Wither codeine?

Abstract

On August 15, 2012, the United States Food and Drug Administration (FDA) issued a safety alert 1 on ‘the use of codeine in certain children after tonsillectomy and/or adenoidectomy’. Six weeks later, the European Medicines Agency (EMA), the equivalent organization in the European Union (EU), announced that it also was starting a review of ‘the safety of codeine containing medications used as pain relief in children 2’. A further update from the FDA, on February 20, 2013, stated that a ‘boxed warning’, the strongest warning they have, would be added to the drug label of codeine containing preparations. It advises healthcare professionals to ‘prescribe an alternative analgesic (to codeine) for postoperative pain control in children who are undergoing tonsillectomy and or adenoidectomy 3’. The two regulatory bodies are both assessing the safety of codeine in children but are taking two different approaches. The FDA has asked healthcare professionals and patients to report any adverse events relating to codeine to their Medwatch Adverse Event reporting program. The EMA has written to ‘marketing authorization holders’ of codeine containing medicinal products asking for answers to a series of questions relating to the safety and efficacy of codeine in all areas of pain relief in children (not just postoperative). Both organizations wish to clarify the balance of benefit vs risk of the use of codeine containing medicines in children. The reviews were triggered by two papers reporting four serious adverse outcomes (three deaths and one near miss) in children who received codeine as postoperative pain relief. Codeine was implicated as the cause of death/morbidity in all the children reported. To understand the link between codeine and these four serious outcomes, we need to understand the pharmacokinetics of codeine. Codeine is a prodrug. It has no analgesic effect itself and weak affinity for mu opioid receptors. The analgesic properties arise from the conversion of the codeine stem to morphine and morphine-6-phosphate. The majority of a dose of codeine (typically 80%) is metabolized to inactive, water soluble metabolites by two routes; glucuronidation by UDP glucuronosyltransferase to form codeine 6 glucuronide, and N-demethylation by the cytochrome P450 3A4 enzymes to norcodeine. A minority of codeine administered is O-demethylated to morphine in the liver by cytochrome P 450 2D6 (CYP2D6) enzymes. The CYP2D6 enzyme system represents only a small percentage (2–4%) of all the CYP hepatic enzymes, but is responsible for metabolism of 25% of all commonly used drugs. It shows a large range of clinical activity due to genetic polymorphism. At least 70 different alleles exist and each allele defines a particular level of enzyme activity (normal function = 1, reduced function = 0.5, and no function = 0). Any individual inherits one allele from their mother and one from their father to form a diplotype, and allele activity scores are additive. Individuals having total activity scores of 1.0–2.0 are called ‘extensive metabolizers’ and represent the majority of a Caucasian population (77–92% of Caucasians). Other individuals may have total activity scores of 0.5 (‘intermediate metabolizers’) or 0 (‘poor metabolizers’). Poor metabolizers form 5–10% of Northern European Caucasian populations, are unable to convert codeine to morphine, and therefore have no analgesia when given codeine. CYP2D6 allele frequencies vary greatly from ethnic group to ethnic group. In some individuals, there are more than two copies of the CYP2D6 gene (gene duplication). The multiple gene copies can result in activity scores of 3 or greater and phenotypically have increased functional enzyme activity. These individuals are termed ‘ultrarapid metabolizers (UM)’, and they may produce potentially high blood concentrations of morphine after the administration of normally therapeutic doses of codeine. The frequency of gene duplication and therefore the percentage of a population that are ultrarapid metabolizers vary greatly between ethnic groups (Table 1). A direct relationship between steady state morphine concentration in blood and respiratory response has been demonstrated. A study in children (2–570 days postnatal age) demonstrated hypercarbia and depressed ventilatory response to carbon dioxide at plasma morphine concentrations above 20 ng·l−1 4. Plasma morphine concentrations were measured in all four children with severe adverse events secondary to codeine. The four children with adverse events linked to codeine need to be considered in detail (Table 2). The first death reported was from Canada in 2009 (Case 1) 5. He was 2 years old, 13 kg in weight who underwent adeno-tonsillectomy (AT) for sleep study proven obstructive sleep apnea. He received pethidine (meperidine) postoperatively and was then discharged home