9.1 Principles of drug therapy: focus on opioids

An understanding of the principles of clinical pharmacology is essential to enable clinicians to prescribe drugs safely and effectively. When prescribing any drug the intention is to gain a therapeutic effect (e.g. the use of opioids to reduce pain), while avoiding harm (e.g. opioid-induced drowsiness). All drugs have the potential to cause side effects, many of which are predictable from knowledge of clinical pharmacology.

Patients with serious or life-threatening illness present challenges to the safe and effective use of medications. Patient-related factors, such as age, gender, and co-morbidities, may influence the choice of drug, starting dose, and route of administration. Most patients receive multiple drugs and clinicians must always be mindful of the risk of drug–drug interactions. Given the risks of polypharmacy in those with advanced illness, options for concurrent non-drug therapy always should be considered. This is certainly important in the case of pain, for which a holistic approach, one that acknowledges that pharmacotherapy is only one strategy among many that may yield favourable outcomes, is often best.

Opioids and other drugs that have the potential for misuse or abuse, addiction, or drug diversion (see Chapter 9.5) are commonly used in palliative care, and these drugs are regulated to a greater or lesser extent in every country. Prescribers may need to make adjustments based on the wider social context implicated in these regulations, and impact of taking medications that may be inconsistently available or arduous to prescribe. In some countries, excessive regulation impedes patient access, particularly in community settings. Clinicians may need to consider whether therapy can be more effectively provided in different care settings (e.g. hospital versus community) or when using specific drugs or formulations, or rates of titration and monitoring. Inclusion of a drug, such as codeine or morphine, on the World Health Organization (WHO) Model List of Essential Medicines (WHO, 2013) increases the likelihood of availability, but does not mean unrestricted access in every country (Junger et al., 2013). Indeed, opioid drugs exemplify the international variation in access that characterizes many types of drugs; some provide their populations with a large number of different opioids and opioid formulations, whereas others fail to ensure access to any. The WHO continues to seek to improve international access to controlled drugs, in particular opioids.

With increasing health-care costs, groups of clinical, pharmacy, and finance professionals are working to promote cost-effective and safer prescribing. In the United Kingdom, efficacy and cost-effectiveness of drugs are assessed at a national level by the National Centre for Health and Care Excellence (NICE). Consideration of cost-effectiveness is now an integral part of the rationale for appropriate prescribing.

Clinical guidelines are recommendations, based on appraisal of the best available evidence. When evidence is limited or absent, expert opinion is necessary to guide and support best practice. For example, most of the evidence-based guidelines for opioid use in palliative care that have been developed by the European Association of Palliative Care are based on weak evidence (Caraceni et al., 2012). In addition to the development of guidelines by professional societies, national guidances (e.g. NICE (2012) or Scottish Intercollegiate Guidelines Network (2008)) have been developed to help inform local, regional, or hospital-specific practices in a manner that considers local variation in resources and practice. A working knowledge of both international guidelines and national guidances is essential.

Formularies are also a useful resource to guide the prescribing and dispensing of medicines. Many provide detailed information about formulations and doses of drugs. If a formulary is promoted or required in practice, it is important to consider the target audience and the context for any recommendations provided. In palliative care, the most widely recognized formulary is the Palliative Care Formulary (Palliativedrugs.com, 2012). This is regularly updated and includes advice about the many drugs that are used off-licence in the palliative care setting. Other generic formularies, such as the British National Formulary (BNF), include a much wider selection of medications (Joint Formulary Committee, 2014). The choice of drugs in local or regional formularies is often restricted due to budget constraints.

Evidence-based medicine (EBM) is centred on the judicious use of current best evidence about the risks and benefits of interventions to inform clinical decision-making. The principles of EBM are discussed in detail in Chapter 19.2. It is important to recognize that the evidence provided by clinical trials evaluates the efficacy of an intervention, such as an opioid, in an ideal/controlled setting. Strict inclusion and exclusion criteria are applied and an enriched or selected sample may be studied, with structured and often short-term follow-up. Conducting high-quality clinical trials in the palliative care setting is especially challenging (Kaasa et al., 2006) and the evidence used to inform prescribing in patients with advanced illness typically originates from studies of relatively healthier populations.

Effectiveness should be distinguished from efficacy. Effectiveness refers to the benefits and burdens of a drug in the wider context, at a population level and as part of everyday practice. Information about efficacy is needed to identify substances that have the potential for clinical benefit; information about effectiveness provides more relevant and actionable information about the likelihood and extent of the therapeutic effect in a given patient.

