Cancer Chemotherapy & Biotherapy: Principles & Practices, 4th Edition


Jerry Collins

Jeffrey G. Supko

It has become generally accepted that the biologic effects of a drug are related to the time course of the concentration of the administered compound or an active metabolite in the bloodstream. The realization of this association has evolved as advances were made in the discipline of pharmacokinetics, which is defined as the study of rate processes involved in the absorption of drug from the administration site into the bloodstream, its subsequent distribution to extravascular regions throughout the body, and its eventual elimination from the body. From a broader perspective, pharmacokinetics may be thought of as the effect that the body has on a drug, whereas the pharmacologic effects that a drug has on the body are the realm of pharmacodynamics.

In anticancer chemotherapy, the general goal of killing tumor cells or inhibiting their proliferation and metastasis is clearly defined. However, in most cases we are severely limited by an inability to deliver drugs in a manner that separates antitumor effects from normal tissue toxicity. Much remains to be learned about the differences between normal and tumor tissues that can be exploited therapeutically. Thus, although pharmacokinetics is a tool that can be used to evaluate the feasibility of a drug delivery strategy based on intended pharmacodynamic effects, it is not a replacement for knowledge of exploitable differences between host and tumor.

Studies to characterize the pharmacokinetic behavior of a drug have become integral to the preclinical and clinical development of new anticancer agents. One group1 has even suggested that “it is now inconceivable to perform clinical research in cancer chemotherapy without obtaining adequate pharmacokinetic data.” Some of the usual objectives for undertaking a pharmacokinetic study in the context of a phase I or II clinical trial in cancer patients are: (a) initial characterization of the pharmacokinetic behavior of new chemotherapeutic agents in humans, (b) assessing whether or not an administration schedule provides a potentially effective pattern of systemic exposure to drug, (c) determining the magnitude of intrapatient and interpatient variability in pharmacokinetic parameters, (d) assessing the influence of patient characteristics on drug disposition, (e) establishing predictive correlations between biologic effects and pharmacokinetic parameters, and (f) determining whether combining drugs results in pharmacokinetic interactions. In addition, pharmacokinetic drug level monitoring has been used to improve therapy through dose individualization, to evaluate patient compliance during chronic therapy, and to assess whether alterations in drug disposition or metabolism are associated with the development of toxicity or the lack of effect.

The fundamental obstacle to greater success in the application of pharmacokinetics and clinical drug level monitoring to anticancer therapy is our limited knowledge of pharmacodynamics. A complete understanding of the actions of a drug necessarily requires discerning the nature of the association between its pharmacokinetic behavior and pharmacodynamic effects. Relationships between pharmacokinetics and the severity of toxicity have been established for many anticancer drugs. However, pharmacokinetic associations accounting for the therapeutic effects of a chemotherapeutic agent are more difficult to establish because of the multiplicity of factors involving the host and tumor that influence response, as noted above, as well as the time lapse from initiating treatment to the first indications of a therapeutic response. Nevertheless, elucidating the pharmacokinetic behavior of an anticancer drug may benefit efforts to determine the dose, route of administration, and schedule that maximizes the potential for therapeutic effectiveness while minimizing the likelihood of serious toxic effects.

The intention of this chapter is to provide readers with a fundamental understanding of clinical pharmacokinetics and its practical application to the development and use of anticancer chemotherapy. Numerous texts with widely varying levels of complexity and focus are available for those interested in a more comprehensive discourse of the subject, ranging from easily understood introductions to the discipline2 to more advanced texts with a mathematical approach.3


Sample Collection and Drug Concentration Measurement

Pharmacokinetic studies involve collecting serial specimens of blood and other biologic fluids, such as urine, at predetermined time intervals from subjects following administration of the drug. Plasma is the blood component in which drugs are most commonly measured during pharmacokinetic studies, although determinations are also made in serum and, less frequently, whole blood. The concentration of drug present in the study samples is measured using an appropriate bioanalytical method. Technical advances in separation and detection methods, especially the maturation of high-performance liquid chromatography coupled to mass spectrometry into a technique suitable for routine use, have provided a greatly improved basis for drug concentration measurement during the past decade. Review articles surveying the current techniques used for assaying drugs in biologic fluids regularly appear in the literature.4

Many anticancer drugs are difficult to measure because of inherent instability, either spontaneously degrading or being degraded by enzymes in blood or tissues. It is therefore important to recognize that the quality of data derived from any pharmacokinetic study ultimately depends on the reliability of the assay used to measure the drug, as well as the manner by which samples were processed and stored prior to analysis. The majority of bioanalytical methods used for pharmacokinetic studies measure the total concentration of drug, that is, free drug plus that which is reversibly associated with plasma proteins. However, the reversible binding of a drug to plasma proteins, such as albumin and α1-acid glycoprotein, needs to be considered in the interpretation of total drug concentrations.5 Only the free or unbound drug is pharmacologically active. Protein binding is usually assessed experimentally by ultrafiltration or equilibrium dialysis.

The Plasma Concentration-Time Profile

Except for cases in which a drug is given by bolus intravenous injection, the plasma concentration-time (CxT) profile of any drug exhibits an initial region of increasing concentration, the achievement of a peak or maximum concentration (Cmax), followed by a continual decline in concentration (Fig. 3.1A). The concentration of drug in plasma increases as long as the rate of input into systemic circulation exceeds the rate of loss due to distribution into other extracellular fluids, intracellular spaces, and tissues throughout the body, and elimination from the body. The Cmax is achieved when the rate of drug input is equivalent to the rate of loss from plasma, which occurs at the instant that an intravenous injection or short infusion is terminated. During a continuous intravenous infusion, plasma levels of the drug increase at a progressively decreasing rate and eventually become constant, indicative of achieving steady-state conditions, if the infusion is continued for a sufficiently long time (Fig. 3.1C).

Figure 3.1 shows the same CxT data plotted on graphs with semilog axes (panel A) and rectangular coordinate axes (panel B). Presenting pharmacokinetic drug CxT profiles on semilog graphs provides a better visual depiction of the entire data set than a coordinate plot because plasma levels of a drug frequently differ by several orders of magnitude during the course of the observation period. Furthermore, the concentration of many drugs in systemic circulation decays in an apparent first-order manner, exemplified by a terminal region in the plasma profile in which the logarithm of the drug concentration is a linear function of time. Thus, a semilog plot provides some immediate inferences regarding the nature of the pharmacokinetic behavior of a drug.

The pattern of decay in the plasma concentration of a drug that exhibits first-order kinetics comprises one or more exponential phase. In the case of a plasma profile with drug concentrations that decline in a single log-linear phase, the entire body appears to be kinetically homogenous, with equilibrium of the drug between plasma and other fluids or tissues into which it distributes being very rapidly achieved, before the first blood specimen has been acquired. Polyexponential behavior results from distinguishable differences in the reversible transfer of drug from plasma to various regions or compartments of the body. Thus, for example, the presence of two exponential decay phases implies that the body behaves as if it is composed of two kinetically distinct compartments, comprising plasma and tissues with which equilibrium is rapidly established, and a second compartment comprising all other regions of the body into which drug distributes more slowly.

