Pharmacokinetics: The Dynamics of Drug Absorption, Distribution, and Elimination

The absorption, distribution, metabolism, and excretion of a drug all involve its passage across cell membranes. Mechanisms by which drugs cross membranes and the physicochemical properties of molecules and membranes that influence this transfer are, therefore, important. The determining characteristics of a drug are its molecular size and shape, degree of ionization, relative lipid solubility of its ionized and nonionized forms, and its binding to tissue proteins.

When a drug permeates a cell, it obviously must traverse the cellular plasma membrane. Other barriers to drug movement may be a single layer of cells (intestinal epithelium) or several layers of cells (skin). Despite such structural differences, the diffusion and transport of drugs across these various boundaries have many common characteristics, since drugs in general pass through cells rather than between them. The plasma membrane thus represents the common barrier.

Cell Membranes

The plasma membrane consists of a bilayer of amphipathic lipids, with their hydrocarbon chains oriented inward to form a continuous hydrophobic phase and their hydrophilic heads oriented outward. Individual lipid molecules in the bilayer vary according to the particular membrane and can move laterally, endowing the membrane with fluidity, flexibility, high electrical resistance, and relative impermeability to highly polar molecules. Membrane proteins embedded in the bilayer serve as receptors, ion channels, or transporters to elicit electrical or chemical signaling pathways and provide selective targets for drug actions.

Most cell membranes are relatively permeable to water either by diffusion or by flow resulting from hydrostatic or osmotic differences across the membrane, and bulk flow of water can carry with it drug molecules. Such transport is the major mechanism by which drugs pass across most capillary endothelial membranes. However, proteins and drug molecules bound to them are too large and polar for this type of transport to occur; thus, transcapillary movement is limited to unbound drug. Paracellular transport through intercellular gaps is sufficiently large that passage across most capillaries is limited by blood flow and not by other factors (see below). As described later, this type of transport is an important factor in filtration across glomerular membranes in the kidney. Important exceptions exist in such capillary diffusion, however, since 'tight' intercellular junctions are present in specific tissues and paracellular transport in them is limited. Capillaries of the central nervous system (CNS) and a variety of epithelial tissues have tight junctions (see below). Although bulk flow of water can carry with it small, water-soluble substances, if the molecular mass of these compounds is greater than 100 to 200 daltons, such transport is limited. Accordingly, most large lipophilic drugs must pass through the cell membrane itself by one or more processes.

Passive Membrane Transport

Drugs cross membranes either by passive processes or by mechanisms involving the active participation of components of the membrane. In the former, the drug molecule usually penetrates by passive diffusion along a concentration gradient by virtue of its solubility in the lipid bilayer. Such transfer is directly proportional to the magnitude of the concentration gradient across the membrane, the lipid:water partition coefficient of the drug, and the cell surface area. The greater the partition coefficient, the higher is the concentration of drug in the membrane and the faster is its diffusion. After a steady state is attained, the concentration of the unbound drug is the same on both sides of the membrane if the drug is a nonelectrolyte. For ionic compounds, the steady-state concentrations will be dependent on differences in pH across the membrane, which may influence the state of ionization of the molecule on each side of the membrane and on the electrochemical gradient for the ion.

Weak Electrolytes and Influence of pH

Most drugs are weak acids or bases that are present in solution as both the nonionized and ionized species. The nonionized molecules are usually lipid-soluble and can diffuse across the cell membrane. In contrast, the ionized molecules are usually unable to penetrate the lipid membrane because of their low lipid solubility.

Therefore, the transmembrane distribution of a weak electrolyte usually is determined by its pK_a and the pH gradient across the membrane. The pK_a is the pH at which half of the drug (weak electrolyte) is in its ionized form. To illustrate the effect of pH on distribution of drugs, the partitioning of a weak acid (pK_a= 4.4) between plasma (pH = 7.4) and gastric juice (pH = 1.4) is depicted in Figure 12. It is assumed that the gastric mucosal membrane behaves as a simple lipid barrier that is permeable only to the lipid-soluble, nonionized form of the acid. The ratio of nonionized to ionized drug at each pH is readily calculated from the HendersonHasselbalch equation. Thus, in plasma, the ratio of nonionized to ionized drug is 1:1000; in gastric juice, the ratio is 1:0.001. These values are given in brackets in Figure 12. The total concentration ratio between the plasma and the gastric juice would therefore be 1000:1 if such a system came to a steady state. For a weak base with a pK_a of 4.4, the ratio would be reversed, as would the thick horizontal arrows in Figure 12, which indicate the predominant species at each pH. Accordingly, at steady state, an acidic drug will accumulate on the more basic side of the membrane and a basic drug on the more acidic sidea phenomenon termed ion trapping. These considerations have obvious implications for the absorption and excretion of drugs, as discussed more specifically below. The establishment of concentration gradients of weak electrolytes across membranes with a pH gradient is a purely physical process and does not require an active transport system. All that is necessary is a membrane preferentially permeable to one form of the weak electrolyte and a pH gradient across the membrane. The establishment of the pH gradient is, however, an active process.

Carrier-Mediated Membrane Transport

While passive diffusion through the bilayer is dominant in the disposition of most drugs, carrier-mediated mechanisms also can play an important role. Active transport is characterized by a requirement for energy, movement against an electrochemical gradient, saturability, selectivity, and competitive inhibition by cotransported compounds. The term facilitated diffusion describes a carrier-mediated transport process in which there is no input of energy and therefore enhanced movement of the involved substance is down an electrochemical gradient. Such mechanisms, which may be highly selective for a specific conformational structure of a drug, are involved in the transport of endogenous compounds whose rate of transport by passive diffusion otherwise would be too slow. In other cases, they function as a barrier system to protect cells from potentially toxic substances.

The responsible transporter proteins often are expressed within cell membranes in a domain-specific fashion such that they mediate either drug uptake or efflux, and often such an arrangement facilitates vectorial transport across cells. Thus, in the liver, a number of basolaterally localized transporters with different substrate specificities are involved in the uptake of bile acids and amphipathic organic anions and cations into the hepatocyte, and a similar variety of ATP-dependent transporters in the canalicular membrane export such compounds into the bile. Analogous situations also are present in intestinal and renal tubular membranes. An important efflux transporter present at these sites and also in the capillary endothelium of brain capillaries is P-glycoprotein, which is encoded by the multidrug resistance-1 (MDR1) gene, important in resistance to cancer chemotherapeutic agents (Chapter 52: Antineoplastic Agents). P-glycoprotein localized in the enterocyte also limits the oral absorption of transported drugs since it exports the compound back into the intestinal tract subsequent to its absorption by passive diffusion.

The lipophilic characteristics of drugs that promote their passage through biological membranes and subsequent access to their site of action hinder their excretion from the body. Renal excretion of unchanged drug plays only a modest role in the overall elimination of most therapeutic agents, since lipophilic compounds filtered through the glomerulus are largely reabsorbed back into the systemic circulation during passage through the renal tubules. The metabolism of drugs and other xenobiotics into more hydrophilic metabolites is therefore essential for the elimination of these compounds from the body and termination of their biological activity. In general, biotransformation reactions generate more polar, inactive metabolites that are readily excreted from the body. However, in some cases, metabolites with potent biological activity or toxic properties are generated. Many of the metabolic biotransformation reactions leading to inactive metabolites of drugs also generate biologically active metabolites of endogenous compounds. The following discussion focuses on the biotransformation of drugs but is generally applicable to the metabolism of all xenobiotics as well as a number of endogenous compounds, including steroids, vitamins, and fatty acids.

Phase I and Phase II Metabolism

Drug biotransformation reactions are classified as either phase I functionalization reactions or phase II biosynthetic (conjugation) reactions. Phase I reactions introduce or expose a functional group on the parent compound. Phase I reactions generally result in the loss of pharmacological activity, although there are examples of retention or enhancement of activity. In rare instances, metabolism is associated with an altered pharmacological activity. Prodrugs are pharmacologically inactive compounds, designed to maximize the amount of the active species that reaches its site of action. Inactive prodrugs are converted rapidly to biologically active metabolites, often by the hydrolysis of an ester or amide linkage. If not rapidly excreted into the urine, the products of phase I biotransformation reactions can then react with endogenous compounds to form a highly water-soluble conjugate.

Phase II conjugation reactions lead to the formation of a covalent linkage between a functional group on the parent compound or phase I metabolite with endogenously derived glucuronic acid, sulfate, glutathione, amino acids, or acetate. These highly polar conjugates are generally inactive and are excreted rapidly in the urine and feces. An example of an active conjugate is the 6-glucuronide metabolite of morphine, which is a more potent analgesic than its parent compound.

Site of Biotransformation

The metabolic conversion of drugs generally is enzymatic in nature. The enzyme systems involved in the biotransformation of drugs are localized in the liver, although every tissue examined has some metabolic activity. Other organs with significant metabolic capacity include the gastrointestinal tract, kidneys, and lungs. Following nonparenteral administration of a drug, a significant portion of the dose may be metabolically inactivated in either the intestinal epithelium or the liver before it reaches the systemic circulation. This first-pass metabolism significantly limits the oral availability of highly metabolized drugs. Within a given cell, most drug-metabolizing activity is found in the endoplasmic reticulum and the cytosol, although drug biotransformations also can occur in the mitochondria, nuclear envelope, and plasma membrane. Upon homogenization and differential centrifugation of tissues, the endoplasmic reticulum breaks up, and fragments of the membrane form microvesicles, referred to as microsomes. The drug-metabolizing enzymes in the endoplasmic reticulum therefore often are classified as microsomal enzymes. The enzyme systems involved in phase I reactions are located primarily in the endoplasmic reticulum, while the phase II conjugation enzyme systems are mainly cytosolic. Often drugs biotransformed through a phase I reaction in the endoplasmic reticulum are conjugated at this same site or in the cytosolic fraction of the same cell.

Cytochrome P450 Monooxygenase System

The cytochrome P450 enzymes are a superfamily of heme-thiolate proteins widely distributed across all living kingdoms. The enzymes are involved in the metabolism of a plethora of chemically diverse, endogenous and exogenous compounds, including drugs, environmental chemicals, and other xenobiotics. Usually they function as a terminal oxidase in a multicomponent electron-transfer chain that introduces a single atom of molecular oxygen into the substrate with the other atom being incorporated into water. In microsomes, the electrons are supplied from NADPH via cytochrome P450 reductase, which is closely associated with cytochrome P450 in the lipid membrane of the smooth endoplasmic reticulum. Cytochrome P450 catalyzes many reactions, including aromatic and side-chain hydroxylation; N-, O- and S-dealkylation; N-oxidation; N-hydroxylation; sulfoxidation; deamination; dehalogenation; and desulfuration. Details and examples of cytochrome P450mediated metabolism are shown in Table 12. A number of reductive reactions also are catalyzed by these enzymes, generally under conditions of low oxygen tension.

Of the approximately 1000 currently known cytochrome P450s, about 50 are functionally active in human beings. These are categorized into 17 families and many subfamilies according to the amino acidsequence similarities of the predicted proteins; the abbreviated term CYP is used for identification. Sequences that are greater than 40% identical belong to the same family, identified by an Arabic number; within a family, sequences greater than 55% identical are in the same subfamily, identified by a letter; and different individual isoforms within the subfamily are identified by an Arabic number. About 8 to 10 isoforms in the CYP1, CYP2, and CYP3 families primarily are involved in the majority of all drug metabolism reactions in human beings; members of the other families are important in the biosynthesis and degradation of steroids, fatty acids, vitamins, and other endogenous compounds. Each individual CYP isoform appears to have a characteristic substrate specificity based on structural features of the substrate; considerable overlap, however, often is present. As a result, two or more CYP isoforms and other drug-metabolizing enzymes often are involved in a drug's overall metabolism, leading to the formation of many primary and secondary metabolites. The various isoforms also have characteristic inhibition and induction profiles, as described later. Additionally, CYP-catalyzed metabolism is often regio- and stereoselective; the latter characteristic may be important if the administered drug is a racemate and the enantiomers have different pharmacological activities.

The relative contributions of the various CYP isoforms in the metabolism of drugs is illustrated in Figure 13. CYP3A4 and CYP3A5, which are very similar isoforms, together are involved in the metabolism of about 50% of drugs; moreover, CYP3A is expressed in both the intestinal epithelium and the kidney. It is now recognized that metabolism by CYP3A during absorption through the intestinal enterocyte is a significant factor, along with hepatic first-pass metabolism, in the poor oral bioavailability of many drugs. Isoforms in the CYP2C family and CYP2D6 subfamily also are involved to a large extent in the metabolism of drugs. Although isoforms such as CYP1A1/2, CYP2A6, CYP2B1, and CYP2E1 are not involved to any major extent in the metabolism of therapeutic drugs, they do, however, catalyze the activation of many procarcinogenic environmental chemicals to the ultimate carcinogenic form. Accordingly, they are considered to be important in susceptibility to various cancers, such as tobacco smokingassociated lung cancer.

Other oxidative enzymes such as dehydrogenases and flavin-containing monooxygenases also are capable of catalyzing the metabolism of specific drugs, but, in general, such enzymes are of minor overall importance.

Hydrolytic Enzymes

The reactions of the major hydrolytic enzymes are illustrated in Table 12. A number of nonspecific esterases and amidases have been identified in the endoplasmic reticulum of human liver, intestine, and other tissues. The alcohol and amine groups exposed following hydrolysis of esters and amides are suitable substrates for conjugation reactions. Microsomal epoxide hydrolase is found in the endoplasmic reticulum of essentially all tissues and is in close proximity to the cytochrome P450 enzymes. Epoxide hydrolase generally is considered a detoxification enzyme, hydrolyzing highly reactive arene oxides generated from cytochrome P450 oxidation reactions to inactive, water-soluble transdihydrodiol metabolites. Protease and peptidase enzymes are widely distributed in many tissues and are involved in the biotransformation of polypeptide drugs. Delivery of such drugs across biological membranes requires the inhibition of these enzymes or the development of stable analogs.

