Opioid use in pregnant women has increased over the last decade. Following birth, infants with in utero exposure demonstrate signs and symptoms of withdrawal known as the neonatal abstinence syndrome (NAS). Infants express a spectrum of disease, with most requiring the administration of pharmacologic therapy to ensure proper growth and development. Treatment often involves prolonged hospitalization. There is a general lack of high-quality clinical trial data to guide optimal therapy, and significant heterogeneity in treatment approaches. Emerging trends in the treatment of infants with NAS include the use of sublingual buprenorphine, transition to outpatient therapy, and pharmacogenetic risk stratification.
- •
All infants with in utero exposure to opioids demonstrate signs and symptoms of withdrawal. Two-thirds of infants require pharmacologic therapy to ensure proper feeding and development.
- •
Opioid replacement is the optimal primary therapy. The current standard is morphine, although there is significant heterogeneity in treatment regimens, with many centers using methadone.
- •
Of predictive factors, lack of polysubstance exposure, prematurity, and maternal use of buprenorphine are most strongly associated with less severe withdrawal symptoms and need for pharmacologic therapy.
- •
Emerging therapies include the use of buprenorphine for primary therapy, and clonidine as an adjunct.
- •
Pharmacogenetic profiling of infants and the use of modeling and simulation to optimize dosing are emerging, but not fully developed, technologies that may change the treatment of the neonatal abstinence syndrome.
Definition of neonatal abstinence syndrome
Neonatal withdrawal symptoms have been noted following prenatal exposure to several drugs. Examples include opioids, benzodiazepines, mood-stabilizing medications, selective serotonin reuptake inhibitors, and nicotine. For all drug classes except opioids, these symptoms are usually self-limited and do not require pharmacologic treatment. Infants born to mothers with opioid abuse or receiving methadone maintenance often develop withdrawal symptoms, following the postpartum cessation of in utero exposure to opioids. This complex is known as the neonatal abstinence syndrome (NAS). The full mechanistic basis for the clinical presentation is unclear. Tolerance induced by long-term exposure to opioids is primarily mediated by receptor downregulation coupled with upregulation in the cyclic adenosine monophosphate (cAMP) pathway. Other putative mechanisms include neuroimmune activation, production of anti-opioid peptides, or activation of the spinal dynorphin system. Symptoms of withdrawal are hypothesized to be due to increased adenylyl cyclase activity and an abrupt increase in norepinephrine following removal of the mu opioid ligand. NAS is characterized by signs of central nervous system (CNS) hyperirritability, gastrointestinal dysfunction, respiratory distress, and vague autonomic symptoms. Common symptoms in order of frequency includes tremor, high-pitched cry, sneezing, increased muscle tone, regurgitation and vomiting, poor sleep, loose stools, sweating, excoriation, mottling, nasal stuffiness, low-grade fever, and tachypnea. Impaired weight gain and seizures are seen with untreated NAS. All infants with prolonged in utero opioid exposure will develop signs and symptoms of withdrawal of varying severity. However, the disorder encompasses a diverse spectrum, and those with milder symptoms respond well to supportive treatments. NAS symptoms severe enough to require pharmacologic treatment occur in 55% to 94% of infants born to opioid-dependent mothers.
Current use of illicit drugs occurs in 4.4% of pregnant women. Heroin use during pregnancy is associated with fetal death and infant morbidity, including intrauterine growth retardation, placental insufficiency, postpartum hemorrhage, preeclampsia, and premature rupture of membranes. In an attempt to counteract these poor outcomes, methadone maintenance in opioid-dependent pregnant women has been used for the past 35 years, and is associated with improved birth weight and improvements in multiple domains. A more recent development is the expansion in the use and abuse of prescription opioids. Whereas the use of heroin decreased by 19% between 1998 and 2008, the abuse of prescription opioids during the same period increased by 41%. In 2010, 5.1 million individuals reported nonmedical abuse of prescription pain medications within the previous month, with 71% of abused pain relievers being obtained from friends or family, and either bought or taken without permission. The societal burden of NAS is difficult to assess, as is evident from the wide variations and implausible rates reported to regional authorities for hospitals in the defined geographic area with similar patient populations. Such variation is due to limited self-reporting of drug abuse, and underreporting of NAS using the International Classification of Diseases. In 1996, a survey by the National Institute of Drug Addiction estimated that 7000 cases occur each year, although the report conceded this is potentially an underestimation. More recently, the rate of NAS in the United States has increased from 1.2 to 3.9 per 1000 live births between 2000 and 2009. A similar incidence rate has been estimated in Australia.