Cost-effectiveness refers to a comparison of effectiveness for a target indication against cost, that is, the ratio of effectiveness to cost. Clearly, it is more efficient to use the cheaper of two drugs that are equally effective and safe, and this information often informs local guidance.

From the clinical perspective, the expectation of benefit must be compared to the expectation of risk to determine whether treatment or a change in treatment is justified. Although it can be difficult to estimate the balance between the potential risks and benefits of a particular treatment in an individual case, it is necessary to do so before action is taken.

The likelihood of benefit or risk in the individual is informed by an understanding of likely outcomes in the population overall. Two broad measures that can be useful as part of this assessment are number needed to treat (NNT) and number needed to harm (NNH). The NNT estimates the number of patients that would need to be given a treatment for one of them to achieve a desired outcome (e.g. 50% pain relief). The NNH is calculated for adverse effects in a similar way. Although these measures have been criticized because they are derived from controlled clinical trials data, they nevertheless provide a useful point of comparison among drugs (Christensen and Kristiansen, 2006).

The cut-offs in pain relief or severity of an adverse effect used to calculate the NNT and NNH, respectively, are accepted by convention. In fact, perceived benefit or risk may be strongly influenced by individual variation or therapeutic context. For example, patients with severe pain may feel that a 30% or smaller reduction in pain relief is clinically meaningful. In addition, some patients may be willing to tolerate mild/moderate side effects to achieve a small improvement in pain control if this translates into improved function or quality of life. Knowledge of the patient’s drug history in terms of previous success or failures of treatment will also guide choice of drug in prescribing. Prescribers also must always check individual patient factors, such as renal function and known allergies, to further inform assessment of risk/harm and potential for drug interactions.

Information about side effects of a drug must continue to be gathered after the drug is licensed. This is particularly important when a drug is used in the palliative care context, which is usually characterized by patients with advanced illness and the use of multiple drugs adapted for off-licence indications. In these situations, the use of a drug may be associated with a side effect liability quite different than that expected based on clinical trials information.

Clinical pharmacology is broadly divided into pharmacokinetics (‘what the body does to the drug’) and pharmacodynamics (‘what the drug does to the body’). In palliative medicine, drugs are not often used to cure or modify underlying disease but are predominantly focused on improving symptoms. They often are administered with the intent to continue treatment until death. Knowledge of pharmacokinetic variation, between patients and across time as disease worsens, combined with an understanding of the basic modes of drug action, underpins the logical selection and use of the most appropriate treatment.

Pharmacokinetics encompasses the absorption, distribution, metabolism, and excretion of drugs (ADME). Detailed descriptions of each of these processes are available in pharmacology textbooks. To some extent, inter-individual variation in kinetics is genetically determined. Each process also is influenced by many other factors, however, and in the the palliative care setting, these also contribute to large inter-individual variation and the potential for significant changes across time (Box 9.1.1).

At the cellular level, absorption occurs across lipid cell membranes and is a passive process along a concentration gradient. For most drugs, this process takes place in the small intestine. As long as the drug is in solution, has a degree of lipid solubility, and there is sufficient surface area and time for diffusion in the small bowel, then problems should not arise.

A reduced rate of absorption may occur if there is delayed emptying of the stomach. This might arise as part of a pathological process or pharmacological agents that slow gastric motility, such as anticholinergic drugs or opioid analgesics.

Many drugs are now formulated as modified-release preparations, which need to remain in the small bowel for a specified period to achieve the expected absorption profile. In patients with either increased or decreased gastrointestinal transit time, there is a risk that the expected time–action relationship may not materialize. In either case, the prolonged duration or extent of therapeutic effects may be lost.

Absorption and bioavailability are not the same. The bioavailability of a drug is the percentage of administered drug that gains access unchanged to the systemic circulation. Bioavailability is of most clinical relevance after oral administration. Extensive first-pass hepatic metabolism results in relatively low bioavailability and/or large inter-individual variability for some drugs. For example, the bioavailability of oral morphine is just 35% on average and the range is 15–64%, whereas oxycodone has a bioavailability of 75% and range of 60–87%. The difference in bioavailability complicates the challenge of safe dose selection when changing between oral and parenteral routes and is one of the main reasons that dose titration is essential to identify an effective opioid dose.

Bioavailability can be altered by disease processes that affect hepatic function, or by exposure to drugs that either induce or inhibit enzymes of the cytochrome P450 (CYP450) system. In patients with chronic liver disease, for example, blood may be ‘shunted’ from portal to systemic vessels; this bypasses hepatic enzymes, reduces the first-pass effect, and increases bioavailability. These changes in hepatic function may have a profound effect on drug levels after oral administration but relatively little effect when the drug is given parenterally.