For some purposes, a mathematical equation or model is necessary to interpret pharmacokinetic data, but often questions may be answered without a formal model construction. Recently, there has been a growing trend toward analyzing pharmacokinetic data by empirical approaches that consider only the concentration of drug in the sampled fluid and require few assumptions about model structure. In these techniques, which include model-independent analysis6and noncompartmental analysis,7 the various exponential decay phases are usually referred to simply as the initial, intermediate, and terminal disposition phase. Regardless of the particular method of analysis employed, the ultimate objective is the same, which is to estimate values of descriptive pharmacokinetic parameters from the CxT data.

Figure 3.1 A. Plasma concentration-time profile for a 175 mg/m2 dose of paclitaxel administered as a 3-hour continuous intravenous infusion shown on a graph with log-linear axes. Pharmacokinetic variables that can be estimated by visual inspection are indicated: maximum drug concentration (Cmax), the time at which the peak concentration (tmax) occurs, and the biological half-life (t1/2, z). B. Presentation of the same data shown in the upper panel on rectangular coordinate axes. The shaded area corresponds to the area under the curve (AUC). C. Time course of paclitaxel in plasma when given as a 96-hour continuous intravenous infusion at a rate of 25 mg/m2 per day. The steady-state plasma concentration of the drug is approximately 40 nM.

Physiologic Pharmacokinetic Models

For pharmacologists interested in developing an understanding of drug disposition in individual tissue compartments, models that incorporate physiologic compartments are of considerable interest. These models require measurements of actual physiologic parameters, such as volumes and blood flow rates, as well as drug concentrations in various compartments, and therefore are based primarily on data from experimental animals. Entry into specific areas such as the central nervous system may be of critical importance in the use of drugs, and physiologic models can allow comparisons of CxT profiles for various schedules and routes of administration. Physiologic models have been constructed for many anticancer drugs. Models have been published for the most important drugs in clinical practice, among which are methotrexate (MTX),8 5-fluorouracil (5-FU),9 cisplatin,10 and doxorubicin.11

In the most general form, physiologic pharmacokinetic models are overly complex and require too large a database for routine clinical use. However, they provide a basis for understanding a drug's kinetic behavior that can be incorporated into simpler models, either physiologic or hybrid, assimilating both empiric observations and physiologic information. Physiologic modeling goes beyond the usual goals of empiric pharmacokinetic modeling to allow for incorporation of data into the model that has been obtained in other species or in vitro. The compartments comprising a physiologic pharmacokinetic model have an anatomic basis, and the transfer processes in the model have a physiologic or pharmacologic identity. Each organ is modeled separately; then, the model connections are provided by blood flow. The structure for the physiologic model for cytarabine is presented in Figure 3.2.12


Area Under the Curve

Noncompartmental analysis is considerably simpler than any equation-defining method of pharmacokinetic data analysis. All calculations and data manipulations can be performed by most spreadsheet software programs. The observed plasma CxT data is numerically integrated, most commonly by the trapezoidal method. In its simplest application, each successive set of data points, beginning with time zero, is used to define a trapezoid, the area of which is readily calculated. The cumulative sum of the areas of all such trapezoids affords an estimation of the area under the CxT curve to the last sample with a measurable drug concentration ([Ct] AUCOt). The slope of the terminal log-linear phase of the CxT profile (-λlz) is then determined by linear regression using log-transformed concentration values (see Fig. 3.1A). The area under the curve from time zero to infinity (AUC) can then be calculated as

Figure 3.2 Physiologic pharmacokinetic model for cytosine arabinoside. GI, gastrointestinal. (Reprinted with permission from Dedrick RL, Forrester DD, Cannon JN, et al. Pharmacokinetics of 1-β-D-arabinofuranosylcytosine (ARA-C) deamination in several species. Biochem Pharmacol 1973;22:2405–2417.)

Although the AUC is not a pharmacokinetic parameter per se, because its magnitude depends on the administered dose of drug, it represents an important quantitative measure of total systemic drug exposure, as illustrated in Figure 3.1B. In addition, knowledge of the AUC is required to calculate values of pharmacokinetic parameters, as described in the following section.

Total Body Clearance

The total body clearance (CL) of a drug is formally defined as the volume of plasma from which drug is completely removed per unit time. It is readily calculated as

where Div is the dose of the drug given by intravenous injection or infusion. CL reflects the combined contribution of all processes by which drug is removed from the body, as represented by the equation

where CLR and CLNR designate renal and nonrenal clearance, respectively.13 Renal clearance is usually the only route of drug elimination that can be directly and quantitatively determined in patients by noninvasive procedures. All other mechanisms of drug elimination that cannot be readily estimated, including biliary excretion of unchanged drug, metabolism, nonenzymatic irreversible reactions with endogenous molecules, and spontaneous chemical degradation, are grouped together as CLNR. CL has units of volume per time (e.g., milliliters per minute, Liter per hour) and is frequently normalized to the body weight or body surface area of subjects (e.g., milliliters per minute per kilogram, liter per hour per square meter) under the presumption of minimizing interpatient variability in the magnitude of the parameter. However, this practice has recently become a topic of considerable controversy because the underlying presumption of a relationship between unnormalized clearance values and body surface area does not exist for a significant number of anticancer drugs.14 The CL values are often compared with glomerular filtration rate and hepatic blood flow, average values of which are approximately 125 mL/min (4.6 liter/hr per square meter) and 1,500 mL/min (56 liter/hr per square meter), respectively, in normal adults.15, 16 Although often informative, these comparisons can be extremely misleading unless the extent of plasma protein binding has been taken into account because only the free fraction of drug that is not bound to plasma proteins is usually subject to organ-mediated excretion or metabolism.

Apparent Volume of Distribution

The total body apparent volume of distribution, Vz, is strictly a proportionality constant relating the total amount of drug in the body to plasma concentration. It may be calculated by the equation

and has units of volume, typically expressed in terms of milliliters or liters normalized to body weight or body surface area (e.g., milliters per kilogram, liters per square meter). Vz is designated as an apparent volume because it is a hypothetical value that is not directly related to any real physiologic space. Nevertheless, it is an informative parameter, providing an indication of the relative extent of drug distribution from plasma. Specifically, for a given amount of drug in the body, the fraction present in plasma decreases as its distribution into peripheral tissues increases, leading to greater values of Vz.17 Therefore, the effective lower limit of Vz is the plasma volume, which is approximately 4.5% of body weight (i.e., 45 mL/kg, 1.7 liter/m2) for a normal adult. There really is no upper limit, as Vz can assume extremely large values in cases where the half-life of the terminal disposition phase is long relative to that of the preceding disposition phase, and drug levels decrease by several orders of magnitude before the terminal phase is achieved. For example, some anticancer agents, such as the anthracyclines, have Vz values exceeding 1,000 liters/m2 (27 times body weight).

Biologic Half-Life

The biologic half-life of a drug (t1/2, z) is the time required for its plasma concentration to decrease by 50% any time during the terminal log-linear phase in the CxT profile (see Fig. 3.1A). It is only applicable to drugs that exhibit apparent first-order pharmacokinetics (see later discussion). As indicated by the relationship

t1/2, z reflects both the ability of the body to eliminate the drug as well as the extent to which the drug distributes throughout the body. Nevertheless, there is a recurrent tendency in the anticancer drug literature to place undue emphasis on the value of t1/2, z as an indicator of drug elimination. The t1/2, z has an important practical application in that steady-state conditions during administration of a drug by continuous intravenous infusion or a multiple dosing regimen are achieved when the duration of treatment exceeds 4 times the value of t1/2, z.