Predictors of NAS severity
The dose of maternal methadone dose as a covariate of the need for NAS treatment length has been examined extensively. Although a meta-analysis that evaluated studies by methodological quality did not identify a statistically significant difference in outcomes between high-dose and low-dose methadone, there is a suggestion of modest maternal dose dependency on outcomes. However, if such an effect does exist, it is small and not relevant in terms of choosing a maternal dose or differential treatment approaches in the treatment of infants. Lower maternal methadone doses have been associated with higher rates of illicit substitution, and a consensus view is that maternal doses of methadone should not be reduced solely to reduce NAS severity. High-quality randomized, controlled trial evidence from the MOTHER study has demonstrated that compared with methadone use, maternal buprenorphine is associated with a decreased need for morphine treatment in NAS and neonatal length of stay. The maternal study population in this study has been convincingly demonstrated to be similar to the patient population at large, strongly supporting the generalizability of results.
Whereas Jansson and colleagues described worse NAS symptoms and pharmacotherapy in males, severity, need for therapy, or length of therapy were not influenced by gender in a cohort study by Holbrook and Kaltenbach. Similarly, there was no sex dependency in the large randomized MOTHER study, which compared use of methadone and buprenorphine in pregnant women. Intrapartum variability or decelerations in fetal heart rate do not predict the need for therapy in NAS. However, alternations in autonomic regulation, as measured by analysis of maternal or infant vagal tone, have been noted to be predictors of worse NAS symptomatology. It is postulated that infants who adapt to maternal methadone–induced autonomic changes are maladapted to more severe NAS following birth. Methadone exposure during pregnancy is associated with an approximately 2.5-fold increase in the rate of preterm birth. Preterm infants have a well-described natural history of NAS and a need for treatment that differs from term infants. The current NAS scoring instruments have not been examined in this population. Need for therapy and length of stay is shorter in the preterm population. The preterm population thus appears to be categorically different in terms of in utero opioid exposure.
Polydrug abuse during pregnancy is associated with impaired fetal markers (heart rate and variability) and greater need for postpartum pharmacologic therapy. A retrospective study by Seligman and colleagues demonstrated that the length of NAS treatment for all opioid only exposed infants between 2000 and 2006 was 31 days, compared with 38 days for polydrug-exposed term infants. Strikingly, a multivariate analysis of infants revealed a significant prolongation of treatment duration for NAS (31 vs 47 days, P< .01) of benzodiazepine versus non–benzodiazepine-exposed infants. Benzodiazepine withdrawal symptoms in adults include anxiety, tremors, anorexia, nausea, postural hypotension, and in severe cases seizures, delirium, and hyperpyrexia. The onset of symptoms is 12 to 24 hours for short-acting agents with a peak at 72 hours, whereas longer-acting agents such as diazepam are associated with an onset of 24 to 72 hours and a peak between 5 and 8 days following the last dose. Benzodiazepines cross the placenta, but maternal confounders have made it difficult to estimate adverse effects specific to in utero exposure of benzodiazepines. Whereas teratogenicity is unlikely in benzodiazepine-exposed infants, decreased birth weight and neonatal withdrawal have been noted, the latter manifested by hypotonia and hypoventilation or tremulousness. The half-life of diazepam in neonates is 31 days. Thus, in contrast to that of adults, initiation of neonatal withdrawal for many benzodiazepines can be delayed with an onset at a week and effects noticeable for weeks. There is no specific treatment for neonatal benzodiazepine withdrawal. Tobacco exposure is associated with worse NAS symptomatology. Analysis of meconium for tobacco, methadone or its metabolites, cocaine, or opioids other than methadone, however, are not predictive of NAS outcomes.
