Our objective was to assess, using clinical trial simulation, the feasibility of a fixed 200-µg dose of darbepoetin alfa (Aranesp) administered every 2 weeks in chemotherapy-induced anemia. A pharmacokinetic/pharmacodynamic
ABSTRACT: Our objective was to assess, using clinical trial simulation, the feasibilityof a fixed 200-µg dose of darbepoetin alfa (Aranesp) administered every 2 weeksin chemotherapy-induced anemia. A pharmacokinetic/pharmacodynamic model wasdeveloped using clinical data from 547 cancer patients who received darbepoetinalfa at various doses and schedules. Monte Carlo simulations were performed forweight-based (3 µg/kg every 2 weeks) and fixed-dose (200 µg every 2 weeks)regimens and were compared with observed clinical data. Mean hemoglobin changesfrom baseline to end of treatment were +1.61 g/dL, +1.83 g/dL, and +1.79 g/dLfor observed data, the weight-based simulation, and the fixed-dose simulation,respectively. The rates of required transfusions (hemoglobin ≤ 8 g/dL)were also similar between groups. For patients between 45 and 95 kg (over 90% ofthe population), the impact of a fixed dose on mean hemoglobin change wasnegligible. There was a slight weight effect at body weight extremes (< 45 kgand > 95 kg). Clinical outcomes from simulations of weight-based and fixeddosing of darbepoetin alfa were similar to those of observed weight-based data.Given the weight distribution of a typical cancer population, the majority wouldbe expected to benefit equally from weight-based and fixed-dose darbepoetin alfain the amelioration of chemotherapy-induced anemia. [ONCOLOGY 16(Suppl 11):37-44,2002]
Anemia is a common complication in patients withcancer.[1,2] Anemia, defined by low hemoglobin or a low red blood cell (RBC) count,can be a consequence of the myelotoxicity of the chemotherapy regimen(especially platinum-containing regimens) or an inherent effect of the canceritself (particularly in multiple myeloma, lymphomas, or metastatic bonedisease). Although approximately 50% of patients receiving chemotherapy canexperience anemia of varying severity, and despite the attendant morbiditieswhichcan include fatigue and cognitive and other central nervous system effects[4,5]anemiarepresents an undertreated complication in cancer patients.
Since its introduction in 1989, recombinant human erythropoietin (rHuEPO) hasbeen a mainstay for the treatment of cancer-related anemia that reduces the needfor RBC transfusions. When anemic cancer patients receiving chemotherapy weretreated with a standard regimen of rHuEPO, hemoglobin response (defined as anincrease in hemoglobin ³ 2 g/dL) was seen in 53% of patients.Additionally, the requirement for blood transfusions was significantly decreasedand quality-of-life indices were significantly improved.
Recommended dosing of rHuEPO is three times weekly, although weekly regimensare commonly utilized in cancer patients. This requirement for frequentdosing is due to therelatively short serum half-life ofrHuEPO. In response to this shortcoming in dosing profile, a novel molecule wasdesigned. Site-directed mutagenesis was employed to modify the amino acidbackbone of human erythropoietin to allow for additional N-linked sialic acid-containing carbohydratechains. These additional sialic acid moieties act to slow the clearance of theglycoprotein and hence prolong the serum half-life.
The resulting molecule, darbepoetin alfa (Aranesp), is a uniqueerythropoietic protein with a greater in vivo potency relative to rHuEPO.This increased biologic activity is due primarily to the slower clearance of themolecule. Following intravenous administration to patients with chronic kidneydisease, the terminal half-life was 25.3 hours, approximately threefold longerthan that of rHuEPO. In subcutaneous usage, the rate of absorption from thesubcutaneous administration site controls the elimination rate. Followingsubcutaneous administration, the terminal half-life of darbepoetin alfa was 48.8hours, approximately twofold longer than that for intravenous administration,with a bioavailability of 37%.