Linear and Nonlinear Pharmacokinetics

The majority of clinically used anticancer agents exhibit linear or first-order pharmacokinetics, whereby plasma concentrations of the drug decline in an exponential manner following intravenous administration. A distinguishing and defining characteristic of linear pharmacokinetics is that the plasma concentration of drug at a given time after dosing is directly proportional to the administered dose. Thus, the AUC increases proportionately with the dose and values of the pharmacokinetic parameters (i.e., CL, Vz) and are independent of the dose. When a drug is predominantly eliminated by a potentially saturable process, such as hepatic metabolism or active tubular secretion, departures from linear pharmacokinetic behavior may become evident if sufficiently high doses can be administered to patients. As illustrated in Figure 3.3, classic nonlinear pharmacokinetics is indicated by a change in the appearance of the plasma profile from exponential character at lower doses to the appearance of a distinct downward curvature in the semilog plot of the plasma profile at higher doses.18 In addition, the apparent CL exhibits a progressive decrease in magnitude as the dose is escalated. A clear example of this phenomenon was reported recently for high-dose cytarabine given by continuous intravenous infusions in which small changes in the infusion rate produced disproportionately large increases in the steady-state drug concentration in plasma.19

Figure 3.3 Plasma profiles of 5-flourouracil determined at doses of 25 mg/m2(▪), 125 mg/m2 (), and 375 mg/m2 (●) illustrating the effect of classic nonlinear pharmacokinetics. Values of the apparent total body clearance decreased progressively from 142 liter/hr per square meter for the 25 mg/m2dose to 47 liter/hr per square meter at 125 mg/m2 and 30 liter/hr per square meter at 375 mg/m2. There would be no significant difference between the clearance determined at different doses if the pharmacokinetic behavior of the drug was linear.



Renal and Hepatic Excretion

Establishing the major pathways of drug elimination in patients is also an important objective of clinical pharmacokinetic studies. Disease states that compromise the function of a major drug-eliminating organ, such as the kidneys or liver, can enhance a patient's sensitivity to the toxic effects of the drug as a result of increased drug exposure. For this reason, patients with significant organ impairment are usually excluded from initial phase I studies to avoid possibly confounding sources of toxicity.

Renal excretion is a quantitatively significant route of elimination for many relatively small compounds, with molecular weights less than about 300, that are also highly to moderately hydrophilic,20 if they are not substantially metabolized. Larger compounds and those with a more lipophilic character tend to be predominantly eliminated by biliary excretion, either directly or after metabolism. Determining CLR involves measuring the amount of unchanged drug present in the urine (Ae) collected during one or more defined time intervals (Δt) following intravenous drug administration. It may be calculated by either of the following equations

depending on whether urine has been continuously collected and pooled from the beginning of dose administration throughout the time that plasma specimens were obtained, or during one or more discrete time intervals after dosing. In the second equation, Cmid is the plasma concentration of drug at the midpoint of the urine collection interval. The amount of unchanged drug in feces cannot be taken as a direct indication of biliary excretion because of the potential for drug metabolism by the gastrointestinal microflora.21

In cases in which renal or biliary excretion are significant pathways of drug elimination, a predictive correlation may exist between clinical indicators of renal or hepatic function, such as serum creatinine and bilirubin levels, respectively, and CL. Establishing these relationships serves as the basis for defining guidelines pertaining to the minimal organ function required for patient eligibility in phase II studies and devising an empirical algorithm for dosage adjustment, including those documented in Table 3.1.22 (See also “Eliminating Organ Dysfunction.”)

Drug Metabolism

Metabolism represents a quantitatively important route of elimination for most anticancer agents. Xenobiotic biotransformation reactions may be broadly categorized into two classes, designated phase I and phase II. The principal phase I reactions are oxidation, reduction, and hydrolysis. Phase II reactions involve the conjugation or coupling of endogenous molecules, including glucuronide, sulfate, amino acid, methyl and glutathione moieties to the parent drug or a precursory phase I metabolite. Hepatic oxidation mediated by the cytochromes P450 (CYP450), a large family of heme-containing isozymes, undoubtedly plays the greatest overall role in drug metabolism among the phase I reactions.23 The CYP450 enzymes are most abundantly expressed in the liver, but they are also present in the kidney, lung, and gastrointestinal epithelium. The predominant enzyme in this family, CYP3A4, catalyzes the oxidation of a multitude of structurally diverse compounds.24, 25, 26 These include imatinib, gefitinib, docetaxel, etoposide, ifosfamide, vincristine, and paclitaxel. In addition to hepatic metabolism, some important phase I reactions are mediated by ubiquitous enzymes found in virtually all tissues of the body, such as dihydropyrimidine dehydrogenase, which catalyzes the reduction of 5-FU, and cytidine deaminase, which inactivates cytarabine.27, 28 Glucuronide conjugation catalyzed by uridine diphosphate glucuronosyl-transferases (UGT) is the most commonly encountered phase II reaction. In contrast to phase I metabolism, which may yield a biologically active product, glucuronidation almost exclusively represents a detoxification mechanism that inactivates a compound and facilitates its excretion through enhanced hydrophilicity and recognition by biliary canicullar efflux proteins.29 Glucuronidation is a clinically important route of elimination for 7-ethyl-10-hydroxycamptothecin (SN-38), the active metabolite of irinotecan, as the extent of its glucuronidation has been associated with the risk of severe diarrhea for the weekly treatment schedule of irinotecan.30


Major Route of Elimination

Anticancer Agent

Dose Adjustment for Organ Dysfunctiona

Renal excretion



Carboplatin, Cisplatin














Hepatic metabolism


















Vinca alkaloids






Ubiquitous enzymes







Nonenzymatic hydrolysis







Biliary excretion





Vinca alkaloids


ab, Decrease dose in proportion to the reduction in creatinine clearance below 60 mL/min. c, Serum bilirubin: 1.5–3.0 mg/100 mL, 50% dose reduction; >3.0 mg/100 mL, 75% dose reduction. d, Patients with S-methyl transferase deficiency. e, Insufficient data to determine if dose reduction is necessary in hepatic dysfunction.
bEnzymatic or spontaneous chemical reactions required for drug activation.

Chemical Degradation

Chemical degradation can be a significant elimination mechanism for drugs that are susceptible to hydrolysis, such as many of the alkylating agents. Nonenzymatic reactions between drugs and endogenous molecules can also contribute prominently to elimination. For example, platinum alkylating agents form covalent adducts with serum albumin.31


Obtaining an indication of interpatient variability in the values of pharmacokinetic parameters and related variables are important objectives of phase I trials. This information has considerable practical utility with regard to clinical drug development. These findings provide the basis for assessing the ability to reliably predict the Cmax and AUC of a drug following the administration of any given dose to patients who have not been previously studied. The recommended dose of cytotoxic anticancer drugs is typically close to the maximum tolerated dose, and dose-limiting toxicities are often related in some manner to the levels of drug achieved in plasma. Thus, the margin of safety of these agents very much depends on the consistency of their pharmacokinetic behavior between patients. Conversely, the existence of a high degree of interpatient pharmacokinetic variability can result in unpredictable episodes of toxicity at the maximum tolerated dose, which may make it difficult to establish a potentially effective and safe dose. Although rarely employed in these circumstances, drug level monitoring to establish the optimal dosing regimen in individual patients may be warranted.

