Chapter 44:  Pharmacology

Membrane Transport

In order to produce cytotoxicity, most anticancer drugs require uptake into the cell. A number of mechanisms exist for the passage of drugs across the plasma membrane including passive diffusion, facilitated diffusion, and active transport systems. Passive diffusion of drugs through the lipid bilayer structure of the plasma membrane is a function of the size, lipid solubility, and charge of the drug molecule. If the extracellular drug concentration is constant, then drug accumulation by the cell will continue until the rate of drug uptake from the extracellular space is equal to the rate of drug efflux from the cell. At this point, a dynamic equilibrium is reached and intracellular and extracellular drug concentrations are equal. As drug is cleared from the extracellular space, intracellular drug levels will decline if the drug is not bound or metabolized intracellularly. An important feature of the passive diffusion process is that it does not saturate. That is, as the extracellular drug concentration increases, influx into the cell increases proportionally and high intracellular drug levels can be achieved. Passive diffusion, however, is a highly inefficient and nonspecific process, although it may be a particularly important mechanism of drug uptake when carrier-mediated processes are nonfunctional, such as occurs in some cases of methotrexate (MTX) resistance.

The passage of physiologically important hydrophilic compounds across the plasma membrane is usually mediated by a specific receptor, or carrier, in the plasma membrane that facilitates the translocation of the substance into or out of the cell. Carrier-mediated transport systems are distinguished from passive diffusion by having a higher degree of specificity and by being saturable at high extracellular drug concentrations due to the presence of a finite number of receptor molecules within the membrane. Once all carrier sites become occupied, further increases in extracellular drug concentration will not produce further increments in drug influx unless a component of passive diffusion comes into play. The affinity of the carrier for the substrate can be estimated from the Michaelis constant (K_m), the drug concentration at which the influx rate is one half maximal; the lower the K_m, the higher the carrier affinity. Although carrier-mediated systems enhance the rate of influx into the cell, not all carriers are able to translocate compounds against electrochemical forces and ultimately develop gradients such that the intracellular concentration exceeds the extracellular drug level. To do so requires the expenditure of energy and the coupling of carrier-mediated transport to an energy-requiring reaction, usually hydrolysis of adenosine triphosphate.

Many antineoplastic drugs, particularly those that are structural analogs of natural compounds, gain entry into the cell by carriermediated mechanisms. The functional and physiologic characteristics of several human nucleoside transporters have been characterized. However, substantial additional information is rapidly emerging as more of these molecules are cloned and their specificities are revealed.^3,4 Naturally occurring nucleosides are transported by both facilitated diffusion (equilibrative) and by concentrative mechanisms. Nucleoside analogs that are important in cancer therapy also use these transporters, but some specificity is emerging.⁵ For instance, ara-C, floxidine, and pentostatin appear to use equilibrative transporters,^6–8 whereas fludarabine, gemcitabine, and cladribine appear to be substrates for concentrative transport systems in addition to equilibrative pathways.⁹ Nucleobase transporters have also been identified, but their role in the entry of useful antimetabolites such as thiopurines and 5-FU into the cell has not been established.¹⁰ Transport of reduced folates and methotrexate is an active energy-dependent process that can be mediated by two distinct mechanisms: a membrane carrier system capable of the rapid transport of reduced folates and of 4-amino analogs of folic acid¹¹ and a group of membrane-bound folate receptors termed the folate binding proteins, which are brought into the cell by endocytosis to release ligand before recycling back to the membrane.^12,13 Candidate cDNAs for this function have now been identified.^14,15 Altered MTX transport features have been described in acute lymphoblastic leukemia blasts and in osteosarcoma as a mechanism of acquired resistance.¹⁶ l-phenylalanine mustard uses at least two amino acid transport systems and its influx can be inhibited by the amino acid substrates specific for these transport carriers.^17–19

The importance of transmembrane movement of a drug to its pharmacologic effect depends on several factors, including the rate of drug delivery to the tissue, the affinity of the transport process, and the nature of the intracellular biochemical events required for drug action. Although membrane transport can be the rate-limiting step in drug action if it governs the rate at which the drug reaches intracellular targets, this is not always the case. If drug delivery to a cell is slow relative to the influx rate, then the drug effect will be limited primarily by extracellular concentration (i.e., blood flow and diffusion of the drug). Similarly, if a drug requires intracellular activation, such as phosphorylation of nucleoside analogs or polyglutamylation of methotrexate, before it can exert a cytotoxic effect, then the rate-limiting step in drug action could be activation rather than transport if the rate of activation is slow relative to the rate of influx into the cell.

Finally, it is important to recognize that membrane transport is frequently bidirectional with the final drug concentration in the cell representing the balance between drug influx and drug efflux. These processes may use different carrier systems and operate at different rates. Several efflux systems that appear to have importance in cancer chemotherapy are the systems that mediate various forms of multidrug resistance.