The volume of distribution (V  _d) is a theoretical volume in which the total amount of drug would need to be uniformly distributed to achieve the blood concentration. For very lipophilic drugs which are taken up into fat stores or muscle, such as fentanyl, the volume may be many times body size.

The V  _d is important as a determinant of half-life (t  _½) and is also of theoretical importance in the calculation of the loading dose of a drug where one is needed. Changes caused by disease, such as cachexia or renal failure, may shift a drug’s concentration–time relationships. Other related processes with potentially profound effects on drug kinetics or dynamics also may occur as a result of alteration in body composition or physicochemical environment. For example, all opioids are weak bases and dissociate into free-base and ionized fractions when dissolved in solution. The ionized form is active at the receptor site, whereas the free-base form is more lipid soluble. The relative proportions of ionized and unionized drug are dependent on pH and pK  _a, and may change with the effects of disease.

All opioids also bind to plasma proteins, such as albumin and glycoproteins, in varying degrees. Opioid molecules which are unbound and unionized are capable of diffusing to the site of action, the proportion of which is known as the diffusible fraction. The concentration of the diffusible fraction and other factors such as lipid solubility determine the speed of onset of the drug. The diffusible fraction of a drug may change with hypoalbuminaemia associated with advanced illness.

A high lipid solubility facilitates diffusion across the blood–brain barrier into the brain and therefore is associated with a rapid onset of action. However this view is simplistic in that it is the ionized form that is active at opioid receptor. Speed of onset is therefore better represented as a complex function of both lipid solubility and percentage of the drug that ionized at physiological pH. Morphine has a high diffusible fraction but low lipid solubility which results in a slow onset of action. Alfentanil, however, has both a high diffusible fraction and a higher lipid solubility, which together explain the more rapid onset of action. Baseline values for both drugs may shift with varied disease-related factors that alter V  _d, protein binding or the proportion of the ionized form of the drug.

Drug biotransformation takes place mainly in the liver and contributes both to the rate of elimination of a drug and its bioavailability. The rate at which metabolism proceeds usually determines the clearance; however, where removal is particularly rapid (high extraction ratio) the rate of delivery of drug to the liver rather than the rate of metabolism, may determine clearance (flow-dependent kinetics). For such drugs, if liver blood flow is markedly reduced, drug accumulation will result.

The biochemical processes of drug metabolism are complex. Two phases of metabolism are usually described. Phase I reactions involve oxidation, reduction, hydrolysis, hydration, dethioacetylation, and isomerization. Of these reactions, oxidation catalysed by members the CYP450 superfamily of enzymes are the most important and best characterized. Phase II reactions usually involves conjugation; this may take the form of glucuronidation, glycosylation, sulphation, methylation, acetylation, or conjugation with glutathione or certain amino acids. All of the reactions involve the production of products which are more water-soluble and amenable to excretion by the kidney. In some circumstances, phase II reactions may take place without a prior phase I reaction. When phase I reactions do occur, they may prepare the drug molecule for a phase II reaction by producing or uncovering a chemically reactive group, which then forms the substrate for a phase II reaction.

Most opioid metabolism, both phase I and II reactions, occurs in the liver. Hydrophilic metabolites are predominantly excreted renally, although a small amount may be excreted in the bile or unchanged in the urine. Opioid metabolites may be active and contribute to both the overall analgesic and side effect profile (Smith, 2011). Metabolism of individual opioids is shown in Table 9.1.1.

The two major organs of elimination are the liver and kidneys, both of which are susceptible to pharmacological and pathophysiological sources of variability. Clearance is defined as the volume of blood which is completely cleared of the drug in a unit of time and reflects the efficiency of the elimination process. It is usually measured in mL/minute or L/hour. It is a major determinant of t  _½ and of the steady-state drug concentration.

This is perhaps the most well-known and commonly used pharmacokinetic parameter. The elimination half-life (t  _½) is a measure of the time taken for half the drug in the body to be removed and generally correlates closely with duration of action. After repeated dosing is initiated, or the dose of an existing regimen is changed, five to six half-lives are required to approach steady-state concentration, irrespective of the route of administration or dosing interval. Drugs with a long t  _½ accumulate for a relatively prolonged period of time, and as a result, the concentration may surpass the effective therapeutic range and build up to toxic levels. In the clinical setting, this accumulation has largely been a problem during methadone therapy. Methadone is exceptionally complex because its slow elimination phase is highly variable (beta t  _½, 15–60 hours) and preceded by a rapid distribution phase (t  _½, 2–3 hours); the overall half-life is relatively long (> 20 hours).

