This action might not be possible to undo. Are you sure you want to continue?
Clinical Pharmacology: The Basics
D Nicholas Bateman is Reader in Clinical Pharmacology and Consultant Physician at the Royal Infirmary, Edinburgh, UK, and Director of the Scottish Poisons Information Bureau. He qualified from Guy’s Hospital, London, and trained at the Royal Postgraduate Medical School, London and in Newcastle upon Tyne. His research interests are pharmaco-epidemiology, clinical toxicology and poisons information systems. James S McLay is Senior Lecturer in the Department of Medicine and Therapeutics at the University of Aberdeen, UK. His research interests include the immunological and growth regulatory functions of the natriuretic peptides and receptors. Clinical pharmacology encompasses an understanding of how drugs work and their appropriate use in humans.
The measurement of the effects of drugs on humans (or, in basic pharmacology, an organ system) is termed ‘pharmacodynamics’. This term encompasses both mechanism of action and end-point (e.g. heart rate, blood pressure).
The actions of most drugs are mediated by the binding and interaction of drug molecules with specific molecular substances or macromolecules located on the cell surface; these are termed ‘receptors’. Some receptor sites are intracellular (e.g. steroid). The drug–receptor interaction leads to a molecular change in the receptor, which triggers a chain of events leading to a response. Receptors tend to be highly specific, interacting with a limited number of structurally related molecules. For some drugs, the receptor is nonspecific in terms of cell function (e.g. an alkalating agent that cross-binds molecules within DNA).
Agonists and antagonists
Agonists are drugs that activate a receptor response. Antagonists are drugs that block receptor response. Examples of such receptor systems include: • adrenergic (agonist – salbutamol, antagonist – atenolol) • dopaminergic (agonist – dopamine, antagonist – haloperidol) • cholinergic (agonist – bethanecol, antagonist – atropine).
The magnitude of the effect following a drug–receptor interaction usually depends on the dose of drug given; this relationship is commonly expressed in the form of a dose– response curve. Onset of response occurs at a threshold dose. For different drugs with similar actions, use of a dose–response curve allows comparison of: • potency (the amount of drug necessary to achieve a certain effect • ED50 (the dose that produces a 50% response, see below) • efficacy (the maximum possible effect of a drug). The relative potency of different drugs may be assessed using ED50. Potency has little clinical relevance, however, because a drug that is more potent than another may also produce more dose-related adverse effects.
The therapeutic index (therapeutic ratio) is the ratio between the toxic dose and the therapeutic dose of a drug. The closer this ratio is to 1, the more difficult the drug is to use in clinical practice. The therapeutic index for digoxin, for example, is very low, whereas that for amoxicillin is extremely high. Medical use of drugs with a narrow therapeutic index has lead to the concept of therapeutic drug monitoring, in which the plasma concentration of drug is measured and the dose adjusted to achieve a desired therapeutic drug concentration (see below).
In the past, toxicology studies in animals involved measurement of the dose of drug
required to kill acutely. The single dose required to kill 50% of a population is called the LD50. This is not a helpful measure in clinical practice, however, and other measures of toxicity are now generally applied, particularly because of animal welfare concerns. ED50 When a drug is given to an animal or human, it has an effect and elicits a measurable response such as increased intracellular calcium level, reduced blood pressure or reduced heart rate. A dose–response curve is created by plotting the response on the y axis and the dose of drug on the x axis. The dose at which the response is 50% of the maximal effect is termed the ED50.
Pharmacokinetics and drug metabolism
The term ‘pharmacokinetics’ refers to the rate and manner in which drugs are absorbed, distributed, metabolized and eliminated within and from the body. Knowledge of drug pharmacokinetics clarifies the relationships between dose, dose frequency, intensity of pharmacological effects, disease and adverse events.
Absorption and bioavailability
Orally administered drugs must be absorbed from the gut (usually in the upper small bowel, where the surface area is greatest). Not all of an orally administered dose may enter the systemic circulation because of insufficient absorption, or because of metabolism in the gut wall or liver before the drug enters the systemic circu-lation. This metabolism is termed ‘first-pass metabolism’ (see page 16). The amount of drug entering the systemic circulation as a proportion of that administered is termed the ‘bioavailability’. Bioavailability can be calculated by comparing areas under the plasma concentration curves for the same dose given orally and intravenously; it is expressed as a percentage. It is possible to avoid first-pass metabolism by giving drugs by other routes; for example, sublingual or dermal administration may occasionally be appropriate for nitrates. Alternatively, the oral dose can be increased appropriately.
To allow simple mathematical modelling of the uptake and distribution of drugs in humans, the body is regarded as a single fluid-filled compartment. The simplest situation is a drug given intravenously that is distributed throughout the body via the bloodstream. • Following intravenous injection, the drug passes through the lungs and heart, and then to dependent organs. • Organs and tissues with the greatest blood supply (e.g. brain, kidneys, liver) are exposed to a greater amount of drug than those with a low blood supply. • The drug equilibrates with these tissues according to blood supply and relative water (hydrophilic) or lipid (lipophilic) solu-bility. As a ‘rule of thumb’, the more lipophilic the drug, the more of it enters lipid-rich tissues; the less lipophilic the drug, the more remains in the plasma. The period during which a drug is distributed through body tissues is termed the ‘distribution phase’.
Apparent volume of distribution (Vd)
Vd is a theoretical volume into which a dose of drug appears to have been dissolved. It is determined by measuring the plasma concentration during the distribution phase before elimination has occurred (Vd = D/C, where D is the total amount of drug in the body and C is the concentration of drug in the plasma). In practice, Vd is determined from a plot of plasma concentration vs time. • Drugs that are highly fat soluble or are taken up rapidly by certain tissues are removed from the circulation quickly and have a low plasma concentration. As a result, Vd is high – possibly greater than the volume of the patient (e.g. the volume of distribution of tricyclic antidepressants and phenothiazine antipsychotics is several thousand litres). • For drugs that are not water soluble, Vd is low, and more closely similar to plasma volume (e.g. that of ethanol is about two-thirds of body weight).
Drugs are eliminated from the body by various processes, of which the most important are renal, biliary and (for volatile compounds such as anaesthetics) respiratory.
Phase I and phase II metabolism: before elimination can occur, lipid-soluble drugs must be converted into more water-soluble compounds by processes known as phase I and phase II enzymatic metabolism. The most important drug-metabolizing enzymes are three families of the cytochrome P450 superfamily of haem protein enzyme isoforms (CYP1A2, CYP2D6 and CYP3A4). These enzymes are responsible for the metabolism of a wide variety of drugs. • In phase I metabolism, reactive groups are introduced into the drug molecule by oxidation, reduction or hydrolysis. • In phase II metabolism, these groups undergo conjugation, usually in the liver with glucuronide or sulphate. Elimination in liver disease: drug-metabolizing enzymes are present in many body tissues, including plasma, but are most active in the liver. Drug metabolism may therefore be impaired in patients with liver disease. Elimination in renal disease: drug excretion is often reduced in patients with renal impairment; this is particularly important for drugs that are excreted unchanged or have active metabolites. An example is codeine; this is metabolized to form morphine, which is further metabolized to a 6-glucuronide compound. Accumulation of this compound can occur in renal failure, leading to excess sedation and respiratory depression. Effects of other drugs: the activity of drug-metabolizing enzymes in the liver may be increased (induced) or reduced (inhibited) by external factors. Changes in metabolism are particularly important for drugs with a low therapeutic index (e.g. warfarin, theophylline, phenytoin). Common examples of compounds that inhibit and induce drug metabolism are shown in Figure 1.
Commonly used drugs and environmental factors that induce or inhibit drug metabolism Enzyme-inducing agents
• • • • • • • • Carbamazepine Phenobarbitone Phenytoin Rifampicin Chronic ethanol consumption Smoking Barbecued meat St John’s wort
• • • • • • Cimetidine Ciprofloxacin Co-trimoxazole Erythromycin Ketoconazole Grapefruit juice
Drugs with which these agents commonly interact
• Carbamazepine • Ciclosporin • Phenytoin • Theophylline • Erythromycin • Warfarin • Low-dose oral contraceptives All of these drugs have relatively narrow therapeutic ranges First-order kinetics: in general, the rate at which drugs are metabolized and eliminated from the body is fixed and is proportional to the dose of drug administered. Such drugs are said to obey first-order kinetics. An effective measure of the rate of elimination of such drugs is the plasma half-life (t½, see below); drugs are considered to be completely eliminated after about five half-lives.
Zero-order kinetics: occasionally, enzyme metabolism is saturable; that is, the body is unable to eliminate more than a certain amount of drug over a fixed period of time. Drugs such as phenytoin and alcohol are said to obey zero-order (saturation) kinetics. Unlike drugs obeying first-order kinetics (for which doubling the dose effectively doubles the plasma concentration), even a small increase in the dose of these drugs produces a disproportionately large increase in plasma concentration, precipi-tating toxicity. Elimination rate constant: the pharmacokinetics of drugs within patients can be further understood by determining the elimination rate constant (K). The simplest method to calculate K is from a natural logarithm plot of the plasma concentration–time curve, which is a straight line. Drug elimination can also be measured in terms of clearance (in a manner analogous to creatinine clearance) by the volume of plasma cleared of drug per unit time. Half-life: a simpler means of understanding and using these eli-mination parameters, and of facilitating patient care, is to express the information in terms of t½. This is defined as the time taken for the plasma concentration of a drug to decrease to 50% of the original value; it can be readily calculated from the equation t½ = 0.693/K. Thus, if the plasma concentration of a drug decreases from 8 mmol/litre to 4 mmol/litre in 4 hours, the elimination t½ of that drug is 4 hours. t½ is clinically important because it enables determination of: • dosing frequency • the time elapsed before a steady-state plasma concentration is reached following repeat dosing (four to five half-lives) • whether use of a loading dose is appropriate. However, it is important to remember that many drugs exhibit a pharmacological action that is longer than the elimination t½.
Several drug-metabolizing enzymes are subject to genetic variation in activity, and this can lead to large differences in the rate of drug clearance from the plasma, unexpected prolongation of t½ and increased adverse effects. Pharmacogenetic variations are relatively common and may vary significantly between races. Examples are debrisoquine, metoprolol and isoniazid. ‘Slow metabolizers’ are more likely to develop adverse effects such as peripheral neuro-pathy with isoniazid. In Western countries, slow metabolizers of drugs such as debrisoquine and metoprolol comprise about 8% of the population.
Therapeutic drug monitoring
Therapeutic drug monitoring may be required for some drugs with a low therapeutic index in which the relationship between dose and plasma concentration is not easily calculated but for which the plasma level/biological effect is well documented. Usually, this process involves measuring the trough level of the drug (i.e. the level immediately before the next dose is administered). Dosing interval and t½ affect the amount of drug in the body at any one time, and knowledge of these factors enables implementation of logical dosing regimens. • A drug such as digoxin with a t½ of 36 hours can be given once per day. • Theophylline has a short t½ and is therefore difficult to use in conventional tablets. The pharmacokinetics of short-t½ drugs may be altered by pharmaceutical means, usually by administering the drug in a slow-release formulation. However, there are few drugs for which this is necessary.
Interactions between drugs may result in changes in their pharmacokinetics (pharmacokinetic interaction) or an increase or decrease in their biological effect (pharmacodynamic interaction).
Pharmacokinetic interactions most commonly involve changes in metabolism in the liver (enzyme inhibition) or excretion by the kidneys. Rarely, protein-binding displacement causes a change in distribution.
In pharmacodynamic interactions, different drugs act at different receptor systems to produce, most commonly, an increased biological effect. A useful pharmacodynamic interaction is the antihypertensive effect of concurrently administered β-blocker and calcium channel antagonist used in hypertensive patients. This interaction may be detrimental, however, when the same drugs result in inappropriate hypotension in patients treated for angina.
Adverse drug reactions
Adverse drug reactions (ADRs) are commonly divided into the following types: • A – predictable (e.g. bradycardia caused by β-adrenoceptor antagonists) • B – bizarre • C – chronic effects occurring only after prolonged treatment (e.g. iatrogenic Cushing’s syndrome caused by corticosteroids) • D – delayed effects occurring many years after treatment, or in the children of treated patients (e.g. second malignancies in patients treated with alkylating agents) • E – end-of-treatment effects, which occur when drugs are stopped suddenly (e.g. myocardial infarction following acute withdrawal of β-adrenoceptor antagonists) Types A and B ADRs are the principal groups.
Type A ADRs
Type A ADRs are common. They are caused by an augmented pharmacological effect, are dose dependent and are seldom fatal. They often arise as a result of altered drug pharmacokinetics caused by disease or concurrent drug therapy.
Type B ADRs
Type B ADRs are uncommon. They are unrelated to drug pharmacology and unrelated to dose. They are unpredictable and often fatal. Such reactions often involve anaphylactoid-type reactions.
Pharmacovigilance is the branch of pharmaco-epidemiology that concentrates on the detection of ADRs. ADRs often mimic common disease states, and uncommon ADRs may be difficult to detect unless they are severe or are temporally related to a specific medication. In the UK, there are three types of pharmaco-vigilance study. • The Yellow Card reporting system is a spontaneous-reporting alerting system that may also be used as a hypothesis-generating process. It relies on health-care professionals reporting any adverse event that they suspect may be caused by a medication. There is no requirement for proof. • Prescription Event Monitoring is a systematic cohort approach that may also act as an hypothesis-generating and testing process. It is typified by the Green Form in the UK, and is based on the monitoring of dispensed prescriptions for new drugs via a central agency. • The third technique involves hypothesis testing, when previous data have suggested that a drug may be responsible for a particular adverse event. Hypothesis testing usually involves case-control studies. Randomized clinical studies and cohort studies are usually less efficient, because large numbers of patients are required to detect rare events.
For most drugs, there is a direct relationship between pharmaco-logical response and concentration at the receptor; thus, to be biologically active, the drug must gain access to the systemic circulation. Plasma drug concentration depends on both drug kinetics and the design of the drug delivery system.
The most commonly used delivery systems involve absorption of drug from the gastrointestinal tract following buccal, sublingual, rectal or, most often, oral administration. Commonly encountered oral forms include:
• solutions • suspensions • capsules • tablets • coated tablets • modified-release tablets. The time taken for the drug to appear in the systemic circulation following oral administration increases in an approximately similar order. Tablets are the most common delivery system. They have advantages of convenience and accuracy of dose. Coated tablets – it is possible to alter the delivery and apparent kinetics of drugs by changing the dissolution characteristics of tablets. Thus, a tablet may be enteric-coated to prevent breakdown in the stomach, ensuring that it remains intact until it reaches the small bowel. This approach is commonly used to protect drugs that are destroyed by gastric acid (e.g. omeprazole). Modified-release tablets – the excipients of tablets may be modified to improve drug delivery by controlling the rate, amount and duration of drug release over a 24hour period. This approach is used for drugs with a short t½, which require frequent dosing to maintain therapeutic levels (e.g. theophylline, verapamil, nifedipine). Pro-drugs – a similar effect may be achieved by using a pro-drug. Pro-drugs are inactive compounds that are activated by biological fluids or metabolizing enzymes following administration (e.g. enalapril is converted to its active form enalaprilat).
Intravenous administration is most commonly used when rapid onset of action and careful control of plasma levels are required. Drugs may be given as a: • bolus injection • slow infusion • continuous infusion. Slow infusion is used when excessively high transient plasma levels are undesirable (e.g. phenytoin). Continuous infusion is used when the drug has a short t½ or when its therapeutic index is narrow and sustained controlled blood levels are required.
Clinical trials Compliance
Clinical trials are discussed elsewhere. If a drug is to achieve the desired therapeutic effect, it is necessary for the patient to be compliant (concordant) with medical therapy. Compliance with therapy is defined as the extent to which patients follow the course of treatment. Factors believed to be important in ensuring patient compliance include the complexity of the therapeutic regimen, and patients’ understanding of their disease and the need for and benefits of treatment.