Intracellular Activation

Many anticancer drugs require activation before they are able to exert a cytotoxic effect. The activation process may involve chemical or enzymatic reactions in either normal or tumor tissues (Table 44.1). Cisplatin, for example, undergoes a chemical reaction with water molecules intracellularly resulting in the generation of a positively charged aquated species that attacks nucleophilic sites on DNA.²⁰ In contrast, the activation of cyclophosphamide is mediated primarily by CYP2B6 (one of the P-450 enzymes), resulting in the release of active alkylating species into the systemic circulation.²¹

Intracellular activation by tumor cells is a critical determinant of effect for virtually all antimetabolites. Nucleoside antimetabolites such as ara-C, fludarabine,²² gemcitabine,²³ and cladribine²⁴ require phosphorylation to active nucleotide triphosphate forms and incorporation into DNA before they are able to exert a cytotoxic effect. Nucleobase analogs such as 6-mercaptopurine and 6-thioguanine undergo phosphoribosylation to the nucleoside monophosphate forms, which are active inhibitors of de novo purine nucleotide synthesis. Amination of 6-mercaptopurine to thioguanine monophosphate followed by phosphorylation, reduction to the deoxynucleotide, and a subsequent phosphorylation results in 2'-deoxythioguanine triphosphate, which is a substrate for incorporation into DNA. Phosphoribosylation also converts 5-FU to the monophosphate, which is then phosphorylated to the diphosphate, reduced to the deoxynucleotide, and dephosphorylated to the active monophosphate F-dUMP, which inhibits thymidylate synthase. Additionally, the drug may be cytotoxic after incorporation of either the ribosyl or deoxyribosyl triphosphate, respectively, into RNA or DNA. Although methotrexate is an effective enzyme inhibitor in its native form, intracellular conversion of the drug to polyglutamate metabolites significantly increases its potency and facilitates its binding to a number of enzymatic sites.^25,26 Consistent with this is the finding of a more favorable clinical outcome in ALL patients whose blasts accumulated higher levels of MTX polyglutamates.^27,28 It is important to note that phosphorylation of nucleic acid analogs and polyglutamylation of MTX produces charged molecules that are unlikely to diffuse or to be transported out of cells.

The rate of formation of the activated drug species in the cell depends on the rate of transmembrane influx of the drug, the amount and affinity of the activating enzyme(s) in the cell, the extent of competition by the naturally occurring substrates of the activating enzymes, and the rate of degradation of the activated drug by catabolic enzymes. For many antimetabolites, membrane transport is rapid relative to enzymatic activation and is therefore not rate limiting. Once inside the cell, antimetabolites must compete with the natural enzyme substrates for binding and activation. Finally, the activated drug then becomes a substrate for catabolic enzymes in the cell that tend to degrade it to the parent compound or to an inactive metabolite. Thus, the concentration of the active cytotoxic drug in the cell is the result of all of these processes.

The pyrimidine nucleoside analog, ara-C, provides an excellent example of these processes. Ara-C gains entry to the cell by a high-capacity equilibrative nucleoside transport system; transport velocity is nearly proportional to ara-C concentration up to 100 mM.²⁹ This process may limit ara-C activation in cells at plasma ara-C concentrations < 1 μM achieved by standard dose rates (< 20 mg/m²/h). However, at higher dose rates that achieve > 10 μM ara-C in plasma (250 mg/m²/h), the transport system provides cellular concentrations of ara-C that saturate the rate of ara-C phosphorylation.³⁰ After gaining entry to the cell, ara-C is metabolized in three successive phosphorylation reactions to ara-C triphosphate (ara-CTP), which, after its incorporation into replicating or repairing DNA by various DNA polymerases, is inhibitory to cell growth. The initial activating enzyme, deoxycytidine kinase, is present at the lowest specific activities in human leukemic blasts³¹ and is believed to be the rate-limiting step in the formation of ara-CTP³² and probably for incorporation of the drug into DNA. At each phosphorylation step, ara-C and its metabolites compete with endogenous deoxycytidine and its nucleotides for enzyme binding. Biochemical modulation strategies that reduce dCTP and thereby activate dCyd kinase result in increased ara-CTP formation³³ and improved clinical response.³⁴ Opposing the activation of ara-C are cytidine deaminase and dCMP deaminase, which convert ara-C and ara-CMP, respectively, to inactive uracil derivatives. In addition, the activity of phosphatases such as 5'-nucleotidase, the activities of which differ among cell types, may be important determinants of the steady-state ara-CTP concentrations and the rate of elimination of the triphosphate at the end of an ara-C infusion. The response of patients with acute leukemia treated with single-drug ara-C, either on an intermittent schedule or by continuous infusion, was strongly correlated with the ability of cells to retain ara-CTP³⁵ or with the steady-state ara-CTP concentrations³⁶ in blasts during therapy. These findings validate the importance of favorable pharmacokinetic characteristics for response to ara-C in particular and provide a basis for pharmacologic modulation strategies with other drugs.

Loss or diminished affinity of an activating enzyme or enhanced activity of a catabolic enzyme may be responsible for drug resistance. Although molecular reagents are now available³⁷ that have permitted the discovery of dCyd kinase deficiencies in selected clinical samples,³⁸ this does not appear to be a major cause of clinical resistance to ara-C because the blasts of patients with resistant disease accumulate ara-CTP levels similar to those of responders.³⁹