In phase II dose-finding trials, clear dose-dependent increases in hemoglobinresponse were seen. In a placebo-controlled phase III study of 320 anemiclung cancer patients receiving platinum-containing chemotherapy, statisticallysignificant improvements favoring darbepoetin alfa vs placebo were seen in theproportion of patients with a hematopoietic response (66% vs 24%), in patientsrequiring transfusions (26% vs 60%), and in the mean number of transfusions perpatient (1.14 vs 2.64). Darbepoetin alfa is approved in the United StatesandEurope for the treatment of anemia associated with chronic renal failure(including dialysis and predialysis patients) and in the United States for thetreatment of chemotherapyinduced anemia in patients with nonmyeloid malignancies.
Due to the longer serum half-life, darbepoetin alfa should be administeredless frequently than rHuEPO. This provides obvious advantages for bothpatients and health-care providers. The current recommendation for rHuEPO dosingin cancer patients with chemotherapy-induced anemia is a starting dose of 150U/kg three times weekly, adjusted in accordance with resulting hemoglobinlevels. Approved dosing for darbepoetin alfa in this same setting is 2.25µg/kg once weekly. However, a comparative study indicated that darbepoetinalfa administered at 1.5 µg/kg once weekly produced a similar mean hemoglobinchange from baseline as the above three-times-weekly regimen of rHuEPO.[11,15]
Alternative dosing strategies for darbepoetin alfa have been tested that morefully exploit the pharmacokinetic profile. These include dosing every 2weeks, where it has been demonstrated that darbepoetin alfa at 3 µg/kg every2 weeks produced the same hematopoietic response as rHuEPO given at 40,000 Uonce weeklya typical weekly dose. Even longer dosing intervals are beingexplored to coincide with the cycle of chemotherapy (viz, every 3 weeks andevery 4 weeks).
An additional dosing paradigm that might serve to enhance the utility ofdarbepoetin alfa is a fixed dose, ie, a dosing regimen independent of patients’body weight. A fixed dose would contribute to the simplicity of darbepoetin alfausage and, by the administration of the entire contents of a vial, eliminatewasted product. However, a thorough understanding of the pharmacokinetic/pharmacodynamicprofile of darbepoetin alfa is necessary to ensure the success of a fixed-dosestrategy. Other relevant considerations would include the distribution of bodyweight in the oncology population, the potential for under- or overdosingpatients, and attendant benefit/risk issues.
Clinical trial simulation that includes the integration of pharmacokineticand pharmacodynamic modeling has been advocated as a means of applying modelingtechniques to drug development.[17-20] The Monte Carlo method is one suchtechnique for performing clinical trial simulation. Pharmacokinetic/pharmacodynamicmodeling based on population analysis with subsequent Monte Carlo simulationswould potentially provide the best and least biased estimate of the anticipatedhemoglobin response both within the treated population and for an individualwith specific characteristics. This methodology permits the incorporation ofmeasures of variability and uncertainty into pharmacokinetic/pharmacodynamicmodeling.[22-26] Monte Carlo simulation uses prior information, such as baselinebody weight and hemoglobin concentration, and parameter estimates (from thepharmacokinetic/pharmacodynamic model), and allows multiple sampling of thesequantitatively defined probability distributions and the subsequent computationof model outputs. This method allows for a more rigorous assessment ofvariability of response than simpler (mean parameter or deterministic) methods.
This paper presents the results of a clinical trial simulation using theMonte Carlo method that assesses the predicted hematopoietic response of a fixeddose of darbepoetin alfa (200 µg every 2 weeks) vs modeled results for aweight-based dose (3 µg/kg every 2 weeks). Model validation was performedby comparing predicted outcomes for weight-based dosing with observed clinicaldata from weight-based dosing.