Mucklow J C, Waller D G. Graduate Therapeutics. A Primer for MRCP and Specialist Training. Oxford: Butterworth-Heinemann, 2001. Page C, Curtis M, Sutter M et al. Integrated Pharmacology. 2nd ed. St Louis: Mosby, 2002. Waller D G, Renwick A G, Hillier K. Medical Pharmacology and Therapeutics. Philadelphia: Saunders, 2001.
• The pharmacodynamic effect of a drug is its effect on a measurable physiological parameter such as heart rate, blood pressure or acid secretion into the stomach • Pharmacokinetics describe the rate and manner in which drugs are absorbed, distributed, metabolized and eliminated within and from the body • Following oral administration, it takes about five half-lives for plasma drug levels to reach a steady state • It takes about five half-lives for a drug to be completely eliminated from the body once administration has stopped • Drug–drug interactions are common and predictable, and may arise as a result of increased pharmacodynamic effect • Drug–drug interactions involving kinetic changes usually result from effects on drugmetabolizing enzymes in the liver; they are often serious when drugs with a narrow therapeutic index are involved • ADRs are common, and most often arise as a result of altered pharmacokinetics in disease states such as renal impairment • Pharmacovigilance is the branch of pharmaco-epidemiology that concentrates on the detection of ADRs • Compliance (concordance) is a measure of the extent to which a patient population complies with a therapeutic medical regimen
Copyright © 2003 The Medicine Publishing Company Ltd
Simon R J Maxwell is Senior Lecturer in Clinical Pharmacology at the University of Edinburgh and Honorary Consultant Physician at the Western General Hospital, Edinburgh, UK. His research interests include the vascular biology of atherosclerosis and endothelial function, and the role of oxidative stress and antioxidants. David J Webb is Christison Professor of Therapeutics and Clinical Pharmacology at the University of Edinburgh and Honorary Consultant Physician at the Western General Hospital, Edinburgh, UK. His research interests include endothelial function and atherogenesis. All clinicians should have an understanding of receptor mechanisms for the following reasons. • Many drugs commonly used in modern practice act on receptors. • Safe drug use requires an understanding of receptor pharmacology. • Some common diseases involve changes in receptor numbers or coupling of receptor stimulation to response. • Future advances in pharmacology and therapeutics are likely to develop from the discovery of further receptors and molecular modelling of drugs to interact with them.
Definitions Receptors and ligands
Receptors are proteins situated in the cell membrane or at an intracellular site that specifically recognize and bind molecules termed ‘ligands’. The ligand may be a neurotransmitter, a hormone, a drug or a growth factor. This interaction initiates a conformational change in the receptor protein that eventually transmits a signal into the cell (Figure 1). The resulting intracellular signal is usually a self-amplifying cascade of biochemical events that ‘couples’ receptor–ligand interaction to a biological response such as contraction or secretion. There are four principal types of receptor. • Channel-linked receptors are usually coupled directly to an ion channel. An example is the nicotinic acetylcholine receptor. • G-protein-coupled receptors are all coupled to intracellular effector mechanisms via a family of closely related ‘G-proteins’ that participate in signal transduction by coupling receptor binding to intracellular enzyme activation or the opening of an ion channel. Examples include the muscarinic cholinergic receptor, adrenoreceptors and opioid receptors. • Kinase-linked receptors are linked directly to an intracellular protein kinase that triggers a cascade of phosphorylation reactions. Examples include receptors for insulin and various growth factors. • Receptors regulating gene transcription are located intracellularly and are also known as ‘nuclear receptors’. They include receptors for steroids and thyroid hormone. Although the term ‘receptor’ is usually restricted to proteins whose only function is to bind a ligand, it is sometimes used more widely in pharmacology to include functional targets such as ion channels and enzymes (Figure 2). For example, voltage-sensitive sodium channels in excitable tissues might be considered as the ‘receptor’ for local anaesthetics, and the enzyme xanthine oxidase the ‘receptor’ for allopurinol. In the case of mechanoreceptors, intracellular signals are generated by mechanical forces rather than specific ligands. Receptor types are usually defined according to their most potent endogenous agonist (e.g. adrenergic, serotoninergic – Figure 2). Receptor subtypes can then be further defined by their relative response to different agonists and antagonists. Although many different kinds of receptor have now been described, relatively few signalling systems have been discovered; thus, there must be ‘cross-talk’ between receptors (e.g. β-adrenergic, histamine H2, dopamine, opioid and γ-amino-butyric acid type B receptors are all linked to modulation of cAMP levels).
Agonists and antagonists
• An agonist can be defined as a ligand, the binding of which to a receptor protein produces a conformational change necessary to initiate a signal that is coupled to a biological response. When all available receptors are occupied, a maximal response is produced.
Free agonist Extracellular Cell membrane G-protein Adenylate cyclase Intracellular
Receptor binding G-protein Phospholipase C Signal transduction
Protein kinase A
Ca2+ release from storage Signal amplification
The interaction of a ligand with a receptor protein induces a conformational change that eventually initiates an intracellular signal. The signalling mechanism may include direct opening of an ion channel or production of a secondary messenger via interaction of the receptor with G-proteins (e.g. cAMP, inositol triphosphate). The signal is amplified intracellularly, coupling the receptor–ligand interaction to a biological response.
Receptors and coupling mechanisms
Receptor type • Receptor-controlled ion channels
Coupling mechanism Entry/exit of ions → Depolarization or hyperpolarization
Examples of ligands Nicotinic acetylcholine γ-aminobutyric acid Muscarinic acetylcholine β-adrenergic Dopamine Serotonin Insulin Atrial natriuretic peptide Steroid hormones
• G-protein receptors Receptor protein associated with a G-protein which may: Activate an enzyme producing a ‘secondary messenger’ Adenylate cyclase → cAMP Phospholipase C → inositol triphosphate, diacylglycerol Activate an ion channel Opioids • Receptor-controlled enzymes • Intracellular receptors to protein synthesis Initiate protein phosphorylation (e.g. tyrosine kinase, guanylate cyclase) Stimulate mRNA synthesis in the cell nucleus, leading Thyroid hormones Vitamin D
• An antagonist is a ligand that binds to a receptor but does not produce the conformational change that initiates an intracellular signal. Occupation of the receptor by an antagonist prevents the binding of any other ligand and so ‘antagonizes’ the biological response to the agonist. • Some ligands have properties intermediate between agonists and antagonists and are known as ‘partial agonists’. They are unable to produce a maximal signalling effect even when all available receptors are occupied. However, partial agonists also block receptor sites that could potentially be occupied by the full agonist and this competition for receptors means that, in some circumstances, partial agonists may also appear to antagonize the effect of full agonists (see below). • Some ligands produce an effect opposite to that of the full agonist when they bind to a receptor; they are termed ‘inverse agonists’. For these to be identified, the relevant receptor must exhibit endogenous activation in the absence of ligand binding.
An understanding of receptor pharmacology requires a basic knowledge of the dose–response curve relating the concentration of an agonist to the biological effect it produces (Figure 3). A sigmoid curve is obtained when the percentage of maximum effect (e.g. decrease in blood pressure, reduction in gastric acid secretion) is plotted against the logarithm (base 10) of the concentration of the agonist. This representation is useful because it visually expands the region over which the drug response changes most rapidly with concentration. It is evident that increased concentration of agonist produces an increased effect, but only over a relatively narrow concentration range – further increases in concentration beyond this range produce little extra effect. The maximum response on this curve is termed ‘Emax’ and the concentration of agonist that produces 50% maximum effect (Emax/2) is the EC50.
Response (% maximum)
10–9 10–8 10–7 10–6 10–5
Concentration of agonist (log scale)
The biological effect (% maximum response) and the concentration of a full agonist are plotted on a logarithmic scale. An equipotent partial agonist has a lower efficacy than a full agonist – it cannot achieve the maximum response even when all available receptors are occupied. EC50 is the concentration of agonist that produces 50% maximum effect.
Increasing drug potency
Response (% maximum)
Concentration of agonist (log scale)
Dose–response curves for four full agonist drugs of different potencies but equal efficacy.
Although it is generally true that the biological effect elicited by an agonist is proportional to the fraction of receptors occupied, the relationship is often not simple because of the phenomenon of ‘spare receptors’. It has been observed in many receptor systems that full agonists can elicit the maximum effect without occupying all available receptors. This apparent excess of receptors allows a full response to occur at lower ligand concentrations than would otherwise be required. Most drugs used in practice are administered in doses that give concentrations close to the top of the dose–response curve. This ensures that most patients respond to the drug without the need for a specific dose–response curve for each individual. However, it is possible, in many cases, to achieve the desired therapeutic response at a lower dose. This is important because drug side-effects are often dose-related in a similar manner. The aim of the prescriber is to provide drugs at doses that provide maximum therapeutic efficacy with minimum side-effects, and this principle guides the licensed dose range. For most drugs, the ratio between these two doses (therapeutic index) is high, but there are notable exceptions (e.g. digoxin, theophylline). Efficacy of an agonist is the extent to which it can produce the maximum response possible when all available receptors are occupied. Full agonists have high efficacy and can produce the maximum response of which a tissue is capable. Partial agonists have a lower efficacy – they can achieve only sub-maximum responses even when all receptor sites are occupied. The dose–response curve for a partial agonist is usually flatter than that of a full agonist and reaches a lower maximum (Figure 3). The term ‘therapeutic efficacy’ is used by clinical pharmacologists to compare drugs that produce different therapeutic effects on a biological system despite having maximum pharmacological efficacy at their target receptor (e.g. thiazide diuretics have a lower therapeutic efficacy than loop diuretics). Potency of an agonist is measured by the amount (weight) required to produce a given level of effect. More potent drugs produce biological effects at lower doses (i.e. they have a lower EC50). It is important to note that a less potent drug may have an efficacy similar to that of a more potent one; the difference in potency can be overcome by giving the less potent drug in higher doses. A common example is the inhibition of acid secretion produced by ranitidine and cimetidine at H2-receptors. Although differences in relative potency can be overcome by a change in dose, it should be remembered that the adverse effects of drugs are often dose-related. If these occur by a mechanism other than receptor–ligand interaction, potency may be relevant because only the more potent drug will be active at concentrations that avoid unwanted side-effects. The potency of a drug is related to its affinity for the receptor – that is, how readily the drug–receptor complex is formed.
Receptor–ligand interactions Affinity
The affinity of a ligand for its receptor can be conveniently assessed by the concentration of ligand required to produce 50% maximum binding. Affinity is a function of both the rate of association and the rate of dissociation of the ligand–receptor complex; the former depends on the ‘goodness of fit’ at a molecular level, whereas the latter depends on how tightly the ligand is bound. Systems requiring rapid, fine modulation (e.g. nerve synapses) cannot have agonists with a high receptor affinity because these would produce unnecessarily prolonged responses. During stimulation, agonist concentration near the receptor must be relatively high, but the agonist is then cleared rapidly by active transport. In contrast, growth factors are often peptides with a very high affinity for their receptors, and achieve their effects at concentrations that are difficult to detect in vivo.
Most natural and synthetic ligands are reversible and eventually dissociate from their receptor when the ligand concentration is reduced. This fact enables biological responses to be modulated and allows competitive interaction between agonists and antagonists for access to the receptor. Thus, the effect of a reversible competitive antagonist can always be overcome by giving the agonist at a sufficiently high concentration (i.e. it is surmountable). Some antagonists are irreversible and do not dissociate from the receptor. Such irreversible antagonists produce indefinite receptor blockade and are less desirable as therapeutic agents. However, there are notable exceptions. Phenoxybenzamine, the irreversible α-antagonist, is valuable when
preparing for surgery in patients with phaeochromocytoma, in whom unpredictable bursts of agonist activity would be life-threatening. Commonly used irreversible inhibitors of functional enzymes include aspirin and omeprazole. The pharmacological effects of irreversible antagonists disappear only when new receptors or enzyme are synthesized. This explains the benefits of aspirin taken intermittently as prophylaxis against cardiovascular events. The relative receptor occupancy of agonists and antagonists depends on their individual concentrations and affinity for the receptor. When an antagonist is introduced into the receptor–agonist model, the dose–response curve for the full agonist is shifted to the right (Figure 4). This occurs because the agonist is now competing for receptor occupation with the antagonist and, for a given percentage of receptor occupation (and hence biological response), a higher agonist concentration is required. The effect of competitive antagonists can be overcome with sufficiently high doses of agonist. When a partial agonist is introduced in the presence of a full agonist, the biological response depends on the availability of full agonist and unoccupied receptors. When the receptors are not fully occupied, the partial agonist can be bound without displacing the full agonist, leading to augmentation of the biological response. When the available receptors are completely occupied by full agonist, the presence of partial agonist competing for receptor sites tends to diminish the response. Therefore, at concentrations at which drugs are present in excess of receptors, the dose–response curve for the full agonist is shifted to the right (Figure 4). This fact may have important clinical consequences. The partial β1-adrenoceptor agonist xamoterol can stimulate cardiac contractility in healthy individuals at rest, but in patients with severe heart failure it may antagonize the response to the full agonist adrenaline (epinephrine) and reduce contractility. Similarly, the partial opioid agonist nalorphine may appear to antagonize the analgesic effect of morphine. Partial agonists have some advantages as therapeutic agents – though they are unable to achieve the same maximum effect as the full agonist, they are less likely to produce limiting toxic side-effects at the top of their dose–response curve (e.g. the partial opioid receptor agonist buprenorphine does not cause as much respiratory depression as morphine when it is used as an analgesic).
Changes in receptor responsiveness
In the short term, most receptor-mediated responses are altered by changing the local concentration of agonist around the receptors. It is now recognized that, in the longer term, changes are possible in the response of the tissue to a given concentration of agonist, and can be attributed to altered receptor function. These changes (loosely termed ‘up-regulation’ and ‘down-regulation’) may involve alterations in the number of receptors on the cell surface or in the coupling of receptor occupation to the intracellular response. Down-regulation is usually seen in response to chronic receptor overstimulation as a result of long-term administration of a drug (exogenous agonist) or a disease process (endogenous agonist). In some cases, this can occur quickly, leading to the phenomenon of ‘tachyphylaxis’, in which repeated administration of a drug is associated with rapidly diminishing efficacy. This phenomenon is typical of drugs that stimulate release of neurotransmitters from nerve terminals, which leads to rapid depletion of stores. The term ‘tolerance’ is used to describe a more gradual decrease in responsiveness to a drug. Tolerance may be related to downregulation of receptors, but may also result from altered drug metabolism and induction of physiological compensatory mechanisms. The tolerance that develops to organic nitrates is probably related to both. A further example in which physiological compensation reduces drug effect is the activation of the renin–angiotensin system that occurs in response to thiazide diuretic treatment in hypertension. There are many clinically relevant examples of altered receptor responsiveness. • In Parkinson’s disease, the term ‘on–off phenomenon’ is used to describe the rapid swing from mobility to immobility that occurs some hours after administration of Ldopa. It has been attributed to down-regulation of dopaminergic receptors in the brain following prolonged L-dopa treatment, which renders the extrapyramidal pathways unusually sensitive to the decreasing concentration of dopamine. • In patients with phaeochromocytoma with high endogenous catecholamine concentrations, receptor numbers are down-regulated, leading to a relative insensitivity to catecholamines. • In autonomic neuropathy, receptor up-regulation renders the cardiovascular system hypersensitive, and exaggerated responses to catecholamines may occur. • In patients with prostate cancer, gonadotrophin-releasing hormone given
Increasing antagonist concentration
Response (% maximum)
10–7M 10–6M 10–5M
Concentration of agonist (log scale)
In the presence of an antagonist, the dose–response curve of the full agonist is shifted to the right because the full agonist competes for receptor binding. The higher the concentration of antagonist, the greater the shift.