Drug Targets

Although cytotoxic anticancer drugs have traditionally been classified based on their mechanisms of action or their origins, they can also be grouped based on the target of drug action. There are essentially four potential targets of drug action: nucleic acids, specific enzymes, microtubules, and hormone/growth factor receptors. When nucleic acids are the target, it is generally DNA rather than RNA that is presumed to cause cell death. There are several mechanisms by which drugs can bind DNA, the most well understood being alkylation of nucleophilic sites within the double helix. Most clinically effective alkylating agents have two moieties capable of developing a charged carbon that binds covalently to negatively charged sites on DNA such as the O⁶ or N⁷ positions of guanine. The cross-linking of the two strands of DNA produced by the bifunctional alkylating agents prevents the use of that DNA as a template for further DNA and RNA synthesis leading to inhibition of DNA replication and cell death.⁴⁰ Although alkylating agents are among the most widely used drugs in clinical oncology, the relationship of pharmacologic parameters to clinical effects has not been well defined for these agents. In part, this has been due to the lack of sensitive and specific techniques to detect drug-DNA binding in clinical specimens. Studies of chlorambucil-DNA binding in the tumor cells of patients with chronic lymphocytic leukemia have demonstrated considerable heterogeneity in drug-DNA binding among patient samples, but no clear correlations between amount of drug bound and disease stage or sensitivity to treatment have been shown,⁴¹ although the drug clearly targets purines.⁴² In contrast, the formation of cisplatin adducts to DNA has been shown to correlate with cell kill in mammalian tumor cell lines.⁴³ Immunologic methods have been used to quantitate platinum-DNA adduct formation in either peripheral white blood cells⁴⁴ after cisplatin therapy or in buccal cells of patients receiving cisplatin with carboplatin chemotherapy.⁴⁵ A subsequent study that used atomic absorption spectroscopy to quantitate total cell platinum in lymphocytes indicated a relationship between the adduct levels after the first dose of either single-drug cisplatin or carboplatin and clinical response in 49 patients with 24 different tumor types.⁴⁶ Although adduct formation in these surrogate cell types was correlated with the response of the tumor to chemotherapy in previously untreated patients, it is difficult to imagine that such determinations will continue to reflect response as the originally platinum-sensitive tumor becomes resistant to treatment. A second mechanism of drug binding to nucleic acids is intercalation, the insertion of a planar ring structure between two adjacent nucleotide bases of DNA. This mechanism is characteristic of many antitumor antibiotics. The antibiotic molecule is non-covalently, although firmly, bound to DNA and distorts the shape of the double helix, resulting in inhibition of RNA or DNA synthesis.^47,48 Many agents capable of classical intercalation, such as doxorubicin and mitoxantrone, are also inhibitors of topoisomerase II and may produce DNA strand breaks by inhibition of the reannealing function of this enzyme.^49,50 Indeed, a direct correlation has been noted between DNA topoisomerase II activity and cytotoxicity in doxorubicin-sensitive and -resistant P388 leukemia cells.⁵¹ A third mechanism of nucleic acid damage is illustrated by the anticancer drug bleomycin. The amino terminal tripeptide of the bleomycin molecule appears to intercalate between guanine-cytosine base pairs of DNA. The opposite end of the bleomycin peptide binds Fe (II) and serves as a ferrous oxidase, able to catalyze the reduction of molecular oxygen to superoxide or hydroxyl radicals that produce DNA strand scission.^52,53 Predictably, the levels of antioxidant enzymes such as catalase, peroxidases, and super oxide dismutase in plasma and blood are inversely correlated with chromosomal damage.⁵⁴

Enzymes represent the second general category of targets for chemotherapeutic agents. Antimetabolites function as inhibitors of key enzymes in the purine or pyrimidine biosynthetic pathways or as inhibitors of DNA polymerases. The triphosphate of fludarabine, for instance, is known to inhibit both ribonucleotide reductase⁵⁵ and DNA ligase I.⁵⁶ After incorporation into DNA, it not only inhibits the function of multiple DNA polymerases,⁵⁷ DNA primase,⁵⁸ DNA ligase I,⁵⁶ but it is also resistant to removal by the proof-reading exonuclease activities associated with DNA polymerases.⁵⁷ Since these enzymes are highly active during DNA replication, antimetabolites tend to be cytotoxic only when present in sufficient concentration during the vulnerable S phase of the cell cycle; these drugs are thus frequently referred to as S phase-specific. Nevertheless, because these enzymes are also required for repair of damaged DNA, it is likely that antimetabolites that inhibit these enzymes will be synergistic with agents that elicit an incision DNA repair response that requires resynthesis of a DNA patch, regardless of cell cycle stage.⁵⁹

The effectiveness of enzyme inhibitors also depends on the amount of the target enzyme, its affinity for the inhibitor, and on the extent of competition by natural substrates for enzyme binding. For example, complete saturation of all dihydrofolate reductase binding sites is required before the enzyme is effectively inhibited. As MTX inhibits enzymatic activity, dihydrofolate, the natural substrate, accumulates behind the metabolic block and is able to effectively compete with MTX for further enzyme binding.⁶⁰ Thus, large amounts of MTX, well in excess of the enzyme binding capacity, are required to effectively inhibit dihydrofolate reductase activity. Similarly, in the case of 5-FU, the dUMP/F-dUMP ratio may be an important determinant of optimal inhibition of the target enzyme thymidylate synthase, and high ratios have been associated with lack of tumor response.⁶¹ Similarly, the amount of thymidylate synthase expression or activity is an important determinant of 5-FU activity and correlates with therapeutic response.^62,63 In addition, high basal levels of thymidine phosphorylase have recently been associated with lack of response to 5-FU.⁶⁴

In addition to the enzymes required for purine and pyrimidine biosynthesis, the topoisomerases are important targets of several antineoplastic agents. Topoisomerase I and II catalyze the passage of DNA strands through single- or double-strand breaks in the DNA molecule, respectively, by nicking and then reannealing the DNA strands. Topoisomerase inhibitors bind to the enzyme and stabilize the reaction intermediate enzyme-DNA cleavable complex. This interference with the DNA breakage-resealing process, which is necessary for both DNA replication or RNA transcription, results in DNA strand breaks that are lethal to the cell. The epipodophyllotoxins, etoposide and teniposide, are potent inhibitors of topoisomerase II, as are a number of DNA intercalating agents including doxorubicin, actinomycin D, and amsacrine.^48,49,65

Two topoisomerase I inhibitors, topotecan and irinotecan, have recently been approved for broad clinical use. Whereas topotecan interacts with topoisomerase I directly, irinotecan requires activation by carboxylesterases for SN-38 in order to affect the target.⁶⁶ It is not yet clear whether irinotecan’s antitumor effects are due primarily to intratumoral or intrahepatic activation, although there is some evidence that higher enzyme activity results in a greater cytotoxic effect.^67–70