The aim of any dosing regimen in an individual patient is to achieve a concentration of drug in the blood that is high enough to give the intended effect without producing side effects. This concentration can never be completely steady as peaks will occur at the point of maximum drug absorption after administration, and troughs will occur immediately before each dose (Fig. 9.1.1). The degree of swing between peak and trough concentrations is determined by the drug’s elimination t  _½ and the frequency of drug administration.

The time taken for a drug to reach steady-state plasma concentration is dependent on the half-life. As noted, five to six half-lives are required to approach steady-state drug concentration if the same dose of drug is given at a constant time interval; four t  _½ yields approximately 95% of this concentration. This applies only to drug where elimination is governed by ‘first-order’ kinetics. Fortunately, this comprises the vast majority of drugs, including opioids. Phenytoin is a notable exception, which involves both ‘first-’ and ‘zero-order’ processes.

The t  _½ of morphine is 2–4 hours. Therefore, when morphine is administered every 4 hours, steady state will be 95% achieved after approximately 16 hours. Titration of the dose on a daily basis ensures that dose changes are occurring at steady state. In contrast, the methadone requires approximately 4–7 days, and occasionally much longer, to achieve steady state. Taking this into consideration a loading dose of methadone is often used followed by a period of cautious titration (after a number of days) to minimize the risk of toxicity.

Drugs produce their effects on the body by binding with receptors, modifying enzyme processes, or by direct chemical or physical actions. Opioids exert their influence by interacting with opioid receptors (primarily the mu opioid receptor).

Receptors are specialized proteins within the cell membrane which are integral for communication between the cell and the outside world. They are highly specific for certain ligands, such as specific hormones, cytokines and/or drugs.

Opioid receptors were originally classified by pharmacological activity and later by molecular sequencing (Pasternak, 2004). There are three types of classical opioid receptor: μ (mu) or MOR, κ (kappa) or KOR, and δ (delta) or DOR. Another opioid-like receptor has been identified named the nociceptin or orphanin FQ peptide receptor or NOR (Mollereau et al., 1994).

The three classical receptors are activated differentially by the endogenous opioids (encephalins, endorphins, and dynorphins) (Table 9.1.2). Exogenous opioids, such as morphine, act primarily at the MOR. Different opioids may show differential binding to sites on the MOR and may also bind to other opioid or non-opioid receptors (Pasternak and Pan, 2011).

The opioid receptors belong to the superfamily of G protein-coupled receptors. Each consists of an extracellular N-terminus, seven transmembrane helices, three extra and intracellular loops, and an intracellular C-terminus. Each receptor type is coded for by a different gene (Meng et al., 1993; Knapp et al., 1994; Wang et al., 1994). The three receptors share a high degree of homology with most variation found in the extracellular loops and N-terminal domains (Minami and Satoh, 1995). The extracellular loops are particularly important as they determine ligand binding. Opioid receptors are widely, yet differentially, distributed in both central and peripheral nervous systems (Table 9.1.2) (Minami and Satoh, 1995).

The µ-opioid receptor is clinically the most important of the family is, as it is responsible for the inhibition of nociceptive pathways and is exploited by all exogenous opioids. Knockout studies in mice show that the MOR is essential for morphine-induced analgesia (Matthes et al., 1996). Many of the unwanted effects of opioids are also related to activity at this receptor (Table 9.1.2). The MOR is expressed on central and peripheral neurons, and the latter are up-regulated in response to inflammatory stimuli. Peripherally, MOR are found pre- and postsynaptically. For example, approximately 70% of MOR receptors in the dorsal horn are expressed on the primary afferent terminations (presynaptic), where they modulate afferent transmission (Stein et al., 2003). At a cellular level, µ-opioid receptor activation results in an overall inhibitory effect (Box 9.1.2).

δ- and κ-opioid receptors also are involved in the modulation of pain. Knock-out studies in mice show that KOR may influence chemical visceral pain and thermal nociception (Simonin et al., 1998). Studies with selective opioid antagonists suggest that oxycodone analgesia depends on binding to the KOR receptor (Smith et al., 2001); other studies indicate that oxycodone is more like morphine and exerts its analgesic effects through MOR activation (Kalso et al., 1990; Chen et al., 1991; Yoburn et al., 1995). Pharmacological studies also suggest a role for KOR in mediating the dysphoric and sedative effects of opioids (Mark, 1990).