Full agonist Full agonist and partial agonist
Response (% maximum)
Concentration of agonist (log scale)
In the presence of a partial agonist, the dose–response curve of the full agonist is shifted to the right.
continuously paradoxically inhibits gonadotrophin release in contrast to the normal pulsatile secretion. An alternative example of receptor up-regulation is the use of 3-hydroxy-3methylglutaryl co-enzyme A inhibitors to lower plasma cholesterol concentration. These drugs inhibit the rate-limiting enzyme in the synthesis of cholesterol in hepatocytes, promoting increased synthesis and expression of receptors for low-density lipoprotein (LDL) particles. As a consequence, the rate of uptake of cholesterol-laden LDL particles from the circulation is increased and the plasma cholesterol concentration is reduced. Sometimes, receptor responses are regulated by changes in the coupling of receptor occupation to intracellular signalling. Chronic β-adrenoceptor blockade potentiates the response of the adenylate cyclase–cAMP system and consequently up-regulates the response to other receptors that are linked to it.
Receptors are usually subtyped on the basis of their selectivity for agonists or antagonists. Agonist selectivity is determined by the ratio of EC50 at the two receptor subtypes. In the case of β-adrenoceptors, the concentration of norepinephrine (noradrenaline) required to cause bronchodilatation (β2) is ten times higher than that required to cause tachycardia (β1); therefore, the selectivity of norepinephrine for β1receptors with respect to β2-receptors is 10.
Antagonist selectivity is measured as the relative shift of the agonist dose–response curves achieved by a single dose of antagonist affecting responses mediated through the two receptors. For example, if the non-selective β-adrenoceptor agonist isoprenaline was used in the situation above, the concentration of atenolol achieved in the bloodstream after administration of a 50 mg dose would shift the dose–response curve at the bronchus by only 10% of that at the heart, giving a selectivity for β1receptors of 10. Thus, a ten times greater concentration of atenolol is required to produce a shift in the agonist response curve for the lung equal to that for the heart. It must be remembered that selectivity for a receptor subtype is only a relative concept and does not equate with specificity. Drugs with selectivity for one receptor subtype can produce a maximum effect at other subtypes if a sufficient quantity is given. This is particularly important when the beneficial effects are activated by one receptor subtype and the unwanted effects by another. For example, atenolol is considered a β -selective adrenoreceptor antagonist but has some effects at β2-receptors, and is 1 therefore absolutely contraindicated in asthmatic patients in whom any reduction in β2-mediated bronchodilatation may be dangerous. Selectivity is useful in clinical practice only when the ratio of the impact of the drug at the two receptor sites is 100 or more. When selectivity is lower, it is difficult to predict drug doses that will exploit the difference in subtype activity. Selectivity is most likely to be achieved at the lowest effective dose.
Advances in molecular biology have enabled cloning and sequencing of receptor proteins. With recombinant synthetic technology providing sufficient quantities of receptor protein, it is now possible to screen novel ligands rapidly without wholeanimal or organ experiments. These studies have uncovered a much higher degree of variability within each receptor family than could have been predicted from pharmacological experiments. From these data, it is becoming possible to design ligands with higher selectivity for individual receptor subtypes. Large-scale DNA sequencing has identified many DNA sequences with a distant relationship to the G-protein-coupled receptor superfamily. The sequence homology is usually insufficient to assign these ‘orphan’ receptors to a particular subfamily. However, cloning and expression of orphan receptors in cells and screening for functional responses is likely to lead to the discovery of many novel receptor ligands.
Assaying receptor binding
Assays for receptors are designed to detect binding sites, not all of which are coupled to a signal. Most assays use radiolabelled ligand and measure the amount of radioactivity bound to cell membranes, whole cells or tissue sections fixed to a microscope slide. Nonspecific binding can be assessed by displacing the radiolabelled ligand by addition of large quantities of unlabelled ligand. This procedure exploits a fundamental property of receptor binding – that it is saturable, unlike nonspecific binding. Mathematical manipulation of the data allows the number of receptors and their affinities to be calculated; however, these calculations are often inaccurate when high-affinity and low-affinity binding sites are present, and it is preferable to use a reiterative computer program to analyse the raw data. In receptor autoradiography, images are obtained by exposing sections of radiographic film after incubation and washing steps. Various image analysis programs allow colour-coded pictures to be obtained after subtraction of the nonspecific image from the total binding. This technique gives anatomical information that is particularly useful in establishing the likely site of action of a neurohumoral factor within a tissue.
Black J V, Stanley N P. Inverse Agonists Exposed. Nature 1995; 374: 214–15. Broach J R, Thorner J. High-throughput Screening for Drug Discovery. Nature 1996; 384: 14–16. Colquhoun D. Binding, Gating, Affinity and Efficacy. Br J Pharmacol 1998; 125: 923–47. Gudermann T, Kalkbrenner F, Schulz C. Diversity and Selectivity of Receptor–G Protein Signalling. Ann Rev Pharmacol Toxicol 1996; 36: 429–59. Lefkowitz R J. G-proteins in Medicine. N Engl J Med 1995; 332: 186. Lerner M R. Tools for Investigating Functional Interactions between Ligands and Gprotein Coupled Receptors. Trends Neurosci 1994; 17: 142–6.
Libert F, Vassart G, Parmentier M. Current Developments in G-protein-coupled Receptors. Curr Opin Cell Biol 1991; 8: 218–23. Mangelsdorf D J, Thummel C, Beato M et al. The Nuclear Receptor Superfamily: The Second Decade. Cell 1995; 83: 835–9. Rang H P, Dale M M, Ritter J M. Pharmacology. Edinburgh: Churchill-Livingstone, 1995. Ross E M, Kenakin T P. Pharmacodynamics: Mechanisms of Drug Action and the Relationship between Drug Concentration and Effect. In: Hardman J G, Limbird L E, Gillman A G, eds. Goodman and Gillman’s The Pharmacological Basis of Therapeutics. New York: McGraw-Hill, 2001: 31–44. Wilson S, Bergsma D J, Chambers J K et al. Orphan G-protein-coupled Receptors: The Next Generation of Drug Targets? Br J Pharmacol 1998; 125: 1387–92.
Copyright © 2003 The Medicine Publishing Company Ltd
Pharmacokinetics for the Prescriber
Howard L McLeod is Associate Professor of Medicine in the Department of Medicine at Washington University, St Louis, USA. He trained in Seattle, Memphis, Philadelphia and Glasgow, UK, and was senior lecturer in medicine at the University of Aberdeen. His research interests include the clinical pharmacology of anticancer drugs, the molecular basis of polymorphic drug metabolism, and the evaluation of cellular targets for chemotherapy within human tumours. Pharmacokinetics is a tool for describing the movement of drugs through the body over time, and deals with the processes of absorption from the site of administration, distribution throughout the body, metabolism or conjugation of the drug, and elimination from the body. Pharmacokinetics can be thought of as what the patient does to the drug (whereas pharmacodynamics is what the drug does to the patient). An understanding of pharmacokinetics should help the clinician to: • appreciate how dosing regimens are devised • tailor a dosing regimen to the individual requirements of the patient (e.g. in renal failure) • determine what may be wrong when a patient fails to respond to treatment • determine why a drug has caused toxicity • elucidate the mechanisms of drug interactions. Clinical pharmacokinetics is particularly valuable with drugs for which the margin between therapeutic and toxic concentrations is narrow. The pharmacokinetic proﬁle of a drug should not be considered in isolation; its pharmacodynamic effects must also be taken into account.
The half-life (t½) of a drug is the time taken for the circulating concentration of the drug to decrease by 50%. A concentration–time curve (Figure 1) can be constructed by administering the drug intravenously and removing blood for assay at frequent intervals. With most drugs, the curve is a straight line when the concentration (vertical axis) is expressed on a logarithmic scale, enabling t½ to be determined easily. Knowledge of t½ is useful as a guide to: • the time required for drug elimination (Figure 2) • the rate of accumulation during repeated dosing (when a drug is given repeatedly, the plasma concentration increases with each dose, Figure 3). However, the rate at which plasma concentration increases reaches a plateau (steady state) when the amount eliminated in a dosing interval is equal to the dose. t½ is the only variable that determines the time required to reach steady state during repeated dosing. More than 90% of the steady-state concentration is achieved after 4–5 half-lives. Lignocaine, for example, has a t½ of about 1 hour; because 4 hours is too long to wait for a clinical effect to occur, an intravenous loading dose is given, followed by a maintenance infusion.
Volume of distribution
Volume of distribution (Vd) is a measure of how widely a drug is distributed in body tissues. It is commonly expressed as the ratio of the total dose administered by bolus injection to the peak plasma concentration; for example, if a dose of gentamicin is 100 mg and the peak concentration is 5 mg/litre, Vd is 20 litres. The concept of Vd is notional; it does not relate to any particular compartment of the body. However, it is useful in determining how extensively a drug is distributed. The Vd of ciclosporin, for example, is 100 litres; it is widely distributed to most body tissues. Knowledge of Vd is also relevant to dosing regimens – when Vd changes, the loading dose must be changed.
The clearance of a drug, not t½, is the best measure of the rate at which it is eliminated from the body. t½ is affected by both clearance and Vd, whereas clearance is independent of Vd. Clearance is calculated by dividing the dose by the area under the plasma concentration–time curve, either after a single dose or during a dosing interval at steady
Log-linear plot of a concentration–time curve following an intravenous dose
Drug concentration (mg/litre)
The elimination half-life (t1/2) is the time taken for the drug concentration to fall by 50%. C(o) is the extrapolated concentration at the time of injection.
Relationship between half-life and drug elimination for a drug with a halflife of 6 hours
Number of half-lives 0 1 2 3 4 5 6 7 8 Time (hours) 0 6 12 18 24 30 36 42 48 Plasma concentration (µg/ml) 100 50 25 12.5 6.25 3.125 1.56 0.78 0.39 Original concentration eliminated (%) 0 50 75 87.5 93.75 96.875 98.44 99.22 99.61
state (Figure 3). Total body clearance is a composite of all available relative elimination (e.g. hepatic metabolism plus renal excretion). Clearance is expressed as the volume of plasma cleared of drug per unit time (e.g. ml/minute, as for creatinine clearance). When calculating clearance after oral administration, an adjustment must be made for incomplete bioavailability. Clearance, Vd and t½ are mathematically interrelated and can be calculated from each other as well as directly from the concentration–time curve (t½ = 0.693 x Vd/clearance).
Although absorption can occur throughout the gastrointestinal tract, drugs are absorbed predominantly in the upper small bowel because of its large surface area. Other medications are inﬂuenced by cellular transporters (Lin & Yamazaki). Lipid-soluble drugs are absorbed rapidly by passive diffusion. Absorption of water-soluble drugs is slower and may be incomplete. Most lipid-soluble drugs reach peak plasma concentrations 30–60 minutes after ingestion, but absorption can be slowed by a heavy meal or by other therapeutic agents that reduce the rate of gastic emptying (e.g. opiates, tricyclic antidepressants, antihistamines,
Plasma drug concentration during repeated dosing
Plasma drug concentration
Mean steady-state concentration
Time (number of half-lives)
Without a loading dose, the blue areas are equal in size
anticholinergics). Reduced absorption rate is important only if the pharmacological effect sought depends on attaining a high peak concentration (e.g. antibiotics, analgesics). Hence, patients with severe migraine and gastric stasis may not obtain pain relief from oral aspirin or paracetamol unless gastric emptying is accelerated. In most drugs at steady state, extent of absorption is more important than rate.
Effect of food
Generally, absorption of lipid-soluble drugs is increased in the presence of food, though the clinical impact is small because these drugs are well absorbed (Schmidt & Dalhoff). Water-soluble drugs are best taken on an empty stomach, at least 1 hour before a meal with a full glass of water. If the drug is likely to produce gastrointestinal irritation (e.g. anti-inﬂammatory agents), it should be taken with food whatever its solubility.
If a drug is inactivated in the gastrointestinal tract or metabolized in the gut wall or liver before it enters the systemic circulation, the amount reaching the site of action is reduced (Figure 4). This effect is called ‘ﬁrst-pass’ metabolism or the ‘ﬁrst-pass’ effect, and usually results from metabolism in the liver. Propranolol is metabolized to an inactive compound; this explains the 20-fold difference between the effective intravenous and oral doses. Grapefruit juice can inhibit enzymes in the duodenum, leading to increased bio-availability of drugs such as felodipine, terfenadine and ciclosporin; in several cases, this interaction has been fatal. It is sometimes possible to bypass the liver using routes of administration other than oral; for example, glyceryl trinitrate may be given sublingually, buccally or transdermally, and other drugs (e.g. morphine) may be given rectally. Other biological factors may inﬂuence the degree of absorption. The P-glycoprotein efﬂux pump protein is located in many tissues, including the duodenum, and acts as a barrier to the absorption of many drugs, including ciclosporin. It functions by pumping drug out of intestinal enterocytes, back into the bowel lumen.
Once a drug reaches the systemic circulation, it may be bound to circulating proteins, usually albumin (basic drugs such as propranolol and disopyramide bind largely to globulins and acute-phase reactants such as α1 acid glycoprotein). Most drugs must be unbound (free) to have a pharmacological effect. Protein binding is clinically important with only a few drugs, of which the most important are phenytoin and warfarin; three criteria must be met: • high protein binding (if the drug is < 90% bound to plasma proteins, changes in protein binding have little impact on the amount of unbound drug in circulation – i.e. the drug
Drug metabolism during first and subsequent passes through the gastrointestinal tract and liver
Extensive inactivation in the liver during the first pass (e.g. propranolol, morphine) Enhanced efflux from enterocytes by P-glycoprotein
Liver Drug Portal vein Metabolites Biliary tract Gut
Inactivation in the gut lumen (e.g. insulin)
Inactivation in the gut wall (e.g. tyramine)
Enterohepatic recirculation (e.g. oestrogens, warfarin)
Deconjugation and reabsorption (e.g. oestrogens)
4 that is available for distribution to the site of action) • low Vd (if the distribution of the drug to the tissues is large, even total displacement of the drug from protein-binding sites would be ‘mopped-up’ by the wide distribution to the tissues) • a change in binding (the action of a drug changes only when the protein binding changes). Factors that alter protein binding include hypoalbuminaemia (albumin < 25 g/litre), lasttrimester pregnancy (partly as a result of hypoalbuminaemia), renal failure (in which the afﬁnity of proteins for drugs appears to be altered) and displacement by other drugs. The change in protein binding is likely to be sudden only when displacement by other drugs has occurred. In the other cases, the changes are gradual and alone would not alter the effects of drugs because the compensatory increase in the rate of clearance from the body prevents the build-up of a high concentration of bound drug in the plasma. However, these effects alter the relationship between the total plasma concentration of a drug and its unbound concentration.
Most drugs are lipid soluble and must be biotransformed to more water-soluble products before excretion in bile and/or urine. Inactivation of a drug is the most common effect of metabolism. The principal site of drug metabolism is the liver; other organs (e.g. gut, lung, kidney) can metabolize drugs to some extent. The metabolism of many substances occurs in two phases – oxidation (I) and conjugation (II) (Figure 5). A few pro-drugs (e.g. ciclosporin, enalapril, levodopa, cyclo-phosphamide, sulindac) have no biological activity and require metabolic activation.
The enzymes responsible for oxidation and similar phase I reactions (mono-oxygenases, mixed function oxidases) are located in the hepatic endoplasmic reticulum and are collectively termed ‘cytochrome P450’. Phase I reactions may produce inert compounds, but many oxidated metabolites have biological activity and occasionally a toxic moiety (e.g. from paracetamol overdose) or a carcinogenic moiety (e.g. from cigarette smoke) is formed. Products of phase I reactions may be excreted directly or further metabolized by conjugation. The activities of phase I enzymes can be increased (induced) by enzyme inducers (phenobarbitone, phenytoin, carbamazepine); this is the basis of some important drug interactions.