The microtubule spindle structure provides a third target for chemotherapeutic agents, classically the Vinca alkaloids, vincristine and vinblastine, but more recently vinorelbine. The Vinca alkaloids exert their cytotoxic effects by binding to specific sites on tubulin, inhibiting assembly of tubulin into microtubules and ultimately dissolution of the mitotic spindle structure.⁷¹ The microtubule system in cells performs a variety of other important functions, including transport of solutes, cell movement, and chromosomal separation, and provides structural integrity, any one of which could potentially be disrupted by tubulin binding agents.⁷² The taxanes are a newer class of agents, consisting of the natural plant alkaloid paclitaxel and a semisynthetic derivative docetaxel. These novel plant alkaloids inhibit cell division by stimulating tubulin polymerization, thus enhancing the formation and stability of microtubules.⁷³ Paclitaxel-treated cells accumulate large numbers of microtubules, free and in bundles, that disrupt microtubule function and that ultimately cause cell death.^74,75 Although docetaxel appears to be more potent than paclitaxel, the drugs appear to have similar toxicity profiles.⁷⁶

The search for specific small molecule inhibitors of hormone and growth factor receptors has been ongoing since the demonstration that antiestrogens can be effective treatment for breast cancers that contain the estrogen receptor. Antiandrogens, such as flutamide and bicalutamide, are also important in the treatment of prostate cancer.⁷⁷ The rapid advances in our understanding of cancer biology has led to an explosion in new targets and potential new drugs. Drugs targeted against angiogenesis⁷⁸ and matrix metalloproteinases⁷⁹ have completed phase I testing, and many other new agents are in late preclinical or early clinical evaluation.^80–84

Repair of Drug-induced Injury

Cells that have been damaged by cytotoxic drugs exhibit a variety of repair mechanisms. Indeed, the cytotoxic effects of a drug often represent the balance between injury and repair, and amplified repair mechanisms may account for cellular resistance to certain drugs. The cytotoxicity of alkylating agents reflects the balance between DNA cross-link formation and removal by cellular repair processes. Many cells contain specific enzymes that remove alkyl moieties from DNA and thereby repair drug damage. A specific example is the protein O⁶-alkylguanyltransferase, which repairs DNA injury produced by chloroethylnitrosoureas. Cells containing large amounts of this protein tend to be relatively resistant to these chemotherapeutic agents. Depletion of alkyltransferase activity by exposure of cells to modified purine bases such as O⁶-benzylguanine may be effective in circumventing this mechanism of resistance.^85–88 It is now clear that mammalian cells possess a family of such enzymes that are capable of repairing alkylation to specific nucleic acid bases, and that the abundance of these in a particular tissue may be responsible for conferring relative sensitivity or resistance to chemotherapeutic alkylating agents.^89,90

The broader task of protecting the genome from a wide variety of adducts that affect replication and transcription is taken up by the nucleotide excision repair system.^91,92 The increased incidence of cancer in individuals who have genetic diseases such as xeroderma pigmetosum, which is characterized by the lack of effective nucleotide excision repair, attests to the key role that this system has in suppressing carcinogenesis due to DNA damage.⁹³ This system has a broad specificity of adducts that it can remove from DNA, ranging from simple methyl groups to bulky adducts, including natural molecules such as psoralen and aminofluorene. Lesions produced by cisplatin and cyclophosphamide are also substrates; an increase in the rate of platinum adduct removal has been associated with drug resistance.⁹⁴ The mechanism of adduct removal is becoming clear, and the various proteins involved are being identified.^95,96 Basically, two incisions are made in the adduct-containing DNA strand about 27 to 29 nucleotides apart. This adduct-containing oligonucleotide is removed as a single piece and new nucleotides are polymerized in the repair patch by the same DNA polymerases involved in replication. This DNA synthesis phase presents a new opportunity to incorporate nucleoside analogs into DNA of cells that are not in S phase that would otherwise not be affected. This possibility has given rise to therapeutic strategies that combine agents or modalities that elicit DNA repair with one of the newer nucleoside analogs, such as fludarabine²² and gemcitabine,⁹⁷ which inhibit DNA synthesis by a number of different mechanisms and subsequently induce cell death by apoptosis.⁹⁸

Cells also contain a variety of free radical scavenging systems that protect them from the effects of ionizing radiation and drugs that generate oxygen free radicals intracellularly. Catalase, superoxide dismutase, and glutathione peroxidase, key enzymes in the detoxification of reactive oxygen species, may be deficient in some tissues, like cardiac muscle, leading to excessive drug toxicity, or increased in others, leading to relative drug resistance.⁹⁹ Some doxorubicin-resistant cells have been shown to have increased activity of superoxide dismutase and sodium-dependent glutathione peroxidase and diminished susceptibility to oxygen radical injury.¹⁰⁰ Other studies suggest that expansion of intracellular reduced glutathione pools or increased expression of glutathione transferase may be important mechanisms of alkylating agent resistance in animal and human tumors.^101–103

Finally, cells may be able to circumvent drug-induced injury by increased production of target enzymes. In experimental models, exposure of cells to MTX or 5-FU can be shown to stimulate production of dihydrofolate reductase or thymidylate synthase, respectively.^104,105 New enzyme production occurs within minutes to hours of drug exposure and is presumed to represent enhanced translation of existing mRNA rather than transcription of additional message. Amplification of DNA also occurs, however, and may be a fundamental mechanism of cellular resistance to antimetabolites and natural products due to increased constitutive production of target enzymes or P-glycoprotein.¹⁰⁶

As mentioned earlier, a prerequisite to drug effect at the target tissue is adequate drug delivery. Pharmacokinetics describes the concentration-time history of a drug in the body and can be used to answer fundamental questions concerning the optimal route and schedule of drug administration. The remainder of this chapter will present the principles of pharmacokinetics and pharmacodynamics and illustrate their importance in cancer chemotherapy.