Predictors of NAS severity
The dose of maternal methadone dose as a covariate of the need for NAS treatment length has been examined extensively. Although a meta-analysis that evaluated studies by methodological quality did not identify a statistically significant difference in outcomes between high-dose and low-dose methadone, there is a suggestion of modest maternal dose dependency on outcomes. However, if such an effect does exist, it is small and not relevant in terms of choosing a maternal dose or differential treatment approaches in the treatment of infants. Lower maternal methadone doses have been associated with higher rates of illicit substitution, and a consensus view is that maternal doses of methadone should not be reduced solely to reduce NAS severity. High-quality randomized, controlled trial evidence from the MOTHER study has demonstrated that compared with methadone use, maternal buprenorphine is associated with a decreased need for morphine treatment in NAS and neonatal length of stay. The maternal study population in this study has been convincingly demonstrated to be similar to the patient population at large, strongly supporting the generalizability of results.
Whereas Jansson and colleagues described worse NAS symptoms and pharmacotherapy in males, severity, need for therapy, or length of therapy were not influenced by gender in a cohort study by Holbrook and Kaltenbach. Similarly, there was no sex dependency in the large randomized MOTHER study, which compared use of methadone and buprenorphine in pregnant women. Intrapartum variability or decelerations in fetal heart rate do not predict the need for therapy in NAS. However, alternations in autonomic regulation, as measured by analysis of maternal or infant vagal tone, have been noted to be predictors of worse NAS symptomatology. It is postulated that infants who adapt to maternal methadone–induced autonomic changes are maladapted to more severe NAS following birth. Methadone exposure during pregnancy is associated with an approximately 2.5-fold increase in the rate of preterm birth. Preterm infants have a well-described natural history of NAS and a need for treatment that differs from term infants. The current NAS scoring instruments have not been examined in this population. Need for therapy and length of stay is shorter in the preterm population. The preterm population thus appears to be categorically different in terms of in utero opioid exposure.
Polydrug abuse during pregnancy is associated with impaired fetal markers (heart rate and variability) and greater need for postpartum pharmacologic therapy. A retrospective study by Seligman and colleagues demonstrated that the length of NAS treatment for all opioid only exposed infants between 2000 and 2006 was 31 days, compared with 38 days for polydrug-exposed term infants. Strikingly, a multivariate analysis of infants revealed a significant prolongation of treatment duration for NAS (31 vs 47 days, P< .01) of benzodiazepine versus non–benzodiazepine-exposed infants. Benzodiazepine withdrawal symptoms in adults include anxiety, tremors, anorexia, nausea, postural hypotension, and in severe cases seizures, delirium, and hyperpyrexia. The onset of symptoms is 12 to 24 hours for short-acting agents with a peak at 72 hours, whereas longer-acting agents such as diazepam are associated with an onset of 24 to 72 hours and a peak between 5 and 8 days following the last dose. Benzodiazepines cross the placenta, but maternal confounders have made it difficult to estimate adverse effects specific to in utero exposure of benzodiazepines. Whereas teratogenicity is unlikely in benzodiazepine-exposed infants, decreased birth weight and neonatal withdrawal have been noted, the latter manifested by hypotonia and hypoventilation or tremulousness. The half-life of diazepam in neonates is 31 days. Thus, in contrast to that of adults, initiation of neonatal withdrawal for many benzodiazepines can be delayed with an onset at a week and effects noticeable for weeks. There is no specific treatment for neonatal benzodiazepine withdrawal. Tobacco exposure is associated with worse NAS symptomatology. Analysis of meconium for tobacco, methadone or its metabolites, cocaine, or opioids other than methadone, however, are not predictive of NAS outcomes.
Long-term and short-term sequelae
Environmental and social factors are more important influences upon childhood development than brief periods of prepartum and peripartum exposure to drugs of abuse. Infants exposed in utero to opioids show low birth weights, increased preterm birth, and reduced fetal growth parameters, but investigations have been hampered by the logistical difficulty of controlling for tobacco and other social factors associated with illicit drug use. Studies linking in utero opioids to impaired neurodevelopment have been criticized for not accounting for confounding of the child’s social and environmental milieu. The database for opioid exposure is less robust than that for cocaine. It is possible that there are subtle neurodevelopmental effects arising from in utero opioid exposure apart from effects caused by environmental and home settings. Even if this is real, however, these associations do not provide guidance about practical therapeutic decisions. For newborns, the benefits of maternal opioid therapy during pregnancy using methadone in a structured program clearly outweigh no therapy. There is evidence that untreated or undertreated women may seek street sources of opioids to treat withdrawal symptoms, which clearly has negative neonatal outcomes. Of importance, there is no evidence of long-term adverse outcomes in children treated with pharmacologic agents in comparison with infants who do not require treatment for NAS, or for treatment with different classes of agents. Although the database of information is smaller, in human and animal studies neonatal outcomes with in utero buprenorphine exposure are generally favorable compared with methadone.