Combination opioid receptor knockout studies suggest that DOR plays a role in modulating mechanical and inflammatory pain (Martin et al., 2003). δ-opioid receptor knock-out mice also do not exhibit analgesic tolerance to morphine (Nitsche et al., 2002).

G-protein-coupled receptors, including MOR, KOR and DOR, have been shown to form different configurations, including homo- and hetero-dimers and oligomers, with unique internalization and activation pathways. Dimerization modulates receptor pharmacology and this process could present targets for novel interventions (Milligan, 2005). The development of MOR-DOR heteromers exemplifies this potential. There is evidence that there is an increased abundance of MOR–DOR heteromers in chronic pain and/or chronic exposure to morphine, and that ligand binding to DOR in the context of MOR-DOR is not associated with the development of opioid tolerance, in contrast to ligand binding to DOR alone. MOR-DOR may represent a new pharmacological target with the potential induce analgesia without tolerance (Costantino et al., 2012).

Genetic polymorphism caused by alternative splicing of mRNA has been shown to give rise to various human MOR receptor subtypes. Much of the variation is found in the intracellular C-terminal domain and includes the creation of potential phosphorylation sites (Pasternak and Pan, 2011). These receptor subtypes have differential expression patterns and demonstrate activation profiles that vary among the various MOR agonists; they are likely to explain some of the clinical variation in opioid response (Pasternak, 2004). Genetic polymorphism also may be involved individual variation in pain responses. For example, the minor T allele of OPRM1 rs563649 is associated with higher expression levels of the MOR-1K isoform and also with high pain sensitivity (Shabalina et al., 2009).

Other cellular adaptations are likely to be involved in the varied responses to chronic opioid exposure, such as variation in the development of tolerance (Ferguson, 2001; Bailey and Connor, 2005). Chronic morphine exposure leads to little change in MOR expression but does seem to produce changes in the non-neuronal population of glial cells, with increased activation provoking central sensitization (Watkins et al., 2005).

Although it is widely accepted that clinically relevant pharmacological tolerance to opioid analgesic effects is not an issue for the majority of patients with cancer-related pain, it is difficult to assess analgesic tolerance clinically (Chang et al., 2007). There does not appear to be a simple correlation between exposure to opioids and induction of analgesic tolerance. A process of adaptation occurs, which is likely to depend on diverse factors, but this process cannot be fully explained on the basis of current knowledge of cellular mechanisms.

Based on their interactions with receptors, opioid compounds can be divided into agonist, agonist–antagonist, and antagonist classes (Table 9.1.3). Most opioids used in the clinical setting are full agonist drugs. Opioid antagonists, such as naloxone and naltrexone, bind to MOR and produce no agonist activity. Agonist–antagonist drugs include a group of mixed agonist–antagonists, which are agonists at one or more opioid receptor subtypes and antagonists at others, and partial agonists. The mixed agonist–antagonists, such as pentazocine, are seldom used in the management of patients with advanced illness. Buprenorphine is a partial agonist that is being used more as an analgesic with the advent of a transdermal delivery system. Buprenorphine is believed to have a ceiling effect at doses of 8–16 mg/day (Walsh and Eissenberg, 2003). However as the recommended analgesic doses are much lower than the ceiling dose (equivalent to up to 3–4 mg/day, or 2 × 70-microgram/hour patches), buprenorphine typically is used with doses in the linear part of the dose-response curve and clinically performs as a full agonist during the management of pain. At the higher doses used to treat heroin addiction, the partial agonist effect may be encountered (Greenwald et al., 2003).

Efficacy is defined by the maximal response induced by administration of the active agent. In practice, this is determined by the degree of analgesia produced following dose escalation through a range limited by the development of adverse effects. Potency, in contrast, reflects the dose–response relationship and is typically defined by the intensity of a specified effect, such as analgesia, associated with a specific dose. Potency is influenced by pharmacokinetic factors (i.e. how much of the drug enters the body’s systemic circulation and then reaches the receptors) and by affinity to drug receptors.

Clinically, the utility of potency measurements is created by comparing drugs using relative potency ratios, or the ratio of doses required to produce the same analgesic effect. The relative potency of each of the commonly used opioids is based upon a comparison with 10 mg of oral morphine. Data from single-and repeated-dose studies in patients with acute or chronic pain have been used to develop ‘equianalgesic’ or dose conversion tables. The term ‘equianalgesic’ is however misleading as there is wide inter-individual differences in response to different opioids and such tables should be used only as a guide when switching between opioids (Riley et al., 2006). In particular, care should be exercised when switching from one opioid to another as part of the management of opioid toxicity. In this situation, conservative conversions should be used followed by individual titration. Opioid switching is discussed in detail in Chapter 9.4.