on the day of surgery with advice to take paracetamol and codeine 10–12.5 mg regularly, 4–6 hourly. He was found dead on the second morning after surgery. Analysis of femoral blood at postmortem showed a blood morphine concentration of 32 ng·l−1. CYP2D6 genotyping revealed functional duplication of the CYP2D6 allele consistent with the ultrarapid metabolizer genotype. The authors concluded that there was increased conversion of codeine to morphine due to ultrarapid metabolism resulted in a toxic accumulation of morphine. Three further adverse outcomes have since been reported by the same Canadian group in a paper in 2012 (Cases 2–4) 6. Case 2 was a 4-year-old (27.6 kg) First Nations (North American Indian) boy who underwent AT for obstructive sleep apnea (diagnostic method not defined). After an overnight stay, he was discharged home on 8 mg doses of codeine and received four doses in total. He was found dead on the second afternoon after surgery. Postmortem morphine concentration was 17.8 ng·ml−1. Genotyping revealed he was an ultrarapid metabolizer due to gene duplication, with death attributed to increased morphine concentration leading to respiratory arrest. Case 3 was a 3-year-old girl (14.4 kg) of ‘Middle Eastern’ descent undergoing tonsillectomy for obstructive sleep apnea (again method of diagnosis undefined). She received two doses of codeine in hospital (15 mg each), stayed overnight and was discharged home on a codeine 15 mg per paracetamol 150 mg combination every 4–6 h. After an unspecified number of hours, she was found unresponsive, with markedly reduced respirations and an oxygen saturation of 65%. She was resuscitated and survived. Blood morphine concentration was 17 ng·ml−1. Her genotype was an extensive metabolizer, but the authors state that overlap of the EM and UM phenotypes accounts for her high morphine levels. Case 4 was a 5-year-old (29 kg) boy undergoing tonsillectomy for recurrent tonsillitis and snoring. He was prescribed paracetamol and codeine 12 mg every 4 h and was discharged home on the day of operation. He was found dead 24 h after surgery. His postmortem morphine concentration was 30 ng·ml−1. Genotyping was not undertaken but analysis of blood codeine, and morphine concentrations suggest the child was an ultrarapid metabolizer. The four children share a number of common features. They: These case reports raise a number of questions for the practicing pediatric anesthetist. They include: The place of codeine as an analgesic agent for children and the limited number of studies assessing the effectiveness of codeine as an analgesic agent was considered in a Pro–Con debate in this journal 3 years ago 7. Our increased understanding of codeine's pharmacokinetics and these adverse events provide additional evidence against the use of codeine in children. There are two key problems Firstly, codeine is a prodrug; it is converted to an effective analgesic by an enzyme system which shows considerable genetic variation. Any dose of codeine will result in a large spread of different morphine concentrations across a population. The morphine concentration resulting will vary substantially not only between poor–intermediate metabolizers and extensive metabolizers, but also within the broad range of ‘normal’ children who are extensive metabolizers. This was demonstrated in a comparative study of codeine vs morphine, in addition to regular paracetamol and ibuprofen as part of a multimodal postoperative analgesic regime in children, who were predominately extensive metabolizers 8. Codeine showed more variable efficacy than morphine. There are no new studies in children in the last 3 years describing either positively or negatively the effectiveness of codeine as an analgesic agent in children. Secondly, high morphine concentrations may occur in a further subgroup of patients (ultrarapid metabolizers) leading to profound respiratory depression in a very small number of cases. The size of this risk is undefined. The two reviews by the FDA and EMA aim to identify any additional cases of significant morbidity, quantify this risk, perhaps calculating a numerator and denominator. Some additional information is already available. A recent review article searched databases for case reports of opioid induced respiratory depression in children less than 12 years of age 9. They identified one potential additional case of severe respiratory depression after an operation 10. This case was a 2-year-old, 14 kg, child of North African descent who underwent tonsillectomy for sleep study proven OSA. He was discharged home as a day case and received four doses of paracetamol with codeine, 12–24 mg per dose after the operation. He was found apneic and unresponsive by his mother on the second night after surgery; emergency medical services were called, and he was successfully resuscitated. The genotype is