Molecular weight < 300
Liver Drug Phase I enzymes
Oxidation Dealkylation Reduction Hydrolysis
Phase II enzymes
Glucuronide Sulphate Glutathione N-acetyl
Urine Conjugates Bile
Molecular weight > 300
Phase I enzymes catalyse the modification of existing functional groups in drug molecules (oxidation reactions) Conjugating enzymes (phase II) facilitate the addition of endogenous molecules such as sulphate, glucuronic acid and glutathione
In the process of conjugation, an endogenous group (e.g. glucuronic acid, sulphate) is added to the parent drug or to its oxidated metabolite. Almost all conjugates are inert; a notable exception is morphine 6-glucuronide. Larger molecules are excreted in the bile, and those with a molecular weight below 300 in the urine. Some metabolites are deconjugated by bacterial ﬂora in the gut lumen and the parent drug can then be reabsorbed. Enterohepatic cycling occurs with some benzodiazepines and the oestrogenic components of the oral contraceptive pill. Interruption of this recycling has been suggested as a basis of the possible interaction of oral contraceptives with some antibiotics.
Factors affecting metabolism
Individuals have their own complement of enzymes with overlapping substrate speciﬁcities. These enzymes are genetically controlled, but they are also affected by factors such as age, nutritional status, disease, smoking (tobacco and marijuana), alcohol consumption and concomitant drugs (Fuhr). Age: drug metabolism is impaired in premature babies. Infants and young adults eliminate drugs more rapidly than adults. Hepatic mono-oxygenase activity decreases gradually into old age, but conjugation enzymes appear to be better maintained in the elderly.
Oxidation reactions are less efﬁcient in the presence of severe liver disease and in malnourished patients. Conjugation mechanisms are often preserved, however, even in hepatic cirrhosis. Dose adjustment in patients with liver disease is crude, because impairment of hepatic function relates poorly to changes in biochemical liver function tests.
Drug metabolism that varies on a genetic basis is often called polymorphic drug metabolism because multiple clinical groups with unique patterns of elimination have been identiﬁed (Evans & McLeod). Early studies with isoniazid identiﬁed slow and fast acetylators in whom the drug has different pharmacological and toxic effects. Slow acetylators have a greater hypotensive response to hydralazine and increased susceptibility to lupus-like syndrome. In the UK, about 8% of individuals are unable to 4-hydroxylate debrisoquine (poor metabolizers) and consequently exhibit an exaggerated hypotensive response to this drug (which is now little used). Impaired metabolism of other drugs (e.g. nortriptyline, perhexiline, metoprolol) is linked to this defect. This enzyme is also responsible for
activating codeine to morphine, and poor metabolizers do not experience pain relief after administration of codeine. Many of these drugs are metabolized by several enzymes; thus, the pharmacological response of poor metabolizers may not differ markedly from that in normal individuals. However, poor metabolizers may produce a different metabolite proﬁle and may be at a greater risk of toxic effects with these agents. For many polymorphic enzymes, the prevalence of poor metabolizers varies between ethnic groups; for example, slow acetylators constitute 22% of Inuits, 91% of Egyptians and 45% of the UK population. The molecular basis has been determined for many poor metabolizer phenotypes, and DNA-based tests using polymerase chain reaction analysis are available at research centres. More recently, evidence has emerged for a clinically important inﬂuence of genetic polymorphism on cellular transporters and cellular targets of drug therapy. Initial studies have focused on therapy for HIV, cancer, depression and epilepsy, but transport and target pharmacogenetics are likely to be important in many areas of medicine.
Drugs are principally excreted via the kidneys. Some drugs are also excreted via the bile, and a few have other routes of elimination. Rifampicin, for example, is excreted in tears and sweat, and may cause these to turn orange; patients should be warned about this, to avoid unnecessary alarm. Alcohol is partly eliminated via the lungs, giving the breath a distinct odour.
Drugs are handled by the kidneys in three main ways. • All drugs are ﬁltered at the glomerulus. • Some are actively secreted in the proximal tubule. • Some are passively reabsorbed in the distal tubule. Some drugs may undergo all three processes. Active secretion increases the rate of renal clearance of drugs and passive reabsorption reduces it; renal clearance is therefore equal to the sum of the rates of elimination by ﬁltration and secretion minus the rate of reabsorption. Only drug molecules that are not protein bound in the plasma can be ﬁltered; the clearance of drug by ﬁltration is therefore equal to the unbound fraction multiplied by the glomerular ﬁltration rate. The renal tubular system can actively secrete drugs. Separate transport systems exist for acids (e.g. penicillins, cephalosporins) and bases (e.g. amiloride, ethambutol). Competition occurs between drugs utilizing these systems. Probenecid increases circulating levels of penicillins and cephalosporins and potentiates methotrexate toxicity by reducing their tubular secretion. Cimetidine can also have a role in blocking tubular secretion. In patients with renal impairment, doses of drugs that are usually excreted at least 50% unchanged by the kidneys should be reduced; these drugs include digoxin, aminoglycoside antibiotics, lithium, procainamide, methotrexate, cisplatin, carboplatin, trimethoprim and methyldopa. Reduction in both the frequency of administration and individual doses may be required (e.g. for aminoglycoside antibiotics, peak and trough plasma concentrations both relate to toxicity); in these situations, therapeutic drug monitoring is often used to tailor the dose.
Drugs are also excreted by hepatocytes into the bile, usually in conjugated form. Metabolites with a molecular weight of more than 300 are likely to follow this route. Biliary excretion provides a back-up pathway when renal function is impaired. Reabsorption of biliary excreted drugs may lead to an enterohepatic cycle, which can prolong their action (e.g. some benzodiazapines). Recycling of warfarin and digitoxin can be interrupted by resins (e.g. cholestyramine), which bind them and thereby increase the rate of elimination.
Therapeutic drug monitoring
Measurement of drug concentrations in human plasma, serum and whole blood is a useful tool for guiding the safe and effective use of many medications (Van Heeswijk). Therapeutic drug monitoring is applied to various therapeutic classes of agents that: • have a high degree of pharmacokinetic variability in the relevant patient population • exhibit a relationship between blood concentration and toxic or therapeutic effect • have a narrow therapeutic index • are detectable by a rapid, accurate and reproducible assay.
At present, the indications for therapeutic drug monitoring focus on the prevention of severe or life-threatening toxicity (e.g. acute renal failure from gentamicin, CNS toxicity from phenytoin). There is a therapeutic basis for its use with most drugs, but it is prevention of toxicity that makes the analysis cost-effective and justiﬁable in the current economic environment. Most commonly monitored medications are antibiotics, including aminoglycosides and vancomycin. Anti-epileptics (phenytoin, carbamazepine, valproic acid, phenobarbitone), lithium, digoxin, ciclosporin and theophylline are also commonly monitored. Plasma or serum is the matrix of choice for most medications, but with some drugs (e.g. ciclosporin) whole blood is used to give a better indication of cellular accumulation. Trough concentrations (immediately before the next dose) are associated with toxicity in many medications. Peak concentrations (usually obtained 1 hour after bolus injection or oral administration) predict efﬁcacy for some antibiotics.
Evans W E, McLeod H L. Pharmacogenomics – Drug Disposition, Drug Targets, and Side Effects. N Engl J Med 2003; 348: 538–49. Fuhr U. Induction of Drug Metabolizing Enzymes: Pharmacokinetic and Toxicological Consequences in Humans. Clin Pharmacokinet 2000; 38(6): 493–504. Lin J H, Yamazaki M. Role of P-glycoprotein in Pharmacokinetics: Clinical Implications. Clin Pharmacokinet 2003; 42(1): 59–98. Schmidt L E, Dalhoff K. Food–Drug Interactions. Drugs 2002; 62(10): 1481–502. Van Heeswijk R P. Critical Issues in Therapeutic Drug Monitoring of Antiretroviral Drugs. Ther Drug Monit 2002: 24(3): 323–31.
Clark B, Smith D A. An Introduction to Pharmacokinetics. Oxford: Blackwell Scientific, 1986. Derendorf H, Hochhaus G. Handbook of Pharmacokinetic/Pharmacodynamic Correlation. Oxford: CRC Press, 1995. Rowland M, Tozer T N. Clinical Pharmacokinetics: Concepts and Applications. 3rd ed. Philadelphia: Lea & Febiger, 1995.
Copyright © 2003 The Medicine Publishing Company Ltd
Gabrielle M Hawksworth is Professor of Molecular Toxicology in the Departments of Medicine and Therapeutics and Biomedical Sciences at the University of Aberdeen, UK. Most pharmacologically active drugs are lipophilic and are metabolized to some extent. To be excreted from the body, lipophilic drugs must be metabolized to water-soluble products, which are not readily reabsorbed by the kidney. In some cases (pro-drugs, e.g. azathioprine, enalapril, L-dopa, zidovudine, cyclophosphamide), metabolism is required for therapeutic effect. For most, however, metabolism leads to loss of therapeutic effect. In some drugs (e.g. paracetamol), metabolism may result in toxicity if the detoxifying pathways are saturated at high drug doses. The cytochrome P450 isoforms and other enzymes involved in major metabolic routes will have been determined for newly licensed drugs, and adverse drug interactions caused by altered metabolism are thus less likely to occur. For existing drugs, however, knowledge of the routes of metabolism is essential. Despite the abundance of papers in the medical literature, predictable adverse metabolic drug interactions continue to occur.
Drug-metabolizing enzymes in humans
Drug metabolism is usually divided into two phases – phase I involves drug oxidation and phase II involves conjugation of drugs with endogenous compounds. The liver is the principal organ involved in drug metabolism; most of phase I metabolism is catalysed by the mixed-function oxidase system, of which cytochrome P450 is the main component. P450-dependent metabolism in the liver is the major source of differences between individuals in response to drugs and is responsible for several adverse drug interactions. Terfenadine–ketoconazole is the most well known P450-related interaction. It was first described in 1990, in a patient on the recommended prescribed dose of terfenadine in addition to cefaclor, ketoconazole and medroxyprogesterone. She developed symptomatic torsade de pointes, and excessive levels of parent terfenadine and proportionately reduced levels of metabolite were found in her serum, suggesting inhibition of terfenadine metabolism. Subsequently, this interaction was shown to be predictable – terfenadine is metabolized by an isoform of P450 that is selectively inhibited by ketoconazole. Other compounds (e.g. a constituent of grapefruit juice) also inhibit this isoform. These findings led to the issue of warnings with terfenadine, and subsequently to withdrawal of the drug; the active metabolite fexofenadine, which has antihistamine activity, is now prescribed instead. Regulatory bodies (e.g. the UK Committee on Safety of Medicines, the US Food and Drug Administration) now require information on which isoforms of cytochrome P450 are involved in the metabolism of drugs submitted for registration. Isoforms of P450 – P450 is a large superfamily of proteins, synthesis of which is controlled by a superfamily of genes. Isoforms of the families CYP1, CYP2 and CYP3 are responsible for drug metabolism. Within these families, six isoforms (CYP1A2, CYP2C9, CYP2C19, CYP2D6, CYP2E1 and CYP3A4) are involved in the metabolism of a large proportion of drugs in humans; CYP3A4 is responsible for most P450mediated metabolism. These isoforms show selectivity for different drugs; some commonly used drugs are metabolized mainly by one or two P450 isoforms (Figure 1). CYP3A4 can metabolize a wide variety of chemically heterogeneous drugs, whereas the substrate requirements for metabolism by other P450s is more restricted.
Metabolism of some drugs is markedly dependent on the route of administration. Drugs administered orally gain access to the systemic circulation via the hepatic portal vein; thus, the entire absorbed dose is exposed first to the intestinal wall and then to the liver. Extensive first-pass metabolism occurs when a high proportion of the parent drug is removed in this first pass through the liver; low bioavailability and marked betweenindividual variation in blood concentrations of drug result. This may be avoided by administering the drug sublingually (e.g. glyceryl trinitrate). In other cases (e.g. lignocaine, propranolol), response to therapy may be highly variable, with problems in clinical use.
Drug substrates, inhibitors and inducers of the major human cytochrome P450 isoforms
P450 isoform CYP1A2 Substrate Caffeine, imipramine, paracetamol, R-warfarin, verapamil, estradiol, theophylline Coumarin, nicotine, testosterone Cyclophosphamide, testosterone, mianserin Carbamazepine, taxol Diazepam, diclofenac, ibuprofen, phenytoin Piroxican, S-warfarin Diazepam, hexobarbital, imipramine, lansoprazole Omeprazole, proguanil, propranolol Amitriptyline, clomipramine, codeine Desipramine Flecainide, fluoxetine, imipramine, metoprolol, mexiletene Mianserin, nortriptyline, paroxetine, propafenone, propranolol, timolol, trifluperidol Caffeine, dapsone Enflurane, ethanol, halothane, theophylline Alfentanil, amiodarone, carbemazepine Cyclophosphamide, ciclosporin, dapsone, digitoxin Sulphaphenazole Sulphinpyrazone Orphenadrine Inhibitor Fluvoxamine Furafylline Inducer Charcoal-broiled beef Cigarette smoke Omeprazole Barbiturates Dexamethasone
CYP2A6 CYP2B6 CYP2C8 CYP2C9 CYP2C19 CYP2D6
Disulfiram Clotrimazole, gestodene , Itraconazole, ketoconazole, miconazole
Ethanol Isoniazid Carbemazepine Dexamethasone
Diltiazem, diazepam, erythyromycin, ethinylestradiol Etoposide, imipramine, lansoprazol Lignocaine, losartan, lovastatin, simvastatin, atorvastatin, midazolam, nifedipine Omeprazole, quinidine, retinoic acid Saquinavir, efavirenz, nevirapine, tacrolimus, taxol, terfenadine Theophylline, triazolam, verapamil, R-warfarin
Omeprazole Phenobarbital Phenytoin Rifampicin
Induction/inhibition of metabolism
Induction is a process by which enzyme activity is enhanced as a result of increased enzyme synthesis. P450 isoforms show selectivity of induction (e.g. rifampicin selectively induces CYP3A4). A practical consequence of enzyme induction is that, when two or more drugs are given simultaneously, the inducer accelerates the metabolism of the other drugs. Both licensed drugs and herbal medicines (e.g. St John’s wort) have been shown to have clinically significant interactions. Inhibition – several drugs (e.g. methotrexate, angiotensin-converting enzyme inhibitors, non-steroidal anti-inflammatory drugs) exert their therapeutic effect by enzyme inhibition. However, inhibition of drug metabolism by a concurrently administered drug can lead to drug accumulation and toxicity. Some inhibitors are selective for P450 isoforms (e.g. ketoconazole), whereas others (e.g. cimetidine) inhibit all cytochrome P450-mediated metabolism. Inhibition of metabolism can also occur when two concurrently administered drugs are substrates for the same isoform of a drug-metabolizing enzyme. Several of these interactions have been described for statins and antiviral drugs that are substrates for CYP3A4.
Genetic polymorphisms lead to between-individual variation in drug response; the population is divided into ‘poor metabolizers’ and ‘extensive metabolizers’, in whom the standard dose of a drug may lead to higher-than-normal blood concentrations or apparent non-response, respectively. Several P450 isoforms (notably CYP2D6, CYP2E1, CYP2C19 and CYP3A5) are expressed polymorphically in humans; this may be important therapeutically when the drug metabolized by these isoforms has a relatively narrow therapeutic window. Ethnic differences are seen in the incidence of polymorphisms. A polymorphism of thiopurine methyl transferase that occurs in about 10% of Caucasian populations affects treatment with azathioprine and mercaptopurine. It is possible to genotype patients for the known polymorphisms of cytochrome P450 and other drug-metabolizing enzymes, and hence predict phenotype; this could be used to individualize therapy, but it is unlikely to occur in clinical practice.
Bolego C, Baetta R, Bellosta S et al. Safety Considerations for Statins. Curr Opin Lipidol 2002; 13: 637–44. Dresser G K, Bailey D G. A Basic and Conceptual and Practical Overview of Interactions with Highly Prescribed Drugs. Can J Clin Pharmacol 2002; 9: 191–8. Garattini S. Drug Metabolism: From Experiments to Regulatory Aspects. Drug Metab Rev 1997; 29(3): 853–86. Ma M K, Woo M H, McLeod H L. Genetic Basis of Drug Metabolism. Am J Health Syst Pharm 2002; 59: 2061–9. Murphy P A. St John’s Wort and Oral Contraceptives: Reasons for Concern? J Midwifery Womens Health 2002; 47: 447–50. Rogers J F, Nafziger A N, Bertino J S. Pharmacogenetics affects Dosing, Efficacy and Toxicity of Cytochrome-P450 Metabolized Drugs. Am J Med 2002; 113: 746–50.