Patient Characteristics

Clinically significant associations between CL and patient characteristics such as age, sex, and race have been identified for many anticancer drugs. For example, it has been shown that the CL of 5-FU in females is significantly lower than in males and that formation of the glucuronide metabolite of SN-38 by UGT is subject to pharmacogenetic variations related to both race and gender.32, 33 These factors are now being examined extensively during the clinical evaluation of new anticancer drugs.

Currently, more than half of all cancers are presented by patients over 60 years old. However, relatively few elderly patients are entered into early-stage clinical trials because of referral patterns and investigator bias. As a consequence, the pharmacokinetic behavior of most anticancer drugs has not been adequately characterized in elderly patients.34 There is a very high degree of heterogeneity in older cancer patients as a result of natural changes in body composition, including decreased muscle mass, increasing adipose tissue, and decreased renal function that occur with advancing age. Normal aging is accompanied by a 25 to 35% decrease in liver volume and a 35 to 40% decrease in hepatic blood flow.35, 36 Thus, the CL of drugs with a high hepatic extraction ratio, which is limited by liver blood flow, may be decreased in the elderly.37, 38Age-associated changes in the function of some drug-metabolizing enzymes have been identified but their clinical significance remains uncertain.39, 40

At the other end of the age spectrum, experience has shown that safe and effective doses of anticancer agents for children very often cannot be based simply on body weight or body surface area scaling of an adult dosage.41 Age-related changes in the enzymatic and excretory systems that are involved in drug elimination can have a profound effect on pharmacokinetics.42 Thus, the rational use of drugs in pediatric patients that are known to be eliminated primarily by hepatic metabolism in adults may depend on thoroughly characterizing its pharmacokinetics and metabolism in children of various ages. Similarly, the potential for interaction of a new agent, that is a potential substrate or inhibitor of hepatic CYP450 enzymes, with other drugs that may be concurrently administered to pediatric patients also warrants careful evaluation.

Eliminating Organ Dysfunction

Physiologic conditions that affect hepatic or renal function, including blood flow to the liver or kidneys, can have a dramatic effect on the pharmacokinetic behavior of a drug in individual patients.43 Powis44 reviewed the effects of both renal and hepatic dysfunction for anticancer drugs. The estimation of creatinine clearance from serum creatinine concentration is a conveniently measured indicator of renal function. Hepatic function is more difficult to quantify. Serum transaminase and bilirubin concentrations provide indirect but somewhat useful information on hepatic function. Empirical guidelines for dose reduction in patients with underlying renal or hepatic dysfunction are devised by establishing relationships between these biochemical parameters and CL. In general, these adjustments would be expected to be less precise than adjustments based on drug-level measurements. Occasionally, there is a close relationship between a renal function indicator and plasma pharmacokinetics. Egorin et al45 have elegantly applied such correlations for dose adjustments of carboplatin (Fig. 3.4) and hexamethylene bisacetamide.46 In fact, individualizing the dose of carboplatin to target a specific AUC value based on estimated creatinine clearance in patients has become a routine clinical practice.47 Table 3.1 summarizes the recommended dose modifications for the standard anticancer drugs.

Figure 3.4 Relationship between thrombocytopenia and plasma levels of carboplatin. AUC, area under the concentration × time curve. (Reprinted with permission from Egorin MJ, Van Echo DA, Olman EA, et al. Prospective validation of a pharmacologically based dosing scheme for the cis-diamminedichloroplatinum(II) analog diamminecyclobutanedicarboxylatoplatinum. Cancer Res 1985; 46:6502–6506.)

Drug Interactions

Essentially all treatment protocols include combinations of drugs, encompassing two or more anticancer drugs, as well as various other drugs related to general symptomatic and supportive therapy of the patient. Many adjuvant medications that are routinely used in the management of cancer patients can potentially affect the pharmacokinetics of chemotherapeutic agents by either inhibiting or enhancing metabolic elimination (Table 3.2). Because cytotoxic anticancer drugs are usually administered at their maximum tolerated doses, there is a substantially greater risk for pharmacokinetic interactions resulting in clinically significant toxicity than exists with drugs for most other indications. Accordingly, the administration of an anticancer agent together with another drug that has the potential to modulate the activity of an enzyme that represents a major pathway of its elimination should be avoided whenever possible. Another consideration that should be recognized is that highly variable pharmacokinetics are frequently exhibited by drugs predominantly eliminated by hepatic metabolism.48 It is becoming increasingly apparent that genetic polymorphisms and mutations affecting key drug-metabolizing enzymes may account for aberrant pharmacokinetics in a minority of patients, or an otherwise high degree of interpatient variability.49

The serious adverse reactions caused by administration of ketoconazole to patients taking terfenadine,50 which had been widely used and considered to be a relatively safe antihistamine, provide a cautionary note for potential interactions with anticancer drugs because of their much narrower therapeutic index. Another common drug, cimetidine, is reported to inhibit the metabolism of cyclophosphamide51 and hexamethylmelamine.52On the other hand, anticancer drugs are reported to interfere with the absorption of noncancer drugs, such as digoxin.53 Balis54 has reviewed the literature of drug interactions related to anticancer drugs. When evaluating drug-drug interactions, recent findings with paclitaxel illustrate the difficulties generated by interspecies differences in metabolic pathways.55


Chemotherapeutic Agent

Interacting Drug

Effect on Clearance of Anticancer Agent

Probable Mechanism



CYP450 enzyme induction


Cyclosporin A

Inhibit biliary excretion



CYP450 enzyme induction





Inhibit xanthine oxidase




Inhibit tubular secretion




Inhibit CYP450 metabolism or biliary excretion



Inhibit tubular secretion



Inhibit CYP450 metabolism

Chemotherapeutic agents that are metabolized by the hepatic CYP450 system, especially members of the CYP3A subfamily, are particularly prone to pharmacokinetic interactions from the multitude of drugs and compounds of dietary origin that are inhibitors or inducers of CYP450.56 A particularly serious example is the use of the dietary supplement, St. John's wort. This product induces drug-metabolizing enzymes and produces lack of drug efficacy.57 Repeated daily administration of glucocorticoids, commonly used as antiemetics, can induce the expression of hepatic CYP450 and thereby enhance the CL of anticancer drugs that are CYP3A4 substrates.58 In addition to hepatic drug-metabolizing enzymes, there are examples of pharmacokinetic interactions resulting from effects directed on other enzyme systems, excretory pathways, and even drug absorption. Salicylates can reduce the renal tubular secretion of MTX.59, 60Morphine and its derivatives can alter the rate and extent of absorption of orally administered cytotoxic drugs by reducing gastrointestinal motility.61 As discussed in a subsequent chapter, some antiseizure drugs that are frequently used in the clinical management of patients with brain tumors have been shown to significantly enhance the clearance of many anticancer agents by inducing CYP450 enzymes.