Source of Clinical Data
Clinical data for 547 patients were utilized in generating the model for thesimulation. These patients had participated in one of three Amgen-sponsoredclinical trials of darbepoetin alfa for chemotherapy-induced anemia (studynumbers 980290, 980291, and 20000174).[11,16,27] These trials were similar innature, all being conducted in a multicycle chemotherapy setting for adultpatients with solid tumors. Patients were required to have a baseline hemoglobin£ 11 g/dL for study eligibility. The trials investigated various doses andschedules of darbepoetin alfa in a randomized sequential- or parallel-cohortstructure. Doses and schedules of darbepoetin alfa that were studied included0.5 through 18 µg/kg given every week, every 2 weeks, every 3 weeks,and every 4 weeks (a total of 18 different regimens) over a 12-weektreatment period.
Table 1 gives descriptive statistics on the demographics of the 547 patientswhose data were used in the model development. Over two-thirds of the populationwere female, due to the prevalence of breast and gynecologic cancer patientstreated in these trials. Across the three trials, the mean age of the patientswas 61 years (range: 20 to 91 years). Mean body weight was 70 kg (range: 39 to129 kg). Tumor types represented in this population includedin descendingorder of prevalencebreast, lung, gastrointestinal, gynecologic, andgenitourinary. Platinum-containing regimens were used in approximately 38% ofpatients across all three studies.
In addition to serial hemoglobin levels representing the pharmacodynamic response, serum drug levels for the pharmacokinetic response weremeasured intensively in five patients given darbepoetin alfa doses of 0.5, 1.5,or 4.5 mg/kg once weekly subcutaneously, and were also measured predose and 48hours postdose in 211 patients at certain dosing time points. Observed clinicaldata from 33 patients in study 980290 who received darbepoetin alfa ina 3.0 µg/kg every-2-week regimen were utilized for comparison with thesimulated fixed dose and the simulated weight-based dose.
Modeling and Simulation Procedures
Pharmacokinetic/pharmacodynamic modeling and clinical trial simulation wereused to evaluate every-2-week dosing of darbepoetin alfa and to assess theimpact of a fixed dose on predicted response and its variability. The modelingand clinical trial simulation steps were as follows: (1) fitting and optimizingthe model to data from the three darbepoetin alfa clinical studies (describedabove), (2) developing a clinical trial simulation platform by incorporatingrelevant clinical study design elements, and (3) performing clinical trialsimulations to evaluate the impact of a fixed dose of darbepoetin alfa on predicted response and its variability.
Darbepoetin Alfa Pharmacokinetic/Pharmacodynamic Structural Model
The basis of the pharmacokinetic/pharmacodynamic model is the relationshipbetween darbepoetin alfa and hemoglobin concentrations, as demonstrated indogs. The erythropoiesis pharmacokinetic/pharmacodynamic structural model isan indirect response model incorporating RBC physiology (Figure 1). Themodel assumes that darbepoetin alfa concentrations stimulate the production rateof RBCs in the bone marrow at the precursor stage. The production rate isincreased by a factor proportional to Emax, with half-maximal stimulationobserved at a serum drug concentration of EC50. The stimulation function occursin the first of a series of compartments with equal transfer coefficients; thismight be considered physiologically equivalent to the bone marrow (Figure1).
The total maturation transit time in the "bone marrow" isconsidered to be similar to the RBC production time. The "bone marrow"is linked to the RBCs in circulation or "hemoglobin compartment" (alinked series of compartments that have equal transfer coefficients withelimination from the terminal compartment to allow for aging of the RBCs). Thetotal transit time of the RBCs in circulation is representative of the RBClifespan.
The pharmacokinetic and pharmacodynamic parameters and associated covariancesfor this population were estimated simultaneously using data from the threedarbepoetin alfa clinical studies cited above with BigNPEM. Pharmacokineticparameter estimates were determined first from available intensivepharmacokinetic profiles, then updated using predose and 48-hour postdose datafrom 211 additional patients. Previous analyses demonstrated that thepharmacokinetic properties of darbepoetin alfa are dose- and time-linear.The pharmacokinetic parameters were then fixed to allow estimation of thepharmacodynamic parameters. Hemoglobin response data were used from the 547patients discussed above. Mean response data for all treatment groups werefitted simultaneously.