Conjugation Reactions

Both an activated form of an endogenous compound and an appropriate transferase enzyme are necessary for the formation of a conjugated metabolite. In the case of glucuronidationthe most important conjugation reaction (Figure 13)uridine diphosphate glucuronosyltransferases (UGTs) catalyze the transfer of glucuronic acid to aromatic and aliphatic alcohols, carboxylic acids, amines, and free sulfhydryl groups of both exogenous and endogenous compounds to form O-, N-, and S-glucuronides, respectively. Glucuronidation also is important in the elimination of endogenous steroids, bilirubin, bile acids, and fat-soluble vitamins. The increased water solubility of a glucuronide conjugate promotes its elimination in the urine or bile. Unlike most phase II reactions, which are localized in the cytosol, UGTs are microsomal enzymes. This location facilitates direct access of phase I metabolites formed at the same site. In addition to the liver, UGTs also are found in the intestinal epithelium, kidney, and skin. About 15 human UGTs have been identified, and, based on amino acid similarity (>50% identity), two main families have been categorized. Members of the human UGT1A family are all encoded by a complex gene, and individual isoforms are produced by alternative splicing of 12 promoters/exon 1 with common exons 2 to 5 to produce multiple different proteins. By contrast, UGT2 contains only three subfamilies: 2A, 2B, and 2C. While it appears that individual UGTs have characteristic substrate specificities, there is considerable overlap, so that multiple isoforms may be responsible for formation of a particular glucuronide metabolite. Cytosolic sulfation also is an important conjugation reaction that involves the catalytic transfer by sulfotransferases (STs) of inorganic sulfur from activated 3^'-phosphoadenosine-5^'-phosphosulfate to the hydroxyl group of phenols and aliphatic alcohols. Therefore, drugs and primary metabolites with a hydroxyl group often form both glucuronide and sulfate metabolites. Two N-acetyltransferases (NAT1 and NAT2) are involved in the acetylation of amines, hydrazines, and sulfonamides. In contrast to most drug conjugates, acetylated metabolites often are less water-soluble than the parent drug, and this may result in crystalluria unless a high urine flow rate is maintained.

Factors Affecting Drug Metabolism

A hallmark of drug metabolism is a large interindividual variability that often results in marked differences in the extent of metabolism and, as a result, the drug's rate of elimination and other characteristics of its plasma concentrationtime profile. Such variability is a major reason why patients differ in their responses to a standard dose of a drug and it must be considered in optimizing a dosage regimen for a particular individual. A combination of genetic, environmental, and disease-state factors affect drug metabolism, with the relative contribution of each depending on the specific drug.

Genetic Variation

Advances in molecular biology have shown that genetic diversity is the rule rather than the exception with all proteins, including enzymes that catalyze drug-metabolism reactions. For an increasing number of such enzymes, allelic variants with different catalytic activities from that of the wild-type form have been identified. The differences involve a variety of molecular mechanisms leading to a complete lack of activity, a reduction in catalytic ability, or, in the case of gene duplication, enhanced activity. Furthermore, these traits are generally inherited in an autosomal, Mendelian recessive fashion and, if sufficiently prevalent, result in subpopulations with different drug-metabolizing abilities, i.e., genetic polymorphism. In addition, the frequency of specific allelic variants often varies according to the racial ancestry of the individual. It is possible to phenotype or genotype a person with respect to a particular genetic variant, and it is likely that such characterization will become increasingly useful in individualizing drug therapy, especially for drugs with a narrow therapeutic index. Accumulating evidence also suggests that individual susceptibility to diseases associated with environmental chemicals, such as cancer, may reflect genetic variability in drug-metabolizing enzymes.

A number of genetic polymorphisms are present in several cytochrome P450s that lead to altered drug metabolizing ability. The best characterized of these is that associated with CYP2D6. About 70 single nucleotide polymorphisms (SNPs) and other genetic variants of functional importance have been identified in the CYP2D6 gene, many of which result in an inactive enzyme while others reduce catalytic activity; gene duplication also occurs. As a result, four phenotypic subpopulations of metabolizers exist: poor (PM), intermediate (IM), extensive (EM), and ultrarapid (UM). Some of the variants are relatively rare, whereas others are more common, and importantly, their frequency varies according to racial background. For example, 5% to 10% of Caucasians of European ancestry are PMs, whereas the frequency of this homozygous phenotype in individuals of Southeast Asian origin is only about 1% to 2%. More than 65 commonly used drugs are metabolized by CYP2D6, including tricyclic antidepressants, neuroleptic agents, selective serotonin reuptake inhibitors, some antiarrhythmic agents, -adrenergic receptor antagonists, and certain opiates. The clinical importance of the CYP2D6 polymorphism is mainly in the greater likelihood of an adverse reaction in PMs when the affected metabolic pathway is a major contributor to the drug's overall elimination. Also, in UMs, usual drug doses may be inefficacious, or in the case where an active metabolite is formed, for example, the CYP2D6-catalyzed formation of morphine from codeine, an exaggerated response occurs. Inhibitors of CYP2D6, such as quinidine and selective serotonin reuptake inhibitors, may convert a genotypic EM into a phenotypic PM, a phenomenon termed phenocopying that is an important aspect of drug interactions with this particular CYP isoform.

CYP2C9 catalyzes the metabolism of some 16 commonly used drugs, including that of warfarin and phenytoin, both of which have a narrow therapeutic index. The two most common allelic CYP2C9 variants have markedly reduced catalytic activity (5% to 12%) compared to the wild-type enzyme. As a consequence, patients who are heterozygous or homozygous for the mutant alleles require a lower anticoagulating dose of warfarin, especially the latter group, relative to homozygous, wild-type individuals. Also, initiating warfarin therapy is more difficult, and there is an increased risk of bleeding complications. Similarly, high plasma concentrations of phenytoin and associated adverse effects occur in patients with variant CYP2C9 alleles. Genetic polymorphism also occurs with CYP2C19, where 8 allelic variants have been identified that result in a catalytically inactive protein. About 3% of Caucasians are phenotypically PMs, whereas the frequency is far higher in Southeast Asians, 13% to 23%. Proton-pump inhibitors such as omeprazole and lansoprazole are among the 18 or so drugs importantly metabolized by CYP2C19 to an extent determined by the gene dose. The efficacy of the recommended 20-mg dose of omeprazole in combination with amoxicillin in eradicating Helicobacter pylori is markedly reduced in patients of the homozygous wild-type genotype compared with the 100% cure rate in homozygous PMs, reflecting differences in the drug's effect on gastric acid secretion. Although CYP3A activity shows marked interindividual variability (>10-fold), no significant functional polymorphisms have been found in the gene's coding region; it is, therefore, likely that unknown regulatory factors primarily determine such variability. Genetic variability also is present with dihydropyrimidine dehydrogenase (DPYD), which is a key enzyme in the metabolism of 5-fluorouracil. Accordingly, there is a marked risk of developing severe drug-induced toxicity in the 1% to 3% of cancer patients treated with this antimetabolite who have substantially reduced DPYD activity compared to the general population.

A polymorphism in a conjugating drug-metabolizing enzyme, namely that in NAT2, was one of the first to be found to have a genetic basis some 50 years ago. This isoform is involved in the metabolism of about 16 common drugs including isoniazid, procainamide, dapsone, hydralazine, and caffeine. About 15 allelic variants have been identified, some of which are without functional effect, but others are associated with either reduced or absent catalytic activity. Considerable heterogeneity is present in the worldwide population frequency of these alleles, so that the slow-acetylator phenotype frequency is about 50% in American whites and blacks, 60% to 70% in North Europeans, but only 5% to 10% in Southeast Asians. It has been speculated that acetylator phenotype may be associated with environmental agentinduced disease such as bladder and colorectal cancer; however, definitive evidence is not yet available. Similarly, genetic variability in the catalytic activity of glutathione S-transferases may be linked to individual susceptibility to such diseases. Thiopurine methyltransferase (TPMT) is critically important in the metabolism of 6-mercaptopurine, the active metabolite of azathioprine. As a result, homozygotes for alleles encoding inactive TPMT (0.3% to 1% of the population) predictably exhibit severe pancytopenia if given standard doses of azathioprine; such patients typically can be treated with 10% to 15% of the usual dose.

Environmental Determinants

The activity of most drug-metabolizing enzymes may be modulated by exposure to certain exogenous compounds. In some instances, this may be a drug, which, if concomitantly administered with a second agent, results in a drug:drug interaction. Additionally, dietary micronutrients and other environmental factors can up- or down-regulate the enzymes, termed induction and inhibition, respectively. Such modulation is thought to be a major contributor to interindividual variability in the metabolism of many drugs.

Inhibition of Drug Metabolism

A consequence of inhibiting drug-metabolizing enzymes is an increase in the plasma concentration of parent drug and a reduction in that of metabolite, exaggerated and prolonged pharmacological effects, and an increased likelihood of drug-induced toxicity. These changes occur rapidly and with essentially no warning and are most critical for drugs that are extensively metabolized and have a narrow therapeutic index. Knowledge of the cytochrome-P450 isoforms that catalyze the main pathway of metabolism of a drug provides a basis for predicting and understanding inhibition, especially with regard to drug-drug interactions. This is because many inhibitors are more selective for some isoforms than others. Often, inhibition occurs because of competition between two or more substrates for the same active site of the enzyme, the extent of which depends on the relative concentrations of the substrates and their affinities for the enzyme. In certain instances, however, the enzyme may be irreversibly inactivated; for example, the substrate or a metabolite forms a tight complex with the heme iron of cytochrome P450 (cimetidine, ketoconazole) or the heme group may be destroyed (norethindrone, ethinylestradiol). A common mechanism of inhibition for some phase II enzymes is the depletion of necessary cofactors.

Inhibition of the CYP3A-catalyzed mechanism is both common and important. Because of the high expression level of CYP3A in the intestinal epithelium and the fact that oral ingestion is the most common route of entry of drugs and environmental agents into the body, inhibition of the isoform's activity at this site is often particularly consequential, even if that in the liver is unaffected. This is because of the potential, large increase in bioavailability associated with the reduction in first-pass metabolism for drugs that usually exhibit this effect to a substantial extent. The antifungal agents ketoconazole and itraconazole, HIV protease inhibitors (especially ritonavir), macrolide antibiotics such as erythromycin and clarithromycin but not azithromycin, are all potent CYP3A inhibitors. Certain calcium channel blockers such as diltiazem, nicardipine, and verapamil also inhibit CYP3A, as does a constituent of grapefruit juice. Many inhibitors of CYP3A also reduce P-glycoprotein function, so that drug-drug interactions may involve a dual mechanism. Also, the disposition of drugs that are not significantly metabolized but are eliminated by P-glycoproteinmediated transport also may be affected by a CYP3A inhibitor. For example, the impaired excretion of digoxin by quinidine and a large number of other unrelated drugs is caused by inhibition of P-glycoprotein. With CYP2D6, quinidine and selective serotonin reuptake inhibitors are potent inhibitors that may produce phenocopying. On the other hand, other drugs are more general inhibitors of cytochrome P450catalyzed metabolism. For example amiodarone, cimetidine (but not ranitidine), paroxitene, and fluoxetine reduce the metabolic activity of several CYP isoforms. Phase I metabolic enzymes other than cytochrome P450 also may be inhibited by drug administration, as exemplified by the potent effect of valproic acid on microsomal epoxide hydrolase, and the inhibition of xanthine oxidase by allopurinol, which can result in life-threatening toxicity in patients concurrently receiving 6-mercaptopurine.

Induction of Drug Metabolism

Up-regulation of drug- metabolizing activity usually occurs by enhanced gene transcription following prolonged exposure to an inducing agent, although with CYP2E1 stabilization of the protein against degradation is the major mechanism. As a result, the consequences of induction take considerable time to be fully exhibited, c.f., inhibition of metabolism. Moreover, the consequences of induction are an increased rate of metabolism, enhanced oral first-pass metabolism and reduced bioavailability, and a corresponding decrease in the drug's plasma concentration, all factors that reduce drug exposure. By contrast, for drugs that are metabolized to an active or reactive metabolite, induction may be associated with increased drug effects or toxicity, respectively. In some cases, a drug can induce both the metabolism of other compounds and its own metabolism; such autoinduction occurs with the anticonvulsant carbamazepine. In many cases involving induction, the dosage of an affected drug must be increased to maintain the therapeutic effect. This is particularly the case when induction is extensive following administration of a highly effective inducer; in fact, women are advised to use an alternative to oral contraceptives for birth control during rifampin therapy because efficacy cannot be assured. The therapeutic risk associated with metabolic induction is most critical when administration of the inducing agent is stopped while maintaining the same dose of a drug that has been previously given. In this case, as the inducing effect wears off, plasma concentrations of the second drug will rise unless the dose is reduced, with an increase in the potential for adverse effects.

Inducers generally are selective for certain CYP subfamilies and isoforms, but at the same time, multiple other enzymes may be simultaneously up-regulated through a common molecular mechanism. For example, polycyclic aromatic hydrocarbons derived from environmental pollutants, cigarette smoke, and charbroiled meats produce marked induction of the CYP1A subfamily of enzymes both in the liver and extrahepatically. This involves activation of the cytosolic arylhydrocarbon receptor (AhR), which interacts with another regulatory protein, the AhR nuclear translocator (Arnt); the complex functions as a transcription factor to up-regulate CYP1A expression. In addition, the expression of phase II enzymes such as UGTs, GSTs, and NAD(P)H:quinone oxidoreductase are simultaneously increased. A similar type of receptor mechanism involving the pregnane X receptor (PXR) is involved in the induction of CYP3A by a wide variety of diverse chemicals, including drugs such as rifampin and rifabutin, barbiturates and other anticonvulsants, some glucocorticoids, and even alternative medicines such as St. John's wort. These latter drugs also can affect other CYP isoforms; for example, rifampin and carbamazepine induce CYP1A2, CYP2C9, and CYP2C19. Chronic alcohol use also results in enzyme induction, especially with CYP2E1; the risk of hepatotoxic adverse effects of acetaminophen is higher in alcoholics because of increased CYP2E1-mediated formation of a reactive metabolite, N-acetyl-p-benzoquinoneimine.