Dose Individualization

The existence of a high degree of interpatient pharmacokinetic variability can result in unpredictable episodes of toxicity and make it difficult to establish a potentially effective and safe dose for the population. For the individual, clinical monitoring and pharmacokinetics offer the possibility of tailoring drug delivery to the particular patient's needs. The standard doses derived from group studies do not allow for interindividual variability. However, doses may be adjusted on the basis of direct measurements of drug concentration in the individual patient, indicators of renal or hepatic dysfunction, or interactions of the anticancer drug with concomitant medications. Under these circumstances, it may also be beneficial to individualize doses of the drug based on plasma levels of the compound afforded by a test dose or a biochemical parameter that is predictive of CL.62 Although this technique is rarely employed, dose individualization has significantly improved the outcome and minimized toxicity for children with B lineage ALL treated with MTX.63


For the average patient, or the general population, pharmacokinetics can help answer the fundamental questions in delivery of drugs: (a) What route of administration? (b) How much to give (dose)? (c) How often to administer (schedule)? These questions are answered using empiric observation (what works best in an experimental or clinical setting) as well as biochemical, cell kinetic, and pharmacokinetic considerations.

Routes of Drug Administration

The choice of drug administration route is based primarily on the ability to formulate an acceptable dose preparation for intravenous, oral, intramuscular, intrathecal, or subcutaneous use and pharmacokinetic assessment of the pattern of systemic drug exposure that they provide. Although current trends point toward the preferential development of orally administered drugs, cytotoxic anticancer drugs are still most commonly given by the intravenous route as this provides complete control over the actual dosage delivered to the systemic circulation and the rate at which it is presented. This results in maximum safety because the variability in systemic drug exposure between and within patients achieved with direct intravenous administration is typically much lower than that resulting from oral administration. Furthermore, for agents given by continuous intravenous infusion, drug delivery can be readily terminated, if necessary, because of the occurrence of an acute adverse reaction during administration.

All routes of administration other than intravenous, including oral, subcutaneous, intramuscular, intraperitoneal, and intrathecal delivery, involve an absorption process whereby dissolved drug molecules are transferred from the site of administration into the vasculature. Accordingly, drug given by an extravascular route is conceptualized as being outside the body until gaining access to the systemic circulation. Oral dosage forms are presently available for an increasing number of anticancer drugs including hydroxyurea, MTX, etoposide, idarubicin, gefitinib and imatinib. In the future, oral administration will undoubtedly attain greater prevalence from the clinical development of cytostatic antiproliferative agents that require chronic dosing for efficacy.

The bioavailability of a drug given by any extravascular route is defined as the rate and extent of absorption into systemic circulation. The absolute systemic availability (F) of a drug is ascertained by determining the AUC in the same patient following intravenous and extravascular administration of the agent, with an adequate time interval period between the two treatments. For the same dose given intravenously and extravascularly,

where D is the dose. In studies where an agent is administered exclusively by the oral route, CL and F are indeterminable, as explicitly indicated by the relationship

Many factors influence oral bioavailability, including release of the drug from the dosage form, dissolution of drug within the gastrointestinal tract, drug stability under conditions encountered in the gastrointestinal tract, transport of dissolved drug across the intestinal epithelium into the vasculature, and the extent of first-pass hepatic metabolism. Mercaptopurine is an example of a drug with very low and erratic bioavailability,64 whereas imatinib is a drug with consistently high bioavailability.65

Absorption through the lipid-bilayer cell membrane of the intestinal mucosa is determined by molecular size, lipid solubility, and the presence of transport systems. As cancer chemotherapy shifts increasingly towards oral drug delivery, the importance of many general carrier systems, such as the “ABC” transporters, is becoming more widely appreciated alongside such specialized carriers as the folate transport mechanisms for antifolates. The physiologic state of the intestinal tract may be affected adversely by disease or by previous drug therapy. Vomiting induced by chemotherapeutic drugs such as cisplatin may lead to loss of a major portion of an oral dose. In addition to intestinal absorption, presystemic metabolism and biliary excretion may prevent orally administered drugs from reaching the systemic circulation in an active form. Presystemic metabolism, also known as the “first-pass effect,” is a unique concern for the oral route because a drug is exposed to metabolism both in the gastrointestinal mucosa and in the liver, which it enters through the portal vein before returning to the heart.66

A tumor may grow in a region of the body, such as the central nervous system, that is not penetrated readily by systemically administered drugs. Accordingly, several unusual routes of administration have been implemented to maximize delivery of drugs to the site of the tumor and to reduce the deleterious effects associated with ordinary systemic administration. At least two of these routes have become accepted therapeutic practice: intrathecal delivery for meningeal leukemia67 and intravesical delivery for transitional-stage bladder carcinoma. As discussed in detail in Chapter 21, intrathecal administration has been used primarily to obtain adequate drug levels in the cerebrospinal fluid to eradicate cancer cells that are otherwise protected from effective therapy. Intra-arterial drug administration, especially hepatic arterial delivery, is another route that has been actively investigated but has not emerged as standard therapy.

Peritoneal dialysis continues to be evaluated as a delivery vehicle for anticancer drugs when disease is localized to the abdomen.68 The pharmacokinetic rationale suggests that tumor tissue may be exposed to high local concentrations, whereas systemic levels are no greater than normally encountered with intravenous therapy. In an analogous fashion to intrathecal delivery, only cells in close contact with the peritoneal fluid will benefit from this mode of drug delivery. The intraperitoneal route has been the subject of many pilot studies and formal phase I and phase II trials by our group and others. Some promising pharmacologic results have been obtained and more definitive therapeutic trials are in progress. Three randomized phase III trials totaling approximately 3,000 patients with ovarian cancer have shown an advantage for intraperitoneal delivery compared with intravenous delivery for both time to disease recurrence/progression69, 70 and survival.69, 71

Pharmacokinetic analysis can help to evaluate the potential usefulness of these approaches. Of course, the pharmacokinetic advantage of achieving greater drug exposure is not always associated with improved responses.


Dose is usually determined by an empiric phase I trial using a fixed treatment schedule, with stepwise evaluation of toxicity at progressively higher doses. In certain circumstances, dose also may be determined by setting pharmacologic objectives, such as a target drug concentration in a specific body compartment such as plasma, cerebrospinal fluid, or ascites. This type of regimen planning requires pharmacokinetic design and verification by drug level monitoring and has been used in only a few clinical oncologic settings, such as intrathecal chemotherapy with MTX and intraperitoneal therapy with MTX and 5-FU. Additional information on the relationship of drug concentration to tumor cell kill, as provided by in vitro assays, may provide a basis for more precise pharmacokinetic adjustment of dosage.

Pharmacologically guided dose escalation was developed as an alternative to the predetermined escalation procedures such as the modified Fibonacci method for phase I trials.72 After the first group of patients has been treated with the starting dose in a phase I clinical trial, the rate of dose escalation is determined by the plasma levels of drug relative to target plasma levels measured in mice at the maximum tolerated dose. With this approach, investigators can estimate the difference between the target concentration and plasma levels produced by the current dose level. Such information provides the opportunity to intervene at an early stage in the phase I trial. Cautious escalation may be indicated if it is determined that plasma levels of the drug are close to the target. If the current plasma levels are substantially below the targeted value, then a more rapid escalation of the dose could generate considerable savings in time and clinical resources, and fewer patients will be exposed to doses that have little potential of being therapeutically effective. Although this procedure is conceptually attractive and found support in Europe and Japan, as well as the United States,73, 74, 75 it has not been widely used, primarily because of logistical difficulties in its implementation.

