity has been shown.14-17 However, no report on the characterization of the relationship between PK and hematologic toxicity for the EPI-DTX combination has been published. Hence, the contribution of each drug to hematologic toxicity in the combination therapy has not yet been quantified.
The aims of the present study were (1) to characterize the PK of both component drugs and (2) to describe the relationship between PK and the dose-limiting hematologic toxicity for the EPI-DTX regimen in breast cancer patients, as this regimen has promising clinical utility.
PATIENTS AND METHODS
Pharmacokinetic Sampling
Blood samples for PK were collected in EDTA-coated Vacutainer tubes (Becton Dickinson, Plymouth, United Kingdom) at five time points within the following sampling windows: immediately before the end of the EPI and DTX infusions, respectively; 30 to 60 minutes after the end of DTX infusion; 5 to 8 hours and 15 to 27 hours after the start of EPI infusion. However, deviations occurred due to practical reasons or patient convenience, but the actual time of sampling was recorded and used in the data analysis. The samples were immediately put in an ice-water bath and were centrifuged at 4°C within 20 minutes to obtain plasma that was stored at –70°C until analysis. Sampling for determination of WBC was scheduled for days 1 (dosing day), 8, 11 or 12, 15, and 22 in association with each treatment occasion. However, for practical reasons, the true sampling times diverged ± 1 to 2 days from those scheduled.
Drug Assays
EPI plasma concentration was determined as previously described,19 with minor modifications: The column used was a Zorbax SB-CN (Agilent Industries, Hgersten, Sweden), and the mobile phase consisted of formiate buffer and acetonitrile (ratio, 675:325). Between-run and within-run coefficients of variation, measured at five concentrations (6.172 to 1,111 ng/mL), were less than 13%. Accuracy was 92% to 104% within the same concentration range. DTX plasma concentrations were analyzed according to a method originally developed for analysis of paclitaxel in plasma.20 Between-run and within-run coefficients of variation, measured at five concentrations (8.1 to 7,800 ng/mL), were less than 12%. Accuracy was 87% to 103% within the same concentration range.
Data Analysis
PK and PK-pharmacodynamic (PD) modeling was carried out using mixed-effects modeling in the NONMEM program, Version V and VI beta, using the FOCE INTER/CENTER methods.21 The model building process was guided by graphical evaluation within the program Xpose, Version 3.0,22 as well as by the objective function value, which is a goodness-of-fit statistic produced by NONMEM. For hierarchical models, a drop in the objective function value by more than 3.84, 6.63, and 10.83 denotes an improved fit at P < .05, .01, and .001, respectively, for a one-parameter difference. Whereas the population model-predicted drug concentrations and WBC are based on the estimated typical values of the model parameters, the individual model predictions are based on the individual parameter estimates, obtained in NONMEM as empirical Bayes (POSTHOC) estimates.
The modeling was performed in steps. First, the PK models were developed using the ADVAN 11 subroutine. In the PK-PD analysis that followed, the individual PK parameter estimates from the final PK models were fixed, and the predicted individual drug concentration-time profiles were used as input functions in the pharmacodynamic model. PK in the individuals from whom PK information was missing was assumed to be the same as in the typical individual in the studied population. For this part of the analysis, the ADVAN 6 was used.
Since it is well known that EPI and DTX show three-compartment PK,19,23 only such models were considered in the PK part of the analysis. During the development of the DTX PK model, variability terms including interoccasion variability were tried sequentially on all PK parameters and where kept in the model only when significant on the P < .01 level. EPI data did not support a three-compartment model. Therefore, EPI data were analyzed using the PRIOR routine within NONMEM.24 Parameter estimates from a previous PK analysis of data from a similar population of 79 breast cancer patients (unpublished data) were used as frequentist priors, where a penalty is added to the objective function on deviation from these priors. The structural and statistical parts of the previously developed model were kept when characterizing the EPI data. The estimate used as prior for total-body clearance (CL) and the intercompartmental clearances (Q2 and Q3) were (with coefficient of variation [CV%] in parentheses) 71.7 (2.7) L/h, 70.6 (3.9) L/h, and 17.8 (10) L/h, respectively; whereas the estimates for central volume of distribution (V1) and the peripheral volumes of distribution (V2 and V3) were 13.1 (7.5) L, 776 (3.7) L, and 14.6 (7.4) L, respectively. The estimates used for interindividual variability in CL, interoccasion variability in V1, and residual error were 15% (39), 73% (16), and 25.7% (7.7), respectively.