Disease Factors

Since the liver is the major location of drug-metabolizing enzymes, dysfunction in this organ in patients with hepatitis, alcoholic liver disease, biliary cirrhosis, fatty liver, and hepatocarcinomas potentially can lead to impaired drug metabolism. In general, the severity of the liver damage determines the extent of reduced metabolism; unfortunately, common clinical tests of liver function are of little value in assessing this. Moreover, even in severe cirrhosis, the extent of impairment is only to about 30% to 50% of the activity in non-liver-diseased patients. However, with drugs that undergo substantial hepatic first-pass metabolism, oral bioavailability may be increased two- to fourfold in liver disease which, coupled with the prolonged presence of drugs in the body, increases the risk of exaggerated pharmacological responses and adverse effects. It appears that cytochrome-P450 isoforms are affected to a greater extent by liver disease than are those that catalyze phase II reactions such as glucuronosyltransferases.

Severe cardiac failure and shock can result in both decreased perfusion of the liver and impaired metabolism. The best example of this is the almost twofold reduction in lidocaine metabolism in cardiac failure, which also is accompanied by a change in distribution to a similar extent. As a result, the loading and maintenance doses of lidocaine used to treat cardiac arrhythmias in such patients are substantially different from those used in patients without this condition.

Age and Sex

Functional cytochrome P450 isoforms and to a lesser degree phase II drug-metabolizing enzymes develop early in fetal development, but the levels, even at birth, are lower than those found postnatally. Both phase I and phase II enzymes begin to mature gradually following the first 2 to 4 weeks postpartum, although the pattern of development is variable for the different enzymes. Thus, newborns and infants are able to metabolize drugs relatively efficiently but generally at a slower rate than are adults. An exception to this is the impairment of bilirubin glucuronidation at birth, which contributes to the hyperbilirubinemia of newborns. Full maturity appears to occur in the second decade of life with a subsequent slow decline in function associated with aging. Unfortunately, few generalizations are possible regarding the extent or clinical importance of such age-related changes in an individual patient. This is particularly true for elderly patients who, because of multiple diseases, may be taking a large number of drugs, many of which may produce drug-drug interactions. In addition, increased sensitivity of target organs and impairment of physiological control mechanisms further complicate the use of drugs in the elderly population. Phase I drug-metabolizing enzymes appear to be affected to a greater extent than are those that catalyze phase II reactions. However, the changes are often modest relative to other causes of interindividual variability in metabolism. On the other hand, for drugs exhibiting a high first-pass effect, even a small reduction in metabolizing ability may significantly increase oral bioavailability. Drug use in the elderly, therefore, generally requires moderate reductions in drug dose and awareness of the possibility of exaggerated pharmacodynamic responsiveness.

A number of examples indicate that drug treatment and/or responsiveness of men and women may be different for certain drugs. Some sex-related differences in drug-metabolizing activity, especially that catalyzed by CYP3A, also have been noted. However, such differences are minor and unimportant relative to other factors involved in interindividual variability in metabolism. One exception to this generalization is pregnancy, where induction of certain drug-metabolizing enzymes occurs in the second and third trimesters. As a result, drug dosage may have to be increased during this period and returned to its previous level postpartum. This situation is particularly important in the treatment of patients with seizures using phenytoin during their pregnancy. Many oral contraceptive agents also are potent irreversible inhibitors of CYP isoforms through a suicide-inactivation mechanism.

A fundamental hypothesis of clinical pharmacokinetics is that a relationship exists between the pharmacological effects of a drug and an accessible concentration of the drug (e.g., in blood or plasma). This hypothesis has been documented for many drugs, although for some drugs no clear or simple relationship has been found between pharmacological effect and concentration in plasma. In most cases, as depicted in Figure 11, the concentration of drug in the systemic circulation will be related to the concentration of drug at its sites of action. The pharmacological effect that results may be the clinical effect desired, a toxic effect, or, in some cases, an effect unrelated to therapeutic efficacy or toxicity. Clinical pharmacokinetics attempts to provide both a quantitative relationship between dose and effect and a framework with which to interpret measurements of concentrations of drugs in biological fluids. The importance of pharmacokinetics in patient care is based on the improvement in therapeutic efficacy that can be attained by application of its principles when dosage regimens are chosen and modified.

The various physiological and pathophysiological variables that dictate adjustment of dosage in individual patients often do so as a result of modification of pharmacokinetic parameters. The four most important parameters are clearance, a measure of the body's efficiency in eliminating drug; volume of distribution, a measure of the apparent space in the body available to contain the drug; elimination half-life, a measure of the rate of removal of drug from the body; and bioavailability, the fraction of drug absorbed as such into the systemic circulation. Of lesser importance are the rates of availability and distribution of the agent.

Clearance

Clearance is the most important concept that needs to be considered when a rational regimen for long-term drug administration is to be designed. The clinician usually wants to maintain steady-state concentrations of a drug within a therapeutic window associated with therapeutic efficacy and a minimum of toxicity. Assuming complete bioavailability, the steady state will be achieved when the rate of drug elimination equals the rate of drug administration:

Dosing rate =CL C_SS   (11)

where CL is clearance from the systemic circulation and C_ss is the steady-state concentration of drug. Thus, if the desired steady-state concentration of drug in plasma or blood is known, the rate of clearance of drug by the patient will dictate the rate at which the drug should be administered.

The concept of clearance is extremely useful in clinical pharmacokinetics, because its value for a particular drug usually is constant over the range of concentrations encountered clinically. This is true because systems for elimination of drugs such as metabolizing enzymes and transporters usually are not saturated, and thus the absolute rate of elimination of the drug is essentially a linear function of its concentration in plasma. A synonymous statement is that the elimination of most drugs follows first-order kineticsa constant fraction of drug in the body is eliminated per unit of time. If mechanisms for elimination of a given drug become saturated, the kinetics approach zero-ordera constant amount of drug is eliminated per unit of time. Under such a circumstance, clearance will vary with the concentration of drug, often according to the following equation:

CL = v_m/(K_m+ C)   (12)

where K_m represents the concentration at which half of the maximal rate of elimination is reached (in units of mass/volume) and _m is equal to the maximal rate of elimination (in units of mass/time). This equation is analogous to the MichaelisMenten equation for enzyme kinetics. Design of dosage regimens for such drugs is more complex than when elimination is first-order and clearance is independent of the drug's concentration (see below).

Principles of drug clearance are similar to those of renal physiology, where, for example, creatinine clearance is defined as the rate of elimination of creatinine in the urine relative to its concentration in plasma. At the simplest level, clearance of a drug is its rate of elimination by all routes normalized to the concentration of drug, C, in some biological fluid:

CL= rate of elimination/C   (13)

Thus, when clearance is constant, the rate of drug elimination is directly proportional to drug concentration. It is important to note that clearance does not indicate how much drug is being removed but, rather, the volume of biological fluid such as blood or plasma from which drug would have to be completely removed to account for the elimination. Clearance is expressed as a volume per unit of time. Clearance usually is further defined as blood clearance (CL_b), plasma clearance (CL_p), or clearance based on the concentration of unbound drug (CL_u), depending on the concentration measured (C_b, C_p, or C_u).

Clearance by means of various organs of elimination is additive. Elimination of drug may occur as a result of processes that occur in the kidney, liver, and other organs. Division of the rate of elimination by each organ by a concentration of drug (e.g., plasma concentration) will yield the respective clearance by that organ. Added together, these separate clearances will equal systemic clearance:

CL_renal+ CL_hepatic+ CL_other= CL   (14)

Other routes of elimination could include that in saliva or sweat, secretion into the intestinal tract, and metabolism at other sites.

Systemic clearance may be determined at steady state by using Equation (11). For a single dose of a drug with complete bioavailability and first-order kinetics of elimination, systemic clearance may be determined from mass balance and the integration of Equation (13) over time.

CL= Dose/AUC   (15)

where AUC is the total area under the curve that describes the concentration of drug in the systemic circulation as a function of time (from zero to infinity).

Examples

In Appendix II, the plasma clearance for cephalexin is reported as 4.3 ml min¹ kg¹, with 90% of the drug excreted unchanged in the urine. For a 70-kg man, the clearance from plasma would be 300 ml/minute, with renal clearance accounting for 90% of this elimination. In other words, the kidney is able to excrete cephalexin at a rate such that it is completely removed (cleared) from approximately 270 ml of plasma per minute. Because clearance usually is assumed to remain constant in a stable patient, the rate of elimination of cephalexin will depend on the concentration of drug in the plasma [Equation (13)]. Propranolol is cleared from the blood at a rate of 16 ml min¹ kg¹ (or 1120 ml/minute in a 70-kg man), almost exclusively by the liver. Thus, the liver is able to remove the amount of drug contained in 1120 ml of blood per minute. Even though the liver is the dominant organ for elimination, the plasma clearance of some drugs exceeds the rate of plasma (and blood) flow to this organ. Often this is because the drug partitions readily into red blood cells, and the rate of drug delivered to the eliminating organ is considerably higher than suspected from measurement of its concentration in plasma. The relationship between plasma and blood clearance at steady state is given by:

Clearance from the blood, therefore, may be estimated by dividing the plasma clearance by the drug's blood to plasma concentration ratio, obtained from knowledge of the hematocrit (H= 0.45) and the red cell to plasma concentration ratio. In most instances the blood clearance will be less than liver blood flow (1.5 liters/minute) or, if renal excretion also is involved, the sum of the two eliminating organs' blood flows. For example, the plasma clearance of tacrolimus, about 2 liters/minute, is more than twofold higher than the hepatic plasma flow rate and even exceeds the organ's blood flow, despite the fact that the liver is the predominant site of this drug's extensive metabolism. However, after taking into account the extensive distribution of tacrolimus into red cells, its clearance from the blood is only about 63 ml/minute, and it is actually a low- rather than high-clearance drug, as might be interpreted from the plasma clearance value. Sometimes, however, clearance from the blood by metabolism exceeds liver blood flow, and this indicates extrahepatic metabolism. In the case of esmolol (11.9 liters/minute), the blood clearance value is greater than cardiac output, because the drug is efficiently metabolized by esterases present in red blood cells.

A further definition of clearance is useful for understanding the effects of pathological and physiological variables on drug elimination, particularly with respect to an individual organ. The rate of presentation of drug to the organ is the product of blood flow (Q) and the arterial drug concentration (C_A), and the rate of exit of drug from the organ is the product of blood flow and the venous drug concentration (C_V). The difference between these rates at steady state is the rate of drug elimination:

Division of Equation (17) by the concentration of drug entering the organ of elimination, C_A, yields an expression for clearance of the drug by the organ in question:

The expression (C_AC_V)/C_A in Equation (18) can be referred to as the extraction ratio for the drug (E).

Hepatic Clearance

The concepts developed in Equation (18) have important implications for drugs that are eliminated by the liver. Consider a drug that is efficiently removed from the blood by hepatic processesmetabolism and/or excretion of drug into the bile. In this instance, the concentration of drug in the blood leaving the liver will be low, the extraction ratio will approach unity, and the clearance of the drug from blood will become limited by hepatic blood flow. Drugs that are cleared efficiently by the liver (e.g., drugs in Appendix II with systemic clearances greater than 6 ml min¹ kg¹, such as diltiazem, imipramine, lidocaine, morphine, and propranolol) are restricted in their rate of elimination, not by intrahepatic processes, but by the rate at which they can be transported in the blood to the liver.

Additional complexities also have been considered. For example, the equations presented above do not account for drug binding to components of blood and tissues, nor do they permit an estimation of the intrinsic ability of the liver to eliminate a drug in the absence of limitations imposed by blood flow, termed intrinsic clearance. In biochemical terms and under first-order conditions, intrinsic clearance is a measure of the ratio of the MichaelisMenten kinetic parameters for the eliminating process, i.e., _m/K_m. Extensions of the relationships of Equation (18) to include expressions for protein binding and intrinsic clearance have been proposed for a number of models of hepatic elimination (seeMorgan and Smallwood, 1990). All of these models indicate that, when the capacity of the eliminating organ to metabolize the drug is large in comparison with the rate of presentation of drug, clearance will approximate the organ's blood flow. In contrast, when the metabolic capability is small in comparison to the rate of drug presentation, clearance will be proportional to the unbound fraction of drug in blood and the drug's intrinsic clearance. Appreciation of these concepts allows understanding of a number of possibly puzzling experimental results. For example, enzyme induction or hepatic disease may change the rate of drug metabolism in an isolated hepatic microsomal enzyme system but not change clearance in the whole animal. For a drug with a high extraction ratio, clearance is limited by blood flow, and changes in intrinsic clearance due to enzyme induction or hepatic disease should have little effect. Similarly, for drugs with high extraction ratios, changes in protein binding due to disease or competitive binding interactions should have little effect on clearance. In contrast, changes in intrinsic clearance and protein binding will affect the clearance of drugs with low intrinsic clearances and, thus, extraction ratios, but changes in blood flow should have little effect (Wilkinson and Shand, 1975).

Renal Clearance

Renal clearance of a drug results in its appearance as such in the urine; changes in the pharmacokinetic properties of drugs due to renal disease also may be explained in terms of clearance concepts. However, the complications that relate to filtration, active secretion, and reabsorption must be considered. The rate of filtration of a drug depends on the volume of fluid that is filtered in the glomerulus and the unbound concentration of drug in plasma, since drug bound to protein is not filtered. The rate of secretion of drug by the kidney will depend on the drug's intrinsic clearance by the transporters involved in active secretion as affected by the drug's binding to plasma proteins, the degree of saturation of these transporters, and the rate of delivery of the drug to the secretory site. In addition, processes involved in drug reabsorption from the tubular fluid must be considered. The influences of changes in protein binding, blood flow, and the number of functional nephrons are analogous to the examples given above for hepatic elimination.