Administration Schedule

The route and frequency of administration evaluated in the initial phase I trial of a cytotoxic anticancer agent is generally derived from the schedule that produces an optimal therapeutic effect against preclinical in vivo tumor models. There is an increasing interest in assessing the use of noncytotoxic compounds, such as cytostatic, differentiation-inducing, and antiangiogenic agents in the treatment of neoplastic diseases. However, accepted preclinical models to evaluate and refine in vivo efficacy for many classes of candidate noncytotoxic antiproliferative drugs do not presently exist. Under these circumstances, it would be reasonable to base the treatment schedule evaluated in initial phase I trials on that required to achieve the pattern of systemic exposure to drug in laboratory animals that best approximates the concentration and duration of exposure necessary for optimal in vitro activity.

Past experience has repeatedly demonstrated that impressive preclinical antitumor activity is not a reliable predictor of clinical efficacy. A reasonable argument can also be advanced to support the hypothesis that a candidate drug has little likelihood of being therapeutically effective unless a clinically tolerable dosing regimen provides a pattern of systemic exposure to the drug that is at least comparable with that required for activity against appropriate in vivo or in vitro preclinical models. Accordingly, when considered together with toxicologic and physiologic response factors, pharmacokinetic data acquired during phase I studies can facilitate efforts to optimize dosing regimens. Alternatively, withdrawing an agent from continued clinical development may be an option that warrants serious consideration in situations in which the plasma concentrations achieved in patients treated at the maximum tolerated dose are considerably lower than target levels, given the availability of limited clinical resources and ethical considerations of entering patients into a phase II trial of a compound that has little prospect of being therapeutically effective.

The schedule of drug administration depends highly on pharmacokinetic considerations and requires a choice of the duration of administration (e.g., bolus intravenous injection versus prolonged intravenous infusion), frequency of repeated dosing, and the sequencing of multiple drugs or drugs and other treatment modalities such as radiation. Bolus intravenous injection provides maximal peak drug levels in plasma but a rapid decline thereafter as the drug is eliminated from the plasma compartment by metabolism or excretion. This very convenient dosing method is appropriate for drugs that are not cell cycle-phase–dependent and therefore do not have to be present during a specific phase of the cell cycle. Examples are the alkylating agents, such as chloroethylnitrosoureas, nitrogen mustards, and procarbazine, as well as other drugs that chemically interact with DNA.

Administration by prolonged intravenous infusion (i.e., 6 to 120 hours) is advantageous for agents that act preferentially in discrete phases of the cell cycle, such as S-phase–specific drugs (e.g., cytarabine, MTX, camptothecins), particularly if the drug is rapidly cleared from systemic circulation. Prolonged infusions have the additional advantage of providing a specific and constant plasma concentration of the drug, a desirable feature if information regarding the chemosensitivity of the tumor is available, as determined experimentally by various in vitro tests. Intermediate-length infusions (i.e., 1 to 4 hours) may provide a means to overcome the acute toxicities that are produced by exposing host organs to high peak drug levels. Particularly for neurotoxic or cardiotoxic compounds, rapid intravenous infusions may present unacceptable dangers, but intermediate-length infusions may reduce peak drug levels adequately while retaining some of the convenience of bolus dosing.

It may be desirable to achieve the steady-state concentration rapidly for a drug given as a continuous intravenous infusion, in which case a loading dose may be given by bolus injection at the same time that the infusion is started. The bolus dose is usually selected to achieve an initial concentration near the steady-state target value. In this way, the time lag to achieve the plateau in the CxT profile, which may be considerable for some drugs, is eliminated. As an alternative to administering a drug by continuous intravenous infusion, it may be possible to maintain reasonably constant plasma levels using a repeated bolus injection dosing regimen. There is an approach to steady-state conditions in which the peak and trough plasma concentrations increase successively during repeated doses before becoming constant. As with the continuous intravenous infusion, steady-state can be reached immediately with the proper choice of loading dose. The most common such schedule targets the peak concentration as twice the trough concentration. This design requires dosing once each half-life. An initial dose of twice the successive (maintenance) doses abolishes the time lag. As the dosing frequency increases, the ratio of peak-to-trough concentrations approaches 1, and the CxT curve appears more like that of a constant infusion. These same scheduling considerations also apply to the timing of oral drug delivery.


The toxicities of anticancer drugs are often better correlated with a pharmacokinetic variable than the administered dose. Relationships between the severity of toxicity and the AUC are most commonly encountered. However, other variables such as the Cmax and duration of time that the drug concentration in plasma exceeds a particular threshold level are also predictive of toxicity. For example, the time interval that plasma levels of paclitaxel remain above 50 nM is better correlated with neutropenia, the principal dose limiting toxicity, than either Cmax or AUC.76 The nature of these relationships can often be described by a sigmoidal Emax model but they may appear linear unless patients have been evaluated across a sufficiently broad range of doses.77

As previously indicated, therapeutic response ultimately depends on the delivery of drug from the bloodstream to the tumor in such a way that malignant cells are exposed to biologically effective concentrations of the active form of the agent for an adequate duration of time. The rate processes associated with drug distribution and elimination depend on the physicochemical properties of the drug and numerous physiologic factors. As is the case with any specific organ or tissue, the time course of the concentration of a compound within a solid tumor cannot be defined from experimental data restricted to measurements made in plasma, serum, or whole blood. Although there is undoubtedly some temporal relationship between drug concentrations in plasma and the tumor, elucidating the tumor CxT profile requires physical measurement of drug levels within the tumor itself. Whereas this cannot be easily accomplished in solid tumors, in most cases, hematologic malignancies are considerably more amenable to such studies because the cancer cells reside within the bloodstream itself, bone marrow or lymphatic tissues, which are considerably more accessible to drug. Consequently, efforts to determine whether adequate concentrations of the active form of a drug are achieved in cancer cells should be considered an important objective of phase I trials to evaluate new anticancer drugs in hematologic malignancies. The availability of this information will better facilitate the rational selection of drugs warranting further clinical evaluation. The emergence of noninvasive imaging techniques may provide some momentum for pharmacokinetic-pharmacodynamic (i.e., PK-PD) relationships in solid tumors.


There are numerous reasons for acquiring pharmacokinetic data during various stages in the clinical development of anticancer drugs. The therapeutic indices of many drugs used in the treatment of cancer are inherently narrow because they are used at doses close to the upper limit of tolerability. Furthermore, cancer patients frequently exhibit increased sensitivity to many medications because of compromised organ function or diminished overall tolerance from their underlying disease state, augmenting the potential for an undesirable pharmacokinetic interaction with the host of concurrent medications used in the clinical management of cancer patients. The chances for an adverse event resulting from inappropriate dosing of a chemotherapeutic agent to a cancer patient are, therefore, considerably greater than experienced with most other patient groups. Since the dose-limiting toxicities of a chemotherapeutic agent are very often related to some measure of systemic exposure to the drug, the margin of safety of a potentially effective dose depends on the consistency of its pharmacokinetic behavior among patients.

The ultimate goal of pharmacokinetics is to assist in the optimization of therapy. Although progress has been made in pharmacokinetic areas, the limiting step for optimization of therapy is inadequate knowledge of the relationship between drug CxT profiles and drug effects. Pharmacokinetics can serve as a useful tool to help elucidate pharmacodynamic relationships by determining which profiles are feasible and by helping design administration strategies. Also, because overall drug effect results from both kinetic and dynamic variables, studies can be designed to adjust doses individually so that kinetic differences between patients can be minimized and attention can be focused solely on drug dynamics. Finally, pharmacokinetics can serve a useful role in the process of drug development by assisting the overall integration of data between preclinical testing and early clinical trials.78 Initial human studies rely heavily on toxicologic and pharmacologic data obtained in mice and dogs, and pharmacokinetics provides a convenient approach to comparative analysis.