Model Describing Hematologic Toxicity
The model that was used in the present PK-PD modeling procedure was a recently developed semiphysiological model shown to be able to describe the WBC concentration-time profile following single-agent chemotherapy,15 and that is now used to describe the effect of two drugs in combination.
The model consists of five compartments, of which one represents proliferating cells. That compartment is linked to a compartment representing circulating leukocytes via a maturation chain that is composed of three sequential compartments (Fig 1). The prediction at any time (t) after dose in the fifth compartment, WBC(t), is hence fit to the WBC observations obtained from the patients.
The rate at which an average cell moves (matures) from one compartment (stage) to another is governed by the first-order maturation rate constants (kMTT). The product of the proliferation rate constant (kprol) and the baseline value of the amount of proliferative cells needs (from mass balance considerations) to equal the product of the elimination rate constant for circulating leukocytes (kcirc) and the baseline WBC. Without loss of generality, kpro1 was set to equal kMTT, resulting in a baseline value of the amount of proliferative cells equaling the baseline leukocyte count.15 The circulating WBC elimination rate constant (kcirc) was, as a default, fixed to be the same value as kMTT. This is a procedure that has been used before to handle the lack of information in this parameter from this type of clinical data.15 The lack of information about kcirc stems from the fact that the rate-limiting step for the temporal aspects of the leucopenia profile is the maturation time, not the half-life of leukocytes in circulation. In addition to the default procedure, kcirc was fixed to a value corresponding to a half-life of 6 hours, a value previously reported in the literature.25 Since the present maturation chain consists of three compartments, the average time it takes for a cell to mature and appear in the systemic circulation (ie, the mean transit time through the chain [MTT]) is equal to 4/kMTT.
Also included in the model is a feedback mechanism that imitates the endogenous colony stimulation factor feedback system. The system responds to a decrease in blood-cell concentration, and it is modeled as a function of the baseline WBC (WBCbase) and WBC(t) according to (WBCbase/WBC[t]){gamma}. The drug-related bone marrow toxicity is a function of the drug plasma concentration (CEPI and CDTX, respectively) and for each drug, a specific slope parameter (SEPI and SDTX) that linearly relates the drug effect in the proliferating compartment to concentration of drug in plasma. Thus, the changes in the amount of proliferative cells (Prol, compartment 1) can be expressed by the following equation:
SDTX and the uncertainty in that estimate, obtained in a previous analysis on data from 601 patients who received DTX as single therapy (7.91 L/mg [SE, 0.49 L/mg]), was used as a prior in the model.15
No data were available from patients receiving EPI as single therapy in the present study. Therefore, to assess whether the assumption of additivity of EPI and DTX effects was appropriate, a model which as well as the effects of EPI and DTX alone also included an effect component from the product of the two drug concentrations was compared with the final model. In addition, the estimated SEPI was evaluated versus previously published data.17 The WBC-time profile after the doses given in that study was simulated, and the model-predicted decrease in WBC 12 days after administered dose was compared with the observations obtained 10 to 14 days after given dose. Doses used in the simulation were hence 40, 60, 90, and 135 mg/m2.
Investigated Correlations
The correlations between the PK parameters for the two drugs were assessed. In addition, relationships between different exposure measurements and nadir WBC, WBC absolute decrease at nadir, as well as survival fraction of WBC, were evaluated. Exposure measurements evaluated were the sum and product of mg- and body surface area–based doses as well as the sum and product of the AUCs. Statistical significance (P < .05) was tested using Pearson’s product moment correlation coefficient.