Distribution

Volume of Distribution

Volume is a second fundamental parameter that is useful in considering processes of drug disposition. The volume of distribution (V) relates the amount of drug in the body to the concentration of drug (C) in the blood or plasma, depending upon the fluid measured. This volume does not necessarily refer to an identifiable physiological volume, but merely to the fluid volume that would be required to contain all of the drug in the body at the same concentration as in the blood or plasma:

V= amount of drug in body/C   (19)

A drug's volume of distribution, therefore, reflects the extent to which it is present in extravascular tissues. The plasma volume of a typical 70-kg man is 3 liters, blood volume is about 5.5 liters, extracellular fluid volume outside the plasma is 12 liters, and the volume of total body water is approximately 42 liters. However, many drugs exhibit volumes of distribution far in excess of these values. For example, if 500 g of digoxin were in the body of a 70-kg subject, a plasma concentration of approximately 0.75 ng/ml would be observed. Dividing the amount of drug in the body by the plasma concentration yields a volume of distribution for digoxin of about 650 liters, or a value almost ten times greater than the total body volume of a 70-kg man. In fact, digoxin distributes preferentially to muscle and adipose tissue and to its specific receptors, leaving a very small amount of drug in the plasma. For drugs that are extensively bound to plasma proteins but that are not bound to tissue components, the volume of distribution will approach that of the plasma volume. In contrast, certain drugs have high volumes of distribution even though most of the drug in the circulation is bound to albumin, because these drugs are also sequestered elsewhere.

The volume of distribution may vary widely depending on the relative degrees of binding to plasma and tissue proteins, the partition coefficient of the drug in fat, and so forth. As might be expected, the volume of distribution for a given drug can differ according to patient's age, gender, body composition, and presence of disease.

Several volume terms commonly are used to describe drug distribution, and they have been derived in a number of ways. The volume of distribution defined in Equation (19) considers the body as a single homogeneous compartment. In this one-compartment model, all drug administration occurs directly into the central compartment and distribution of drug is instantaneous throughout the volume (V). Clearance of drug from this compartment occurs in a first-order fashion, as defined in Equation (13); that is, the amount of drug eliminated per unit of time depends on the amount (concentration) of drug in the body compartment. Figure 14A and Equation (110) describe the decline of plasma concentration with time for a drug introduced into this compartment.

C= (dose/V) exp(kt)   (110)

where k is the rate constant for elimination that reflects the fraction of drug removed from the compartment per unit of time. This rate constant is inversely related to the half-life of the drug (k= 0.693/t_1/2).

The idealized one-compartment model discussed above does not describe the entire time course of the plasma concentration. That is, certain tissue reservoirs can be distinguished from the central compartment, and the drug concentration appears to decay in a manner that can be described by multiple exponential terms (seeFigure 14B). Nevertheless, the one-compartment model is sufficient to apply to most clinical situations for most drugs.

Rate of Drug Distribution

The multiple exponential decay observed for a drug that is eliminated from the body with first-order kinetics results from differences in the rates at which the drug equilibrates with tissues. The rate of equilibration will depend upon the ratio of the perfusion of the tissue to the partition of drug into the tissue. In many cases, groups of tissues with similar perfusion/partition ratios all equilibrate at essentially the same rate, such that only one apparent phase of distribution (rapid initial fall of concentration, as in Figure 14B) is seen. It is as though the drug starts in a 'central' volume, which consists of plasma and tissue reservoirs that are in rapid equilibrium with it, and distributes to a 'final' volume, at which point concentrations in plasma decrease in a log-linear fashion with a rate constant of k (seeFigure 14B).

If the pattern or ratio of blood flows to various tissues changes within an individual or differs among individuals, rates of drug distribution to tissues also will change. However, changes in blood flow also may cause some tissues that were originally in the 'central' volume to equilibrate sufficiently more slowly so as to appear only in the 'final' volume. This means that central volumes will appear to vary with disease states that cause altered regional blood flow. After an intravenous bolus dose, drug concentrations in plasma may be higher in individuals with poor perfusion (e.g., shock) than they would be if perfusion were better. These higher systemic concentrations may, in turn, cause higher concentrations (and greater effects) in tissues such as brain and heart, whose usually high perfusion has not been reduced by the altered hemodynamic state. Thus, the effect of a drug at various sites of action can be variable, depending on perfusion of these sites.

Multicompartment Volume Terms

Two different terms have been used to describe the volume of distribution for drugs that follow multiple exponential decay. The first, designated V_area, is calculated as the ratio of clearance to the rate of decline of concentration during the elimination (final) phase of the logarithmic concentration versustime curve:

The estimation of this parameter is straightforward, and the volume term may be determined after administration of a single dose of drug by intravenous or oral routes (where the dose used must be corrected for bioavailability). However, another multicompartment volume of distribution may be more useful, especially when the effect of disease states on pharmacokinetics is to be determined. The volume of distribution at steady state (V_ss) represents the volume in which a drug would appear to be distributed during steady state if the drug existed throughout that volume at the same concentration as that in the measured fluid (plasma or blood). Following intravenous dosing, estimation of V_ss is more complicated than Equation (111), but feasible (Benet and Galeazzi, 1979). It is more difficult to estimate V_ss following oral dosing. Although V_area is a convenient and easily calculated parameter, it varies when the rate constant for drug elimination changes, even when there has been no change in the distribution space. This is because the terminal rate of decline of the concentration of drug in blood or plasma depends not only on clearance but also on the rates of distribution of drug between the 'central' and 'final' volumes. V_ss does not suffer from this disadvantage. When using pharmacokinetics to make drug dosing decisions, the differences between V_area and V_ss usually are not clinically significant. Nonetheless, both are quoted in the table of pharmacokinetic data in Appendix II, depending upon availability in the published literature.

Half-Life

The half-life (t_1/2) is the time it takes for the plasma concentration or the amount of drug in the body to be reduced by 50%. For the simplest case, the one-compartment model (Figure 14A), half-life may be determined readily and used to make decisions about drug dosage. However, as indicated in Figure 14B, drug concentrations in plasma often follow a multiexponential pattern of decline; two or more half-life terms may thus be calculated.

In the past, the half-life that was usually reported corresponded to the terminal log-linear phase of elimination. However, as greater analytical sensitivity has been achieved, the lower concentrations measured appeared to yield longer and longer terminal half-lives. For example, a terminal half-life of 53 hours is observed for gentamicin (versus the more clinically relevant 2- to 3-hour value in Appendix II), and biliary cycling is probably responsible for the 120-hour terminal value for indomethacin (as compared with the 2.4-hour half-life listed in Appendix II). The relevance of a particular half-life may be defined in terms of the fraction of the clearance and volume of distribution that is related to each half-life and whether plasma concentrations or amounts of drug in the body are best related to measures of response. The single half-life values given for each drug in Appendix II are chosen to represent the most clinically relevant half-life.

Early studies of pharmacokinetic properties of drugs in disease were compromised by their reliance on half-life as the sole measure of alterations of drug disposition. It is now appreciated that half-life is a derived parameter that changes as a function of both clearance and volume of distribution. A useful approximate relationship between the clinically relevant half-life, clearance, and volume of distribution at steady state is given by:

t 0.693 V_SS/CL   (112)

Clearance is the measure of the body's ability to eliminate a drug; thus, as clearance decreases, due to a disease process, for example, half-life would be expected to increase. However, this reciprocal relationship is valid only when the disease does not change the volume of distribution. For example, the half-life of diazepam increases with increasing age; however, it is not clearance that changes as a function of age, but the volume of distribution (Klotz et al., 1975). Similarly, changes in protein binding of the drug may affect its clearance as well as its volume of distribution, leading to unpredictable changes in half-life as a function of disease. The half-life of tolbutamide, for example, decreases in patients with acute viral hepatitis, exactly the opposite from what one might expect. The disease alters the drug's protein binding in both plasma and tissues, causing no change in volume of distribution but an increase in clearance, because higher concentrations of unbound drug are present (Williams et al., 1977).

Although it can be a poor index of drug elimination, half-life does provide a good indication of the time required to reach steady state after a dosage regimen is initiated or changed (i.e., four half-lives to reach approximately 94% of a new steady state), the time for a drug to be removed from the body, and a means to estimate the appropriate dosing interval (see below).

Steady State

Equation (11) indicates that a steady-state concentration eventually will be achieved when a drug is administered at a constant rate. At this point, drug elimination [the product of clearance and concentration; Equation (13)] will equal the rate of drug availability. This concept also extends to intermittent dosage (e.g., 250 mg of drug every 8 hours). During each interdose interval, the concentration of drug rises and falls. At steady state, the entire cycle is repeated identically in each interval. Equation (11) still applies for intermittent dosing, but it now describes the average drug concentration (C_ss) during an interdose interval. Steady-state dosing is illustrated in Figure 15.

Extent and Rate of Bioavailability

Bioavailability

It is important to distinguish between the rate and extent of drug absorption and the amount of drug that ultimately reaches the systemic circulation, as discussed above. The amount of the drug that reaches the systemic circulation depends not only on the administered dose but also on the fraction of the dose, F, which is absorbed and escapes any first-pass elimination. This fraction often is called bioavailability. Reasons for incomplete absorption have been discussed above. Also, as noted previously, if the drug is metabolized in the intestinal epithelium or the liver, or excreted in bile, some of the active drug absorbed from the gastrointestinal tract will be eliminated before it can reach the general circulation and be distributed to its sites of action.

Knowing the extraction ratio (E_H) for a drug across the liver [seeEquation (18)], it is possible to predict the maximum oral availability (F_max), assuming hepatic elimination follows first-order processes:

F_max= 1 E_H= 1 (CL_hepatic/Q_hepatic)

Thus, if the hepatic blood clearance for the drug is large relative to hepatic blood flow, the extent of availability will be low when it is given orally (e.g., lidocaine). This reduction in availability is a function of the physiological site from which absorption takes place, and no modification of dosage form will improve the availability under conditions of linear kinetics. Incomplete absorption and/or intestinal metabolism following oral dosing will, in practice, reduce this predicted maximal value of F.

When drugs are administered by a route that is subject to first-pass loss, the equations presented previously that contain the terms dose or dosing rate [Equations (11), (15), (110), and (111)] also must include the bioavailability term F, such that the available dose or dosing rate is used. For example, Equation (11) is modified to:

F dosing rate =CL C_SS   (114)

Rate of Absorption

Although the rate of drug absorption does not, in general, influence the average steady-state concentration of the drug in plasma, it may still influence drug therapy. If a drug is absorbed rapidly (e.g., a dose given as an intravenous bolus) and has a small 'central' volume, the concentration of drug initially will be high. It will then fall as the drug is distributed to its 'final' (larger) volume (seeFigure 14B). If the same drug is absorbed more slowly (e.g., by slow infusion), it will be distributed while it is being given, and peak concentrations will be lower and will occur later. Controlled-release preparations are designed to provide a slow and sustained rate of absorption in order to produce a less fluctuating plasma concentrationtime profile during the dosage interval compared to more immediate-release formulations. A given drug may act to produce both desirable and undesirable effects at several sites in the body, and the rates of distribution of drug to these sites may not be the same. The relative intensities of these different effects of a drug may thus vary transiently when its rate of administration is changed.

Nonlinear Pharmacokinetics

Nonlinearity in pharmacokinetics (i.e., changes in such parameters as clearance, volume of distribution, and half-life as a function of dose or concentration of drug) usually is due to saturation of protein binding, hepatic metabolism, or active renal transport of the drug.

Saturable Protein Binding

As the molar concentration of drug increases, the unbound fraction eventually also must increase (as all binding sites become saturated). This usually occurs only when drug concentrations in plasma are in the range of tens to hundreds of micrograms per milliliter. For a drug that is metabolized by the liver with a low intrinsic clearance/extraction ratio, saturation of plasma-protein binding will cause both V and clearance to increase as drug concentrations increase; half-life may thus remain constant [seeEquation (112)]. For such a drug, C_ss will not increase linearly as the rate of drug administration is increased. For drugs that are cleared with high intrinsic clearances/extraction ratios, C_ss can remain linearly proportional to the rate of drug administration. In this case, hepatic clearance would not change, and the increase in V would increase the half-time of disappearance by reducing the fraction of the total drug in the body that is delivered to the liver per unit of time. Most drugs fall between these two extremes, and the effects of nonlinear protein binding may be difficult to predict.

Saturable Elimination

In this situation, the MichaelisMenten equation [Equation (12)] usually describes the nonlinearity. All active processes are undoubtedly saturable, but they will appear to be linear if values of drug concentrations encountered in practice are much less than K_m. When they exceed K_m, nonlinear kinetics is observed. The major consequences of saturation of metabolism or transport are the opposite of those for saturation of protein binding. When both conditions are present simultaneously, they may virtually cancel each others' effects, and surprisingly linear kinetics may result; this occurs over a certain range of concentrations for salicylic acid.

Saturable metabolism causes oral first-pass metabolism to be less than expected (higher F), and there is a greater fractional increase in C_ss than the corresponding fractional increase in the rate of drug administration. The latter can be seen most easily by substituting Equation (12) into Equation (11) and solving for the steady-state concentration:

As the dosing rate approaches the maximal elimination rate (_m), the denominator of Equation (115) approaches zero and C_ss increases disproportionately. Because saturation of metabolism should have no effect on the volume of distribution, clearance and the relative rate of drug elimination decrease as the concentration increases; therefore, the log plasma level-time curve is concave-decreasing until metabolism becomes sufficiently desaturated and first-order elimination is present. Thus, the concept of a constant half-life is not applicable to nonlinear metabolism occurring in the usual range of clinical concentrations. Consequently, changing the dosing rate for a drug with nonlinear metabolism is difficult and unpredictable, since the resulting steady state is reached more slowly, and, importantly, the effect is disproportionate to the alteration in the dosing rate.

Phenytoin provides an example of a drug for which metabolism becomes saturated in the therapeutic range of concentrations (seeAppendix II). K_m (5 to 10 mg per liter) is typically near the lower end of the therapeutic range (10 to 20 mg per liter). For some individuals, especially children, K_m may be as low as 1 mg per liter. If, for such an individual, the target concentration is 15 mg per liter, and this is attained at a dosing rate of 300 mg per day, then, from Equation (115), _m equals 320 mg per day. For such a patient, a dose 10% less than optimal (i.e., 270 mg per day) will produce a C_ss of 5 mg per liter, well below the desired value. In contrast, a dose 10% greater than optimal (330 mg per day) will exceed metabolic capacity (by 10 mg per day) and cause a long and slow but unending climb in concentration until toxicity occurs. Dosage cannot be controlled so precisely (less than 10% error). Therefore, for those patients in whom the target concentration for phenytoin is more than tenfold greater than the K_m, alternating inefficacious therapy and toxicity is almost unavoidable. For a drug like phenytoin that has a narrow therapeutic index and exhibits nonlinear metabolism, therapeutic drug monitoring (seeTherapeutic Drug Monitoring) is most important.