Definitions

Pharmacokinetics is the study of drug absorption, distribution, metabolism, and excretion (Fig. 44.1). A fundamental concept in pharmacokinetics is drug clearance (i.e., elimination of drugs from the body), analogous to the concept of creatinine clearance. In clinical practice, clearance of a drug is rarely measured directly but is calculated as either

The area under the concentration-time curve (AUC) represents the total drug exposure integrated over time and is an important parameter for both pharmacokinetic and pharmacodynamic analyses. As indicated in equation 1, the clearance is simply the ratio of the dose to the AUC, so that the higher the AUC for a given dose, the lower the clearance. If a drug is administered by continuous infusion and steady-state is achieved, the clearance can be estimated from a single measurement of the plasma drug concentration (C_ss) as per equation 2.

Clearance can conceptually be considered to be a function of both distribution and elimination. In the simplest pharmacokinetic model,

V is the volume of distribution and K is the elimination constant. V is the volume of fluid in which the dose is initially diluted; thus, the higher the V, the lower the initial concentration. K is the elimination constant, which is inversely proportional to the half-life, the period of time that must elapse to reach a 50% decrease in plasma concentration. When the half-life is short, K is high and plasma concentrations decline rapidly. Thus, both a high V and a high K result in relatively low plasma concentrations and a high clearance.

Linear Pharmacokinetic Models

Although pharmacokinetic analysis can be conducted without specifying any mathematic models (noncompartmental methods), it is helpful to use such models as guides in therapeutic decision making. There are several important properties of drugs that have linear pharmacokinetics (Table 44.2). The key feature of a linear pharmacokinetic model is that

This indicates that the instantaneous rate of change in drug concentration depends only on the current concentration. The half-life will remain constant no matter how high the concentration.

One implication of this principle is that the drug exposure (AUC) is not affected by changes in drug schedule. For example, the AUC after a 60 mg/m² bolus dose of doxorubicin equals the total AUC for three daily (or weekly) bolus doses of 20 mg/m², which equals the AUC for the same dose administered as a 96-hour infusion. A second implication is that the AUC is proportional to the dose. Thus, if one measures the AUC for a 60 mg/m² dose, one can estimate the AUC for a 90 mg/m²dose in the same patient as being 50% higher.

The simplest linear pharmacokinetic model is shown graphically in Figure 44.3. This model assumes that the drug is administered as an instantaneous bolus, and that complete distribution of the drug is also instantaneous.

These assumptions are often not valid. If the drug is administered as a slow bolus or infusion, the model must be corrected for the infusion duration. During the administration of the drug the concentration is increasing:

After the infusion is terminated, the drug concentration decays at the same rate as if it had been administered as an instantaneous bolus. Thus, if T represents the infusion time, then the post-infusion drug concentrations can be represented as

^{Often, the pharmacokinetic data are more complex than those shown in Figure 44.2, and may be optimally fitted to a multicompartment model, usually two or three compartments (see Figure 44.3). It must be emphasized that the compartments are theoretical, and do not necessarily correlate with any anatomic space or physiologic process.}

A large variety of computer software are available for pharmacokinetic analysis.^107-109 The interested reader is likely to benefit from “hands-on” experience with such programs. Several caveats need to be emphasized for the casual reader. The validity of pharmacokinetic modeling depends to a large extent on the quality of the data entered into the model. Thus, drug infusions must be precisely timed, plasma samples must be obtained on schedule, and analytical methods must be sensitive and specific. The data must be properly weighted to avoid bias due to the increased probability of analytical errors at drug concentrations near the detection limit of the assay. Results obtained using a specific model should be compared to those using noncompartmental methods. Extrapolation of models outside the known time points must be done with great caution.

Nonlinear Pharmacokinetic Models

Nonlinear pharmacokinetic models imply that some aspect of the pharmacokinetic behavior of the drug is saturable. The mathematics of nonlinear models are beyond the scope of this chapter, but the principles are very relevant to several anticancer agents.^110,111 In contrast to drugs with linear pharmacokinetics, alteration of the schedule of administration of drugs that display nonlinear kinetics may markedly affect the AUC and potentially alter clinical effects.

Nonlinear pharmacokinetic behavior commonly occurs when there is saturation of a major metabolic pathway. This results in decreased clearance at higher doses, with a greater than proportional increase in the AUC. The AUC will also increase if the infusion duration is shortened, due to slower clearance at the higher peak plasma concentrations. This is clearly the case for 5-FU, probably due to saturation of its conversion to dihydrofluorouracil by the enzyme dihydropyrimidine dehydrogenase.^112–115 Schaaf et al. demonstrated that doubling of the 5-FU dose from approximately 7.5 mg/kg to 15 mg/kg (by IV bolus) resulted in a 135% increase in the mean AUC.¹¹⁴ Since 5-FU is used on a variety of schedules, its nonlinear pharmacokinetic behavior may be one factor in its highly schedule-dependent effects. Paclitaxel has also been demonstrated to have nonlinear pharmacokinetics.^116,117 Thus, the AUC is higher, for a fixed dose, when administered by shorter (3-hr vs. 24-hr) infusion, although this does not result in enhanced toxicity.¹¹⁸

The opposite situation arises when a drug’s absorption from the gastrointestinal tract (or renal tubular reabsorption) is saturable. In this case, an increase in dose results in a less than proportional increase in the AUC. Gastrointestinal absorption of drugs that resemble natural compounds is frequently mediated by active transport processes in the gastrointestinal tract that display saturable kinetics. Folate analogues such as MTX or leucovorin and amino acid analogues such as melphalan are examples of drugs with saturable absorption.^119–121 Cisplatin appears to have nonlinear pharmacokinetics due to saturation of its renal tubular reabsorption.^122,123 Forastiere et al. demonstrated that free plasma platinum is increased by 42% when the drug is given as a 24-hour continuous infusion, rather than as a 20-minute infusion.¹²² Prolonged infusion was also associated with a greater than three-fold increase in the free platinum half-life.