The mean parameter vector was optimized for the 3 µg/kg every-2-week dosingparadigm to make model predictions of responses for this treatment regimen. Aposterior predictive check on model performance was performed using bothsimulated patients and observed clinical trial results. The predicted-vs-observedplots following the generation of maximal a posteriori Bayesian probabilityestimates for each regimen showed that the pharmacokinetic/pharmacodynamic modelaccurately predicted the observed hemoglobin profiles across all 18 treatmentregimens (r2 ³ 0.95), demonstrating the utility of this model in describinghemoglobin response following administration of various regimens of darbepoetinalfa.
Clinical Trial Simulation
Development of the clinical trial simulation platform encompassed thefollowing elements:
Clinical trial simulation (5,000 patients/cohort) was performed in order topredict outcomes and associated variability for each of the dose cohorts(weight-based vs fixed dose of darbepoetin alfa). These simulations wereperformed by sampling from the distributions of each of the parameters andbaseline characteristics to enable assessment of predicted outcomes. Allstatistics for the simulated cohorts were derived directly from theclinical trial simulations.
Hemoglobin End Points
Mean hemoglobin change from baseline over time is plotted in Figure 2 for thesimulation of a weight-based dose (3 µg/kg every 2 weeks) and the simulation ofa fixed dose (200 µg every 2 weeks), vs the observed data of 33 patients whoreceived darbepoetin alfa at 3 µg/kg every 2 weeks. The two simulated curves arevirtually superimposed, with mean hemoglobin changes from baseline increasingover time throughout the 12-week treatment period. The observed data for3 µg/kg every 2 weeks compared closely with the simulated data. Standarddeviations were similar in magnitude across all three groups, indicative ofcomparable variability in the respective populations. Overall, these hemoglobinprofiles demonstrated close similarity between the observed data and thesimulations and provide validation for the model.
Mean changes in hemoglobin concentration between baseline and variousposttreatment time points (Table 2) also illustrate the similarity betweenthe three data sets. At day 84, mean hemoglobin change from baseline forthe observed data cohort was +1.61 g/dL (standard deviation [SD] 1.7 g/dL),compared with 1.83 g/dL (SD 1.5 g/dL) for the weight-based dose simulation and+1.79 g/dL (SD 1.5 g/dL) for the fixed-dose simulation.
Additional hemoglobin end points for the three cohorts are summarized in Table 3; these include the proportion of patients achieving a hemoglobinresponse (eg, an increase of ³ 2 g/dL from baseline) and patientsrequiring one or more RBC transfusions. For each end point, the two simulatedcohorts gave comparable results. The proportion of patients who demonstrated ahemoglobin response was 60% in the observed data, compared with 77% and 76% inthe simulations; however, the 95% confidence intervals in the former overlappedthe simulated values.
The RBC transfusion trigger was reached in 21% and 22% of simulated patientsin the weight-based and fixed-dose cohorts, respectively. These rates werehigher than the rate of actual transfusion usage in the observed data (16%),possibly due to the subjectivity of clinicians in choosing to transfuse realpatients, whereas in the simulation a "transfusion" was automatic at ahemoglobin concentration ≤ 8 g/dL. However, the simulated transfusionrates were lower than the rate of 26% reported from patients receivingdarbepoetin alfa at 2.25 µg/kg once weekly in the phase IIIplacebo-controlled trial.
Effects of Body Weight
Figure 3 provides a profile of body weight distribution using 1,180 patientswho received darbepoetin alfa, rHuEPO, or placebo across four oncology trials.Mean and median body weights were 70 and 69 kg, respectively, with a range of 35to 165 kg. Fifty percent of the patients in this population fell within bodyweights of 59 and 79 kg, and 90% were between 48 and 96 kg.