RESULTS
One individual showed unexplainable diverging DTX PK compared with the remaining patients during the third and last occasion (Fig 2D). The data from that patient at that occasion were therefore excluded from the data set and analyzed separately. The DTX clearance was estimated to 6.3 L/h compared with the estimates from the first and second occasion (26.9 and 24.1 L/h, respectively). The EPI PK parameters did not differ, and the EPI clearance for that individual was estimated to 74 L/h, which is near the population parameter estimate of 69.6 L/h.
Hematologic Toxicity
A total of 805 WBC observations at 187 occasions from 42 individuals were used in the PK-PD modeling (Fig 3A). Twenty-one of the patients received G-CSF support at one or several occasions; one patient was treated with G-CSF during all dosing occasions. Data from those occasions were not included in the analysis and are not shown.
The leukocyte count at baseline (ie, observed WBC at the start of therapy during the first occasion, was on average (CV and range within parentheses) 6.8 x 109/L (27%; 3.9 to 12 x 109/L). WBC nadir (ie, the lowest observed count between dosing occasions, occurred on average at 10 days (23%; 7 to 14 days) after treatment and averaged 1.2 x 109/L (47%; 0.3 to 3.3 x 109/L).
Exposure-Toxicity Relationships
In all correlation analyses, the survival fraction (SF) of WBC at nadir, defined as the ratio between the measured WBC at nadir and measured WBC before the start of the first course of therapy, showed the highest correlation with exposure measures. Correction for fraction unbound drug, 0.2326,27 and 0.0328,29 for EPI and DTX, respectively, did not improve the fits.
No significant correlation between the sum of EPI and DTX doses and SF was found when considering either all dosing occasions or the first occasion only.
Relationships between different exposure measurements and SF were evaluated separately for the dosing occasions from which PK was available. For those occasions, regardless of considering all doses or the first dose only, no significant relationship was found between SF and any dose measurement, whereas a significant correlation was obtained when instead the sum of the AUCs was considered as a predictor for SF.
The PK/PD Model for Hematologic Toxicity
The parameter estimates in the final model describing development of hematological toxicity after administration of EPI in combination with DTX are presented in Table 3. Only small changes in parameter estimates were observed when only individuals with pharmacokinetic information (n = 16; relative change in population mean parameters < 8%) were included in the analysis, or when kcirc was fixed to 0.12 hours–1 (relative change in population mean parameters < 22%). Predictions from the final model versus observations are apparent in Figure 3B–C. The model predicts a WBC of 1.1 * 109/L at nadir, occurring 9.5 days after treatment, after the combination of the median doses administered in the study (ie, 160 mg and 140 mg of EPI and DTX, respectively). The hematologic toxicity caused by each milligram of DTX is, according to the final model, predicted to be less pronounced than that caused by EPI, considering not only the differences in steepness of the estimated slopes but also the differences in pharmacokinetics. In Figure 4 the model predicted WBC-time profiles after administrations of the actual median dose of each drug, either alone or in combination, are shown.
DISCUSSION
WBC count at nadir, and other toxicities, have previously been suggested as biologic markers for the adjuvant chemotherapy efficacy in breast cancer patients.35-37 Accordingly, the dosing strategy in the present patient cohort was to treat the patients with individualized doses that resulted in equivalent hematological toxicity in terms of a nadir WBC of ≤ 1.0 * 109/L. This was carried out by stepwise dose escalation/reduction, based on the hematologic toxicity. Using this principle, it was previously demonstrated that individually tailored and dose escalated FEC (fluorouracil, epirubicin, cyclophosphamide) resulted in statistically significantly fewer breast cancer relapses compared with standard FEC followed by marrow supported high-dose therapy.33 However, the potential disadvantage with the stepwise strategy used in the present study is that patients showing good tolerability will not receive their individual optimal dose until the fourth dosing occasion (ie, 9 weeks after start of therapy). This delayed delivery of a potentially optimal dose may facilitate the development of chemotherapy resistant tumor cell clones.