Interpatient Pharmacokinetic Variability

In describing a drug’s pharmacokinetics, it is important to consider the extent of interpatient variability, often represented as the coefficient of variation (ratio of standard deviation to mean). Cancer patients may have significant hepatic or renal dysfunction, as well as other abnormalities that lead to alterations in pharmacokinetic parameters (Table 44.3). Identifying genetic differences in drug metabolism may be particularly fruitful in understanding pharmacokinetic variability.¹²⁴ Such pharmacogenetic variation has been demonstrated to be important in explaining variability observed following administration of 6-mercaptopurine,^125,126 5-FU, amonafide,^127–129 and irinotecan.^130–133

Studies of interpatient pharmacokinetic variability are potentially of great importance for optimizing antineoplastic therapy. Variability in gastrointestinal absorption is generally not considered in the use of orally administered antineoplastic agents even though drugs such as cyclophosphamide, chlorambucil, melphalan, and etoposide are commonly administered orally for a variety of malignancies.² The percentage of a drug absorbed is referred to as its bioavailability (i.e., the ratio of the plasma AUC after oral administration to the plasma AUC after intravenous administration of the same dose). Bioavailability may be influenced by drug metabolism in the gastrointestinal tract or liver as well as by absorption. The (6S) isomer of leucovorin, for example, has limited bioavailability due primarily to its rapid conversion to 5-methyltetrahydrofolate prior to reaching the systemic circulation.¹³⁴ By contrast, the bioavailability of (6R) leucovorin is limited primarily by absorption. Bioavailability is often highly variable and unpredictable,^25,135–138 and may be accentuated by concomitant administration of other chemotherapeutic agents, particularly those that produce toxicity to the gastrointestinal mucosa.¹²⁰

Variability in drug distribution may be attributed to changes in body size or to the ratio of fat to total mass.¹³⁹ In the latter case, there may be altered distribution of lipophilic drugs, which includes most of the natural product anticancer drugs and their analogs. The most well-described example of abnormal drug distribution is delayed clearance of MTX due to accumulation and slow release of the drug from ascites or pleural effusions.¹⁴⁰ The terminal elimination half-life of doxorubicin, cyclophosphamide, and ifosfamide is prolonged in obese patients.^141,142 In the case of doxorubicin and cyclophosphamide, this appears to be due to a reduction in clearance, whereas in the case of ifosfamide, it is related to an increased volume of distribution of the drug.¹⁴¹

Many patients with advanced cancer have abnormalities of liver function tests or known mass lesions within the liver, often in association with significant malnutrition. Given that many antineoplastic agents are metabolized or excreted by the liver, recognizing altered elimination by the liver becomes important in the optimization of chemotherapy dosing. Unfortunately, altered hepatic elimination or metabolism of drugs is not easily predictable. Clearly, patients with severe hyperbilirubinemia due to parenchymal replacement or obstruction are likely to have altered elimination.¹⁴³ However, it is not often recognized that many patients with normal serum bilirubin levels may have a low drug clearance, resulting in a high AUC and corresponding toxicity. A decrease in serum albumin (in patients with normal serum bilirubin concentrations) has been associated with a decrease in the hepatic elimination of antipyrine—a commonly used marker drug—and of vinblastine and trimetrexate.^144–147 Thus, patients with a serum albumin less than 2.5 g/dL may be at increased risk of toxicity and are potentially candidates for dose reduction of agents requiring hepatic metabolism or excretion. At present, there are few firm guidelines useful for accurate dosing of antineoplastics in the setting of obvious hepatic disease.^148–150

In contrast, alterations in renal function generally correlate with renal clearance of drugs, since renal drug clearance tends to correlate with creatinine clearance. This has been well established for carboplatin, where a firm relationship exists between renal function and carboplatin clearance that can be used prospectively to modify the carboplatin dose and avoid excessive toxicity.^151–153 In addition, a recent study suggested that dose reduction of topotecan is required in patients with moderate renal dysfunction.¹⁵⁴

Abnormalities of protein binding are common but rarely impact upon clinical outcome. Many anticancer drugs, such as the Vinca alkaloids and etoposide, are highly protein bound.^143,155,156 Changes in protein binding may affect drug clearance.¹⁵⁷ Most importantly, abnormal protein binding must be considered in the interpretation of measured total plasma drug concentrations, since a decrease in protein binding will result in a relative increase in the pharmacologically active free drug.^143,158

Intrapatient Pharmacokinetic Variability

Although it is well established that interpatient pharmacokinetic variability may be significant, the importance of intrapatient variability (within a single patient) is less clear.¹⁵⁹ Oncologists are commonly faced with the clinical situation of increasing myelosuppression after repetitive dosing. This is generally assumed to be due to the cumulative effects of chemotherapy, making the patient more sensitive to subsequent doses. However, it is also possible that the patient’s clearance of the drug(s) may have decreased, resulting in increased drug exposure.