1. Donelli MG, D'Incalci M, Garattini S. Pharmacokinetic studies of anticancer drugs in tumor-bearing animals. Cancer Treat Rep 1984;68:381–400.

2. Notari, R. E. Biopharmaceutics and Clinical Pharmacokinetics. New York: Marcel Dekker, 1987.

3. Gilbaldi M, Perrier D, eds. Pharmacokinetics. 2nd Ed. New York: Marcel Dekker,1982.

4. Timmerman PM, de Vries R, Ingelse BA. Tailoring bioanalysis for PK studies supporting drug discovery. Curr Top Med Chem 2001;1:443–462.

5. Wright JD, Boudinot FD, Ujhelyi MR. Measurement and analysis of unbound drug concentrations. Clin Pharmacokinet 1996;30: 445–462.

6. Dunne A. An iterative curve stripping technique for pharmacokinetic parameter estimation. J Pharm Pharmacol 1986;38:97–101.

7. Gillespie WR. Noncompartmental versus compartmental modelling in clinical pharmacokinetics. Clin Pharmacokinet 1991;20: 253–262.

8. Dedrick RL, Myers CE, Bungay PM, et al. Pharmacokinetic rationale for peritoneal drug administration in the treatment of ovarian cancer. Cancer Treat Rep 1978;62:1–11.

9. Speyer JL, Sugarbaker PH, Collins JM, et al. Portal levels and hepatic clearance of 5-fluorouracil after intraperitoneal administration in humans. Cancer Res 1981;41:1916–1922.

10. Farris FF, King FG, Dedrick RL, et al. Physiologic model for the pharmacokinetics of cis-dichlorodiammineplatinum(II)(DDP) in the tumored rat. J Pharmacokinet Biopharm 1985;13:13–39.

11. Chan KK, Cohen JL, Gross JF, et al. Prediction of adriamycin disposition in cancer patients using a physiologic, pharmacokinetic model. Cancer Treat Rep 1978;62:1161–1171.

12. Dedrick RL, Forrester DD, Cannon JN, et al. Pharmacokinetics of 1-β-D-arabinofuranosylcytosine (ARA-C) deamination in several species. Biochem Pharmacol 1973;22:2405–2417.

13. Rowland M, Benet LZ, Graham GG. Clearance concepts in pharmacokinetics. J Pharmacokinet Biopharm 1973;1:123–136.

14. Sawyer M, Ratain M. Body surface area as a determinant of pharmacokinetics and drug dosing. Invest New Drugs 2001;19: 171–177.

15. Carlisle KM, Halliwell M, Read AE, et al. Estimation of total hepatic blood flow by duplex ultrasound. Gut 1992;33:92–97.

16. Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron 1976;16:31–41.

17. Gibaldi M, McNamara PJ. Apparent volumes of distribution and drug binding to plasma proteins and tissues. Eur J Clin Pharmacol 1978;13:373–380.

18. Collins JM, Dedrick RL, King FG, et al. Nonlinear pharmacokinetic models for 5-fluorouracil in man: intravenous and intraperitoneal routes. Clin Pharmacol Ther 1980;28:235–246.

19. Donehower RC, Karp JE, Burke PJ. Pharmacology and toxicity of high-dose cytarabine by 72-hour continuous infusion. Cancer Treat Rep 1986;70:1059–1065.

20. Besseghir K, Roch-Ramel F. Renal excretion of drugs and other xenobiotics. Ren Physiol 1987;10:221–241.

21. Ilett KF, Tee LB, Reeves PT, et al. Metabolism of drugs and other xenobiotics in the gut lumen and wall. Pharmacol Ther 1990; 46:67–93.

22. Balis FM, Holcenberg JS, Bleyer WA. Clinical pharmacokinetics of commonly used anticancer drugs. Clin Pharmacokinet 1983;8: 202–232.

23. Glue P and Clement RP. Cytochrome P450 enzymes and drug metabolism—basic concepts and methods of assessment. Cell Mol Neurobiol 1999;19:309–323.

24. von Moltke LL, Greenblatt DJ, Schmider J, et al. Metabolism of drugs by cytochrome P450 3A isoforms. Implications for drug interactions in psychopharmacology. Clin Pharmacokinet 1985; 29:33–43.

25. Gillum JG, Israel DS, Polk RE. Pharmacokinetic drug interactions with antimicrobial agents. Clin Pharmacokinet 1993;25:450–482.

26. Kivisto KT, Kroemer HK, Eichelbaum M. The role of human cytochrome P450 enzymes in the metabolism of anticancer agents: implications for drug interactions. Br J Clin Pharmacol 1995;40:523–530.

27. Chabot GG, Bouchard J, Momparler RL. Kinetics of deamination of 5-aza-2′-deoxycytidine and cytosine arabinoside by human liver cytidine deaminase and its inhibition by 3-deazauridine, thymidine or uracil arabinoside. Biochem Pharmacol 1983;32: 1327–1328.

28. Milano G, McLeod HL. Can dihydropyrimidine dehydrogenase impact 5-fluorouracil-based treatment? Eur J Cancer 2000;36: 37–42.

29. Clarke DJ, Burchell B. The uridine diphosphate glucronosyltransferase multigene family: function and regulation. In: Kauffman FC, ed. Handbook of Experimental Pharmacology, Conjugation-Deconjugation Reactions in Drug Metabolism and Toxicity. Berlin: Springer-Verlag, 1994:3–43.

30. Ratain MJ. Insights into the pharmacokinetics and pharmacodyamics of irinotecan. Clin Cancer Res 2000;6:3393–3394.

31. Ivanov AI, Christodoulou J, Parkinson JA, et al. Cisplatin binding sites on human albumin. J Biol Chem 1998;273:14721–14730.

32. Milano G, Etienne MC, Cassuto-Viguier E, et al. Influence of sex and age on fluorouracil clearance. J Clin Oncol 1992;10:1171–1175.

33. Innocenti F, Iyer L, Ratain MJ. Pharmacogenetics of anticancer agents: lessons from amonafide and irinotecan. Drug Metab Dispos 2001;29:596–600.

34. Lichtman SM, Skirvin JA. Pharmacology of antineoplastic agents in older cancer patients. Oncology 2000;14:1743–1752.

35. Geokas M, Haverback B. The aging gastrointestinal tract. Am J Surg 1969;117:881–892.

36. Bender A. the effect of increasing age on the distribution of peripheral blood flow in man. J Am Geriatr Soc 1965;13:192–198.

37. Bach B, Hansen J, Kampmann J, et al. Disposition of antipyrine and phenytoin correlated with age and liver volume in men. Clin Pharmacokinet 1981;6:389–396.

38. Durnas C, Loi C, Cusack, BJ. Hepatic drug metabolism and aging. Clin Pharmacokinet 1990;19:359–389.

39. Baker SD, Grochow LB. Pharmacology of cancer chemotherapy in the older person. Clin Geriatr Med 1997;13:169–183.

40. Kinirons MT, O'Mahony MS. Drug metabolism and ageing. Br J Clin Pharmacol 2004;57:540–544.

41. Anderson GD. Children versus adults: pharmacokinetic and adverse-effect differences. Epilepsia 2002;43(Suppl 3):53–59.