not clearly described, and on reassessment, he was probably not an ultrarapid metabolizer 11. Blood morphine and codeine levels were not obtained. A number of adverse outcomes linked to codeine in nonoperative settings have also been reported. Three infants, all less than 4 months of age, requiring resuscitation for severe respiratory depression, after receiving codeine as a cough linctus, have been reported. A case is also reported of death of a newborn infant who was breastfeeding. The baby's mother was an ultrarapid metabolizer receiving codeine for episiotomy pain and she converted significant amounts of codeine to morphine which subsequently passed into breast milk. The National Confidential Enquiry into Patient Outcome and Death (NCEPOD) 12 reviewed all deaths in children within 30 days of planned and unplanned surgery in England and Wales between April 1, 2008 and March 31, 2010 11. No unexplained cluster of deaths following AT was reported. We know that 26 000 tonsillectomy operations were undertaken in England in 2011/12 in children under 14 years of age (Hospital Episode Statistics for England, Health and Social Care Information Centre HSCIS, Leeds, UK). A survey of APA members in 2009 reported 39% used codeine to provide postoperative pain relief following tonsillectomy, and so, we have some measure of a denominator. Unfortunately, we do not know the numerator. The HSCIS declined to undertake a database run to identify the number of deaths within 30 days within this cohort despite a freedom of information request. In summary, we continue to lack detailed information on both the effectiveness and scale of risks associated with using codeine in children, but the increased areas of concern mean alternatives require active consideration. It is well recognized that a subgroup of children with SDB are acutely sensitive to the respiratory depressant effects of opioids. In children with obstructive sleep apnea (OSA) the opioid concentration–ventilator response relationship is moved to the left 13. The time after surgery before the opioid–respiratory response relationship returns to the right is unknown. The children most likely to develop significant respiratory problems in the initial perioperative period are well described 14. Risk factors include age under 2 years, weight <15 kg, failure to thrive, obesity (z score >2.5), and severe OSA on sleep study. In addition, we know that in at least 25% of children with OSA, the operation does not reverse the abnormal respiratory findings on overnight sleep study (PSG) and therefore is not curative. Cure rates (defined by respiratory variables on a PSG) are even lower in obese children undergoing AT for OSA. None of these deaths occurred in the first 24 h after surgery. Although a number of the children were discharged home on a day case basis, some had received a dose of codeine in hospital prior to discharge without incident. There is no evidence in the literature to suggest that witnessed doses of codeine in hospital increase the safety of subsequent home administration. Taking these factors together all the children so far reported who developed severe respiratory depression and/or death postoperatively had an operation for a condition associated with increased sensitivity to opioids, and all fell into ‘at-risk’ groups for perioperative respiratory complications. It is interesting that no similar deaths have been reported as yet from UK, Europe, or Australasia. This cannot be accounted for differences in the genetic makeup of populations. Spain for instance has a high percentage of UMs in its population and so would be expected to report problems. In addition, major international cities such as London have children from multiple ethnic backgrounds, and so, problems would be expected in children of Ethiopian, North African, or Arabian descent. It is important to recognize that genotype is not the same as phenotype. Genotype is the genetic makeup of an individual. The characteristic the individual displays (phenotype) is the result of the interaction of environmental factors on the genotype. Many factors including, the presence of other drugs, renal function, coexisting conditions, impact on the activity of an enzyme system. Looking at genotype alone may be simplistic. The fact that all reports come from North America may reflect a difference in management of postoperative pain in children following AT in North America from other continents. In UK, the most common approach is to send children home after AT with regular prescriptions of paracetamol and nonsteroidal anti-inflammatory (NSAIDs) drugs, with codeine prescribed as an ‘as required’ medicine for breakthrough pain. The North American approach may be slightly different, with avoidance of NSAIDs, because of concerns of postoperative bleeding, and discharge home on regular codeine and paracetamol. NSAIDs