Figure 4 shows mean change in hemoglobin concentration from baseline to day84 (end of treatment period) by body weight class for the two simulated dosinggroups and for the observed group. As might be predicted, the profile for theweight-based dosing simulation showed little variation in hemoglobin by bodyweight class, with mean hemoglobin changes between +1.7 and +1.9 g/dLacross all weight classes. The fixed-dose simulation showed some weight effecttrend as expected, with a higher mean hemoglobin change from baseline at thelowest (< 45 kg) body weight class (+2.11 g/dL [SD 1.55 g/dL]), slightdecreases with increasing body weight, and the lowest change in hemoglobin atthe highest (> 95 kg) body weight class (+1.15 g/dL [SD 1.46 g/dL]).Importantly, however, at the central portion of the weight distribution (ie,between 45 and 95 kg body weight, representing over 90% of the population), theweight-based dose and the fixed-dose simulations were within 0.2 g/dL of eachother in terms of hemoglobin change from baseline.
Comparing the weight distribution in Figure 3 with the hemoglobin responsesby weight class in Figure 4, it is clear that the majority of patients would beexpected to benefit equally from a weight-based and a fixed dose of darbepoetinalfa.
Table 4 illustrates the weight-based dose equivalent of a 200-mg fixed dose of darbepoetin alfa for body weights from 40 to 100 kg. Dosing a 40-kg patient with 200 µg of darbepoetin alfa equates to 5.0 µg/kg; for a 100-kg patient, the weight-based equivalent is 2.0 µg/kg. These doses have been studied in clinical trials and have been shown to be safe and to have biologic activity. In practice, some patients will require titration of their dose of darbepoetin alfa in order to achieve optimal hematopoietic response. In general, however, the model predicts that a fixed darbepoetin alfa dose of 200 µg every 2 weeks and a weight-based dose of 3 µg/kg every 2 weeks would provide similar efficacy for the majority of cancer patients with chemotherapy-induced anemia.
Previous work has demonstrated that darbepoetin alfa administered at3 µg/kg every 2 weeks was as efficacious as a standard regimen of rHuEPO ineliciting hemoglobin increases in cancer patients with chemotherapy-induced anemia. The Monte Carlo clinical trial simulation presented in thispaper demonstrated the comparability between these observed clinical data fromweight-based dosing of darbepoetin alfa and a simulation of weight-based dosing.Additionally, simulated hematopoietic outcomes from a fixed dose of darbepoetinalfa were comparable with those from both simulated and actual weight-baseddosing.
In considering the feasibility and utility of a fixed dose for darbepoetinalfa (eg, a regimen independent of a patient’s body weight) for the treatmentof cancer patients with chemotherapy-induced anemia, the following questionsmust be addressed:
Simulated hematopoietic response to darbepoetin alfa by patient weight classindicated only a slight weight effect with a fixed-dose regimen in terms ofchange in hemoglobin from baseline. Importantly, given the data on weightdistribution from a typical oncology patient population, the majority ofpatients would be appropriately initiated on treatment with a fixed darbepoetinalfa dose of 200 µg every 2 weeks. Regardless of the use of weight-based orfixed dosing for erythropoietic agents, due to the multiple variables that causeanemia and influence erythropoiesis in these patients, standard practicerequires dose titration to the desired hemoglobin concentration.
The minor role that body weight plays in determining hematopoietic responsefor any given individual is not unexpected given the overall complexity of thedisease state. On a population level, responsiveness to erythropoietic factors(quantified by the EC50) appears to be determined primarily by disease state.For example, patients with chronic kidney disease are generally more responsiveto erythropoietic factors than are cancer patients with underlying anemia, whoin turn are more responsive than cancer patients with chemotherapy-inducedanemia; this phenomenon is manifested in varying dose requirements for theseagents. Chronic kidney disease is characterized by a deficiency in endogenouserythropoietin that requires simple replacement therapy, whereas cancer patientsreceiving chemotherapy have compromised bone marrow that interferes with theirability to respond normally to erythropoietic agents.