42. Hammerlein A, Derendorf H, Lowenthal DT. Pharmacokinetic and pharmacodynamic changes in the elderly. Clinical implications. Clin Pharmacokinet 1998;35:49–64.

43. Barre J, Houin G, Brunner F, et al. Disease-induced modifications of drug pharmacokinetics. Int J Clin Pharmacol Res 1983;3: 215–226.

44. Powis G. Effect of human renal and hepatic disease on the pharmacokinetics of anticancer drugs. Cancer Treat Rev 1982;9:85–124.

45. Egorin MJ, Van Echo DA, Olman EA, et al. Prospective validation of a pharmacologically based dosing scheme for the cis-diamminedichloroplatinum(II) analogue diamminecyclobutanedicarboxylatoplatinum. Cancer Res 1985;45:6502–6506.

46. Egorin MJ, Sigman LM, Van Echo DA, et al. Phase I clinical and pharmacokinetic study of hexamethylene bisacetamide (NSC95580) administered as a five-day continuous infusion. Cancer Res 1987;47:617–623.

47. van den Bongard HJ, Mathot RA, Beijnen JH, et al. Pharmacokinetically guided administration of chemotherapeutic agents. Clin Pharmacokinet 2000; 39:345–367.

48. Shimada T, Yamazaki H, Mimura M, et al. Interindividual variations in human liver cytochrome P-450 enzymes involved in the oxidation of drugs, carcinogens and toxic chemicals: studies with liver microsomes of 30 Japanese and 30 Caucasians. J Pharmacol Exp Ther 1994;270:414–423.

49. Watters JW, McLeod HL, Cancer pharmacogenomics: current and future applications. Biochim Biophys Acta 2003;1603:99–111.

50. Peck CC, Temple R, Collins JM. Understanding consequences of concurrent therapies. JAMA 1993;269:1550–1552.

51. Dorr RT, Soble MJ, Alberts DS. Interaction of cimetidine but not ranitidine with cyclophosphamide in mice. Cancer Res 1986;46: 1795–1799.

52. Hande K, Combs G, Swingle R, et al. Effect of cimetidine and ranitidine on the metabolism and toxicity of hexamethylmelamine. Cancer Treat Rep 1986;70:1443–1445.

53. Bjornsson TD, Huang AT, Roth P, et al. Effects of high-dose cancer chemotherapy on the absorption of digoxin in two different formulations. Clin Pharmacol Ther 1986;39:25–28.

54. Balis FM. Pharmacokinetic drug interactions of commonly used anticancer drugs. Clin Pharmacokinet 1986;11:223–235.

55. Jamis-Dow CA, Klecker RW, Katki AG, et al. Metabolism of taxol by human and rat liver in vitro: a screen for drug interactions and interspecies differences. Cancer Chemother Pharmacol 1995; 36:107–114.

56. van Meerten E, Verweij J, Schellens JH. Antineoplastic agents. Drug interactions of clinical significance. Drug Saftey 1995;12: 168–182.

57. Markowitz JS, Donovan JL, DeVane CL, et al. Effect of St John's wort on drug metabolism by induction of cytochrome P450 3A4 enzyme. JAMA 2003;290:1500–1504.

58. McCune JS, Hawke RL, LeCluyse EL, et al. In vivo and in vitro induction of human cytochrome P4503A4 by dexamethasone. Clin Pharmacol Ther 2000;68:356–366.

59. Bannwarth B, Pehourcq F, Schaeverbeke T, et al. Clinical pharmacokinetics of low-dose pulse methotrexate in rheumatoid arthritis. Clin Pharmacokin 1996;30:194–210.

60. Evans WE, Christensen ML. Drug interactions with methotrexate. J Rheumatol 1985;12(Suppl 12):15–20.

61. Wood M. Pharmacokinetic drug interaction in anaesthetic practice. Clin Pharmacokinet 1991;21:285–307.

62. Kerr IG, Jolivet J, Collins JM, et al. Test dose for predicting high-dose methotrexate infusions. Clin Pharmacol Ther 1983;33:44–51.

63. Evans WE, Relling MV, Rodman JH, et al. Conventional compared with individualized chemotherapy for childhood acute lymphoblastic leukemia. N Engl J Med 1998;338:499–505.

64. Zimm S, Collins JM, Riccardi R, et al. Variable bioavailability of oral mercaptopurine: Is maintenance chemotherapy in acute lymphoblastic leukemia being optimally delivered? N Engl J Med 1983;308:1005–1009.

65. Peng B, Dutreix C, Mehring G, et al. Absolute bioavailability of imatinib (Glivec) orally versus intravenous infusion. J Clin Pharmacol 2004;44:158–62.

66. Rubin GM, Tozer TN. Theoretical considerations in the calculation of bioavailability of drugs exhibiting Michaelis-Menten elimination kinetics. J Pharmacokinet Biopharm 1984;12:437–450.

67. Blasberg R, Patlak CS, Fenstermacher JD. Intrathecal chemotherapy: brain tissue profiles after ventriculocisternal perfusion. J Pharmacol Exp Ther 1975;195:73–83.

68. Myers CE, Collins JM. Pharmacology of intraperitoneal chemotherapy. Cancer Invest 1983;1:395–407.

69. Markman M, Bundy BN, Alberts DS, et al. Phase III trial of standard-dose intravenous cisplatin plus paclitaxel versus moderately high-dose carboplatin followed by intravenous paclitaxel and intraperitoneal cisplatin in small-volume stage III ovarian carcinoma: an intergroup study of the Gynecologic Oncology Group, Southwestern Oncology Group, and Eastern Cooperative Oncology Group. J Clin Oncol 2001;19:921–923.

70. Alberts DS, Markman M, Armstrong D, et al. Intraperitoneal therapy for stage III ovarian cancer: a therapy whose time has come! J Clin Oncol 2002;20:3944–3946.

71. Alberts DS, Liu PY, Hannigan EV, et al. Intraperitoneal cisplatin plus intravenous cyclophosphamide versus intravenous cisplatin plus intravenous cyclophosphamide for stage III ovarian cancer. N Engl J Med 1996;335:1950–1955.

72. Collins JM, Zaharko DS, Dedrick RL, et al. Potential roles for preclinical pharmacology in phase I trials. Cancer Treat Rep 1986; 70:73–80.

73. EORTC Pharmacokinetics and Metabolism Group. Pharmacokinetically guided dose escalation in phase I clinical trials. Eur J Cancer Clin Oncol 1987;23:1083–1087.

74. Fuse E, Kobayashi S, Inaba M, et al. Application of pharmacokinetically guided dose escalation with respect to cell cycle phase specificity. J Natl Cancer Inst 1994;86:989–996.

75. Collins JM, Grieshaber CK, Chabner BA. Pharmacologically guided phase I trials based upon preclinical development. J Natl Cancer Inst 1990;82:1321–1326.

76. Gianni L, Kearns CM, Giani A, et al. Nonlinear pharmacokinetics and metabolism of paclitaxel and its pharmacokinetic/pharmacodynamic relationships in humans. J Clin Oncol 1995;13: 180–190.

77. Holford NH. Clinical pharmacokinetics and pharmacodynamics of warfarin. Understanding the dose-effect relationship. Clin Pharmacokinet 1986;11:483–504.

78. Collins JM. Pharmacology and drug development. J Natl Cancer Inst 1988;80:790–792.