Design and Optimization of Dosage Regimens

Following administration of a dose of drug, its effects usually show a characteristic temporal pattern (Figure 16). Onset of the effect is preceded by a lag period, after which the magnitude of the effect increases to a maximum and then declines; if a further dose is not administered, the effect eventually disappears. This time-course reflects changes in the drug's concentration as determined by the pharmacokinetics of its absorption, distribution, and elimination. Accordingly, the intensity of a drug's effect is related to its concentration above a minimum effective concentration, whereas the duration of this effect is a reflection of the length of time the drug level is above this value. These considerations, in general, apply to both desired and undesired (adverse) drug effects and, as a result, a therapeutic window exists reflecting a concentration range that provides efficacy without unacceptable toxicity. Similar considerations apply after multiple dosing associated with long-term therapy; therefore they determine the amount and frequency of drug administration to achieve an optimal therapeutic effect. In general, the lower limit of the therapeutic range appears to be approximately equal to the drug concentration that produces about half of the greatest possible therapeutic effect, and the upper limit of the therapeutic range is such that no more than 5% to 10% of patients will experience a toxic effect. For some drugs, this may mean that the upper limit of the range is no more than twice the lower limit. Of course, these figures can be highly variable, and some patients may benefit greatly from drug concentrations that exceed the therapeutic range, while others may suffer significant toxicity at much lower values.

For a limited number of drugs, some effect of the drug is easily measured (e.g., blood pressure, blood glucose), and this can be used to optimize dosage, using a trial-and-error approach. Even in this ideal case, certain quantitative issues arise, such as how often to change dosage and by how much. These usually can be settled with simple rules of thumb based on the principles discussed (e.g., change dosage by no more than 50% and no more often than every three to four half-lives). Alternatively, some drugs have very little dose-related toxicity, and maximum efficacy is usually desired. For these drugs, doses well in excess of the average required will both ensure efficacy (if this is possible) and prolong drug action. Such a 'maximal dose' strategy typically is used for penicillins and most -adrenergic receptor antagonists.

For many drugs, however, the effects are difficult to measure (or the drug is given for prophylaxis), toxicity and lack of efficacy are both potential dangers, and/or the therapeutic index is narrow. In these circumstances, doses must be titrated carefully, and drug dosage is limited by toxicity rather than efficacy. Thus, the therapeutic goal is to maintain steady-state drug levels within the therapeutic window. For most drugs, the actual concentrations associated with this desired range are not and need not be known. It is sufficient to understand that efficacy and toxicity are generally concentration-dependent, and how drug dosage and frequency of administration affect the drug level. However, for a small number of drugs, where there is a small (two- to threefold) difference between concentrations resulting in efficacy and toxicity (e.g., digoxin, theophylline, lidocaine, aminoglycosides, cyclosporine, and anticonvulsants), a plasma-concentration range associated with effective therapy has been defined. In this case, a target level strategy is reasonable, wherein a desired (target) steady-state concentration of the drug (usually in plasma) associated with efficacy and minimal toxicity is chosen, and a dosage is computed that is expected to achieve this value. Drug concentrations are subsequently measured, and dosage is adjusted if necessary to approximate the target more closely (see alsoChapter 3: Principles of Therapeutics).

Maintenance Dose

In most clinical situations, drugs are administered in a series of repetitive doses or as a continuous infusion to maintain a steady-state concentration of drug associated with the therapeutic window. Thus, calculation of the appropriate maintenance dosage is a primary goal. To maintain the chosen steady-state or target concentration, the rate of drug administration is adjusted such that the rate of input equals the rate of loss. This relationship was defined previously in Equations (11) and (114) and is expressed here in terms of the desired target concentration:

Dosing rate = target C_p CL/F    (116)

If the clinician chooses the desired concentration of drug in plasma and knows the clearance and bioavailability for that drug in a particular patient, the appropriate dose and dosing interval can be calculated.

Example

Oral digoxin is to be used as a maintenance dose to gradually 'digitalize' a 69-kg patient with congestive heart failure. A steady-state plasma concentration of 1.5 ng/ml is selected as an appropriate target. Based on the fact that the patient's creatinine clearance (CL_CR) is 100 ml/min, digoxin's clearance may be estimated from data in Appendix II.

Equation (116) is then used to calculate an appropriate dosing rate knowing that the oral bioavailability of digoxin is 70% (F= 0.7)

In practice, the dose rate would be rounded to the closest dosage size, either 0.375 mg/24 hr, which would result in a steady- state plasma concentration of 1.65 ng/ml (1.5 x 375/340), or 0.25 mg/24 hr, which would provide a value of 1.10 ng/ml (1.5 x

Dosing Interval for Intermittent Dosage

In general, marked fluctuations in drug concentrations between doses are not desirable. If absorption and distribution were instantaneous, fluctuation of drug concentrations between doses would be governed entirely by the drug's elimination half-life. If the dosing interval (T) was chosen to be equal to the half-life, then the total fluctuation would be twofold; this is often a tolerable variation.

Pharmacodynamic considerations modify this. If a drug is relatively nontoxic, such that a concentration many times that necessary for therapy can be tolerated easily, the maximal dose strategy can be used, and the dosing interval can be much longer than the elimination half-life (for convenience). The half-life of amoxicillin is about 2 hours, but it is often given in large doses every 8 or 12 hours.

For some drugs with a narrow therapeutic range, it may be important to estimate the maximal and minimal concentrations that will occur for a particular dosing interval. The minimal steady-state concentration C_{ss, min} may be reasonably determined by the use of Equation (117):

where k equals 0.693 divided by the clinically relevant plasma half-life and T is the dosing interval. The term exp(kT) is, in fact, the fraction of the last dose (corrected for bioavailability) that remains in the body at the end of a dosing interval.

For drugs that follow multiexponential kinetics and that are administered orally, the estimation of the maximal steady-state concentration C_{ss, max} involves a complicated set of exponential constants for distribution and absorption. If these terms are ignored for multiple oral dosing, one may easily predict a maximal steady-state concentration by omitting the exp(kT) term in the numerator of Equation (117) [see Equation (118), below]. Because of the approximation, the predicted maximal concentration from Equation (118) will be greater than that actually observed.

Example

In the patient with congestive heart failure discussed above, an oral maintenance dosing of 0.375 mg/24 hr of digoxin was calculated to achieve an average plasma concentration of 1.65 ng/ml during the dosage interval. Digoxin has a narrow therapeutic index, and plasma levels between 0.8 and 2.0 ng/ml are usually associated with efficacy and minimal toxicity. It is, therefore, important to know what maximum and minimum plasma concentrations would be predicted with the above regimen. This first requires estimation of digoxin's volume of distribution based on data in Appendix II.

Combining this value with that of digoxin's clearance provides an estimate of digoxin's elimination half-life in the patient [Equations (11) through (112)].

Accordingly, the fractional rate constant of elimination is equal to 0.01136 hr¹ (0.693/61 hr). Maximum and minimum digoxin plasma concentrations may then be predicted depending upon the dosage interval. If this were every 2 days:

Accordingly, the plasma concentrations would fluctuate about twofold, consistent with the similarity of the dosage interval to digoxin's half-life. Also, the peak concentration would be above the upper value of the therapeutic range, exposing the patient to possible adverse effects, and at the end of the dosing interval the concentration would be above but close to the lower limit. By using the same dosing rate but decreasing the frequency of dosing, a much smoother plasma concentration versus time profile would be obtained while still maintaining an average steady-state value of 1.65 ng/ml. For example, with a dose every 24 hours, the predicted maximum and minimum plasma concentrations would be 1.90 and 1.44 ng/ml, respectively, which are in the upper portion of the therapeutic window. By contrast, administering a more conservative dosing rate of 0.25 every 24 hours would produce peak and trough values of 1.26 and 0.96 ng/ml, respectively, that would be associated with a steady-state value of 1.10 ng/ml. Of course the clinician must balance the problem of compliance with regimens that involve frequent dosage against the problem of periods when the patient may be subjected to concentrations of the drug that could be too high or too low.

Loading Dose

The 'loading dose' is one or a series of doses that may be given at the onset of therapy with the aim of achieving the target concentration rapidly. The appropriate magnitude for the loading dose is:

Loading dose = target C_p V_SS/F   (120)

A loading dose may be desirable if the time required to attain steady state by the administration of drug at a constant rate (four elimination half-lives) is long relative to the temporal demands of the condition being treated. For example, the half-life of lidocaine is usually 1 to 2 hours. Arrhythmias encountered after myocardial infarction obviously may be life threatening, and one cannot wait 4 to 8 hours to achieve a therapeutic concentration of lidocaine by infusion of the drug at the rate required to attain this concentration. Hence, use of a loading dose of lidocaine in the coronary care unit is standard.

The use of a loading dose also has significant disadvantages. First, the particularly sensitive individual may be exposed abruptly to a toxic concentration of a drug. Moreover, if the drug involved has a long half-life, it will take a long time for the concentration to fall if the level achieved was excessive. Loading doses tend to be large, and they are often given parenterally and rapidly; this can be particularly dangerous if toxic effects occur as a result of actions of the drug at sites that are in rapid equilibrium with plasma. This occurs because the loading dose calculated on the basis of V_ss subsequent to drug distribution is initially constrained within the initial and smaller 'central' volume of distribution. It is, therefore, usually advisable to divide the loading dose into a number of smaller fractional doses that are administered over a period of time. Alternatively, the loading dose should be administered as a continuous intravenous infusion over a period of time. Ideally this should be given in an exponentially decreasing fashion to mirror the concomitant accumulation of the maintenance dose of the drug, and this is now technically feasible using computerized infusion pumps.

Example

'Digitalization,' in the patient described above, is gradual if only a maintenance dose is administered (for at least 10 days based on a half-life of 61 hours). A more rapid response could be obtained (if deemed necessary by the physician; seeChapter 34: Pharmacological Treatment of Heart Failure) by using a loading dose strategy and Equation (120):

To avoid toxicity, this oral loading dose, which also could be intravenously administered, would be given as an initial 0.5-mg dose followed by a 0.25-mg dose at 6 to 8 hours later along with careful monitoring of the patient. It also would be prudent to give the final 0.25-mg fractional dose, if necessary, in two 0.125-mg divided doses separated by 6 to 8 hours to avoid overdigitalization, particularly if there were a plan to initiate an oral maintenance dose within 24 hours of beginning digoxin therapy.

Individualizing Dosage

A rational dosage regimen is based on knowledge of F, CL, V_ss, and t_1/2, and some information about rates of absorption and distribution of the drug. Recommended dosage regimens generally are designed for an 'average' patient; usual values for the important determining parameters and appropriate adjustments that may be necessitated by disease or other factors are presented in Appendix II. This 'one size fits all' approach, however, overlooks the considerable and unpredictable interpatient variability that usually is present in these pharmacokinetic parameters. For many drugs, one standard deviation in the values observed for F, CL, and V_ss is about 20%, 50%, and 30%, respectively. This means that 95% of the time the C_ss that is achieved will be between 35% and 270% of the target; this is an unacceptably wide range for a drug with a low therapeutic index. Individualization of the dosage regimen to a particular patient is, therefore, critical for optimal therapy. The pharmacokinetic principles, described above, provide a basis for modifying the dosage regimen to obtain a desired degree of efficacy with a minimum of unacceptable adverse effects. In situations where the drug's plasma concentration can be measured and related to the therapeutic window, additional guidance for dosage modification is obtained. Such measurement and adjustment are appropriate for many drugs with low therapeutic indices (e.g., cardiac glycosides, antiarrhythmic agents, anticonvulsants, theophylline, and others).

Therapeutic Drug Monitoring

The major use of measured concentrations of drugs (at steady state) is to refine the estimate of CL/F for the patient being treated [using Equation (114) as rearranged below]:

CL/F(patient) = dosing rate/C_SS(measured)   (121)

The new estimate of CL/F can be used in Equation (116) to adjust the maintenance dose to achieve the desired target concentration.

Certain practical details and pitfalls associated with therapeutic drug monitoring should be kept in mind. The first of these relates to the time of sampling for measurement of the drug concentration. If intermittent dosing is used, when during a dosing interval should samples be taken? It is necessary to distinguish between two possible uses of measured drug concentrations to understand the possible answers. A concentration of drug measured in a sample taken at virtually any time during the dosing interval will provide information that may aid in the assessment of drug toxicity. This is one type of therapeutic drug monitoring. It should be stressed, however, that such use of a measured concentration of drug is fraught with difficulties because of interindividual variability in sensitivity to the drug. When there is a question of toxicity, the drug concentration can be no more than just one of many items that serve to interpret the clinical situation.

Changes in the effects of drugs may be delayed relative to changes in plasma concentration because of a slow rate of distribution or pharmacodynamic factors. Concentrations of digoxin, for example, regularly exceed 2 ng/ml (a potentially toxic value) shortly after an oral dose, yet these peak concentrations do not cause toxicity; indeed, they occur well before peak effects. Thus, concentrations of drugs in samples obtained shortly after administration can be uninformative or even misleading.

When concentrations of drugs are used for purposes of adjusting dosage regimens, samples obtained shortly after administration of a dose are almost invariably misleading. The purpose of sampling during supposed steady state is to modify the estimate of CL/F and thus the choice of dosage. Early postabsorptive concentrations do not reflect clearance; they are determined primarily by the rate of absorption, the 'central' (rather than the steady-state) volume of distribution, and the rate of distribution, all of which are pharmacokinetic features of virtually no relevance in choosing the long-term maintenance dosage. When the goal of measurement is adjustment of dosage, the sample should be taken well after the previous doseas a rule of thumb just before the next planned dose, when the concentration is at its minimum. There is an exception to this approach: some drugs are nearly completely eliminated between doses and act only during the initial portion of each dosing interval. If it is questionable whether or not efficacious concentrations of such drugs are being achieved, a sample taken shortly after a dose may be helpful. On the other hand, if a concern is whether or not low clearance (as in renal failure) may cause accumulation of drug, concentrations measured just before the next dose will reveal such accumulation and are considerably more useful for this purpose than is knowledge of the maximal concentration. For such drugs, determination of both maximal and minimal concentrations is thus recommended.

