Codeine toxicity via breast Milk: Can this occur and implications for opiate therapy in children

Abstract

Codeine is commonly used in combination analgesic products. The use of codeine during breastfeeding can rarely be associated with serious adverse effects related to genetic polymorphisms affecting codeine's metabolism and disposition. Clinicians treating infants of breastfeeding mothers should be aware of this rare but potentially serious complication.

In 2006, a case of codeine toxicity via breast milk with fatal consequences was published in Lancet. ¹ This manuscript described a 13 day old infant whose mother was prescribed a codeine‐acetaminophen analgesic after a traumatic delivery who died of morphine and codeine toxicity. This raised the issue of codeine and morphine transmission via breast milk as part of the broader consideration of the role—or lack thereof—of codeine in paediatric and perinatal care. Although this was the first fatal case published with reference to codeine, it does draw on literature describing cases of lethargy and poor feeding of breastfeeding infants from 1985 on whose mothers were using codeine containing products for analgesia. ² While there are many details lacking in some of the earlier literature, a Canadian study of 111 codeine‐treated mothers who were breastfeeding demonstrated that 26 (23%) of infants demonstrated signs of central nervous system depression in association with maternal symptoms of central nervous system depression in 21 (19%) of mothers. ³ Signs of central nervous system depression were associated with higher doses of codeine and with genetic polymorphisms in cytochrome P450 2D6, UDP‐glucuronosyltransferase 2B7, the ABCB1 transporter, the mu‐opioid receptor and catechol O‐methyltransferase. ^{* 3} Further, a recent study using the EudraVigilance data base identified 15 infants whose deaths were related to opiate exposure via breastfeeding. ⁴ In this case, methadone and diamorphine were the opiates used by the breastfeeding mothers, which suggests both habitual drug use and long‐term therapy, the latter of which may be germane to the Lancet case.

This and other cases highlighted the potential risk to breast fed infants of opiate transmission via breast milk and the potential impact of genetic variations on risk of untoward outcomes. In the years since this manuscript was published, further research has helped to uncover issues that relate to the metabolic disposition of codeine, the impact on potential risk to the infant and the safety of codeine in children in general. ⁵ , ⁶ , ⁷ This has included a manuscript detailing thoughtful pharmacokinetic modelling of maternal codeine therapy that found that toxicity in infants was likely occur but also was likely to be rare (Figure 1). ⁷ These findings have impacted regulatory guidance on the use of codeine‐containing products in a number of countries. This has not been without debate, and the original Lancet case has recently had a Letter of Concern noted calling for its retraction (https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(26)00245-X/fulltext). This in part related to a 2020 mini‐review which questioned the plausibility of codeine toxicity via breast milk. ⁸ Although the review raises important questions, there are issues in the data supporting this assertion.

FIGURE 1.

Coupled model of codeine biotransformation and breast milk transfer with intrahepatic formation of morphine from codeine via cytochrome P450 2D6 (CYP2D6) and then to morphine‐6‐glucuronide (M6G) via uridine 5′‐diphospho‐glucuronsyltransferase 2B7 (UBG2B7) with coupling via breast milk ingestion over the course of daily feedings followed by absorption into the systemic circulation. Not shown but relevant is the potential for enterohepatic recycling of glucuronides back to their parent compounds. Figure modified from Willmann et al. Clin Pharmacol Ther 2009; 86, pp. 634–643 ⁷

To address if this is accurate, it is important to first consider the pharmacology of codeine. Of note, essentially all codeine containing products used for analgesia in the context of breastfeeding also contain paracetamol (acetaminophen); in the context of opiate toxicity, only the pharmacokinetics of codeine will be considered. Codeine is a pro‐drug, 3‐methyl morphine, with a methyl group on the 3‐carbon hydroxyl position. ⁹ The bulk of codeine's analgesic activity appears to be via conversion to morphine and codeine 6‐glucuronide, as morphine has a 200‐fold greater affinity to the opioid mμ receptor compared with codeine and the 6‐glucuronide appears to be more potent than morphine. ⁹ , ¹⁰ , ¹¹ Codeine is also metabolized to norcodeine. Demethylation of codeine to morphine is mediated by CYP2D6, an enzyme known to have polymorphic expression as depending on genotype one can be a poor metabolizer, an extensive metabolizer or an ultra‐rapid metabolizer. ¹¹ , ¹² The phenotype is not uniformly distributed, as ultra‐rapid metabolizers are rare in northern European populations (1% or less) but common in the Mediterranean littoral, the Middle East, the Horn of Africa and South India. ¹¹ , ¹² , ¹³  Of note, Kirchheiner et al. demonstrated overlap in central nervous system depression as evidence by sedation between ultra‐rapid and extensive metabolizers, an issue that the modelling work of Willmann et al. raised. ⁷ , ¹²

There is an important further consideration as drug concentrations in the infant are a sum of pharmacological activity across the coupled mother‐baby dyad (Figure 1). Simply put, drug toxicity is typically due to more drug/metabolite being generated than cleared. In the context of maternal therapy, drug metabolism occurs not only in the mother but also in the infant, which is to say that both maternal and infant pharmacokinetics impact drug concentrations and the link between. How does this apply in the case of codeine?