If highest dose possible were the aim of the therapy, the ideal would be to rapidly determine the MTD for an individual patient, if not possible already before the first dosing occasion, at least during the first course of therapy. The variability in the pharmacokinetic parameter clearance was for both drugs in the present study found to be less within than between individuals. Hence, therapeutic drug monitoring (TDM) is a possible method to use, provided that a target concentration or exposure profile has been defined. TDM has been successfully carried out in anticancer therapy.13 However, the fact that anticancer regimens often include drugs with overlapping toxicity and that models that accounts for the contribution of more than one drug to the pharmacodynamic profile has been lacking, have precluded the use of TDM for most drug combinations. Development and implementation of PK/PD models, as the model proposed in the present study, might increase the chance of successfully performing TDM. On the other hand, measuring WBC only and obtaining individualized PK/PD parameter estimates may be sufficient for predicting the outcome of future courses under joint escalation/reduction of DTX and EPI. However, no evaluation of various prospective strategies for dose individualization was performed on the present data and any possible benefit of TDM using this model needs to be established in future studies.
The superior correlation between the sum of the respective drug’s AUC and SF compared with the between sum of doses and SF in the patients from which PK was available, indicates that part of variability in hematologic toxicity is explained by variability in PK. This result is in accordance with previous studies in which EPI and DTX were administered as single drugs.14,16,17
Estimation of the effect of each drug was not possible without using data from another study as prior information in the modeling process. The fractional decreases in WBC 12 days after dosing of EPI alone were, by the model, predicted to be 0.3, 0.42, 0.56, and 0.72, after the doses 68, 102, 153, and 230 mg, respectively, which could be compared with those reported previously (ie, 0.45, 0.52, 0.60, and 0.70, respectively),17 for comparative exposures. The estimated value for docetaxel slope was 7.74 L/mg, which is close to the value of 7.91 L/mg which has been previously shown to adequately describe the leukopenia profile in 601 patients.15 In addition, the model in which multiplicative effects between the drug concentrations were allowed did not result in an improved fit. Thus, the developed model, with additive effects between epirubicin and docetaxel, can describe not only combination treatment, but also monotherapy with epirubicin and with docetaxel. However, for the combination treatment, only a limited dose range was investigated and caution in extrapolation of the results to combinations of the two drugs at other doses should be exercised.
The lower variability within than between individuals in clearance of the studied drugs found in the present study and, in addition, the significant correlation between the sum of the respective drug’s AUC and fractional decrease of leukocyte count at nadir, suggests that individual clearance estimates of the component drugs in the studied regimen may be used to individualize therapy. By in addition estimating the SEPI and SDTX in an individual patient, the chance of defining the MTD early in drug therapy may increase.
The presently proposed model describes the whole time-course of WBC after combination therapy consisting of the two drugs EPI and DTX. The delay in the observed effect is well captured by the model and furthermore, the estimated time at which nadir occurs is in agreement with the observations. Hence, the model might offer the benefit of being predictive of the duration of leukopenia and thus it might be used to calculate the doses of the drugs in the combination to administer.
Authors’ Disclosures of Potential Conflicts of Interest
Acknowledgment
We are grateful to the patients who kindly participated in the study. We also thank Britt Jansson, Ingrid Fallenius, Jessica Nilsson, Carina Andersson, Rita Grnberg, Tina Fornbrandt, Birgitta Ohlander, Gunvor Svensson, and Clementine Molin for technical assistance.
NOTES
Supported by the Swedish Cancer Society.