and paracetamol have a powerful additive analgesic effect potentially reducing the frequency of administration of codeine prescribed on an as required basis 15. The North American population may receive codeine more frequently after AT than children in other continents. Codeine is certainly not the ideal analgesic agent. It will be ineffective in a minority of individuals and potentially life-threatening in an unquantified number of individuals. It is theoretically possible to identify individuals with the UM genotype preoperatively by commercial gene testing. This is increasingly undertaken for a number of conditions where the main drug treatments are active drugs broken down to inactive metabolites by the CYP4502D6 enzyme systems. Many tricyclic antidepressants (amitriptyline in part, desipramine, nortriptyline) and antipsychotic medications (fluoxetine, haloperidol, venlafzaxine) are metabolized to nonactive metabolites by CYP2D6 enzymes and so are ineffective as treatments or lead to high levels of metabolites causing side effects in ultrametabolizers. A number of testing kits are available to identify an individual's CYP4502D6 genotype (e.g., Roche. AmpliChip CYP450 test.) It is therefore perfectly feasible to test all individuals to match future drug therapy to their genetic makeup and avoid ineffective or toxic drug–patient combinations. In practice, it is unlikely to become routine practice in the foreseeable future, and even if undertaken, will not identify all individuals at potential risk of high morphine concentrations after standard doses of codeine. This is in part because of the effect of environment on genotype as previously described. Codeine is commonly used both in hospital and at home in two situations: firstly, on an ‘as required’ (prn) basis as step-up analgesia if pain relief from simple analgesics has been inadequate, and secondly, on a regular basis in combination with paracetamol. The effectiveness and safety of alternative agents in both situations particularly at home requires assessment. In many respects, codeine administration is equivalent to prescribing small amounts of morphine, and so, an alternative approach may be to prescribe weak morphine solution rather than codeine. A weak concentration of morphine (morphine 10 mg per 5mls—Oramorph) is available in the United Kingdom and falls under Schedule 5 of the Misuse of Drugs Regulations (2005)—the same schedule as codeine. It may be stored, prescribed, and dispensed without the tighter regulations required for stronger preparations (concentrations) of morphine. The pharmacokinetics and dynamics of morphine are predictable. It has a long history of use as a postoperative agent in hospitals. It has been shown to provide more predictable postoperative analgesia than codeine 15. It is therefore an attractive alternative to codeine. Replacing codeine with morphine for ‘take-home’ analgesia either on a regular or as required prescription is more problematic. It means releasing multiple bottles of morphine solution into the community with all the potential for individual overdose and accidental ingestion by other family members. It may not represent a safer alternative to codeine. Oxycodone is a potent semisynthetic opioid analgesic. Unlike codeine, it is not a prodrug. The parent drug provides a significant amount of the analgesic effect and so is less reliant on metabolism for efficacy. It undergoes N-demethylation by the CYP450 3A4 enzyme system predominately to inactive metabolites. A minority of oxycodone is metabolized by the CYP450 2D6 enzyme system to an active metabolite noroxymorphone. Sadly, oxycodone has enormous potential for street abuse as is happening in Australia and already exists in the USA 16. It is an effective analgesic but limited information is available on its use in children as a postoperative analgesic agent. Tramadol may represent the most realistic alternative to codeine, although its mode of action and pharmacokinetics are complex. It is a centrally acting synthetic analgesic of moderate potency. It appears to provide analgesia by a number of mechanisms. These are binding to mu opioid receptors, inhibition of reuptake of noradrenaline, and increased release plus decreased reuptake of serotonin. It is a racemic mixture of optical isomers with the two parent drugs providing the majority of its analgesic effect. Metabolism in the liver is by N-demethylation by the cytochrome CYP3A4 enzyme system to an inactive metabolite and to a lesser degree O-demethylation to desmethyltramadol (M1) by the cytochrome 2D6 enzyme. M1 is an active metabolite, binding strongly to mu opioid receptors. A study of 24 children from South Africa showed that tramadol administered orally as drops was rapidly absorbed with peak serum concentrations reached within 30 min of administration. In addition, the peak concentration