The modelling work of Willmann et al. concluded that codeine toxicity via breastfeeding would be rare but possible notably in the case of mothers with the CYP2D6 ultra‐rapid metabolizer phenotype. ⁷ They also concluded that warnings should be extended to normal/extensive metabolizers given overlap in serum concentrations between the two phenotypes. ⁷ The mini‐review by Zipursky and Juurlink noted that codeine metabolism was complex—which is unquestionably true—and that a central issue was the infant's metabolism of codeine to morphine was a central part of toxicity. ⁷ As well, the contribution of the maternal dyad was by transfer of codeine via passage through breastmilk, and that reduced CYP2D6 activity is early after birth renders toxicity implausible. ⁷ The work of Stevens et al. suggesting that CYP2D6 activity in infants is submaximal using in vitro using human microsomes is cited as a support for this position. ¹⁴ However, there are important limits to in vitro data, and the use of human microsomes as a tool for studying in vitro to model issues of drug metabolism in vivo has been shown to be much less than optimal. ¹⁵ In this context, there is elegant in vivo work by Allegaert's group in Belgium that quite clearly demonstrates the early development of robust phenotypic‐specific CYP2D6 activity in infants using tramadol—a CYP2D6 substrate that has demonstrated breast milk transfer—can be metabolized by the infant, supporting the potential for infant metabolism of codeine contributing to the generation of morphine in a setting where excretion is suboptimal (Figure 1). ¹⁶ This is an important consideration not cited in the mini‐review by Zipursky and Juurlink in that the modelling work of Willmann et al. assumed CYP2D6 activity infants being in the range of 10% of adult values, whereas in vivo work using metabolic probes in infants demonstrates in fact that there is a rapid concordance between CYP2D6 genotype and phenotype. ¹⁷

Further to this issue, if we return to the concept of the mother–infant dyad, while it appears that maternal CYP2D6 phenotype is a key element in codeine toxicity via breastmilk, there is an additional consideration with respect to the circumstances of pregnancy that bears consideration. Over the course of pregnancy, CYP2D6 activity has been found to increase, with a nearly 50% increase over baseline by the last weeks of gestation. ¹⁸  Thus, it is likely that maternal production of morphine with transmission via breast milk is considerably greater than might be otherwise predicted.

The final issue with respect to metabolism relates to the potential for route switching. As noted, one of metabolic products of codeine is codeine 6‐glucuronide, the formation of which is mediated by UDP‐glucuronosyltransferase‐2B7 (UGT2B7). Zipursky and Juurlink again cite in vitro data suggesting that the infant in the Lancet case should have had robust capacity to glucuronate codeine. ⁸ However, the studies cited were in vitro studies using a small number of samples spanning the first year of life to 25 years of age. ¹⁹ , ²⁰ Data derived in vivo from infants in contrast demonstrated that there was limited capacity for opiate glucuronidation in the first 2 weeks of life, highly germane as limitation of metabolic capacity via glucuronidation could lead to route switching to demethylation and hence higher than anticipated morphine concentrations. ²¹ , ²² Reduced capacity for glucuronidation has been associated with other drug toxicities in neonates, a classical example being the chloramphenicol Grey Baby Syndrome, in which developmental limitations in capacity for glucuronidation led to accumulation of chloramphenicol producing disruption of oxidative phosphorylation and cardiac failure, frequently fatal. ²³ In addition to this, enterohepatic recycling via deglucuronidation of the 6‐glucuronide can occur in the infant's gastrointestinal tract, also contributing to increasing the infant's codeine and hence by metabolism morphine concentrations. ⁷

Although these issues have addressed drug metabolism notably in terms of the potential to demethylate codeine to morphine, the other part of the equation in terms of toxicity relates to clearance. As noted above, toxicity can often be simply considered as a situation where the amount of toxin generated exceeds the capacity for toxin elimination.