A second important aspect of the timing of sampling is its relationship to the beginning of the maintenance dosage regimen. When constant dosage is given, steady state is reached only after four half-lives have passed. If a sample is obtained too soon after dosage is begun, it will not accurately reflect this state and the drug's clearance. Yet, for toxic drugs, if sampling is delayed until steady state is ensured, the damage may have been done. Some simple guidelines can be offered. When it is important to maintain careful control of concentrations, the first sample should be taken after two half-lives (as calculated and expected for the patient), assuming no loading dose has been given. If the concentration already exceeds 90% of the eventual expected mean steady-state concentration, the dosage rate should be halved, another sample obtained in another two (supposed) half-lives, and the dosage halved again if this sample exceeds the target. If the first concentration is not too high, the initial rate of dosage is continued; even if the concentration is lower than expected, it is usually reasonable to await the attainment of steady state in another two estimated half-lives and then proceed to adjust dosage as described above.

If dosage is intermittent, there is a third concern with the time at which samples are obtained for determination of drug concentrations. If the sample has been obtained just prior to the next dose, as recommended, concentration will be a minimal value, not the mean. However, as discussed above, the estimated mean concentration may be calculated by using Equation (114).

If a drug follows first-order kinetics, the average, minimum, and maximum concentrations at steady state are linearly related to dose and dosing rate [seeEquations (114), (117), and (118)]. Therefore, the ratio between the measured and the desired concentrations can be used to adjust the dose, consistent with available dosage sizes:

In the previously described patient given 0.375 mg digoxin every 24 hours, for example, if the measured steady-state concentration was found to be 1.65 ng/ml rather than a desired level of 1.3 ng/ml, an appropriate, practical change in the dosage regimen would be to reduce the daily dose to 0.25 mg digoxin.

Compliance

Ultimately, therapeutic success is dependent on the patient's actually taking the drug according to the prescribed dosage regimen'Drugs don't work if you don't take them.' Noncompliance with the dosing schedule is a major and often unappreciated reason for therapeutic failure, especially in the long-term treatment of disease using antihypertensive, antiretroviral, and anticonvulsant agents. When no special efforts are made to address this issue, only about 50% of patients follow the prescribed dosage regimen in a reasonably satisfactory fashion; approximately one-third only partly comply; and about 1 in 6 patients is essentially noncompliant. Missed doses are more common than too many doses. The number of drugs does not appear to be as important as the number of times a day doses must be remembered (Farmer, 1999). Reducing the number of required dosing occasions will improve adherence to a prescribed dosage regimen. Equally important is the need to involve patients in the responsibility for their own health, using a variety of strategies based on improved communication regarding the nature of the disease and the overall therapeutic plan.

Acknowledgment

The author wishes to acknowledge Drs. Leslie Z. Benet, Deanna L. Kroetz, and Lewis B. Sheiner, authors of this chapter in the ninth edition of Goodman and Gilman's The Pharmacological Basis of Therapeutics, some of whose text has been retained in this edition.

The effects of most drugs result from their interaction with macromolecular components of the organism. These interactions alter the function of the pertinent component and thereby initiate the biochemical and physiological changes that are characteristic of the response to the drug. The term receptor denotes the component of the organism with which the chemical agent was presumed to interact. The statement that the receptor for a drug can be any functional macromolecular component of the organism has several fundamental corollaries. One is that a drug potentially is capable of altering the rate at which any bodily function proceeds. Another is that drugs do not create effects, but instead modulate intrinsic physiological functions.

Drug Receptors

At least from a numerical standpoint, proteins form the most important class of drug receptors. Examples are the receptors for hormones, growth factors, and neurotransmitters, the enzymes of crucial metabolic or regulatory pathways (e.g., dihydrofolate reductase, acetylcholines-terase), proteins involved in transport processes (e.g., Na⁺,K⁺-ATPase), or structural proteins (e.g., tubulin). Specific binding properties of other cellular constituents also can be exploited. Thus, nucleic acids are important drug receptors, particularly for cancer chemotherapeutic agents.

A particularly important group of drug receptors are proteins that normally serve as receptors for endogenous regulatory ligands (e.g., hormones, neurotransmitters). Many drugs act on such physiological receptors and are often particularly selective, because physiological receptors are specialized to recognize and respond to individual signaling molecules with great selectivity. Drugs that bind to physiological receptors and mimic the regulatory effects of the endogenous signaling compounds are termed agonists. Other drugs bind to receptors without regulatory effect, but their binding blocks the binding of the endogenous agonist. Such compounds, which may still produce useful effects by inhibiting the action of an agonist (e.g., by competition for agonist binding sites), are termed antagonists. Agents that are only partly as effective as agonists are termed partial agonists, and those that stabilize the receptor in its inactive conformation are termed inverse agonists. (See 'Quantitation of DrugReceptor Interactions and Elicited Effect.')

The binding of drugs to receptors can involve all known types of interactionsionic, hydrogen bonding, hydrophobic, van der Waals, and covalent. In most interactions between drugs and receptors, it is likely that bonds of multiple types are important. If binding is covalent, the duration of drug action is frequently, but not necessarily, prolonged. Noncovalent interactions of high affinity also may appear to be essentially irreversible.

StructureActivity Relationship and Drug Design

Both the affinity of a drug for its receptor and its intrinsic activity are determined by its chemical structure. This relationship is frequently quite stringent. Relatively minor modifications in the drug molecule may result in major changes in pharmacological properties.

Exploitation of structureactivity relationships has on many occasions led to the synthesis of valuable therapeutic agents. Because changes in molecular configuration need not alter all actions and effects of a drug equally, it is sometimes possible to develop a congener with a more favorable ratio of therapeutic to toxic effects, enhanced selectivity among different cells or tissues, or more acceptable secondary characteristics than those of the parent drug. Therapeutically useful antagonists of hormones or neurotransmitters have been developed by chemical modification of the structure of the physiological agonist. Minor modifications of structure also can have profound effects on the pharmacokinetic properties of drugs.

Given adequate information about both the molecular structures and the pharmacological activities of a relatively large group of congeners, it should be possible to identify those chemical properties that are required for optimal action at the receptorsize, shape, the position and orientation of charged groups or hydrogen bond donors, and so on. Advances in computational chemistry, structural analysis of organic compounds, and the biochemical measurement of the primary actions of drugs at their receptors have enriched the quantitation of structureactivity relationships and its use in drug design (Kuntz, 1992; Schreiber, 1992). The importance of specific drugreceptor interactions can be evaluated further by analyzing the responsiveness of receptors that have been selectively mutated at individual amino acid residues. Such information is increasingly allowing the optimization or design of chemicals that can bind to a receptor with improved affinity, selectivity, or regulatory effect. Similar structure-based approaches also are used to improve pharmacokinetic properties of drugs (see Chapter 1: Pharmacokinetics: The Dynamics of Drug Absorption, Distribution, and Elimination). Knowledge of the structures of receptors and of drug-receptor complexes, determined at atomic resolution by X-ray crystallography or nuclear magnetic resonance (NMR) spectroscopy, is even more helpful in the design of ligands.

Ironically, advances in molecular biology that contribute to structure-motivated drug design also have spawned powerful but entirely random searches for new drugs. In this approach, huge libraries of randomly synthesized chemicals are generated either by synthetic chemistry or by genetically engineered microbes. A library then is screened for pharmacologically active agents using mammalian cells or microorganisms that have been engineered to express the receptor of therapeutic interest and the associated biochemical machinery necessary for detection of the receptor's response. Active compounds initially discovered by such random screens then can be modified and improved using knowledge of their structurefunction relationships.

Cellular Sites of Drug Action

Because drugs act by altering the activities of their receptors, the sites at which a drug acts and the extent of its action are determined by the location and functional capacity of its receptors. Selective localization of drug action within the organism is therefore not necessarily dependent upon selective distribution of the drug. If a drug acts on a receptor that serves functions common to most cells, its effects will be widespread. If the function is a vital one, the drug may be particularly difficult or dangerous to use. Nevertheless, such a drug may be clinically important. Digitalis glycosides, important in the treatment of heart failure, are potent inhibitors of an ion transport process that is vital to most cells. As such, they can cause widespread toxicity, and their margin of safety is dangerously low. Other examples could be cited, particularly in the area of cancer chemotherapy. If a drug interacts with receptors that are unique to only a few types of differentiated cells, its effects are more specific. Hypothetically, the ideal drug would cause its therapeutic effect by such an action. Side effects would be minimized, but toxicity might not be. If the differentiated function were a vital one, this type of drug also could be very dangerous. Some of the most lethal chemical agents known (e.g., botulinum toxin) show such specificity. Note also that, even if the primary action of a drug is localized, the consequent physiological effects of the drug may be widespread.

Receptors for Physiological Regulatory Molecules

The term receptor has been used operationally to denote any cellular macromolecule to which a drug binds to initiate its effects. Among the most important drug receptors are cellular proteins, whose normal function is to act as receptors for endogenous regulatory ligandsparticularly hormones, growth factors, and neurotransmitters. The function of such physiological receptors consists of binding the appropriate ligand and, in response, propagating its regulatory signal in the target cell.

Identification of the two functions of a receptor, ligand binding and message propagation, correctly suggests the existence of functional domains within the receptor: a ligand-binding domain and an effector domain. The structure and function of these domains often can be deduced from high-resolution structures of receptor proteins and/or by analysis of the behavior of intentionally mutated receptors. Increasingly, the mechanism of intramolecular coupling of ligand binding with functional activation also can be learned. The biological importance of these functional domains is further indicated by the evolution both of different receptors for diverse ligands that act by similar biochemical mechanisms and of multiple receptors for a single ligand that act by unrelated mechanisms.

The regulatory actions of a receptor may be exerted directly on its cellular target(s), effector protein(s), or may be conveyed by intermediary cellular signaling molecules called transducers. The receptor, its cellular target, and any intermediary molecules are referred to as a receptor-effector system or signal transduction pathway. Frequently, the proximal cellular effector protein is not the ultimate physiological target, but rather is an enzyme or transport protein that creates, moves, or degrades a small molecule metabolite or ion known as a second messenger. Second messengers can diffuse through a cell and convey information to a wide variety of targets, which can respond simultaneously to the output of a single receptor.

Receptors and their associated effector and transducer proteins also act as integrators of information as they coordinate signals from multiple ligands with each other and with the metabolic activities of the cell (see below). This integrative function is particularly evident when one considers that the different receptors for scores of chemically unrelated ligands utilize relatively few biochemical mechanisms to exert their regulatory functions, and that even these few pathways may share common signaling molecules.

An important property of physiological receptors, which also makes them excellent targets for drugs, is that they act catalytically and hence are biochemical signal amplifiers. The catalytic nature of receptors is obvious when the receptor itself is an enzyme, but all known physiological receptors are formally catalysts. For example, when a single agonist molecule binds to a receptor that is an ion channel, many ions flow through the channel. Similarly, a single steroid hormone molecule binds to its receptor and initiates the transcription of many copies of specific mRNAs, which in turn can give rise to multiple copies of a single protein.

A Framework for Considering Agonist Activity

If two drugs bind to the same receptor at the same site, why can one be an agonist and the other an antagonist? This question, more broadly stated, is also central to understanding the dynamics of protein structure and proteinligand interactions. At a detailed level, the atomic interactions that allow a bound ligand to alter the structure of the protein to which it binds are increasingly well understood. High-resolution structures of liganded and unliganded proteins combined with increasingly accurate theoretical structural modeling and functional studies of mutant proteins are clarifying how we can think about induced conformational changes.

From a functional perspective, however, the question of agonist action can be restated as: How does a receptor use the free energy of ligand binding to shift its conformation to the active state? A receptor, by definition, exists in at least two conformational states, active (a) and inactive (i).

These conformations might correspond to the open and closed states of an ion channel, the active and inactive states of a protein tyrosine kinase, or the productive versus nonproductive conformations of a receptor for coupling to G proteins. If these states are in equilibrium and the inactive state predominates in the absence of drug, then the basal signal output will be low. The extent to which the equilibrium is shifted toward the active state is determined by the relative affinity of the drug for the two conformations (Figure 21). A drug that has a higher affinity for the active conformation than for the inactive conformation will drive the equilibrium to the active state and thereby activate the receptor. Such a drug will be an agonist. A full agonist is sufficiently selective for the active conformation that, at a saturating concentration, it will drive the receptor essentially completely to the active state (see Figure 21). If a different but perhaps structurally similar compound binds to the same site on R but with only moderately greater affinity for R_a than for R_i, its effect will be less, even at saturating concentrations. A drug that displays such intermediate effectiveness is referred to as a partial agonist. Note that, in an absolute sense, all agonists are partial; selectivity for R_a over R_i cannot be total. A drug that binds with equal affinity to either conformation will not alter the activation equilibrium and will act as a competitive antagonist. Last, a drug with preferential affinity for R_i will actually produce an effect opposite to that of an agonist; examples of such inverse agonists do exist (see Chapters 11: 5-Hydroxytryptamine (Serotonin): Receptor Agonists and Antagonists and 17: Hypnotics and Sedatives). If the preexisting equilibrium for unliganded receptors lies far in the direction of R_i, negative antagonism may be difficult to observe and to distinguish from simple competitive antagonism.

Careful biochemical studies of receptordrug interactions, coupled with the analysis of receptors in which the intrinsic R_a/R_i equilibrium has been shifted by mutation, have supported this general model of drug action. The model is readily applicable to experimental data through the use of appropriate computer-assisted analysis and is frequently used as a guide to understanding drug action.

Physiological Receptors: Structural and Functional Families

The last two decades have witnessed both an explosion in our appreciation of the number of physiological receptors and, in parallel, the development of our understanding of the fundamental structural motifs and biochemical mechanisms that characterize them. Molecular cloning has identified both completely novel receptors (and their regulatory ligands) and numerous isoforms of previously known receptors. There now exist data banks devoted exclusively to structures of single classes of receptors. Members of various classes of receptors, transducers, and effector proteins have been purified, and their mechanisms of action are understood in considerable biochemical detail. Receptors, transducers, and effectors can be expressed via molecular genetic strategies and studied in cultured cells. Alternatively, they can be expressed in large amounts in cells of convenience (bacteria, yeast, etc.) to facilitate their purification.