• Be aware of the major routes of metabolism of drugs prescribed • Be aware of metabolic interactions in patients taking two or more drugs • With complex drug regimens (e.g. transplant patients), leave dose adjustment in drug therapy to specialist centres
Copyright © 2003 The Medicine Publishing Company Ltd
D John Reynolds is Consultant Physician and Clinical Pharmacologist at Oxford Radcliffe Hospitals NHS Trust, Oxford, UK, and Honorary Senior Clinical Lecturer in the University of Oxford. Until recently in the UK, prescribing of drugs to patients was the sole responsibility of doctors and dentists, but that is changing as other health-care professionals are given statutory powers to prescribe. The process of prescribing is surprisingly complex and is underpinned by several principles that all prescribers must understand. To highlight the more important steps involved, this contribution considers a simple scenario of the type that is seen in general practice thousands of times every day.
Mrs G is a 78-year-old woman who has noticed that, over the last year, she has developed pain in both knees when she walks. The right is worse than the left, and occasionally she has mild swelling in that knee. Establishing the diagnosis – before any decisions about treatment are made, the doctor must establish what condition the patient has and what other conditions and factors may be relevant. On examination, you find that Mrs G has moderate osteoarthritis, but before you reach for your prescription pad you must establish more information. Patient expectations – what does Mrs G want from the consultation? Is she sufficiently troubled by pain to want active treatment, or does she simply want reassurance that her problem is caused by ‘wear and tear’ and is nothing more serious? Many patients are satisfied just to know what is wrong, and may initially decide against active treatment once they are reassured. If Mrs G wants treatment, how does she feel about taking tablets? Would exercise or physiotherapy be more appropriate and acceptable? Is she realistically able to lose weight to reduce the burden on her joints? It is all too easy to end the consultation by prescribing a drug without exploring other ways of managing the problem. What else has the patient tried? – remember to ask whether the patient has used any over-the-counter medicines to alleviate the symptoms. Ask directly about herbal and traditional remedies. Patients often fail to mention treatments of which they think the doctor might not approve. It transpires that Mrs G has tried a friend’s glucosamine tablets, with some benefit. How severe is the problem? – do not assume that, by coming to see you, Mrs G has decided that her problem needs drug treatment. You must determine how severe the pain is and how much it restricts her daily activities. Remember that, with symptoms such as pain, the severity is what the patient tells you it is – not what you judge it to be. Is Mrs G’s pain an acute flare-up or a chronic problem? For how long is any treatment likely to be necessary?
What are the risks and benefits of drug treatment?
To make a balanced judgement about risks and potential benefits, you need further information about the individual patient. Co-morbidity – are there any features that might render the patient more susceptible to drug side-effects (Figure 1)? You learn that Mrs G has a history of heart failure secondary to hypertension. Her symptoms are well controlled. She suffered a duodenal ulcer 25 years ago, but has never required admission to hospital.
Factors that predispose to adverse drug reactions
• Age – adverse reactions are more common in the elderly and the very young • Gender – women are at greater risk • Race – pharmacogenetic variations result in many adverse drug reactions and interactions • History of allergy or previous adverse drug reaction • Renal impairment • Hepatic impairment • Heart failure • Nutritional status • Multiple drug therapy – predisposes to drug interactions 1
Other medications – what other medications is Mrs G taking? The potential for drug interactions is significant, though the most serious usually involve drugs for which the margin between therapeutic and toxic effects is narrow (e.g. warfarin, theophylline, digoxin, anticonvulsants). If you are in any doubt or are unfamiliar with a drug, get further information. The British National Formulary contains an excellent appendix on drug interactions. Mrs G is taking furosemide, 40 mg once daily, lisinopril, 20 mg once daily, and atenolol, 25 mg once daily. Patient’s previous experience – many patients express a preference for drugs that have helped them in previous, similar circumstances, and have no faith in others that were not effective (“paracetamol doesn’t work for me, doctor”). What has Mrs G preferred to use for analgesia in the past? Drug allergies and intolerances – always ask whether the patient is allergic to or intolerant of any medicines. Mrs G should be asked specifically about previous exposure to aspirin and non-steroidal anti-inflammatory drugs (NSAIDs), if you are considering these for her arthritis. She reports that she felt light-headed when she took codeine phosphate for recent back pain.
Choosing the right drug
If you and Mrs G agree that drug treatment is indicated, you must make several decisions based on what you know about the patient, her condition and the drugs available. In effect, you are calculating a benefit:harm ratio, for which you need to judge the evidence in each of the following areas. • How serious is the condition? • How effective are the drugs available? • What are the side-effects of the drugs available, including their incidence and severity? • Are there any ways in which this patient might differ significantly from the population studied in published trials? Choosing the most appropriate class of drug – Mrs G suffers pain and needs an analgesic agent. The main classes of drugs to consider are aspirin and the NSAIDs, paracetamol, and paracetamol combined with a minor opioid. The NSAIDs can be further subdivided into non-selective and COX-2-selective agents. There is good evidence that NSAIDs are more effective for the relief of acute pain than both paracetamol alone and paracetamol-combination drugs, but in chronic pain the evidence is less clear. The main difference between these drugs is their associated side-effects. • Paracetamol is well tolerated and has no major common interactions with other drugs, with the notable exception of warfarin. (Although paracetamol is commonly perceived as a safe drug to use with warfarin, it is clear that, when taken chronically, it augments the anticoagulant effect, often requiring warfarin dose adjustment.) The principal problem with paracetamol is that it must be taken four times daily for maximal effect. • Aspirin and the NSAIDs have significant side-effects including upper gastrointestinal haemorrhage and perforation, which may be fatal. The excess risk of death in patients taking a non-selective NSAID for 2 months or longer is about 1/2000, but is substantially higher in certain groups. In the UK, about 2000 patients per year die from haemorrhage or perforation as a direct result of NSAID use. Mrs G is in a high-risk group because she is over 65 years of age and has a history of peptic ulcer disease. In addition, her heart disease means that she is at high risk of a less favourable outcome if she develops bleeding or perforation. • COX-2-selective NSAIDs were developed in response to these problems with NSAIDs, and in the short term at least are associated with significantly fewer upper gastrointestinal side-effects. In 2002, the UK National Institute for Clinical Excellence (NICE) recommended use of COX-2-selective NSAIDs in patients at high risk of bleeding or perforation. However, NSAIDs may also impair renal function and can exacerbate heart failure, and recent evidence suggests that significant episodes of NSAID-related renal and heart failure may be more common than upper gastrointestinal bleeding and perforation. COX-2-selective agents are not free from these effects, and their use may be inadvisable even in the high-risk groups defined by NICE. Mrs G has heart failure and use of an NSAID is therefore best avoided. • Paracetamol with a minor opioid (e.g. codeine, dextropropoxyphene) is less effective than NSAIDs in acute pain, but when used chronically the plasma concentration of the opioid rises and the efficacy increases. Caution is required in the elderly, however, because opioid drugs may cause unsteadiness, confusion and drowsiness. They also tend to cause constipation.
If Mrs G is happy to take a drug three or four times each day, paracetamol seems the safest initial treatment; NSAIDs are the least attractive. If paracetamol proves insufficient, it is probably better to change to paracetamol combined with a minor opioid than to an NSAID, though care will be needed because Mrs G has been intolerant of codeine in the past. If an NSAID is used, a COX-2-selective drug is likely to be safer, but careful monitoring will be needed to ensure that Mrs G does not develop renal impairment or worsening heart failure. Choosing a specific drug – the specific agent to advise must now be chosen. Most prescribers familiarize themselves with only one or two drugs from each class of agents. Dose – in general, the dose prescribed should be in line with that recommended in the product licence. It is a good ‘rule of thumb’ to start any new drug at the lowest dose available and titrate upwards depending on efficacy and side-effects. In certain situations, dose adjustments must be made to avoid side-effects. Particular care is needed in patients with hepatic and/or renal impairment, in the elderly and in children. Route of administration (Figure 2) – most drugs are taken orally as tablets, but alternative approaches are sometimes required. Mrs G could consider use of a topical NSAID instead of oral paracetamol. This would reduce the risks of treatment, but probably at the expense of lower efficacy and greater cost. Duration of therapy – Mrs G’s treatment is likely to be required long term, but drugs are often given for a specific duration. Always set a stop date, or make an early review appointment for appropriate monitoring and re-evaluation. Mrs G should be given no more than 1 month’s supply of treatment initially.
Routes of administration of drugs
Route Gastrointestinal tract • Buccal, sublingual • Oral • Rectal By injection • Subcutaneous • Intradermal • Intramuscular • Intravenous • Intrathecal • Intracavity Examples Glyceryl trinitrate Most agents Metronidazole Morphine Rationale Rapid absorption, and to avoid first-pass metabolism Ease of use; must be lipid soluble to ensure reliable absorption Good absorption as an alternative to oral route if patient cannot swallow (e.g. postoperative) Swift onset of action
Prochlorperazine As an alternative to intravenous injection or for depot preparations Cefuroxime For drugs unreliably absorbed orally and/or when high plasma concentrations needed rapidly Methotrexate For local effect avoiding systemic toxicity
Topical (skin, eye, ear, nose, inhaled) – given for local effect, and to avoid systemic toxicity; can also use drugs with low systemic bioavailablity
Monitoring drug therapy
Benefits – it is easy to start a drug and then assume that the problem has been solved, particularly in asymptomatic conditions such as moderate hypertension and hyperlipidaemia. However, the true end-point of stroke or heart disease may not become apparent for years, and it is therefore essential to monitor a measurable surrogate end-point (e.g. blood pressure, cholesterol concentration) to ensure that the drug is achieving the desired effect. Mrs G can monitor her treatment by assessing the degree of pain she suffers or whether there is any functional improvement (e.g. in the distance she can walk). If she is uncertain of any benefit, she should be encouraged to discontinue therapy and determine whether her symptoms worsen. There is no point in Mrs G taking a drug from which she derives no benefit, yet patients are often prescribed multiple drugs for no clear continuing indication.
Side-effects – all prescribers must ensure that they minimize potential adverse effects and monitor for them when appropriate. Mrs G has impaired left ventricular function secondary to hypertension and is taking three drugs. The safest option for her pain relief is paracetamol, with which significant adverse effects are very unlikely. If paracetamol proves inadequate and an NSAID is considered, however, there are several potential problems for which both Mrs G and her prescriber must be vigilant. She should be warned about the potential for fluid retention and worsening heart failure, so that if she becomes more breathless or develops oedema she can seek help promptly. Before prescribing an NSAID in this situation, it would be prudent to check Mrs G’s renal function and electrolytes, and to repeat the blood tests after 2 weeks.
Prescribing in special situations
The pharmacokinetics and pharmacodynamics of many drugs are significantly affected by pathophysiological changes in renal failure and liver diease, and by the physiological changes of pregnancy. Use of drugs in pregnancy and in lactating mothers carries specific risks for the developing fetus and newborn child. Always refer to a reference text before using any drug with which you are not entirely familiar in any such setting.
Cost and cost-effectiveness
Doctors have a legal obligation to provide the treatments that their patients need, but they are obliged to do so within available financial resources. NICE is a source of guidance for the NHS, based on clinical effectiveness and cost-effectiveness. In addition, most Primary Care Trusts and Hospital Trusts provide local clinicians with upto-date information to guide them in choosing medicines. The most expensive medicine may be the one that the patient does not take, and involving patients in informed decision-making and subsequent monitoring is central to good prescribing.
British National Formulary. London: British Medical Association, Royal Pharmaceutical Society of Great Britain. Firth J D, Reynolds D J M. Scientific Background to Medicine 2. Oxford: Blackwell Science, 2001. Grahame-Smith D G, Aronson J K. Oxford Textbook of Clinical Pharmacology and Drug Therapy. 3rd ed. Oxford: Oxford University Press, 2002. National Institute for Clinical Excellence. www.nice.org.uk
Copyright © 2003 The Medicine Publishing Company Ltd
Adverse Drug Reactions
R E Ferner is Consultant Physician at City Hospital, Birmingham, UK, Honorary Senior Lecturer in Clinical Pharmacology at the University of Birmingham, and Director of the West Midlands Centre for Adverse Drug Reaction Reporting.
What’s new ?
• Thioridazine, cisapride and droperidol have been identified as causes of QT prolongation and torsade de pointes • Antiretroviral protease inhibitors (e.g. indinavir, ritonavir, saquinavir) are associated with a ‘metabolic syndrome’ of abnormal fat distribution and diabetes • Serotonin syndrome can be caused by selective serotonin re-uptake inhibitors, particularly in combination with other drugs (e.g. pethidine, tramadol)
Adverse drug reactions (ADRs) are important but often difficult to diagnose. They should be considered in the differential diagnosis of a wide range of conditions. Detecting and reporting ADRs makes prescribing safer and more likely to achieve its aims.
Adverse drug reaction is ‘a response to a drug that is noxious and unintended and which occurs in doses normally used for the treatment, prophylaxis, or diagnosis of disease, or the modification of physiological function’ (WHO). Adverse drug event is any adverse event that occurs while a patient is taking a given drug. It is not necessary to determine whether the event was a response to the drug. Side-effect is any effect caused by a drug other than the intended therapeutic effect, whether beneficial, neutral or harmful. The term is sometimes taken to be synonymous with ‘adverse drug reaction’, and is sometimes used to describe ‘minor’ and predictable ADRs (e.g. constipation with opiates).
ADRs affect about 7% of medical in-patients; one-half of reactions occur before admission. One study suggested that, in the USA, more than 100,000 patients per year die from ADRs.
The mechanisms of many ADRs are unknown. Known mechanisms are predominantly pharmacological or immunological. Pharmacological ADRs Direct effects – pharmacological ADRs are sometimes an inevitable consequence of the therapeutic action of the drug. An example is the occurrence of hypoglycaemia with insulin; a patient’s usual dose can provoke hypoglycaemia if his or her food intake is reduced or exercise increased, or if the absorption of insulin is increased by, for example, an increase in blood flow. Vasodilatation from artificial tanning devices can lead to ‘sun-bed hypoglycaemia’. Collateral effects – adverse pharmacological consequences of drug treatment can arise when a drug exerts its therapeutic action at a receptor or locus that is also present elsewhere. For example, β-adrenoceptor antagonists given after myocardial infarction act on the heart to reduce heart rate and the risks of arrhythmia, but also act on β-adrenoceptors in peripheral arteries and can consequently worsen peripheral vascular disease. ‘Collateral damage’ from broad-spectrum antibacterial agents (particularly cephalo-sporins and clindamycin) can alter the bowel flora and increase the risk of overgrowth by toxigenic Clostridium difficile; as a result, the patient develops pseudomembranous colitis. Many drugs act at more than one receptor type. For example, the therapeutic effect of tricyclic antidepressants results from stimulation of monoaminergic pathways by a reduction in monoamine re-uptake, but the drugs also have antimuscarinic activity.
This is responsible for adverse effects such as dry mouth, constipation and retention of urine. Advances in molecular biology have revealed the existence of many more receptor subtypes than previously imagined, so more specific drugs may become available in the future. Immunological ADRs are grouped into the four classical types of Gell and Coombs. Important examples are listed in Figure 1. Anaphylactoid reactions are caused by inflammatory mediators released in response to pharmacological rather than immunological stimuli, and can appear similar to anaphylactic reactions. Iodinated contrast media, acetylcysteine infusion and modified gelatin plasma expanders can all cause anaphylactoid reactions. Angiotensin-converting enzyme (ACE) is responsible for degradation of the inflammatory mediator bradykinin, and it is thought that ACE inhibitors predispose patients to angioedema by inhibiting bradykinin breakdown.