The assumption made in the mini‐review that renal excretion would be expected to be robust based on the observation of five wet diapers on the day before death is that urine flow is a surrogate for renal function. ⁸ However, renal drug elimination is a complex process which even assessments such as glomerular filtration rate fail to completely capture. ²⁴ , ²⁵ While glomerular filtration rate at birth at term is typically 30% of adult surface‐area corrected values, there are other key elements in renal drug administration whose ontogeny is much less understood. ²¹ , ²² As well, morphine is a substrate of the OCT1 transporter, and while the ontogeny remains unclear, it appears that expression may be low at birth as well as subject to genetic variation, which would have the effect of reducing elimination, notably in rare cases where there is increased accumulation. ²⁶ , ²⁷ , ²⁸ , ²⁹ Specifically, Tzetkov et al. have found that polymorphisms in OCT1 resulting in loss of function variants are associated with significantly higher morphine concentrations as expressed by area under the curve. ²⁹ This observation was made in 2013, and it would have been illuminating if these variations could have been studied in the Lancet case in 2006. ²⁹

A final potential issue relates to the pharmacodynamics of morphine in infants. Anand et al. have demonstrated that there appear to be ontological differences in morphine responses, which may contribute to adverse neurological outcomes. ³⁰

The review also notes the lack of fatal cases in the literature. However, as noted above, there are cases and case series of adverse effects in breastfed infants of mothers being treated with codeine dating back to 1985 and in the case of other opiates from 1885. ² , ³ In addition, as modelled by Willmann et al., adverse events are likely to be rare and serious ones even rarer. ⁷ There are data suggesting that short (3 day or less) courses of codeine therapy for breastfeeding mothers had no negative impacts on outcomes such as hospital admission versus mothers who were not prescribed codeine‐containing products. ³¹ This is germane as the infant reported in the Lancet case died after on the 13th day of life, during which his mother was taking a codeine‐acetaminophen analgesic while the infant was breastfeeding robustly every 3 h. ¹ To return to our simple paradigm, this suggests that toxin accumulation was exceeding the capacity to excrete.

It could also be noted that although the tragic fact of opiate deaths is infants is real, the circumstances in the Lancet case are not consistent with what is typically seen in these unfortunate cases. ³² Specifically, opiate deaths in infants often occur in the context of drug‐using homes and in a setting where child protection agencies have had involvement, with direct administration being the commonest route of exposure. ³² In a letter published in response to the manuscript of Heerfordt et al., Juurlink and Zipursky caution that in cases of fatal opiate overdose in infants, direct administration must always be considered. ³³ We concur, and this is a prudent consideration that mandates involvement of the respective civil authority, as in the Lancet case which was investigated by the Office of the Chief Coroner of Ontario. ¹

Willmann et al. conclude that serious adverse events due to maternal codeine therapy in the context of breastfeeding are likely to be uncommon and due to unique combinations of risk factors. ⁷ One must concur. This is not uncommon for serious adverse drug events—one example being gentamicin ototoxicity. ³⁴ However, the view that it is implausible as stated in the mini‐review by Zipursky and Juurlink is not tenable notably given the considerable in vivo evidence, which supports the conclusions of Willmann et al. ⁷ , ⁸ There are very few absolutes in medicine, and respectfully, believing the codeine toxicity via breastfeeding is impossible is not one of them—very rare, but not impossible.

In addition to the cases noted above in association with codeine use during breastfeeding, there have been a number of serious adverse effects and deaths associated with codeine therapy in children. ³⁵ , ³⁶ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ While toxicity is observed with all opiates, the observation has been made that severe respiratory depression and death in the context of codeine monotherapy is primarily but not exclusively seen in children. ⁴⁰ The mechanism[s] responsible may relate to age‐related difference in codeine disposition in children versus adults, suggesting that perhaps where there are therapeutic alternatives they may be preferable. ⁴¹

So what is the clinician to take from this going forward? First, that serious and life‐threatening events in the context of maternal therapy with codeine while breastfeeding are rare but can occur and that clinicians need to be aware of this. The Lancet case highlights the risk factors for this, which include longer duration of therapy, maternal symptoms of central nervous system depression, genetic polymorphisms favouring increased generation of morphine and developmental ontogeny limiting both alternate metabolic pathways and drug clearance. Second, it is also essential that careful consideration should be made when selecting potential codeine alternatives, as some of the potential alternatives such as oxycodone may in fact be associated with a higher risk of adverse neonatal effects. ⁴² Recommendations on analgesia during breastfeeding have been made by a variety of authorities including the UK National Health Service and the American College of Obstetrics and Gynaecology, which recommend paracetamol (acetaminophen) or ibuprofen (which has relatively poor breast milk transfer) as first line analgesic therapy followed by selective opiate use at the lowest effective dose for the shortest duration necessary (https://www.acog.org/womens-health/faqs/postpartum-pain-management, https://www.cuh.nhs.uk/patient-information/pain-relief-during-pregnancy-and-after-birth/). Short courses of codeine therapy (less than 3 days) are likely to be safe, while consideration should be made for the use of alternate therapy should longer term therapy be necessary. Given that there are alternate analgesics available, the choice of analgesics for women who will breastfeed after delivery should consider all possible therapeutic choices balancing benefit and risk. Additionally, consideration should be given to the prudent recommendation of Willmann et al. that extensive metabolizers are at risk of high morphine concentrations and that should be included in labelled warnings. ⁷ , ¹²

CONFLICT OF INTEREST STATEMENT

M.R. has no conflicts of interest related to the content of this manuscript.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.