Receptors for physiological regulatory molecules can be assigned to a relatively few functional families whose members share both common mechanisms of action and similar molecular structures (Figure 22). For each receptor family, there is now at least a rudimentary understanding of the structures of ligand-binding domains and effector domains and of how agonist binding influences the regulatory activity of the receptor. The relatively small number of biochemical mechanisms and structural formats used for cellular signaling is fundamental to the ways in which target cells integrate signals from multiple receptors to produce additive, synergistic, or mutually inhibitory responses. Figure 21 provides a schematic diagram of various receptor families and their transducer and effector molecules.

Receptors as Enzymes: Receptor Protein Kinases

The largest group of receptors with intrinsic enzymatic activity are cell surface protein kinases, which exert their regulatory effects by phosphorylating diverse effector proteins at the inner face of the plasma membrane. Phosphorylation can alter the biochemical activities of an effector or its interactions with other proteins. Most receptor protein kinases target tyrosine residues in their substrates; these include receptors for insulin, many cytokines and diverse peptides, and proteins that direct growth or differentiation. A few receptor protein kinases also phosphorylate serine or threonine residues. The structurally most simple receptor protein kinases are composed of an agonist-binding domain on the extracellular surface of the plasma membrane, a single membrane-spanning element, and a protein kinase domain on the inner membrane face. Many variations on this basic plan exist, including obligate oligomerization and the addition of multiple regulatory or protein-binding domains to the intracellular protein kinase domain.

Another family of receptors that are functionally protein kinases contains a modification of the structure described above. Protein kinaseassociated receptors lack the intracellular enzymatic domains but, in response to agonists, bind and/or activate distinct protein kinases on the cytoplasmic face of the plasma membrane. Receptors of this group include several receptors for neurotrophic peptides and the multisubunit antigen receptors on T and B lymphocytes. The antigen receptors also involve tyrosine protein phosphatases in their cellular regulatory activity, and it is plausible that other receptors that apparently lack cytoplasmic effector domains may recruit still other effector proteins.

Receptors with Other Enzymatic Activity

The domain structure just described for cell surface protein kinases is varied in other receptors to utilize other signaling outputs. A family of protein tyrosine phosphatases has extracellular domains with a sequence reminiscent of cellular adhesion molecules. Although the extracellular ligands for many of these phosphatases are not yet known, the importance of their enzymatic activity has been demonstrated through genetic and biochemical experiments in multiple cell types. In the receptors for atrial natriuretic peptides and the peptide guanylin, the intracellular domain is not a protein kinase but a guanylyl cyclase, which synthesizes the second messenger, cyclic GMP. Receptors with guanylyl cyclase activity also serve as pheromone receptors in invertebrates. There may be other variations on this transmembrane topology.

Ion Channels

Receptors for several neurotransmitters form agonist-regulated, ion-selective channels in the plasma membrane, termed ligand-gated ion channels, which convey their signals by altering the cell's membrane potential or ionic composition. This group includes the nicotinic cholinergic receptor; the GABA_A receptor for gamma-aminobutyric acid; and receptors for glutamate, aspartate, and glycine (see Chapters 9: Agents Acting at the Neuromuscular Junction and Autonomic Ganglia, 12: Neurotransmission and the Central Nervous System, and 17: Hypnotics and Sedatives). They are all multisubunit proteins, with each subunit predicted to span the plasma membrane several times. Symmetrical association of the subunits allows each to form a segment of the channel wall and to cooperatively control channel opening and closing.

G ProteinCoupled Receptors

A large family of receptors use distinct heterotrimeric GTP-binding regulatory proteins, known as G proteins, as transducers to convey signals to their effector proteins. G proteincoupled receptors include those for many biogenic amines, eicosanoids, and other lipid signaling molecules as well as numerous peptide and protein ligands. G proteinregulated effectors include enzymes such as adenylyl cyclase and phospholipase C and plasma membrane ion channels selective for Ca²⁺ and K⁺ (Figure 22). Because of their number and physiological importance, G proteincoupled receptors are widely used targets for drugs; it has been estimated that about half of all nonantibiotic prescription drugs are directed toward these receptors.

G proteincoupled receptors span the plasma membrane as a bundle of seven alpha helices. Agonists bind to a cleft within the extracellular face of the bundle or to a globular ligand-binding domain sometimes found at the amino terminus. G proteins bind to the cytoplasmic face of the receptors. Receptors in this family respond to agonists by promoting the binding of GTP to the G protein. GTP activates the G protein and allows it in turn to activate the effector protein. The G protein remains active until it hydrolyzes bound GTP to GDP, which does not activate. G proteins are composed of a GTP-binding subunit, which confers specific recognition by receptor, and an associated dimer of and subunits. Activation of the G subunit by GTP allows it to both regulate an effector protein and drive the release of G subunits, which can regulate their own group of effectors.

A cell may express as many as twenty G proteincoupled receptors, each with distinctive specificity for one or several of its half-dozen G proteins. Each G can regulate one or more effectors. Receptors for multiple ligands can thus integrate their signals through a single G protein. A receptor also may generate multiple signals by activating more than one G protein species. Similarly, a G subunit can regulate the activities of more than one effector. Thus, the receptorG protein effector systems are complex networks of convergent and divergent interactions that permit extraordinarily versatile regulation of cell function (Ross, 1992).

Transcription Factors

Receptors for steroid hormones, thyroid hormone, vitamin D, and the retinoids are soluble DNA-binding proteins that regulate the transcription of specific genes (Mangelsdorf et al., 1994). They are part of a larger family of transcription factors whose members may be regulated by phosphorylation, association with other cellular proteins, or by binding to metabolites or cellular regulatory ligands. These receptors act as dimerssome as homodimers and some as heterodimerswith homologous cellular proteins, but they may be regulated by higher-order oligomerization with other regulating molecules. They provide striking examples of conservation of structure and mechanism, in part because they are assembled as three largely independent domains. The region nearest the carboxyl terminus binds hormone and serves a negative regulatory role; that is, removal of this domain leaves a constitutively active fragment that may be nearly as effective in regulating transcription as is the intact hormone-liganded receptor. Hormone binding presumably also relieves this inhibitory constraint. The central region of the receptor mediates binding to specific sites on nuclear DNA to activate or inhibit transcription of the nearby gene. These regulatory sites in DNA are likewise receptor-specific: the sequence of a 'glucocorticoid-responsive element,' with only slight variation, is associated with each glucocorticoid-responsive gene. The function of the amino-terminal region of the receptor is less well defined, but its loss decreases the receptor's regulatory activity. The activity of each domain is largely independent, a phenomenon best demonstrated by the construction of chimeric receptors that reflect the hormone binding or DNA regulatory activity characteristic of the 'parent' receptor contributing that domain.

Cytoplasmic Second Messengers

Cyclic AMP

Physiological signals also are integrated within the cell as a result of interactions between second-messenger pathways. Compared with the number of receptors and cytosolic signaling proteins, there are relatively few recognized cytoplasmic second messengers. However, their synthesis or release and degradation or excretion reflect the activities of many pathways. Well-studied second messengers include cyclic AMP and cyclic GMP, Ca²⁺, inositol phosphates, diacylglycerol, and nitric oxide; our list of this diverse group of molecules is still growing. Second messengers influence each other both directly, by altering the other's metabolism, and indirectly, by sharing intracellular targets. This superficially confusing pattern of regulatory pathways allows the cell to respond to agonists, singly or in combination, with an integrated array of cytoplasmic second messengers and responses. Cyclic AMP, the prototypical second messenger, remains a good example for understanding the regulation and function of most second messengers. It is synthesized by adenylyl cyclase under the control of many G proteincoupled receptors; stimulation is mediated by G_s and inhibition by G_i.

There are at least ten tissue-specific adenylyl cyclase isozymes, each with its unique pattern of regulatory responses. Several adenylyl cyclase isozymes are either stimulated or inhibited by G protein subunits, which allows G proteins other than G_s to modulate cyclase activity. Some isozymes are stimulated by Ca²⁺ or Ca²⁺calmodulin complexes. Each of the isozymes has its own pattern of enhancement or attenuation by phosphorylation and other regulatory influences, providing a broad array of regulatory features to individual target cells. Cyclic AMP is eliminated by a combination of hydrolysis, catalyzed by cyclic nucleotide phosphodiesterases, and extrusion by several plasma membrane transport proteins. Phosphodiesterases form yet another family of important signaling proteins whose activities are regulated by controlled transcription as well as by second messengers (cyclic nucleotides and Ca²⁺) and interactions with yet other signaling proteins.

In most cases, cyclic AMP functions by activating cyclic AMPdependent protein kinases, a relatively small group of closely related proteins. However, these protein kinases phosphorylate both final physiological targets (metabolic enzymes or transport proteins) and numerous protein kinases and other regulatory proteins. This latter group includes transcription factors that allow cyclic AMP to regulate gene expression in addition to more acute cellular events. In addition to activating a protein kinase, cyclic AMP also directly regulates the activity of plasma membrane cation channels, which are particularly important in olfactory neurons. Cyclic AMP signals are thus propagated throughout the biochemical behavior of the target cell.

Calcium

Intracellular Ca²⁺, another particularly well-studied second messenger, offers several interesting comparisons with cyclic AMP. The release of Ca²⁺ into the cytoplasm is mediated by diverse channels: plasma membrane channels regulated by G proteins, membrane potential, K⁺ or Ca²⁺ itself, and channels in specialized regions of endoplasmic reticulum that respond to the second messenger inositol trisphosphate (IP₃), or, in muscle, to cell depolarization. Ca²⁺ is removed both by extrusion and by reuptake into the endoplasmic reticulum. Ca²⁺ propagates its signals through a much wider range of proteins than does cyclic AMP, including metabolic enzymes, protein kinases, and Ca²⁺-binding regulatory proteins that regulate still other ultimate and intermediary effectors.

Regulation of Receptors

Receptors not only initiate regulation of physiological and biochemical function but also are themselves subject to many regulatory and homeostatic controls. These controls include regulation of the synthesis and degradation of the receptor by multiple mechanisms, covalent modification, association with other regulatory proteins, and/or relocalization within the cell. Transducer and effector proteins similarly are regulated. Modulating inputs may come from other receptors, directly or indirectly, and receptors are almost always subject to feedback regulation by their own signaling outputs.

Continued stimulation of cells with agonists generally results in a state of desensitization (also referred to as refractoriness or down-regulation), such that the effect that follows continued or subsequent exposure to the same concentration of drug is diminished (Figure 23). This phenomenon is very important in therapeutic situations; an example is attenuated response to the repeated use of -adrenergic agonists as bronchodilators for the treatment of asthma (see Chapter 10: Catecholamines, Sympathomimetic Drugs, and Adrenergic Receptor Antagonists).

Feedback inhibition of signaling may be limited to output only from the stimulated receptor, a situation known as homologous desensitization. Attenuation also may extend to the action of all receptors that share a common signaling pathway, heterologous desensitization (Figure 23). Homologous desensitization indicates feedback directed to the receptor molecule itself (phosphorylation, proteolysis, decreased synthesis, etc.), whereas heterologous desensitization may involve inhibition or loss of one or more downstream proteins that participate in signaling from other receptors. Mechanisms involved in homologous and heterologous desensitization of specific receptors and signaling pathways are discussed in greater detail in later chapters related to individual receptor families.

Predictably, supersensitivity to agonists also frequently follows chronic reduction of receptor stimulation. Such situations can result, for example, from the long-term administration of -adrenergic antagonists such as propranolol (see Chapter 10: Catecholamines, Sympathomimetic Drugs, and Adrenergic Receptor Antagonists).

Diseases Resulting from Receptor Malfunction

In addition to variability among individuals in their responses to drugs (see Chapter 3: Principles of Therapeutics), several definable diseases arise from disorders in receptors or receptoreffector systems. The loss of a receptor in a highly specialized signaling system may cause a relatively limited phenotypic disorder, such as the genetic deficiency of the androgen receptor in the testicular feminization syndrome (Griffin et al., 1995). Deficiencies of more widely used signaling systems have a broader spectrum of effects, as are seen in myasthenia gravis or some forms of insulin-resistant diabetes mellitus, which result from autoimmune depletion of nicotinic cholinergic receptors (see Chapter 9: Agents Acting at the Neuromuscular Junction and Autonomic Ganglia) or insulin receptors (see Chapter 61: Insulin, Oral Hypoglycemic Agents, and the Pharmacology of the Endocrine Pancreas), respectively. A lesion in a component of a signaling pathway that is used by many receptors can cause a generalized endocrinopathy. Heterozygous deficiency of G_s, the G protein that activates adenylyl cyclase in all cells, causes multiple endocrine disorders (Spiegel and Weinstein, 1995). Homozygous deficiency in G_s presumably would be lethal.

The expression of aberrant or ectopic receptors, effectors, or coupling proteins potentially can lead to supersensitivity, subsensitivity, or other untoward responses. Among the most interesting and significant events is the appearance of aberrant receptors as products of oncogenes, which transform otherwise normal cells into malignant cells. Virtually any type of signaling system may have oncogenic potential. The erbA oncogene product is an altered form of a receptor for thyroid hormone, constitutively active because of the loss of its ligand-binding domain (Mangelsdorf et al., 1994). The ros and erbB oncogene products are activated, uncontrolled forms of the receptors for insulin and epidermal growth factor, respectively, both known to enhance cellular proliferation (Yarden and Ulrich, 1988). The mas oncogene product (Young et al., 1986) is a G proteincoupled receptor, probably the receptor for a peptide hormone. Constitutive activation of G proteincoupled receptors due to subtle mutations in receptor structure has been shown to give rise to retinitis pigmentosa, precocious puberty, and malignant hyperthyroidism (reviewed in Clapham, 1993). G proteins can themselves be oncogenic when either overexpressed or constitutively activated by mutation (Lyons et al., 1990).