Authors’ disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
1. Airoldi M, Cattel L, Pedani F, et al: Clinical and pharmacokinetic data of a docetaxel-epirubicin combination in metastatic breast cancer. Breast Cancer Res Treat 70:185-195, 2001
2. Wenzel C, Locker GJ, Schmidinger M, et al: Combined analysis of two phase II trials in patients with primary and advanced breast cancer with epidoxorubicin and docetaxel+granulocyte colony stimulating factor. Anticancer Drugs 13:67-74, 2002
3. de Matteis A, Nuzzo F, D’Aiuto G, et al: Docetaxel plus epidoxorubicin as neoadjuvant treatment in patients with large operable or locally advanced carcinoma of the breast: A single-center, phase II study. Cancer 94:895-901, 2002
4. Venturini M, Michelotti A, Papaldo P, et al: Identification of the highest dose of docetaxel associable with active doses of epirubicin: Results from a dose-finding study in advanced breast cancer patients. Ann Oncol 12:1097-1106, 2001
5. Viens P, Roche H, Kerbrat P, et al: Epirubicin-docetaxel combination in first-line chemotherapy for patients with metastatic breast cancer: Final results of a dose-finding and efficacy study. Am J Clin Oncol 24:328-335, 2001
6. Milla-Santos A, Milla L, Rallo L, et al: High-dose epirubicin plus docetaxel at standard dose with lenograstim support as first-line therapy in advanced breast cancer. Am J Clin Oncol 24:138-142, 2001
7. Mavroudis D, Alexopoulos A, Ziras N, et al: Front-line treatment of advanced breast cancer with docetaxel and epirubicin: A multicenter phase II study. Ann Oncol 11:1249-1254, 2000
8. Pagani O, Sessa C, Nole F, et al: Epidoxorubicin and docetaxel as first-line chemotherapy in patients with advanced breast cancer: A multicentric phase I-II study. Ann Oncol 11:985-991, 2000
9. Kouroussis C, Xydakis E, Potamianou A, et al: Front-line treatment of metastatic breast cancer with docetaxel and epirubicin: A multicenter dose-escalation study. The Greek Breast Cancer Cooperative Group (GBCCG). Ann Oncol 10:547-552, 1999
10. Pagani O, Sessa C, Martinelli G, et al: Dose-finding study of epidoxorubicin and docetaxel as first-line chemotherapy in patients with advanced breast cancer. Ann Oncol 10:539-545, 1999
11. Esposito M, Venturini M, Vannozzi MO, et al: Comparative effects of paclitaxel and docetaxel on the metabolism and pharmacokinetics of epirubicin in breast cancer patients. J Clin Oncol 17:1132, 1999
12. Ceruti M, Tagini V, Recalenda V, et al: Docetaxel in combination with epirubicin in metastatic breast cancer: Pharmacokinetic interactions. Farmaco 54:733-739, 1999
13. Evans WE, Relling MV, Rodman JH, et al: Conventional compared with individualized chemotherapy for childhood acute lymphoblastic leukemia. N Engl J Med 338:499-505, 1998
14. Bissett D, Setanoians A, Cassidy J, et al: Phase I and pharmacokinetic study of taxotere (RP 56976) administered as a 24-hour infusion. Cancer Res 53:523-527, 1993
15. Friberg LE, Henningsson A, Maas H, et al: Model of chemotherapy-induced myelosuppression with parameter consistency across drugs. J Clin Oncol 20:4713-4721, 2002
16. Extra JM, Rousseau F, Bruno R, et al: Phase I and pharmacokinetic study of Taxotere (RP 56976; NSC 628503) given as a short intravenous infusion. Cancer Res 53:1037-1042, 1993
17. Jakobsen P, Bastholt L, Dalmark M, et al: A randomized study of epirubicin at four different dose levels in advanced breast cancer: Feasibility of myelotoxicity prediction through single blood-sample measurement. Cancer Chemother Pharmacol 28:465-469, 1991