Other potential factors include baseline hemoglobin and endogenouserythropoietin concentrations, concurrent medications, and nutritional factors.The lower the EC50 of an agent, the more responsive the patient. Sensitivityanalyses have verified that the variability in EC50 of darbepoetin alfa (rangingfrom 0.062 to 9.023 ng/mL) is far more important with regard topharmacodynamic effect than are body weight differences in the population.
Safety issues, of theoretical concern with a fixed dose at lower body weightextremes, are minimal with darbepoetin alfa. Commonly reported adverse events incancer patients receiving darbepoetin alfa were consistent with those in anemicpatients receiving cytotoxic chemotherapy, and included fatigue, edema, nausea,vomiting, diarrhea, fever, and dyspnea.
In terms of overall cost efficiency of a fixed dose for darbepoetin alfa, theweight distribution of a typical cancer population indicates that most patientswould receive a dose similar to the 3 µg/kg every-2-week target, and those whowould receive more drug at a fixed dose of 200 µg every 2 weeks vs theweight-based dose are approximately balanced by those who would receive less.Importantly, by using the entire contents of a 200-µg vial, no drug would bediscarded; furthermore, labor costs associated with measuring an exactweight-based dose of drug would be eliminated.
In summary, the clinical trial simulation presented in this paper indicatedthat a fixed 200-µg dose of darbepoetin alfa administered every 2 weeks would beas effective as a weight-based dose of 3 µg/kg every 2 weeks in amelioratinganemia in patients with solid tumors who are receiving chemotherapy. Adefinitive demonstration of this inference would require a prospective clinicaltrial. There are minimal safety concerns for patients at a lower body weightextreme based on cumulative clinical experience with darbepoetin alfa. Forpatients who display an inadequate (or excessive) hematopoietic response to thisstarting fixed-dose regimen, standard practice calls for titration of doseaccording to resulting hemoglobin concentrations.
1. Groopman JE, Itri LM: Chemotherapy-induced anemia in adults: Incidence andtreatment. J Natl Cancer Inst 91:1616-1634, 1999.
2. Silver RT: New approaches to the treatment of cancer-related anemia. SeminOncol 21:1-2, 1994.
3. Mercadante S, Gebbia V, Marrazzo AS, et al: Anaemia in cancer:Pathophysiology and treatment. Cancer Treat Rev 26:303-311, 2000.
4. Cella D: Factors influencing quality of life in cancer patients: Anemiaand fatigue. Semin Oncol 25(suppl 7):43-46, 1998.
5. Sabbatini P: The relationship between anemia and quality of life in cancerpatients. Oncologist 5(suppl 2):19-23, 2000.
6. Henry DH, Abels RI: Recombinant human erythropoietin in the treatment ofcancer and chemotherapy-induced anemia: Results of double-blind and open-labelfollow-up studies. Semin Oncol 21:21-28, 1994.
7. Glaspy J, Bukowski R, Steinberg D, et al: Impact of therapy with epoetinalfa on clinical outcomes in patients with nonmyeloid malignancies during cancerchemotherapy in community oncology practice. J Clin Oncol 15:1218-1234, 1997.
8. Gabrilove JL, Cleeland CS, Livingston RB, et al: Clinical evaluation ofonce-weekly dosing of epoetin alfa in chemotherapy: Improvements in hemoglobinand quality of life are similar to three-times-weekly dosing. J Clin Oncol19:2875-2882, 2001.
9. Egrie JC, Browne JK: Development and characterisation of novelerythropoiesis stimulating protein (NESP). Br J Cancer 84 (suppl 1):3-10,2001.
10. Macdougall IC, Gray SJ, Elston O, et al: Pharmacokinetics of novelerythropoiesis stimulating protein compared with epoetin alfa in dialysispatients. J Am Soc Nephrol 10:2392-2395, 1999.