The clinical utility of an opioid therapy is determined by a favourable balance between analgesic efficacy and side effects. Many variables may influence whether a dose exists that yields this balance. These include intensity of pain; prior opioid exposure in terms of drug, duration, and dose (and the degree of cross-tolerance that this confers); age; route of administration; level of consciousness and metabolic abnormalities; and genetic polymorphism in the expression of relevant enzymes or receptors (Droney et al., 2012).

The clinical use of combinations of different opioids is increasing with the aims to (a) improve analgesia, (b) reduce side effects, and (c) limit the development of opioid tolerance. The rationale behind this practice is to utilize the inherent differences in the pharmacodynamic and pharmacokinetic properties of this group of drugs to maximize potential benefit and minimize adverse effects (Fallon and Laird, 2011). Hypotheses of the pharmacodynamic mechanisms include splice variation in opioid receptors, the receptor activation versus endocytosis (RAVE) theory, and the formation of opioid receptor homo- and heterodimers which results in changes to G-protein signalling cascades (Davis et al., 2005).

In animal studies, there is some evidence of analgesic synergism between methadone and other µ-agonist opioids (Bolan et al., 2002). In addition two small retrospective case series of patients with uncontrolled cancer-related pain have reported that low dose methadone in combination with existing opioid improved analgesia (McKenna and Nicholson, 2011; Haughey et al., 2012).

A combination product of oxycodone with morphine in a fixed-dose ratio 3:2 is currently being evaluated in clinical trials (MoxDuo^®). It has been trialled in phase II and phase III studies for the management of acute postoperative pain and results suggest that there may be a difference in side effect profile when compared with the individual opioids (Webster et al., 2010; Richards et al., 2011; Webster, 2012). There are currently no trials published in the chronic pain or cancer pain setting.

Further research into the use of opioid combinations in warranted. The potential benefits from this strategy must be weighed against other factors, such as poor patient compliance, confusion over dosing, and prescriber dosing errors, along with potentially unanticipated increased side effects.

Some drugs exert effects by affecting enzyme processes, rather than by binding to receptors. Many act by inhibition of enzyme actions. For example, non-steroidal anti-inflammatory drugs block the effect of the enzyme cyclooxygenase and thereby interfere with the synthesis of prostaglandins and exert anti-inflammatory activity.

Other drugs produce intended effects through a direct chemical or physical action. Antacids are an example of drugs with a direct chemical action; they are bases which neutralize gastric acid. Drugs with a physical mode of action include the bulk laxatives, such as ispaghula husk.

Patients with palliative care needs may already be receiving drugs for a variety of conditions. Some may still be beneficial but others may no longer contribute to improving prognosis or symptoms. Rationalization of the therapeutic regimen always should be considered. If further drugs to relieve symptoms are added, this adds to the potential for drug interaction. Interaction is adverse if it causes therapeutic failure or toxicity from any one drug. Remembering all the possible drug interactions is virtually impossible, but knowledge of the underlying mechanisms of drug interaction can put the prescriber on guard together with frequent consultation with prescribing information is important.

An adverse drug reaction can be defined as an unwanted or harmful reaction experienced following administration of a drug, or combination of drugs, under normal conditions of use that is suspected of being related to the drug. For example opioids act via the µ-opioid receptor to slow gut transit and cause constipation. Opioid-related side effects are summarized in Box 9.1.3 and can be transient or persistent. Chapter 9.4 describes in more detail the principles of opioid switching which can reduce individual adverse drug reactions.

Opioid analgesics are one of the drugs most frequently associated with adverse drug events. A study of 3695 inpatient adverse drug reactions found that 16% were attributable to opioids (Davies et al., 2009). Risk of opioid drug reactions increases in older patients, in those with underlying cardiac or respiratory disease, and when co-prescribed with other sedative medications such as benzodiazepines (Bernard and Bruera, 2000).

Pharmacokinetic interaction arises through alterations in the rate and extent of absorption and changes in metabolism (both pre-systemic and elimination), distribution, and renal excretion. The clinical impact of theoretical interactions can be difficult to predict.

Drugs such as metoclopramide and anticholinergics, which alter the rate of gastric emptying, may affect the speed of absorption of other agents. Some drugs bind others in the gastrointestinal tract and affect their bioavailability. For example, care is necessary if antacid preparations, iron salts, or cholestyramine are used concurrently with certain drugs.

Drug interactions resulting from changes in the rate of metabolism by the liver will result both in changes of bioavailability for those drugs with a significant first-pass effect, and decreased clearance. Steady-state concentrations of drug may be profoundly affected.