Adverse drug reactions with an immunological basis
Immunological reaction Type I – immediate hypersensitivity (e.g. interaction between drug and native protein produces a new, antigenic molecule) Type II – cytotoxic antibody Drug alters a membrane-bound protein; cytotoxic antibodies bind to the altered protein and complement-mediated cell lysis occurs Type III – immune complex Persistent foreign protein molecules cause antibody synthesis; immune complexes are formed and damage tissues Type IV – delayed hypersensitivity Delayed cell-mediated reactions Examples Anaphylaxis and urticaria with penicillins
Haemolysis with methyldopa Thrombocytopenia with quinine
Serum sickness with horse-derived antitetanus serum and snake antivenins
Erythema nodosum with sulphonamides Stevens–Johnson syndrome with carbamazepine Phenytoin hypersensitivity
Individuals can be susceptible to rare ADRs because of a genetic abnormality that alters either their handling of the drug or their physiological response to it. Abnormal drug metabolism: drug metabolism is important for two reasons. • The effects of a drug are related to the amount present at the site of action, so changes that alter the rate of metabolism also alter the extent and duration of effects and adverse effects. • The route of metabolism can influence the toxicity of a compound, by affecting the production of potentially harmful metabolites. Cytochrome P450 system – the most important drug metabolic enzymes are the hepatic microsomal oxidases (the cytochrome P450 system). About sixty such enzymes are known; the most important cytochrome P450s for drug metabolism are CYP2D6 and CYP3A4. The ‘poor-metabolizer’ phenotype CYP2D6 is present in about 8% of the population in the UK, in about 1% of Arabs and in 30% of Hong Kong Chinese. Poor metabolizers exhibit reduced rates of metabolism of debrisoquine (an obsolete antihypertensive), metoprolol, propranolol, haloperidol, amitriptyline, nortriptyline, flecainide and propafenone. CYP3A4 is the major enzyme responsible for the metabolism of nifedipine, amiodarone and carbamazepine. Poor metabolizers can suffer adverse effects as a result of the increased concentration of the drug, or the increased duration of exposure. With drugs such as carbamazepine, impaired metabolism increases the concentration of toxic metabolites and the poor-metabolizer phenotype may be associated with reactions such as hepatitis or Stevens–Johnson syndrome, which are related to metabolite concentrations.
Abnormal responses to drugs Glucose-6-phosphate dehydrogenase (G6PD) deficiency – genetically abnormal responses to drugs include X-linked deficiency of G6PD. This enzyme is responsible for the oxidation of glucose-6-phosphate to the corresponding gluconolactone, generating reduced NADPH in the process. NADPH protects RBCs from oxidative damage, and in G6PD-deficient individuals, drugs that are oxidizing agents can cause haemolysis, which can be fatal. More than 200 genotypes of the enzyme have been described; the phenotype varies from a small increase in susceptibility to exquisite sensitivity to oxidizing agents. Examples of relevant drugs are listed in Figure 2.
Some drugs that can cause haemolysis in patients with reduced glucose6-phosphate dehydrogenase activity
Antimalarials and related drugs • Chloroquine (‘acceptable in acute malaria’ – British National Formulary) • Quinine (‘acceptable in acute malaria’ – British National Formulary) • Quinidine • Primaquine • Dapsone Sulphonamides • Co-trimoxazole • Sulfamethoxazole • Sulfasalazine Other antibacterials • Nitrofurantoin • Ciprofloxacin • Ofloxacin • Other quinolones • Nalidixic acid Vitamins • Vitamin C (ascorbic acid) • Vitamin K as menadiol sodium phosphate Analgesics and local anaesthetics • Aspirin • Paracetamol • Benzocaine Others • Isosorbide dinitrate • Methylene blue dye 2 Porphyria – acute intermittent porphyria is an autosomal dominant disorder characterized by recurrent attacks of abdominal pain, neurological disturbance and excessive amounts of δ-aminolevulinic acid and porphobilinogen in the urine. It is often provoked by drug therapy. An extensive list of drugs that can precipitate the painful and life-threatening crises is given in the British National Formulary. Other factors: susceptibility to ADRs is also a function of age, sex, coexistent disease and co-ingestion of other drugs, all of which can affect both the pharmacokinetics and pharmacodynamics of drugs.
Detection and diagnosis
When an adverse event occurs during or after drug treatment, it may be a result of the drug, or of the disease for which the drug was taken, or it may have another, unrelated cause. In the simplest case, the reaction is closely linked to the drug in time, and the problem resolves when the drug is stopped (‘de-challenge’) and re-emerges when
the drug is restarted (‘re-challenge’); this seldom occurs in practice. Some reactions (e.g. teratogenesis) occur long after exposure, some (including fatal reactions) are irreversible, and some occur only when there is a conjunction of a particular set of circumstances and do not recur at re-challenge. It is helpful to know whether the event is likely to occur with the underlying disease, whether it has previously been reported with the drug in question, and, if not, whether it might be predicted from what is known of the drug’s structure and mode of action. It is impossible to establish the large randomized trials that would be needed to explore all likely and relevant adverse effects. Their detection depends heavily on spontaneous reporting (e.g. the UK Yellow Card scheme), stimulated post-marketing surveillance (e.g. the Drug Safety Research Unit’s Green Card scheme) and casecontrol studies of the proportion exposed and unexposed to the drug in groups with and without the condition.
Common and important syndromes induced by drugs
CNS: acute toxic confusion is a common clinical problem that is manifest as disorientation, hallucinations and altered consciousness. The risk of acute toxic confusion (‘delirium’) is greater in patients who already have cognitive impairment, notably the elderly and those with incipient hepatic encephalopathy. It can occur as a consequence of drug administration, and as part of the withdrawal syndromes for ethanol, benzodiazepines and other sedative drugs. With ethanol, symptoms of psychomotor agitation, tremor, sympathetic activation, hallucination and seizure are manifest within 1 or 2 days of cessation of drinking. Major causes of drug-induced acute toxic confusion are listed in Figure 3.
Some drugs that can cause acute toxic confusion
Administration • Anticholinergic drugs, including tricyclic antidepressants, oxybutynin and tolterodine • Antiparkinsonian agents, including those containing L-dopa, and dopamine agonists • Cannabis and nabilone • Corticosteroids in supraphysiological doses (‘steroid psychosis’) • Digoxin • Histamine H2-antagonists and proton pump inhibitors • Opioids, including dihydrocodeine, pentazocine and tramadol • Stimulant drugs of abuse, including amfetamines, cocaine and ecstasy Withdrawal • Barbiturates • Benzodiazepines • Ethanol 3 Respiratory system Asthma – drugs can cause both acute and chronic respiratory disease. Asthma induced by β-adrenoceptor antagonists remains a concern, and these drugs should not be used in patients with moderate or severe airways obstruction. The benefits after MI may outweigh the risks in patients with mild asthmatic symptoms. Aspirin for secondary prevention of cardiovascular disease and non-steroidal anti-inflammatory drugs (NSAIDs) for analgesia may worsen asthma. About 10% of asthmatic patients develop bronchospasm after challenge with aspirin, but most such patients know that they are susceptible. It is therefore necessary to ask whether the patient is ‘allergic’ to aspirin or similar drugs, but not to withhold treatment for those with asthma who lack a history of aspirin or NSAID sensitivity. Pulmonary fibrosis is classically associated with cytotoxic agents (particularly busulfan and bleomycin) and the antibacterial agent nitrofurantoin. Lung disease develops in about 5% of patients treated with the anti-arrhythmic amiodarone; this drug can cause early hypersensitivity pneumonitis, and a more chronic fibrotic reaction associated with dense nodules in the lung periphery that is thought to result from deposition of iodine from the drug in clusters of pulmonary macrophages (Figure 4). Some patients (particularly those in whom pneumonitis predominates) recover with
4 High-resolution CT scan of the thorax in a patient who suffered increasing shortness of breath while taking amiodarone, 200 mg daily, to suppress ventricular tachycardia. There is widespread interstitial fibrosis (‘honeycomb lung’). corticosteroids, but amiodarone-induced pulmonary fibrosis can be fatal. There is increasing recognition of pulmonary fibrosis associated with ergot alkaloids and derivatives, including bromocriptine and pergolide, and with methotrexate. Gastrointestinal tract NSAIDs – drugs that non-selectively inhibit prostaglandin synthesis increase the rate of peptic ulcer bleeding and complications in treated patients. Ibuprofen in standard doses is less likely to cause peptic ulcer disease than diclofenac or naproxen, but these are substantially safer than azapropazone or piroxicam. Selective cyclo-oxygenase 2 (COX-2) inhibitors have been developed in an attempt to reduce the incidence of ulcer complications while maintaining analgesic efficacy. A randomized controlled trial in patients with rheumatoid disease suggested that use of rofecoxib rather than naproxen reduces bleeding or perforation by about 1 event per 100 patient-years of treatment, but patients were given no anti-ulcer prophylaxis. In a later trial, patients who had already suffered a gastrointestinal haemorrhage fared no better with celecoxib than with diclofenac plus omeprazole. COX-2 inhibitors lack the antiplatelet effects of non-selective inhibitors, and overall serious adverse events may be greater in patients denied the cardiovascular benefits. Cardiovascular system: torsade de pointes is a broad-complex tachycardia in which the QRS axis changes sinusoidally from 0° to 180° to 360° (Figure 5). It is often druginduced, and is more common in women and in patients with hypomagnesaemia or hypokalaemia. The arrhythmia is prefigured by prolongation of the QT-interval on resting ECG, which represents an abnormality of repolarization. Torsade de pointes occurs with many drugs, including the conventional anti-arrhythmics disopyramide and quinidine. Sotalol (a β-adrenoceptor antagonist with class 3 anti-arrhythmic actions) and amiodarone are also potential causes. Phenothiazines, butyrophenones and some tricyclic antidepressants carry a risk of torsade de pointes; droperidol is no longer marketed, and use of thioridazine has recently been restricted because the risks are greater than with similar drugs. The risk is also greater with the nonsedating antihistamines astemizole and terfenadine (a pro-drug of the active metabolite fexofenadine) and the prokinetic agent cisapride. Diuretic agents that cause hypokalaemia or hypo-magnesaemia can cause secondary torsade de pointes. Bone marrow: bone marrow suppression or aplasia, agranulocytosis and aplastic anaemia are important adverse effects of several commonly used treatments, and drugs are an important cause of the syndrome. Patients can present with mouth ulceration or neutropenic sepsis in agranulocytosis, petechiae or haemorrhage from thrombocytopenia, or anaemia. Common or important causes include clozapine, carbimazole and other antithyroid drugs, carbamazepine, sulphonamides and cytotoxic agents. Endocrine: drugs can alter glucose homeostasis by effects on insulin secretion, insulin action and hepatic glucose production. The antiprotozoal agent pentamidine, used to treat Pneumocystis carinii pneumonia, is toxic to pancreatic β-cells and can cause hypoglycaemia by stimulating release of preformed insulin, followed by hyperglycaemia as a consequence of impaired insulin production. Thiazides (including the antihypertensive diazoxide) and the growth hormone analogue somatostatin inhibit insulin secretion and increase blood glucose concentration; quinine and sulfamethoxazole (in co-trimoxazole) stimulate insulin secretion and reduce blood glucose concentration. Glucocorticoids inhibit post-receptor signalling and cause insulin resistance.
Torsade de pointes
Torsade de pointes is a broad-complex tachycardia in which the QRS axis varies sinusoidally. Drugs that can cause torsade de pointes include phenothiazines, cisapride, some selective histamine H2-antagonists and drugs causing hypomagnesaemia.
A syndrome of subcutaneous fat wasting in the face, limbs and buttocks associated with abdominal visceral obesity, dyslipid-aemia and insulin resistance has been described in patients with HIV infection treated with protease inhibitors such as indinavir, ritonavir and saquinavir. The protease inhibitors may reduce the effective concentration of sterol regulatory element-binding protein-1c (a key regulator of lipoprotein metabolism), thereby causing this ‘metabolic syndrome,’ which leads to diabetes mellitus and premature cardiovascular disease. Neuropsychiatric function Serotonin syndrome was first described in experimental animals that developed hyperpyrexia, myoclonus and autonomic dysfunction when given toxic doses of Ltryptophan, a precursor of serotonin. It was subsequently recognized in patients treated with drugs that enhance serotoninergic activity in the brain (Figure 6). Diagnostic features in response to a serotoninergic agent include: • changes in cognition or behaviour (e.g. agitation, confusion) • autonomic disturbances, manifest as fever, sweating or diarrhoea • neuromuscular dysfunction (e.g. tremor, myoclonus). Other causes must be excluded. Serotonin syndrome usually occurs soon after serotoninergic treatment is instituted or increased.
Some causes of serotonin syndrome
• Increased serotonin synthesis • Reduced serotonin metabolism • Increased serotonin release • Reduced serotonin re-uptake
Monoamine oxidase inhibitors type A and B Ecstasy, cocaine Selective serotonin re-uptake inhibitors, tricyclic antidepressants, mirtazepine, nefazodone, venlafaxine, tramadol, pethidine (meperidine) Buspirone
• Serotonin receptor agonists • Dopaminergic agents 6
Neuroleptic malignant syndrome is associated with dopamine antagonists such as haloperidol and chlorpromazine, and tends to occur over days or weeks of treatment. It is associated with rigidity and extrapyramidal signs in addition to the features of serotonin syndrome.
Bombardier C, Laine L, Reicin A et al. Comparison of Upper Gastrointestinal Toxicity of Rofecoxib and Naproxen in Patients with Rheumatoid Arthritis. VIGOR Study Group. N Engl J Med 2000; 343: 1520–8. British National Formulary. London: British Medical Association, Royal Pharmaceutical Society of Great Britain. (www.bnf.org) Chan F K, Hung L C, Suen B Y et al. Celecoxib versus Diclofenac and Omeprazole in Reducing the Risk of Recurrent Ulcer Bleeding in Patients with Arthritis. N Engl J Med 2002; 347: 2104–10. Davies D M, Ferner R E, de Glanville H. Davies’s Textbook of Adverse Drug Reactions. 5th ed. London: Chapman & Hall, 1998. Dukes M N G, Aronson J K. Meyler’s Side Effects of Drugs. 14th ed. Amsterdam: Elsevier, 2000. Evans W E, McLeod H L. Pharmacogenomics – Drug Disposition, Drug Targets, and Side Effects. N Engl J Med 2003; 348: 538–49. (www.pneumotox.com/index.html) Lo J C, Mulligan K, Tai V et al. ‘Buffalo Hump’ in Men with HIV-1 Infection. Lancet 1998; 351: 867–70.
Copyright © 2003 The Medicine Publishing Company Ltd
Drug Development and Regulation
Duncan Richards is Clinical Lecturer in the Department of Clinical Pharmacology at the University of Oxford, UK. He qualified from the University of Oxford, and trained in London and Oxford. His research interests include development of novel techniques to assess pharmacodynamic activity in early-phase clinical trials.
What’s new ?
• The UK Medicines Control Agency and Medical Devices Agency have merged to form the Medicines and Healthcare products Regulatory Agency • The European Clinical Trials Directive will have a significant impact on the conduct of clinical trials in the UK • Molecular biology techniques are changing the way in which drugs are developed • Many new drugs (e.g. monoclonal antibodies) cannot be developed according to the traditional scheme, and are creating new challenges for drug developers and regulators
Who develops drugs?
To many, the phrase ‘drug development’ conjures an image of large, multinational pharmaceutical companies conducting phase III therapeutic trials involving several thousand patients. Although pharmaceutical companies are the force behind the most commonly prescribed new drugs in Western countries, there are se-veral other key stakeholders in the process. Universities and other research establishments are often the source of original research from which ideas for new therapies emerge. In the past, ‘blue-sky’ research and the applied techniques required for drug development were segregated, but this is changing. Increasingly, the drug development process is managed by a partnership between pro-fessional researchers and professional drug developers (usually the pharmaceutical industry). Some drugs are not commercially attractive because the disease they treat is very rare or, more commonly, because the disease is prevalent only in countries unable to afford the high cost of new drugs. In countries where annual health-care spending is about $1 per capita, use of a drug costing $30 per month is impossible. There are two principal approaches to this problem. • In the orphan drug programme, a government (usually the US Government acting through the Food and Drug Administration) subsidizes the cost of drug development undertaken by a commercial company. This has been most effective in rare diseases. • The problem of developing inexpensive treatments for developing countries is less easily solved. There are several programmes sponsored by governments and international agencies such as the WHO. Drug development in these circumstances is difficult because of limited infrastructures for clinical assessment and data capture, and the complexities of international politics.