of the M1 metabolite was approximately one-third the peak parent drug concentration 17. As Tramadol is not a prodrug, it will be an effective analgesic in children who lack the CYP2D6 enzyme, but the importance of the active M1 metabolite, concentrations reached and implications in ultrarapid metabolizers is unclear. The concentration of M1 in children who are ultrarapid metabolizer has not been studied. Tramadol is licensed for use in children over 1 year of age in many European countries, although UK has no license below 12 years of age. There is extensive clinical experience of use in children from New Zealand where it has been used in a similar manner to codeine to provide both regular and rescue postoperative pain relief for more than a decade. The dose used is 1–2 mg·kg−1. The smallest tablet available is 50 mg; so, this is used in children over 25 kg. For children weighing 10–25 kg, an oral solution (100 mg·ml−1) with dropper is available. Doses of either 12.5 or 25 mg are prescribed (Anderson BJ, personal communication). Despite this extensive experience, few studies describing the use of tramadol for postoperative pain relief in children exist in the literature. Postoperative nausea and vomiting may be a problem, although this appears to be dose related. It would be extremely valuable if the New Zealand experience could be published. One hopeful agent on the horizon is tapentadol. This is an analgesic agent working via two mechanisms, opioid receptor binding, and inhibition of noradrenaline uptake. In adults, it is conjugated in the liver and then renally excreted. There is limited experience in children to date. When considering alternatives to codeine, two further factors require consideration. Synthetic alternatives to codeine (e g., tramadol) are more expensive, and wholesale change will put increasing pressures on already strained drug budgets. Secondly, one of the commonest factors blocking the provision of effective pain relief in children is a failure of administration of drugs by parents 18. Any alternative agents must be available in a preparation which is palatable to the child and easily administered by the parent. We should only continue to use codeine if it represents the most effective and safest agent available for children. From the narrow perspective of providing postoperative pain relief, codeine has been perceived as an effective agent for many years. Whether perception matches reality is unclear. The fact that codeine is a prodrug that does not form an active compound in a significant minority of children would suggest that other agents are likely to be more effective and should replace it. Concentration on genotype alone may be misplaced. We must also consider a number of other factors which may alter efficacy. Tramadol appears a more appropriate agent than codeine, but it has not replaced codeine in many countries due to licensing constraints, lack of pediatric friendly preparations, possible side effects, and the limited published information of effectiveness. As anesthetists, we are fortunate in using a relatively small number of drugs which give visible and predictable results in short time scales in most patients. We therefore may have unreasonable and unsupported faith in the effectiveness of the drug treatments we use. It is well recognized that ‘most major drugs are effective in only 25–60% of patients’ 19. It is important that we do not have unrealistic expectations of what any drug can deliver for us. Before we discard codeine, we need to know considerably more about the effectiveness of the alternative agents in clinical practice. The case reports do raise important questions on the safety of codeine in children. All the adverse events were in children with sleep-disordered breathing and therefore at increased susceptibility to opioid induced respiratory depression. Alternative agents may or may not represent safer alternatives. Children with OSA have reduced requirements for opioid analgesia in the immediate postoperative period due to ‘upregulation’ of opioid receptors 20. There are no case series of the use of regular simple analgesia alone, or with low dose morphine, oxycodone or tramadol for breakthrough pain in this group of patients as take-home analgesia. In summary, on the basis of the limited evidence available, the FDA has advised that codeine should in future not be used to provide postoperative pain control after tonsillectomy and or adenoidectomy in children 3. It does not suggest any alternative agents and provide assessment of the efficacy and safety of alternatives. Based on both the limited published evidence and alternatives agents available, it would seem appropriate to. The author has no financial interest in any company or organization directly or indirectly related to the content of this editorial. No conflicts of interest declared.