A number of drugs (particularly phenobarbital, carbamazepine, phenytoin and rifampicin) are capable of inducing the cytochrome P450 and glucuronidase enzymes in the liver. There are a myriad of substrates for this interaction, including methadone, warfarin, corticosteroids, and anticonvulsant drugs. Increased pre-systemic metabolism may result in the need to increased doses to achieve therapeutic levels. Conversely some drugs may inhibit CYP enzymes, individually or as an entire superfamily (Table 9.1.4). Certain foodstuffs may also induce or inhibit hepatic enzyme systems; for example, grapefruit contains furanocoumarins which inhibit CYP3A (Hanley et al., 2011).

The most important drug interactions in the kidney involve competition between agents for active tubular secretion. Active tubular secretion is used by organic acids, and the most frequent interactions are caused by the loop diuretics and some non-steroidal anti-inflammatory drugs. Although renal excretion of some drugs is pH dependent, in general this has minor implications in normal therapeutics. There are a few exceptions, however; for example, methadone’s renal clearance is considerably enhanced by concurrent use of urinary acidifiers such as acetazolamide (Bellward et al., 1977).

Some drug–drug interactions occur at a receptor level. For example, buprenorphine is a partial agonist and morphine is a full agonist, and morphine-induced analgesia may be reversed or limited by competition at the receptor level if buprenorphine is added.

The preferred route of administration for many drugs including opioids is oral. Oral formulations of opioids include immediate-release (IR) syrups, tablets or capsules, and modified-release (MR) tablets or capsules. MR formulations slowly release the drug into the gut, allowing treatment once or twice daily depending on the formulation (e.g. morphine, oxycodone). Care must be taken in prescribing, such that the correct formulation is dispensed, and when more than one formulation is given (e.g. IR and MR) the patient understands how and when to use each preparation.

Opioid drugs can also be given by other routes, including, transdermal, transmucosal (sublingual, buccal, nasal, rectal), and parenteral by injection either intravenously or more commonly subcutaneously. Fentanyl and buprenorphine products have been formulated for the transmucosal and transdermal routes. In addition specialist pain services may use epidural or intrathecal opioids in the palliative setting where appropriate. Prescriptions of these drugs should specify the exact formulation, as formulations differ in systemic availability and may not be interchangeable.

IR preparations are absorbed in the stomach or proximal small bowel, so that absorption is complete within a few hours on ingestion. For example, IR morphine and oxycodone reach peak effects within 1 hour. Modified- or sustained-release formulations allow a drug to be released over 12–24 hours, resulting in a smoother concentration profile of the drug in the blood, extended duration of action, and reduction in tablet burden for the patient.

Drugs that are absorbed through the buccal, nasal, or rectal mucosa avoid first-pass metabolism in the liver by uptake into veins that drain directly into the systemic circulation. This results in higher bioavailability and often a faster onset of action when compared to the oral route. IR transmucosal fentanyl products may be useful in treatment of breakthrough/incident pain. The rapid speed of onset of action (approximately 10 minutes) and short duration of action (≥ 1 hour) may sometimes be more suited to the temporal characteristics of breakthrough/incident pain than conventional IR opioid (Twycross et al., 2012; Davies et al., 2013). There does not, however, appear to be a meaningful relationship between background opioid dose and the effective dose of transmucosal fentanyl, and therefore, titration is essential (Zeppetella, 2011).

Some lipid-soluble drugs are well absorbed through the skin, and their transdermal delivery via ‘patches’ allows controlled release over many hours or days. Opioid examples include fentanyl and buprenorphine. Different formulations have been developed and are not interchangeable and preparations may last for 3–7 days depending on drug/formulation. Clinical trials suggest good patient satisfaction with this mode of delivery. Care must be taken, given the wide variability in drug absorption, especially in cachexic or pyrexial patients (Heiskanen et al., 2009).

The preferred parenteral route of administration in palliative patients is subcutaneous and opioids may be given as stat injections or as a continuous subcutaneous infusion. When continuous subcutaneous infusions are used and multiple drugs combined, care must be taken to ensure drug interactions are avoided, as precipitation of one drug in solution will clearly limit therapeutic effect (e.g. cyclizine and oxycodone are not compatible) (Dickman et al., 2002).

Combination products are attractive and may aid compliance by reducing tablet burden. To be effective, the frequency of administration of the two drugs should be the same. Combination products to not allow for titration of one drug without the other, and this may be a concern with some combinations. For example, concerns have been raised regarding combinations of opioid analgesics with paracetamol because of the need to limit titration of the paracetamol and risk of liver damage in overdose.