Such a situation may arise when either hepatic or renal function changes. Renal function may change due to progressive disease (ureteral obstruction), complications of therapy (volume depletion), or as a direct toxic effect of therapy (cisplatin). Similarly, renal function may improve over time, reducing the actual drug exposure. Hepatic function may also change, producing changes in drug clearance that may result in the appearance of increased toxicity over time, as is the case for VLB administered by prolonged continuous infusion.¹⁴⁶ Thus, clinicians should carefully review the outcome of prior doses to minimize the risk of an undesirable outcome due to intrapatient pharmacokinetic variability.

Another potential source of intrapatient pharmacokinetic variability is an individual’s circadian rhythm. The best studied drugs in this regard are 5-FU and 5-fluorodeoxyuridine.¹⁶⁰ Petit et al. evaluated circadian variability of 5-FU plasma concentrations during a 5-day infusion at a constant dose and demonstrated a greater than two-fold difference between maximum and minimum values.¹⁶¹ Similar results were obtained by Harris et al., who demonstrated an inverse correlation between plasma 5-FU concentration and the activity of dihydropyrimidine dehydrogenase, the major catabolic enzyme for 5-FU.¹⁶²

Drug-Drug Interactions

Despite the fact that anticancer drugs are almost always given as combination chemotherapy, often in conjunction with antiemetics and/or putative modulators, there have been relatively few studies in this area. One well-studied combination is paclitaxel and cisplatin, an important regimen for ovarian cancer, in which cisplatin reduces paclitaxel clearance if given first.¹⁶³

Studies of modulators of drug reactions have also demonstrated that inhibition of clearance may be an unexpected outcome. Such results have been demonstrated for cyclosporine A’s effects on etoposide¹⁶⁴ and doxorubicin^165,166 clearance, and interferon-alpha’s effect on 5-FU clearance.¹⁶⁷

Definitions

In a general sense, pharmacodynamics is the study of dose-response relationships.¹⁶⁸ Thus, any laboratory or clinical study employing different doses of an agent is addressing a pharmacodynamic question. Examples include the exposure of tumor cells in vitro to varying doses of a new agent to evaluate its dose-response relationship, or a phase I clinical trial to define the maximally tolerated dose and dose-limiting toxicities in patients.

In the clinical setting, the results of treatment depend on both pharmacokinetics and pharmacodynamics (Fig. 44.1). A patient may have excessive toxicity at the “standard” dose for one of two reasons. If the patient’s pharmacokinetics are different from those of the typical patient (e.g., decreased renal clearance of carboplatin), there may be decreased total body clearance, resulting in a higher than expected drug exposure. The second possibility is that the patient might simply be more sensitive to an average drug exposure—due to prior therapy, poor nutrition, or other less well-defined reasons. It is important to distinguish between these two possibilities. In the first case, lowering the dose, will result in an “average” drug exposure, whereas in the second case lowering the dose will result in a lower than average drug exposure. Therefore, in the setting of dose reduction, there is a greater possibility of a response in the patient with abnormal pharmacokinetics than in the “sensitive” patient with abnormal pharmacodynamics.

General Pharmacodynamic Principles

In the most general sense, any drug may be considered to have a maximal effect and a median dose (i.e., that required for 50% of the maximal effect). Wagner proposed a generalized sigmoidal model of drug effect (Figure 44.5), derived from the hypothesis that all drug effects require an initial interaction with a receptor.¹⁶⁹

Most studies addressing pharmacodynamic modeling of anticancer agents have addressed phase-specific agents separately.^170,171 It may be adequate to use a simple log-linear model for non-phase- specific agents:^170,172

This may be referred to as a steep dose-response curve, since the effect continues to increase proportionally as the concentration (C) increases. For any K (in equation 8), an increase in C by 2.3/K will result in a 1-log increase in antitumor effect (Figure 44.5A).

The dose-response relationships for phase-specific agents, such as the antimetabolites, are much more complicated. By definition, some cells are out of “phase” and therefore not sensitive (or relatively insensitive) to the effects of the drug during the period of drug exposure. This is not necessarily overcome by increasing the dose but could be overcome by increasing the duration of drug exposure. The result is the appearance of a plateau in the dose-response curve (Figure 44.5B).

The effects of some antineoplastic agents depend on both the drug concentration and the duration of exposure to that concentration. For some agents, the effect is a function of the product of the concentration and exposure time, analogous to the AUC.¹⁷³ However, for antimetabolites and other phase-specific agents, the mathematic relationships are much more complex.^170,171,174 Drug effect tends to be related to duration of exposure above a threshold concentration.

Plasma concentrations may be an inadequate predictor of clinical effect for those agents that undergo intracellular anabolism to active metabolites such as the case for ara-C.¹⁷⁵ Plasma ara-C concentrations do not appear to correlate with the rate of cellular ara-CTP accumulation or peak ara-CTP concentration in leukemic cells, although the intracellular concentration of ara-CTP is an important determinant of treatment outcome. Thus, knowledge of the plasma pharmacokinetics of ara-C is not likely to be a useful predictor of treatment outcome for individual patients. Pharmacogenetic evaluation may be potentially useful for modeling relationships between 6-mercaptopurine pharmacokinetics and clinical effects, as this drug’s conversion to active intracellular 6-thioguanine metabolites by thiopurine methyltransferase is genetically determined.¹²⁶ Studies in children with ALL suggest that intracellular levels of 6-thioguanine nucleotides may be an independent predictor of remission duration.¹²⁵

Pharmacodynamic Modeling of Cancer Chemotherapy

The introduction of pharmacodynamic modeling into clinical oncology has been a slow process. The relationship between toxicity subsequent to high-dose MTX and delayed MTX clearance has led to the routine use of therapeutic drug monitoring of plasma MTX concentrations to guide leucovorin dosing.¹⁷⁶ However, studies of other drugs have not clearly resulted in a change in clinical practice, although there has been a recent increase in clinical research in this area.¹⁷⁷

Most early pharmacodynamic studies addressed relationships between measurements of drug exposure (AUC, C_ss) and toxicity. More recently, investigators have used novel pharmacokinetic parameters to model toxicity, such as time above a threshold concentration for etoposide¹⁷⁸ and paclitaxel.^{116,117,179,180} Other investigators have addressed the importance of active metabolites. This is of particular importance for irinotecan, a drug with both complex metabolism and toxicity patterns. However, recent studies have suggested that irinotecan-induced diarrhea is secondary to relative deficiency in the glucuronidation of SN-38, its active metabolite.^181,182 Hematologic toxicity has been easier to model than nonhematologic toxicity as illustrated in Tables 44.4 and 44.5, respectively.