Mutation of receptors can alter either acute responsiveness to drug therapy or its continuing efficacy. For example, a mutation of -adrenergic receptors, which mediate airway smooth muscle relaxation and bronchial airflow, accelerates desensitization to -adrenergic agonists used to treat asthma (Turki et al., 1995; see Chapter 28: Drugs Used in the Treatment of Asthma). As the mutations that mediate these pathologies are discovered, they can be replicated using the cloned genes, so as to allow development of suitable drugs to target them specifically.

Classification of Receptors and Drug Effects

Traditionally, drug receptors have been identified and classified primarily on the basis of the effect and relative potency of selective agonists and antagonists. For example, the effects of acetylcholine that are mimicked by the alkaloid muscarine and that are selectively antagonized by atropine are termed muscarinic effects. Other effects of acetylcholine that are mimicked by nicotine are described as nicotinic effects. By extension, these two types of cholinergic effects are said to be mediated by muscarinic or nicotinic receptors. Although it frequently contributes little to delineation of the mechanism of drug action, such categorization provides a convenient basis for summarizing drug effects. A statement that a drug activates a specified type of receptor is a succinct summary of its spectrum of effects and of the agents that will regulate it. However, the accuracy of this statement may be altered when new receptors or receptor subtypes are identified or additional drug mechanisms or side effects are revealed.

Significance of Receptor Subtypes

As the diversity and selectivity of drugs increased, it became clear that multiple subtypes of receptors exist within many previously defined classes of receptors. Molecular cloning further accelerated discovery of novel receptor subtypes, and their expression as recombinant proteins has facilitated discovery of subtype-selective drugs. Distinct but related receptors may, but need not, display distinctive patterns of selectivity among agonist or antagonist li-gands. When selective ligands are not known, the receptors are more commonly referred to as isoforms rather than as subtypes. Receptor subtypes may display different mechanisms of signal output. For example, M₁- and M₃-muscarinic receptors activate G_q to initiate Ca²⁺ signaling, and M₂- and M₄-muscarinic receptors activate G_i to activate other signaling pathways. The distinction between classes and subtypes of receptors is, however, often arbitrary and/or historical. The -, and -adrenergic receptors differ from each other both in selectivity among drugs and in their choice of G protein transducers (G_i, G_q, and G_s, respectively), yet and are considered receptor classes and and are considered subtypes. The _{1A 1B}, and _1C receptor isoforms differ little in their biochemical properties; the same is nearly true for the , and subtypes.

Pharmacological differences among receptor subtypes are exploited therapeutically through the development and use of receptor-selective drugs. Such drugs may be used to elicit different responses from a single tissue when receptor subtypes initiate different intracellular signals, or they may serve to differentially modulate different cells or tissues that express one or another receptor subtype. Increasing the selectivity of a drug among tissues or among responses elicited from a single tissue may determine whether the drug's therapeutic benefits outweigh its unwanted effects.

The molecular biological search for novel receptors has moved well beyond the search for isoforms of known receptors toward the discovery of hundreds of genes for completely novel human receptors. Many of these receptors can be assigned to known families based on sequence, and their functions can be confirmed with appropriate ligands. However, many are 'orphans,' the designation give to receptors whose ligands are unknown. Discovery of the endogenous ligands and physiological functions of orphan receptors is widely hoped to lead to new drugs that can modulate currently intractable disease states.

The discovery of numerous receptor isoforms raises the question of their importance to the organism, particularly when their signaling mechanisms and specificity for endogenous ligands are indistinguishable. Perhaps this multiplicity of genes facilitates the independent, cellspecific, and temporally controlled expression of receptors according to the developmental needs of the organism. Regardless of their mechanistic implications (or lack thereof), discovery of isoform-selective ligands may substantially improve our targeting of therapeutic drugs.

Actions of Drugs Not Mediated by Receptors

If one restricts the definition of receptors to macromolecules, then several drugs may be said not to act upon receptors as such. Some drugs specifically bind small molecules or ions that are normally or abnormally found in the body. One example is the therapeutic neutralization of gastric acid by a base (antacid). Another example is the use of mesna, a free radical scavenger rapidly eliminated by the kidneys, to bind to reactive metabolites associated with some cancer chemotherapeutic agents and thus minimize their untoward effects on the urinary tract (see Chapter 52: Antineoplastic Agents). Other agents act according to colligative effects without a requirement for highly specific chemical structure. For example, certain relatively benign compounds, such as mannitol, can be administered in quantities sufficient to increase the osmolarity of various body fluids and thereby cause appropriate changes in the distribution of water (see Chapter 29: Diuretics). Depending on the agent and route of administration, this effect can be exploited to promote diuresis, catharsis, expansion of circulating volume in the vascular compartment, or reduction of cerebral edema.

Certain drugs that are structural analogs of normal biological chemicals may be incorporated into cellular components and thereby alter their function. This property has been termed a 'counterfeit incorporation mechanism,' and has been particularly useful with analogs of pyrimidines and purines that can be incorporated into nucleic acids; such drugs have clinical utility in antiviral and cancer chemotherapy (see Chapters 50 and 52).

Receptor Pharmacology

The major aim of receptor pharmacology is to understand and quantify the effects of chemicals (drugs) on biological systems. This is important in the therapeutic arena because drugs are nearly always used therapeutically in systems different from those in which they were discovered and tested. Biological systems interpret the effects of drugs in different ways, and these interpretations can be confusing. What is needed is a standard scale of drug activity that transcends biological systems and can be used to predict the effects of the drug in all systems. Receptor pharmacology strives to furnish the tools to accomplish this goal.

The basic currency of receptor pharmacology is the doseresponse curve, a depiction of the observed effect of a drug as a function of its concentration in the receptor compartment. Figure 24A shows a typical doseresponse curve; it reaches a maximal asymptote value when the drug occupies all of the receptor sites. The range of concentrations needed to fully depict the doseresponse relationship usually is too wide to be useful in the format shown in Figure 24A. Most doseresponse curves are therefore plotted with the logarithm of the concentration as the x axis (see Figure 24B). Doseresponse curves have three basic properties: threshold, slope, and maximal asymptote; these parameters characterize and quantitate the activity of the drug.

In general, drugs can do two things to receptors: (1) bind to them and (2) possibly change their behavior toward the host cell system. The first function is governed by the chemical property of affinity, ruled by the chemical forces that cause the drug to associate with the receptor. The second is governed by a quantity referred to as efficacy. Efficacy is the information encoded in a drug's chemical structure that causes the receptor to change accordingly when the drug is bound. Historically, efficacy has been treated operationally as a proportionality constant that quantifies the extent of functional change imparted to a receptor upon binding a drug.

Classical Receptor Theory

Receptor occupancy theory, in which it is assumed that response emanates from a receptor occupied by a drug, has its basis in the law of mass action, with modifying constants added to accommodate experimental findings. Agonism was described by modification of this model by Arins (1954), Stephenson (1956), and Furchgott (1966). Stephenson introduced another important concept, stimulus, which is the initial effect of drug upon the receptor itself; stimulus is then processed by the systEM To yield the observable response. Antagonism was modeled by Gaddum (1937, 1957) and Schild (1957) to determine the affinity of antagonists.

The basic components of drug receptormediated response are shown in Figure 25. Affinity is measured by the equilibrium dissociation constant of the drugreceptor complex (denoted K_d); the fraction of receptors occupied by the drug is determined by the concentration of drug and K_d, as shown (see Figure 25). Intrinsic efficacy is a proportionality constant (denoted ) that defines the power of the drug to induce response. The product of occupancy, intrinsic efficacy, and receptor number yields the total receptor-mediated stimulus given to the system. Stimulus is conveyed to physiological effectors by biochemical reactions to produce the response. It should be noted that efficacy is a function of occupancy and the stimulusresponse function (comprising all of the biochemical reactions that take place to translate agonist binding into response) and amplifies stimulus. Therefore, the location of the dose-response curves for response is shifted to the left of the receptor occupancy curve (Figure 25). Before discussion of the quantitation of drugreceptor effects, it is worth considering this amplification process further because it can control the observed response to a drug.

Transmission of Receptor Stimulus by the Target Tissue

The activation of a receptor by a drug can be thought of as an initial signal that is then amplified by the cell. Different cells have different amplification properties; thus a weak receptor signal may produce no visible response in one cell type and a powerful signal in another. The amplification properties of the cell (referred to as the stimulus-response capability) control the observed outcome of drugreceptor interaction, as shown in Figure 26 for three hypothetical drugs and three different cell types. In cell I, which amplifies a stimulus relatively weakly, drug A produces a full tissue response and would be labeled a full agonist. Drug B produces a partial (submaximal) tissue response and would be a partial agonist. Drug C produces no response, but nevertheless occupies the receptor and therefore would antagonize the effects of either drug A or drug B; it would be referred to as an antagonist. When these same drugs are tested on cell II, which has a more efficiently coupled stimulus-response mechanism, drug A remains a full agonist, drug B is now also a full agonist, and drug C, which had insufficient efficacy to cause a physiological response in cell I, is now a partial agonist. The properties of these drugs have not changed; only the efficiency of the signaling system has changed. Thus, the labels for these drugs change as well. Drug B goes from a partial agonist to a full agonist and drug C goes from an antagonist to a partial agonist. This progression continues when these drugs are tested in cell III, which has even more efficient signaling machinery. Now all three drugs act as full agonists (Figure 26). This example illustrates the potential fallacy of classifying drugs on the basis of what they do rather than what they are. What drugs do depends on the receptor and its associated signaling proteins; classification by magnitude of physiological effect can be seriously misleading when drugs are tested in one cellular format for therapeutic use in another. The alternative is to classify drugs according to the magnitude of their two molecular properties: affinity for the receptor and efficacy once bound. By quantifying these system-independent properties, drug activity can be predicted in all systems as long as the identity of the receptor is known.

Quantifying Agonism

Drugs have two observable properties in biological systems: potency and magnitude of effect (when a biological response is produced). Potency is controlled by four factors: two relate to the biological system containing the receptors (receptor density and efficiency of the stimulusresponse mechanisms of the tissue) and two relate to the interaction of drug with its receptor (affinity and efficacy). When the relative potency of two agonists of equal efficacy is measured in the same biological system, downstream signaling effects cancel and the comparison yields a relative measure of the affinity and efficacy of the two agonists (see Figure 27A). Thus, measuring agonist potency ratios is one method of measuring the capability of different agonists to induce a response in a test system and for predicting comparable activity in another. Another method of estimating agonist activity is to compare maximal asymptotes in systems where the agonists do not produce system maximal response (Figure 27B). The advantage of using maxima is that this property is solely dependent upon efficacy, whereas potency is a mixed function of both affinity and efficacy.

Quantifying Antagonism

Characteristic patterns of antagonism are associated with certain mechanisms of blockade of receptors. One is simple competitive antagonism, whereby a drug that lacks intrinsic efficacy but retains affinity competes with the agonist for the binding site. The characteristic pattern of such antagonism is the concentration-dependent production of a parallel shift to the right of the agonist doseresponse curve with no change in the maximal asymptotic response (Figure 28A). The magnitude of the rightward shift of the curve depends only upon the concentration of the antagonist and its affinity for the receptor. The affinity of a competitive antagonist for its receptor can therefore be determined according to its concentration-dependent ability to shift the doseresponse curve for an agonist rightward, as first noted by Schild (1957). Note that a partial agonist similarly can compete with a 'full' agonist for binding to the receptor. However, increasing concentrations of a partial agonist will inhibit response to a finite level characteristic of the drug's intrinsic efficacy; a competitive antagonist will reduce the response to zero. Partial agonists thus can be used therapeutically to buffer a response by inhibiting untoward stimulation without totally abolishing the stimulus from the receptor.

An antagonist may dissociate so slowly from the receptor as to be essentially irreversible in its action. Under these circumstances, the maximal response to the agonist will be depressed at some antagonist concentrations (Figure 28B). Operationally, this is referred to as noncompetitive antagonism, although the molecular mechanism of action really cannot be unequivocally inferred from the effect. An irreversible antagonist competing for the same binding site as the agonist also can produce the pattern of antagonism shown in Figure 28B.

Noncompetitive antagonism can be produced by another type of drug, referred to as an allosteric antagonist. This type of drug produces its effect by binding a site on the receptor distinct from that of the primary agonist and thereby changing the affinity of the receptor for the agonist (see Figure 28). In the case of an allosteric antagonist, the affinity of the receptor for the agonist is decreased by the antagonist (see Figure 28C). In contrast, some allosteric effects potentiate the effects of agonists (Figure 28D). Thus, in cases where the pathology may involve a failing agonist system (i.e., myasthenia gravis, Alzheimer's disease), an allosteric potentiator of the endogenous response would strengthen the signal and, importantly, preserve the natural pattern of response.

By allowing both the overexpression of wild-type receptors and the creation (and discovery) of constitutively active mutant receptors, molecular genetic technology has facilitated the study of a novel class of functional antagonists, the inverse agonists. As discussed above, receptors spontaneously can adopt active conformations that produce a cellular response. The fraction of unoccupied receptors in the active conformation usually is too low to allow observation of their agonist-independent activity; but this activity can be observed readily, either when the receptor is expressed at heterologously high levels or when mutation shifts the conformational equilibrium toward the active form. In these situations, the tissue behaves as if there were an agonist present, and a conventional competitive antagonist has no effect. However, because inverse agonists selectively bind to the inactive form of the receptor and shift the conformational equilibrium toward the inactive state, these agents are capable of inhibiting agonist-independent or constitutive signaling. In systems that are not constitutively active, inverse agonists will behave exactly like competitive antagonists, which in part explains why the properties of inverse agonists and the number of such agents previously described as competitive antagonists were not appreciated until recently.

It is not known to what extent constitutive receptor activity is a pathologically important phenomenon, and it is therefore unclear to what extent inverse agonism is a therapeutically relevant property. In some cases, however, the preferability of an inverse agonist over a competitive antagonist is obvious. For example, the human herpesvirus KSHV encodes a constitutively active chemokine receptor that generates a second messenger that drives cell growth and viral replication (Arvanitakis et al., 1997). Clearly, in such a case a conventional antagonist would not be useful as the chemokine agonist is not involved, and an inverse agonist would be the only viable intervention.