Compliance or adherence is the extent to which a patient follows a prescribed drug regimen. It is important that decisions regarding treatment are jointly made by the prescriber and patient. Allowing adequate time to explain principles and goals of therapy, expected benefit and possible side effects, and plans for review and follow-up is essential. In addition, exploration of patient (and family) concerns regarding addiction, tolerance, side effects, or fear that treatment implies the final stages of life are essential when using opioids for patients with advanced illness (NICE, 2012). The proactive management or prevention of side effects, for example, provision of laxatives for opioid-induced constipation, can improve compliance. In general, more complex regimens with high frequency of administration and/or multiple drugs reduce compliance.

Pharmacogenomics is the study of how genetic variation influences response to drugs. It is the cornerstone of personalized medicine, which aims to tailor treatment to the individual to maximize efficacy and minimize adverse reactions. Two techniques have been used to study pharmacogenomics: the candidate gene approach and genome-wide association. The candidate gene approach targets single nucleotide polymorphisms (SNPs) in genes already known to be important in pharmacokinetic (e.g. drug metabolizing enzymes, drug transporters) and pharmacodynamic (e.g. receptors, ion channels, enzymes) pathways. Genome-wide association studies cast a much wider net examining millions of SNPs across the entire genome at a time and may therefore provide new biological insights into mechanisms (Wilke et al., 2008).

In recent years, the field of pharmacogenomics has exploded to provide a wealth of information to inform personalized prescribing across the medical specialties. In oncology, response to certain chemotherapy agents can now be predicted, for example, variation in UGT1A1 is associated with severe neutropenia from irinotecan (Innocenti et al., 2004). In cardiology, response to warfarin, statins, and clopidogrel have all been associated with genetic factors (Johnson and Cavallari, 2013). In HIV medicine, screening programmes have been used to reduce the risk of hypersensitivity reactions to abacavir by testing for HLA B*5701, which is associated with the condition (Mallal et al., 2008). Work continues on how this knowledge may best be translated into clinical practice (Johnson et al., 2012).

Study of the CYP enzyme 2D6 (CYP2D6) gene has provided perhaps the best examples of how pharmacokinetics and ultimately opioid response is linked to genetic variation. CYP2D6 is involved in the metabolism of several opioids including codeine, tramadol, and oxycodone. Over 70 CYP2D6 alleles have been described which directly affect the final protein; these include SNPs, deletions, insertions, and copy number variation (Leandro-Garcia et al., 2009). The sum functional effect of this variation has been classified into four main phenotypes: poor, intermediate, extensive, and ultrarapid metabolizers.

Codeine is partially (10%) metabolized to morphine by CYP2D6 (Lotsch, 2005). Approximately 10% of Caucasians are poor metabolizers and experience little analgesia from codeine (Sindrup et al., 1990; Persson et al., 1995). Conversely 3% of Caucasians are ultrarapid metabolizers and have a higher incidence of codeine-related adverse reactions (Kirchheiner et al., 2007). There have been case reports of fatal neonatal opioid toxicity in children breastfed by mothers who are ultrarapid metabolizers following ingestion of codeine (Madadi et al., 2007). The CYP2D6 phenotype has also been suggested to affect response to tramadol and oxycodone by altering ratios of the parent opioid to the more active metabolites (Samer et al., 2003, 2010b), although the clinical relevance of this is debated (Gronlund et al., 2010; Samer et al., 2010a, 2010b; Andreassen et al., 2012).

Pharmacodynamic candidate gene studies in palliative care patients suggest that opioid receptor SNPs, for example, OPRM1 A118G, influence patients’ requirements for opioids (Klepstad et al., 2004; Campa et al., 2008; Walter and Lotsch, 2009). Individual pain susceptibility also influences analgesic response, and therefore, many more candidate genes from pain signalling and modulatory pathways, for example, COMT (Rakvag et al., 2005; Rakvag et al., 2008), also may be important in opioid responsiveness.

Pain experience and opioid response are complex traits and therefore influenced by a myriad of gene–gene and gene–environment interactions. Recently, genetic association studies have begun to explore interactions between variants from more than one gene. This has thus far been limited to two candidate SNPs at a time (Reyes-Gibby et al., 2007; Campa et al., 2008). The concept of gene–gene/environment interactions or epistasis provides a huge challenge for the future of opioid pharmacogenetics, both practical and analytical. Further work needs to be done to unpick the complexities behind opioid response to be able to develop a useful predictive tool to inform clinical practice.