Plants and animals: substances extracted from plants and ani-mals remain among the most potent and widely used drugs (e.g. morphine from the opium poppy, digitalis from the foxglove, curare from the bark of a South American tree – Figure 1). Raw extracts from plants and animals often contain a cocktail of active ingredients. Early drug development focused on identifying the most important ingredients and purifying them. It is usually easier to predict the effects of a drug, both beneficial and adverse, when dealing with a single molecule rather than a complex mixture. For example, a series of experiments on dogs showed that removal of the pancreas rendered them diabetic. Further work identified and purified insulin from the pancreas, and the first insulin extracted in this manner was given to a boy with diabetes in 1922. Drug development from animals and plants is continuing by two means: • investigation of traditional remedies, seeking evidence of a drug effect • screening of extracts against batteries of biological and genomic test systems, seeking a drug action (e.g. paclitaxel, the first of the taxane class of chemotherapeutic drugs, was originally extracted from the bark of the Pacific yew, Taxus brevifolia – Figure 1).
O O O N H OH HO
O O H O O
O X CH 3 H3C HO HO O H3C O HO H3C O O H3C HO O O H H H
O O O
Digoxin X = OH Digitoxin X = H
The foxglove (Digitalis purpurea) is the source of the cardiac glycoside digitoxin. (Source: Bodleian Library, Oxford.)
The bark of the Pacific yew (Taxus brevifolia) is the original source of paclitaxel. (By courtesy of Natural Resources Canada, Canadian Forest Service.)
Molecules extracted from plants and animals are often large and complex, and are therefore difficult to manufacture. Obtaining an adequate supply is often a limiting factor in early drug development; for example, 2000 mature yew trees would be required to produce 1 kg of paclitaxel if the drug were produced by extraction. Empirical chemistry and applied pharmacology have been a fertile source of drug development. If the active components of a molecule can be identified, a simpler molecule with similar pharmacological properties can be designed. The molecule may then be further modified to improve, for example, its receptor specificity or how well it is absorbed. This is a laborious process involving synthesis of a range of related compounds, purification, and testing in a pharmacological model (e.g. an isolated muscle preparation). The new technique of combinatorial chemistry enables relatively rapid production of a wide range of related chemical structures. The rate-limiting factor is now the development of suitable systems in which to test the compounds. Physiology: improvements in understanding of physiology and pathophysiology over the last 75 years have led to important drug discoveries. For example, identification of the pivotal role of dopamine in Parkinson’s disease has driven drug development in that field (Figure 2). Rational drug design: a wide range of technologies is available for drug design. If a target has been characterized, its properties can be investigated using computers that can predict the structure of the binding site of a receptor in three dimensions from its amino acid sequence. Using this information, the sorts of molecules that will bind with greatest affinity to that site can be inferred.
Pharmacology and physiology in the treatment of Parkinson’s disease
Pharmacology Physiology Clinical effect
1950s – dopamine identified as a neurotransmitter
Administration of a precursor of dopamine (L-dopa) to patients with Parkinson’s disease
1960s – dopamine is deficient in the brains of patients with Parkinson’s disease
Dopa-decarboxylase inhibitors that do not enter the brain
Adverse effects from peripheral actions of dopamine
Contribution of acetylcholine to basal ganglia function
Combination medicines – co-beneldopa and co-careldopa
1990s – catechol-o-methyl transferase inhibitors
Recognition of other pathways of dopamine metabolism
Large pharmaceutical companies now have many thousands of molecules available to test against potential targets. The rate-limiting factor is the generation of potential and validated targets. Information from the human genome project suggests that only a few traditional transmembrane receptors remain to be identified. New drugs will have to target other molecules such as enzymes and second-messenger molecules. These are more challenging targets because they are likely to have many actions and are found in many cells of the body. Pharmacogenomics: genomic technologies could potentially identify patients who are most likely to respond to new drugs and those most at risk from adverse effects, but such individually tailored treatment remains a distant goal. Genes that influence the disease process may not be suitable drug targets. The process of identification of gene products and their pattern of expression and influence on disease biology is protracted and rewards are not guaranteed. Molecular biological techniques are achieving more immediate benefits. Human epidermal growth factor receptor 2 (HER2) is over-expressed in one-third of breast cancers. Trastusuzmab is a monoclonal antibody directed against these receptors and is effective only in women who over-express HER2. Identification of HER2 status using molecular techniques is now helping to direct treatment appropriately.
Traditional phases of drug development
Drug development is a long process, commonly taking 12–15 years from discovery to licence. A further 3–5 years may be required to determine the drug’s optimal clinical role. The nature and amount of information required for licensing is well established for chem-ical compounds aimed at specific receptors. The traditional model shown in Figure 3 is divided into distinct phases. One of the greatest challenges facing regulatory authorities and drug developers is defining what must be known about new drugs (e.g. growth factors, antibodies) that do not fit neatly into this scheme. Preclinical toxicology and pharmacology: every potential new drug must undergo a period of testing before it can be given to humans. Traditionally, this involves toxicology studies in rodent and non-rodent species to broadly determine the maximum tolerated dose (following single and multiple administration) and likely areas of toxicity. Other studies include effects on fertility and reproduction, teratogenicity, mutagenicity and
Traditional phases of drug development
Permission to proceed from regulatory authority
Submission for approval
Pharmacological evaluation and toxicology
Evaluate safety and pharmacokinetics in humans
Evaluate safety and pharmacokinetics in target population May include some evaluation of therapeutic drug action
Evaluate effectiveness in selected population Evaluate safety in longer-term use
Evaluation of safety and efficacy by regulatory authority Evaluation of quality of manufactured product
Post-marketing studies New indications New formulations
20–80 healthy volunteers
100–300 patient volunteers
1000–3000 patient volunteers
2000–10,000 patient volunteers
Compound patent for 17 years 4 years 5000 compounds 1 year
12 years in development 2 years 5 compounds enter trials 4 years 1 year
5 years exclusive marketing
1 compound approved
This is a general guide only; the nature of the stages can vary considerably according to the drug under development
3 carcinogenicity. However, information from animals can be of only limited relevance to humans. Large numbers of animals are required for preclinical toxicology testing, and considerable efforts are made to find other models that will yield useful information. These models (e.g. cultured human cell lines) may be more relevant when investigating drug pharmacology, but whole-body models are still required to examine how a drug is absorbed and where it goes in the body. Phase I: a risk greater than minimal is not acceptable in a healthy volunteer study. The purpose of preclinical testing is to pro-vide sufficient information to allow administration to humans. A ‘first time in man’ (FTIM) study is usually designed to assess the safety and tolerability of a single dose of the new drug, linked to pharmacokinetic data. The starting dose is usually one-hundredth to one-tenth of a measure of pharmacological or biological acti-vity in the most relevant animal species, from which the dose is increased. The number of volunteers is small and they are monitored closely. When appropriate, other early studies examine the safety and tolerability of the drug after multiple dosing, and the effects of food and other drugs. The principal purpose of these studies is to establish safety, but they also provide the first opportunity to study markers of drug action in humans. A problem with the traditional approach to clinical trials of new drugs is they are conducted only in men, because of concerns about teratogenicity. There is now a move towards including many more women in all clinical studies, not just those for diseases such as breast cancer. Phase II trials are concerned with assessment of the safety and tolerability of the new drug in the target patient population. Correct selection of the study population is
important – the tolerability of a new drug for Alzheimer’s disease may differ between 20-year-olds and patients aged 60–70 years. Some drugs (e.g. cytotoxic drugs) cannot safely be given to healthy volunteers, and the information about the drug that is usually collected from healthy volunteers in phase I must be collected in phase II studies. Many of these studies are in vulnerable groups (e.g. those with cancer), and it is important that these patients understand that they are unlikely to derive any personal benefit from these studies. Phase II trials may evaluate some measure of therapeutic action, but their principal purpose is to define the dose or dose range to be used in phase III. Phase III studies are those with which health-care professionals are most familiar. They are usually large scale and expensive (each costing many millions of pounds). Assessment of safety is a key part of these studies, but they are also the first opportunity to determine whether the drug has beneficial clinical effects in a patient population. Careful patient selection is essential, but limits the applicability of the findings of these studies. The results of phase III trials usually form the basis of marketing and advertising information. It should be remembered that the patients involved were selected to show the maximum effect of the drug, and doctors should consider whether the findings of these studies apply to patients in their clinic. A large phase III clinical trial may involve several hundred investigators and several thousand patients. There are strict, legally enforceable rules regarding the responsibilities of investigators in clinical trials (Figure 4), and clinicians should not enrol patients into clinical trials unless they are able to comply with these rules.
Responsibilities of investigators and other staff according to Good Clinical Practice
The designated Investigator (usually a senior member of staff) is bound to ensure that the conduct of the trial is compliant with Good Clinical Practice at that site. Space does not permit inclusion of the complete guidelines, but if you are asked to help with the conduct of a clinical trial, consider the following points. • Before the start of the trial, an independent ethics committee must have approved the protocol and associated documents. Check that this permission has been received before you enrol a patient. • The Investigator should maintain a list of appropriately qualified persons to whom he or she has delegated significant trial-related duties. You should not undertake significant duties (e.g. enrolment of patients) unless your name is on that list. • The Investigator should ensure that all individuals assisting with the trial are adequately informed about the protocol, the investigational product and their trial-related duties. Do not enrol patients into trials if you are not familiar with the investigational drug or protocol. You cannot obtain informed consent from a patient if you are not familiar with the trial yourself. • Follow the protocol precisely. Do not deviate from the protocol unless there is immediate danger to the patient. If you have to make a permanent change to the protocol, the sponsor and the ethics committee must approve this. • Take care to complete the case record form. Missing or inaccurate data may invalidate the trial, exposing patients to risk without any benefit from the increase in knowledge. If you have to alter the case record form, strike it out with a single line and sign and date the change. • There are special rules regarding rapid reporting of serious adverse events during clinical trials to the regulatory authorities. If your patient suffers an adverse event, contact the sponsor quickly and file a report, regardless of whether you think it was related to the drug.
Phase IV trials are undertaken after a drug has been given some form of approval for use by the appropriate regulatory body. They are usually of larger scale than phase III trials. A drug may have only one licensed indication, such as a β-blocker for hypertension, but a commercial company may sponsor other trials to evaluate its use in myocardial infarction and heart failure. The licensing authority may consider these trials and, if the results are accepted, extend the licence to cover these indications. A phase IV trial may be required for a new formulation of a drug (e.g. a fast-melt sublingual tablet). Commercial companies are often wary of trials comparing their drug with a
competitor, fearing that their drug will be found inferior. Such trials may be sponsored by non-commercial organ-izations (e.g. UK Medical Research Council, Wellcome Trust).
What is a product licence?
To sell a medicine in any country, the sponsor must convince the licensing authority of that country (see below) that it is safe, effective and manufactured to a satisfactory quality. This does not mean that the licensing authority or the sponsor know everything about that medicine. New drugs have been given to, at most, only a few thousand individuals before they are granted a licence. Rare adverse effects, even serious ones, are often not recognized before a licence is issued. In the British National Formulary, new drugs are identified by a black triangle; clinicians should report any adverse effects in patients given these drugs, even if minor and regardless of whether they think the drug was responsible. Do not assume that the licensing authority and manufacturer are already aware of effects observed. To obtain a licence, a new drug must be effective for its indication; it does not have to be better than alternatives, nor does it have to be cost-effective. It may take years and several phase IV trials before the place of a new drug is established. Initially, a new drug should be used only by specialists familiar with the alternatives; as its role becomes established, wider use may be appropriate. Medicines are granted a licence for a given indication only. A company cannot promote a drug for indications for which it does not have a licence. Physicians may use approved drugs outside their licence, and in many cases this is appropriate. However, it is good medical practice to use drugs for their licensed indications. Widespread use outside the licence does not encourage commercial sponsors to evaluate the drug’s use in robust clinical trials.
Licensing new medicines
In the UK, the licensing of drugs is based on the Medicines Act (1968), which provides a comprehensive system of licensing affecting the manufacture, sale, supply and importation of medicinal products. The Medicines and Healthcare products Regulatory Agency (MHRA) (formerly the Medicines Control Agency) is an executive agency of the Department of Health and is responsible for implementation of the provisions of the Act. The MHRA is supported by seven specialist advisory bodies (Figure 5). Licences to manufacture, market, distribute, sell and supply medicinal products are granted by Ministers (‘the Licensing Authority’), who are responsible to Parliament. In practical terms, the licensing authority is the MHRA. Working with the Committee on Safety of Medicines (CSM), it assesses the safety, efficacy and quality of each product (Figure 6). Until recently, companies had to apply for licences in each country separately, and the standard and nature of information required varied considerably between licensing authorities. In 1995, a new European Community (EC) system for the author-ization of medicinal products came into operation. Its purpose was to allow medicines marketed in one EC country to be made available in the same manner, and subject to the same conditions, in other EC countries. The EC licensing system uses the experience of the individual Member States’ authorization through a system of mutual recognition. In addition, there is a centralized system for authorization. The European Agency for the Evaluation of Medicinal Products has been established to provide additional support and to coordinate Member States’ work. EC decisions agreed under the new system are binding on Member States.
Post-marketing evaluation and surveillance
In addition to licensing new medicines, the MHRA is responsible for: • monitoring medicines and acting on safety concerns after they have been placed on the market • checking standards of pharmaceutical manufacture and wholesaling • enforcing requirements • medicines control policy. The CSM evaluates concerns about the safety of medicines. Acting through the MHRA, it can enforce the suspension or withdrawal of a licence. The bulletin Current Problems in Pharmacovigilance is produced four times per year and sent to all doctors, dentists, pharmacists and coroners in the UK. It alerts them to problems with medicines and provides advice on their safe use.
Bodies created as a result of the Medicines Act (1968) and their roles
Executive agency Advisory body Medicines Commission Terms of reference Advises the Health and Agriculture Ministers of the UK ‘on matters relating to the execution of this Act or the exercise of any powers conferred by it, or otherwise relating to medicinal products, where either the Commission consider it expedient, or they are requested by the Minister or Ministers in question to do so’.
Committee on Safety of Medicines
• Provides advice to the Licensing Authority on whether new products (new active substances) submitted to the MHRA should be granted a marketing authorization. These responsibilities require close collaboration with the MHRA’s Licensing Division. • Monitors the safety of marketed medicines, in close association with the MHRA’s Post-Licensing Division, to ensure that medicines meet acceptable standards of safety and efficacy.
British Pharmacopoeia Commission
Medicines and Healthcare products Regulatory Agency (MHRA)
Responsible for preparing new editions of the British Pharmacopoeia and the British Pharmacopoeia (Veterinary) and for keeping them up to date. (These are different from the British National Formulary, which is a joint publication of the British Medical Association and the Royal Pharmaceutical Society of Great Britain. The Department of Health is represented on the Joint Formulary Committee that produces the British National Formulary.)
Veterinary Products Committee
• Gives advice on safety, quality and efficacy relating to the veterinary use of any substance or article (not instruments, apparatuses or appliances) to which any provision of the Medicines Act is applicable. • Promotes collection of information on suspected adverse reactions for the purpose of giving such advice.
Advisory Board on the Registration of Homoeopathic Products
Gives advice on safety and quality relating to any homeopathic medicinal product for human use for which a certificate of registration could be granted, and any homeopathic veterinary product that satisfies the conditions in Article 7 of the Veterinary Homoeopathics Directive (92/74/EEC).
Independent Review Panel for Advertising
• Considers written representations from pharmaceutical companies regarding the conformity of their advertising and promotional material with the Regulations. • Advises Health Ministers on the conformity of advertising and promotional material with the Regulations before a final decision is made by Health Ministers.