One of the best characterized drugs is carboplatin, an analog of cisplatin. Unlike cisplatin, the dose-limiting toxicity of carboplatin is thrombocytopenia, which is a function of drug dose, renal function, pretreatment platelet count, and prior therapy.¹⁵¹ The platelet nadir produced by a dose of carboplatin is related to the carboplatin clearance, which is directly proportional to creatinine clearance. Thus, patients at high risk of severe thrombocytopenia following carboplatin therapy can be identified prospectively and the drug doses can be modified by monitoring creatinine clearance.

Etoposide has also been the subject of extensive evaluation. Pharmacodynamic modeling of etoposide is complicated by the need to either measure free etoposide directly or to estimate the free etoposide concentration on the basis of measured total plasma etoposide concentration, albumin, and/or bilirubin.^156,158 Many studies have now demonstrated that the extent of leukopenia/neutropenia is correlated with etoposide exposure.^{158,183–190} Furthermore, interpatient pharmacodynamic variability may be significant and needs to be considered in future modeling of etoposide and potentially in other drugs.¹⁵⁸

There is an expanding interest in trying to optimize cancer chemotherapy by individualizing dosing on the basis of measurements of plasma or tissue drug concentrations. One recent example is the titration of carboplatin dosing discussed above. Other investigators have attempted to optimize the dosing of etoposide,^158,191 teniposide,¹⁹² hexamethylene bisacetamide,^193,194 etanidazole,¹⁹⁵ melphalan,¹⁹⁶ and 5-FU¹⁹⁷ by monitoring plasma drug concentrations during treatment, then using the information obtained to modify the total dose of chemotherapy administered in an attempt to avoid severe toxicity.

An important recent study from St. Jude Children’s Research Hospital demonstrated that individual dosing of combination chemotherapy can improve survival in children with ALL.¹⁹⁸ A total of 182 children received standard induction therapy followed by postremission therapy with ara-C, MTX, and teniposide. Patients were randomized to standard versus individualized dosing. The latter was based on adjusting the doses to achieve plasma AUCs in the 50 to 90^th percentile (based on historic controls). Those receiving individualized therapy had an improvement in the rate of continuous complete remission (76 vs 66%) at 5 years.

Future Role of Anticancer Pharmacodynamics

Should the clinical oncologist care about pharmacodynamics? Will therapeutic drug monitoring of antineoplastics be as useful as monitoring of theophylline or aminoglycoside dosing? How will these studies improve the therapeutic index? These are important issues that are currently being addressed.

Our true understanding of dosing of most antineoplastic drugs is primitive. Body surface area is generally the only value used to determine initial dosing, and even this has recently been questioned.^199,200 Prior toxicity may be used to adjust dosing for subsequent cycles, although doses are more often reduced than escalated and the magnitude of dose changes is determined empirically and often arbitrarily.

For drugs with a relatively broad therapeutic index and/or minimal interpatient pharmacokinetic or pharmacodynamic variability, these strategies may not be necessary. As an example, therapeutic drug monitoring of tamoxifen in breast cancer is unlikely to be useful. In contrast, therapeutic drug monitoring of doxorubicin in the adjuvant treatment of breast cancer may potentially help to ensure adequate drug exposure and minimize the risk of life-threatening toxicity, since there is an association between dose and relapse rate.²⁰¹

An achievable future goal is the individualization of dosing of drugs with polymorphic metabolism. Important examples are 5-FU,^202,203 6-mercaptopurine,^125,126 amonafide,^127,128 and irinotecan.¹³³ Most of the pharmacodynamic studies to date have focused on toxicity as an end point, primarily due to the patient populations studied (i.e., patients with refractory tumors enrolled in phase I clinical trials). The potential usefulness of this area is underscored by the St. Jude study demonstrating that the adjustment of dose based on plasma concentration decreases the rate of relapse.¹⁹⁸ Prospective evaluation of plasma¹²⁹ or intratumoral²⁰⁴ concentrations in conjunction with phase II clinical trials may improve our understanding of the relationship between clinical pharmacology and drug efficacy for other tumors as well. Although most studies to date have used continuous infusion chemotherapy, recent strategies have been developed with the aim of optimizing dosing after conventional bolus administration.^{175,192,196,205–207} It may eventually become possible to dose routinely toward a target AUC or C_ss (or time above a threshold) using principles of therapeutic drug monitoring to guide dosing.¹⁹⁸

The next challenge will be optimizing the use of combination chemotherapy. As studies of the pharmacodynamics of single agents are completed, it will become possible to evaluate the pharmacodynamics of drug combinations. As an example, it has been demonstrated that carboplatin plus paclitaxel results in less thrombocytopenia than carboplatin as a single agent.²⁰⁸

In conclusion, it is hoped that a better understanding of the clinical pharmacology of antineoplastics will improve the care of patients with cancer. At a minimum, clinicians should understand the basic principles, realizing the limitations of our current approaches.