11. Glaspy JA, Jadeja JS, Justice G, et al: Darbepoetin alfa given every 1 or2 weeks alleviates anaemia associated with cancer chemotherapy. Br J Cancer87:268-276, 2002.
12. Pirker R, Vansteenkiste J, Gateley J, et al: A phase III, double-blind,placebo-controlled randomised study of novel erythropoiesis-stimulating protein(NESP) in patients undergoing platinum treatment for lung cancer. Eur J Cancer37(6):264, 2001.
13. Amgen Inc. Aranesp (darbepoetin alfa) prescribing information. ThousandOaks, California, 2002.
14. Ortho Biotech Products, LP. Procrit (epoetin alfa) prescribinginformation. Raritan, New Jersey, 2000.
15. Scott S: Dose conversion from recombinant human erythropoietin todarbepoetin alfa: recommendations from clinical studies. Pharmacother Suppl 22(9pt 2):160S-165S, 2002.
16. Kotasek D, Albertsson M, Mackey J, et al: Randomized, double-blind,placebo-controlled, dose-finding study of darbepoetin alfa administered onceevery 3 (Q3W) or 4 (Q4W) weeks in patients with solid tumors. Proc Am SocClin Oncol 21:356a, 2002.
17. Peck CC, Barr WH, Benet LZ, et al: Opportunities for integration ofpharmacokinetics, pharmacodynamics, and toxicokinetics in rational drugdevelopment. Clin Pharmacol Ther 51:465-473, 1992.
18. Hale M, Gillespie WR, Gupta SK, et al: Clinical trial simulation:Streamlining your drug development process. Appl Clin Trials 5:35-40, 1996.
19. Holford NHG, Kimko HC, Monteleone JPR, et al: Simulation of clinicaltrials. Annu Rev Pharmacol Toxicol 40:209-234, 2000.
20. Sheiner LB, Steimer JL: Pharmacokinetic/pharmacodynamic modeling in drugdevelopment. Annu Rev Pharmacol Toxicol 40:67-95, 2000.
21. Bonate PL: A brief introduction to Monte Carlo simulation. ClinPharmacokinet 60(1):15-22, 2001.
22. Farrar D, Allen B, Crump K, et al: Evaluation of uncertainty in inputparameters to pharmacokinetic models and the resulting uncertainty in output.Toxicol Lett 49:371-385, 1989.
23. Bois F, Woodruff T, Spear R: Comparison of three physiologically basedpharmaco-
kinetic models of benzene disposition. Toxicol Appl Pharmacol 110:79-88, 1991.
24. Hattis D, White P, Marmorstein L, et al: Uncertainties in pharmacokineticmodeling of perchloroethylene. I. Comparison of model structure, parameters, andpredictions for low-dose metabolism rates for models derived by differentauthors. Risk Analysis 10:449-458, 1990.
25. Gearhart JM, Mahle DA, Greene RJ, et al: Variability of physiologicallybased pharmacokinetic (PBPK) model parameters and their effects on PBPK modelpredictions in a risk assessment for perchlorethylene (PCE). Toxicol Lett(Shannon) 68:131-144, 1993.
26. Krewski D, Wang Y, Bartlett S, et al: Uncertainty, variability, andsensitivity analysis in physiological pharmacokinetic models. Biopharmacol Stat5:245-271, 1995.
27. Glaspy J, Jadeja J, Justice G, et al: Optimizing the management of anemiain patients with cancer: A randomized active-controlled study investigating thedosing of darbepoetin alfa. Proc Am Soc Clin Oncol 21:362a, 2002.
28. Heatherington AC, Cosenza ME, Watson A, et al: Establishment of a PK-PDrelationship for novel erythropoiesis stimulating protein (NESP) in dogs.Pharmaceut Sci Suppl 1(4), 1999.
29. Schumitzky A, Jelliffe R, van Guilder M: NPEM. A program forpharmacokinetic population analysis. Clin Pharmacol Ther 55:163, 1994.