Independent Review Panel on the Classification of Borderline Products
• Considers written and oral representations from companies against a provisional determination by the MHRA, on behalf of the Licensing Authority, that a product is a medicinal product. • Advises the Licensing Authority whether the MHRA’s provisional determination should be confirmed.
European Clinical Trials Directive
The European Clinical Trials Directive (2001/20/EC) has implications for every healthcare professional involved in clinical re-search. It was agreed in February 2001, and the UK is currently preparing to transpose it into domestic law. The scope of the legislation is wide and every clinical trial involving medicinal products will be covered. The Directive sets standards for pro-tecting clinical trial subjects, establishes ethics committees on a legal basis, provides legal status for certain procedures, establishes
Framework for the licensing of new medicinal products in the UK
European legislation UK Parliament Secretary of State Department of Health Executive agency
European Agency for the Evaluation of Medicinal Products
Medicines and Healthcare products Regulatory Agency
Committee on Safety of Medicines
Centralized authorization procedure and mutual recognition
6 standards for the manufacture, import and labelling of investi-gational medicinal products, and provides for quality assurance of clinical trials. The Directive does not distinguish between commercial and non-commercial clinical trials, and there is considerable concern that non-commercial academic research will be hampered by its requirements. In addition, the definition of a medicinal product can be wide; for example, vitamins given for therapeutic use could be included. A small clinical investigation could require full submission to the licensing authority, similar to that required for a new potential commercial product. It remains to be seen how the individual Member States interpret the provisions of the Directive, but the potential effect on academic research is great.
Committee on Safety of Medicines. www.mhra.gov.uk/aboutagency/regframework/csm/ csmhome.htm (Advice on reporting suspected adverse events and information on current problems in pharmacovigilance.) Declaration of Helsinki as amended by the World Medical Association. www.wma.net (The Declaration of Helsinki is a statement of ethical principles guiding physicians and other participants in medical research involving human subjects.) Grahame-Smith D G, Aronson J K. Oxford Textbook of Clinical Pharmacology and Drug Therapy. Oxford: Oxford University Press, 2002. (Includes an overview of the phases of drug development.) International Conference on Harmonization. Guidelines for Good Clinical Practice. www.ich.org/ Medicines and Healthcare products Regulatory Agency. www.mhra.gov.uk (Background information on the structure and function of the MCA, with contact details and application forms for those wishing to conduct clinical trials.) Research on Healthy Volunteers. A Report of the Royal College of Physicians. J R Coll Physicians Lond 1986; 20: 243–57. (Guidelines on the conduct of studies involving healthy volunteers.)
Copyright © 2003 The Medicine Publishing Company Ltd
How to Appraise Clinical Trials
Y K Loke is Lecturer in Pharmacology at the University of Oxford, UK. He qualified from the University of London, and trained in clinical pharmacology and general medicine in Oxford. His research interest is systematic review of adverse drug reactions.
What’s new ?
• Organizations such as the UK National Institute for Clinical Excellence use trial data to support decisions on the usefulness and cost-effectiveness of drugs • Several important components of the assessment of trial validity have now been defined • In view of the current profusion of trial results, doctors must be able to judge for themselves how relevant the findings are to their practice In evidence-based medicine, trial data are of paramount importance when choosing appropriate treatments for patients. Doctors rely on clinical trials to direct, or change, their practice; thus, the ability to appraise a trial report critically is a useful skill. It is important that the evaluation of a trial goes beyond a cursory look at the abstract and results sections (as most busy clinicians are prone to do). Critical assessment of a trial begins at the design phase, long before the first patient is recruited into the study. It is necessary to ask the following questions, both at this point and later. • What is the trial for? Are there clear therapeutic questions that must be addressed? • Can the design and conduct of the trial reliably answer those questions? These issues are of particular interest to pharmaceutical phy-sicians, research ethics committees, funding bodies and regulatory authorities. They are also useful starting points for clinicians, in assessing the relevance or value of the trial’s findings. Doctors should consider the following points, in particular. • What do these results mean for me, and for my patients? • What influence, if any, should the findings have on medical practice? Individual clinicians must judge whether the trial data are of any relevance to the patients they treat in ‘real-life’ practice. At a higher level, medicines advisory committees, including the UK National Institute for Clinical Excellence, use trial results as essential supporting evidence when deciding on the usefulness and costeffectiveness of drugs in the national health setting. Bearing in mind the above, readers of trial reports should systematically evaluate the following areas: • design and conduct of the trial • applicability of trial findings to clinical practice.
Design and conduct of controlled clinical trials
Most pharmacological treatments are now tested in controlled clinical trials before they are licensed for use. The essence of a controlled trial is that, to assess the effects of therapy, two or more patient groups with similar characteristics are exposed to different treatments. Such trials can be considered as scientific experiments on humans and should therefore be conducted according to rigorous methodological standards. As in any experiment, the scientific integrity of the study, and the reliability of the results, can be undermined by the presence of bias. In simple terms, bias is any process (conscious or un-conscious) that causes results to systematically deviate from the true values. There are several important areas in which bias may occur in a controlled clinical trial: • selection of patients • delivery of the treatments that are being evaluated • assessment of treatment outcomes • loss of patients to follow-up. Readers should therefore pay special attention to these aspects of trial reports and should check that the trialists have minimized the possibility of bias.
Selection of patients
Ideally, the patient groups under comparison in a controlled clinical trial should have identical characteristics and differ only with respect to the treatment arms to which they belong. To achieve such a balance, each trial participant must have an equal chance of being assigned to any one of the treatment groups. This is achieved through the process of random allocation – neither the doctor nor the patient knows, or has any influence on, the treatment group to which the patient is allocated. There are two important steps in ensuring that the randomization process is adequate: • generating a truly random sequence, often using a computer or random number tables (failing this, drawing numbers from a hat or flipping a coin must suffice) • ensuring that neither the trialists nor the patients can determine the sequence, to prevent them influencing the treatment allocation process (can be achieved by using a remote telephone randomization centre, or simply with sealed, opaque envelopes). Most trial reports now list the baseline characteristics of the patients in their treatment groups, and it is worth checking that the groups are indeed similar. There are many studies in which inadequate randomization techniques led to an imbalance, in-cluding the following. • Women in a labour ward were asked to choose whether they wanted to use a birthing pool instead of the conventional delivery room. However, the birthing pool group thereby contained more women of higher socioeconomic status who were interested in this innovative method. It was impossible to tell whether the better health of the babies in that group resulted from use of the birthing pool or from their mothers’ higher socioeconomic status. • Patients were openly enrolled into treatment groups depending on the day of admission. However, those admitted on a Sunday, for example, could have differed from those referred on a weekday, and this could lead to an imbalance in the groups. Furthermore, the trialist or the patient could choose his or her preferred treatment by arranging hospital admission for a specific day; for example, frail patients may have preferred what appeared to be the ‘gentler’ treatment arm.
Delivery of the treatments under comparison
Ideally, patients in each group should be managed in exactly the same manner, the only difference being the therapeutic agents under evaluation. This may not always be the case, however, and readers should check the trial report. • In one study, an experimental drug was administered in the coronary care unit, whereas patients in the conventional therapy arm were cared for on general medical wards. Improved outcomes in the experimental group may have been a consequence of closer supervision in coronary care, rather than the drug treatment. • Patients in a study of a new endoscopic device were treated by a consultant who had undergone special training in the technique. Meanwhile, other patients in the trial underwent the conventional procedure performed by a registrar.
Measuring treatment outcomes
Bias may not be a problem when measuring ‘hard’ outcomes such as numbers of patients dead or alive at the end of the trial. However, it may occur with outcomes that are subject to human interpretation such as deciding the cause of death, reading an echocardiogram or assessing symptomatic change. In a trial of the benefit of compression stockings in the prevention of travel-related thrombosis, calf vein clots were monitored by ultrasonographers who knew which patients had used stockings. These technicians may have believed that patients without stockings were at higher thrombotic risk, leading to more thorough scanning of these individuals and highlighting of borderline abnormalities. Blinding or masking of treatments has been introduced to remove this type of bias. In double-blind studies, neither the trialist nor the participant knows which treatment regimen is given. It is worth checking that blinding of treatment is feasible, however; for example, the reader should be sceptical about blinding in a study comparing a rectally administered drug with another drug that must be given through a central venous catheter.
Loss of patients to follow-up
There are many reasons why patients leave trials. Some develop adverse effects; others give up because they feel no better on the trial treatment. The results of trials may be misleading if these drop-outs are not accounted for (Figure 1). This is because
Hypothetical example of a new anti-obesity drug being evaluated in 1000 patients
1000 obese patients start new drug n = 1000 450 patients drop out after 2 months because they experienced no beneficial effect on their weight n = 550 200 experience adverse effects and decide not to continue n = 350 Remaining 350 are happy taking the drug and feel that it is effective in helping them lose weight
Final mean weight loss is based on the 350 patients who complete the trial
Conclusion: the new drug is both effective and well tolerated
the remaining patients are not representative of those who originally started on treatment. To remove this type of bias, ‘intention-to-treat’ analysis is undertaken. All randomized patients are included in the analysis, regardless of whether they completed the trial. If such analysis is impossible, the trial should report the numbers who left and the reasons why. Some trials use a ‘run-in’ period, in which all eligible patients are briefly treated with the study drug before the randomization phase. In one study, for example, 10,500 patients were initially given ramipril, 2.5 mg, but 1000 of these were subsequently excluded from formal participation in the trial for various reasons including noncompliance, adverse effects and deteriorating serum biochemistry. Doctors must be cautious when interpreting adverse effects data from trials that use a run-in phase. Many medical journals now use a standardized reporting structure for randomized controlled trials, with a flowchart (Figure 2) that provides a quick, simple means of assessing the trial’s design and conduct.
Are trial findings relevant to the patients in everyday clinical practice?
For busy clinicians, this is perhaps the most important question to answer when reading a trial report. A trial may be a paragon of scientific integrity, but the patients may have been so highly selected, and the therapy so expensive and complex, that no hospital could hope to deliver such care in real-life practice. When deciding whether the study findings have any validity outside the trial setting, it is helpful to consider the following: • nature of the patients selected to participate • treatment regimens • health-care setting • type and definition of outcome measures.
Selection of participants
The entry and exclusion criteria for clinical trials are often set in such a manner that they select participants who will fare well and exclude those at substantial risk of
CONSORT flowchart showing the details that are required in a randomized controlled trial
Assessed for eligibility (n = …) Excluded (n = …) • Did not meet inclusion criteria (n = …) • Refused to participate (n = …) • Other reasons (n = …) Randomized (n = …)
Allocated to intervention (n = …) • Received allocated intervention (n = …) • Did not receive allocated intervention (give reasons) (n = …)
Allocated to intervention (n = …) • Received allocated intervention (n = …) • Did not receive allocated intervention (give reasons) (n = …)
Lost to follow-up (give reasons) (n = …) Discontinued intervention (give reasons) (n = …)
Lost to follow-up (give reasons) (n = …) Discontinued intervention (give reasons) (n = …)
Analysed (n = …) • Excluded from analysis (give reasons) (n = …)
Analysed (n = …) • Excluded from analysis (give reasons) (n = …)
2 adverse effects. For example, a trial of warfarin in atrial fibrillation excluded patients aged over 70 years and those judged unlikely to be compliant. Stricter entry criteria may create a better scientific experiment, but the trial’s participants may be unrepresentative of patients in the community. Can doctors safely extrapolate the findings of such a trial to clinical practice, in which most patients with atrial fibrillation are over 70 years of age and in which the benefit:harm ratio is likely to be significantly different? Before recommending a new treatment, it is important to check that the trials were performed in relevant groups and, if not, to consider carefully how patients may differ from those in the trials in response to treatment.
Treatment regimens in the trial
The dose or route of administration of drugs used in clinical trials may differ from that in clinical practice. For example, patients in trials of cholesterol-lowering agents are given near-maximum doses (pravastatin, 40 mg, or simvastatin, 40–80 mg), to make the benefit of treatment as clear as possible. In real-life medicine, however, it is unclear whether the doctor should continue with dose-titration according to response or simply adhere to the trial protocol. This is an important area that must be considered when appraising a trial report. It can also create difficulty in the assessment of the costeffectiveness of treatment.
Health-care settings and provision
Clinical trials are often conducted in well-staffed and highly equipped academic centres. Special arrangements are usually made for follow-up, and trial participants are monitored closely. It is therefore unsurprising that patients in trials often experience better outcomes, even when allocated to the control arm. In practice, the monitoring or expertise required to achieve the trial results may not be readily available outside specialist referral centres. For example, trial data show that thrombolysis is beneficial
in patients with echocardiographically detectable right ventricular dysfunction following acute pulmonary embolism, but it is unlikely to be widely used in clinical practice because few, if any, centres can offer immediate echocardiography in the emergency department. The model of care in trials warrants close scrutiny to determine whether it is viable outside research settings.
Selection of outcome measures
There are two issues involved when considering outcome measures. Surrogate outcome measures are often used in clinical trials, but may or may not accurately reflect true benefit. • Two antihypertensive agents may reduce blood pressure (a surrogate outcome) by equal amounts, but in practice one may be better at reducing the risk of adverse cardiac events (the clinical outcome of interest). • Drugs such as donepezil, given for dementia, may improve the ADAS-Cog score by a few points, but what do these extra points mean for quality of life? • A trial of nicorandil in ischaemic heart disease showed significant benefit in a composite outcome measure (cardiac death, non-fatal myocardial infarction and unplanned admission with angina). However, no significant benefit was found when cardiac death and myocardial infarction were analysed in isolation. The main advantage of this drug, if used in clinical practice, may therefore be in terms of admission to hospital rather than serious myocardial infarction. These examples demonstrate the value of scrutinizing end-points carefully to determine whether the purported benefits are important clinical outcomes. Relative risk and odds ratios – trials often present their data in terms of relative risk (RR) or odds ratios (OR). However, clinicians and patients find it easier to understand and work with measures of absolute benefit such as the number needed to treat (NNT) and absolute risk reduction. The NNT is the estimated number of patients who must be given the therapeutic agent for a given period of time, to achieve one additional beneficial outcome. It is a measure of how much effort is needed to yield a useful endpoint. These measures of absolute benefit change with the underlying baseline risk of the patient, though the relative risk often remains constant (Figure 3). It is worth reading beyond superficially impressive RR or OR figures, and converting these into meaningful absolute measures such as NNT.
Examples of how the same relative risk can yield different numbers needed to treat (NNTs) depending on the baseline risk in the patient, based on the hypothetical drug X, which produces a ‘huge’ relative reduction of 40% in the risk of stroke
Baseline annual stroke risk in a population of patients 1% 5% 20%
Annual stroke risk with drug X 0.6% 3.0% 12%
Absolute reduction in stroke risk with drug X 0.4% 2.0% 8.0%
NNT per year1 250 50 13
NNT = 1/absolute risk reduction
Cook R J, Sackett D L. The Number Needed to Treat: A Clinically Useful Measure of Treatment Effect. BMJ 1995; 310: 452–4. (An important guide to interpreting trial data in a clinically relevant manner.) Juni P, Altman D G, Egger M. Systematic Reviews in Health Care: Assessing the Quality of Controlled Clinical Trials. BMJ 2001; 323: 42–6. (A useful article describing the quality indicators by which a trial can be assessed.) Moher D, Schulz K F, Altman D G. The CONSORT Statement: Revised Recommendations for Improving the Quality of Reports of Parallel Group Randomized Trials. BMC Med Res Methodol 2001; 1: 2. www.biomedcentral.com/1471-2288/1/2 (Widely adopted guidelines on how to report trial data in a structured manner.)
• Treatment decisions should be guided by evidence obtained from clinical trials • The reliability of trial results should be evaluated by looking for potential sources of bias in the design and conduct of the trial • The relevance of trial results to everyday clinical practice should be considered before making treatment decisions
Copyright © 2003 The Medicine Publishing Company Ltd
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.