18. Cockroft D, Gault M: Prediction of creatinine clearance from serum creatinine. Nephron 16:31-41, 1976
19. Sandstrom M, Freijs A, Larsson R, et al: Lack of relationship between systemic exposure for the component drug of the fluorouracil, epirubicin, and 4-hydroxycyclophosphamide regimen in breast cancer patients. J Clin Oncol 14:1581-1588, 1996
20. 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 621:231-238, 1993
21. Beal S, Sheiner L: NONMEM Users Guides. San Fransicso, CA, University of California, 1999
22. Jonsson EN, Karlsson MO: Xpose: An S-PLUS based population pharmacokinetic/pharmacodynamic model building aid for NONMEM. Comput Methods Programs Biomed 58:51-64, 1999
23. Bruno R, Vivler N, Vergniol JC, et al: A population pharmacokinetic model for docetaxel (Taxotere): Model building and validation. J Pharmacokinet Biopharm 24:153-172, 1996
24. Gisleskog PO, Karlsson MO, Beal SL: Use of prior information to stabilize a population data analysis. J Pharmacokinetics Pharmacodynamics 29:473-505, 2002
25. Cartwright GE, Athens JW, Wintrobe MM: The kinetics of granulopoiesis in normal man. Blood 24:780-803, 1964
26. Chassany O, Urien S, Claudepierre P, et al: Comparative serum protein binding of anthracycline derivatives. Cancer Chemother Pharmacol 38:571-573, 1996
27. Robert J: Clinical pharmacokinetics of epirubicin. Clin Pharmacokinet 26:428-438, 1994
28. Urien S, Barre J, Morin C, et al: Docetaxel serum protein binding with high affinity to alpha 1-acid glycoprotein. Invest New Drugs 14:147-151, 1996
29. Loos WJ, Baker SD, Verweij J, et al: Clinical pharmacokinetics of unbound docetaxel: Role of polysorbate 80 and serum proteins. Clin Pharmacol Ther 74:364-371, 2003
30. Henderson IC, Hayes DF, Gelman R: Dose-response in the treatment of breast cancer: A critical review. J Clin Oncol 6:1501-1515, 1988
31. Colleoni M, Rocca A, Sandri MT, et al: Low-dose oral methotrexate and cyclophosphamide in metastatic breast cancer: Antitumor activity and correlation with vascular endothelial growth factor levels. Ann Oncol 13:73-80, 2002
32. Fossati R, Confalonieri C, Torri V, et al: Cytotoxic and hormonal treatment for metastatic breast cancer: A systematic review of published randomized trials involving 31,510 women. J Clin Oncol 16:3439-3460, 1998
33. Bergh J, Wiklund T, Erikstein B, et al: Tailored fluorouracil, epirubicin, and cyclophosphamide compared with marrow-supported high-dose chemotherapy as adjuvant treatment for high-risk breast cancer: A randomised trial—Scandinavian Breast Group 9401 study. Lancet 356:1384-1391, 2000
34. Benefit of a high-dose epirubicin regimen in adjuvant chemotherapy for node-positive breast cancer patients with poor prognostic factors: 5-year follow-up results of French Adjuvant Study Group 05 randomized trial. J Clin Oncol 19:602-611, 2001
35. Colleoni M, Price K, Castiglione-Gertsch M, et al: Dose-response effect of adjuvant cyclophosphamide, methotrexate, 5-fluorouracil (CMF) in node-positive breast cancer: International Breast Cancer Study Group. Eur J Cancer 34:1693-1700, 1998
36. Saarto T, Blomqvist C, Rissanen P, et al: Haematological toxicity: A marker of adjuvant chemotherapy efficacy in stage II and III breast cancer. Br J Cancer 75:301-305, 1997
37. Poikonen P, Saarto T, Lundin J, et al: Leucocyte nadir as a marker for chemotherapy efficacy in node-positive breast cancer treated with adjuvant CMF. Br J Cancer 80:1763-1766, 1999(M. Sandstrm, H. Lindman, )