Current therapies
Variability of Current Practice Patterns
Few studies have examined NAS prevalence and treatment patterns. Nandakumar and Sankar published a survey of 17 neonatal units in the Northwest Region of the United Kingdom, revealing not only conflicting practices in scoring, identification, and management but also a deficit of reliable data to assist practitioners in determining the best regimen to treat NAS. In a survey conducted by Sarkar and Donn in United States neonatal intensive care units (NICU), the focus was primarily on determining the percentage of respondents using an abstinence scoring system, those with access to formal written policies or educational programs for NAS management, and practitioners using customary pharmacologic agents for withdrawal. The 13-question survey of 102 accredited fellowship programs (which had 75 respondents) did not include questions on NAS incidence or length of hospital stay. O’Grady and colleagues conducted a 15-question questionnaire of 235 neonatal units that sought to survey current NAS practices in the United Kingdom and Ireland. The survey assessed first-line and second-line agents, attitudes to breastfeeding by women on methadone, and the safety of infants discharged on medication. Crocetti and colleagues assessed the number of opiate-exposed neonates identified as having NAS, as well as policies and procedures for treatment in 27 hospitals throughout Maryland. In aggregate, these surveys demonstrate significant heterogeneity in diagnosis and treatment patterns. There are clear information gaps for identification, treatment, and length of hospital stay. Moreover, there is a lack of pharmacoeconomic analyses on costs and cost-effectiveness of treatment for NAS.
Framework for Treatment
The therapeutic framework of treatment begins with the identification of infants at risk for NAS. NAS is graded using a standard checklist that identifies and stratifies severity of disease based on signs and symptoms in multiple domains. Of several scoring systems used to gauge symptom severity and titrate drug dose, the Finnegan score (or modifications of it) is the most commonly used. A modification of the Finnegan score used in the multicenter MOTHER study of buprenorphine use for pregnant women is the standard instrument used in other randomized NAS research trials. The Finnegan instrument was created to assess severity of disease in those with known opioid exposure. On day 2 of life a score of 7 corresponds with the 95th percentile for nonexposed infants, meaning any score of 8 or greater is highly suggestive of in utero opioid exposure even in those denying opioid use during pregnancy. Nonpharmacologic therapies should be used for all infants with in utero opioid exposure. These treatments include swaddling, the use of small calorically dense formulas, rooming in, breastfeeding, and minimization of excessive external stimulation. Infants with mild symptoms should be observed in the hospital for at least 4 days. For infants with severe symptomatology manifested by seizures, poor weight gain, and elevated values in a NAS-specific scoring instrument, pharmacologic therapy is indicated. Ideal treatment uses a protocol-driven use of drug titration to control symptoms. Both symptom-driven (ie, weight-independent fixed dose titration based on severity of NAS scores) and weight-based dosing regimens have been used, with neither being identified as the standard approach. Regardless of the manner of dose titration, infants who do not have control of symptoms despite high doses of the initial therapy are treated with a secondary drug. After stabilization, symptom scores are used to gradually wean the controlling drug or drugs; this occurs typically in an inpatient setting, as it allows careful observation and dose titration of infants. Some institutions stabilize an infant in an inpatient setting, with terminal weaning done as an outpatient. The use of outpatient treatment in highly selected patients is associated with shorter inpatient stays, but extended total duration of therapy.
The rationale to use pharmacologic therapy is to ensure proper feeding and development, and foster the maternal infant bond. The ideal specific drug used would safely achieve these therapeutic goals, while at the same time minimizing the total duration of therapy and length of hospitalization. The most commonly used initial therapy is an opioid, while phenobarbital or clonidine are the primary adjunctive agents. Although used more commonly, phenobarbital has not been demonstrated to have improved safety or efficacy in comparison with clonidine as an adjunctive therapy. A comparison of these agents as adjuncts is currently being investigated ( NCT01175668 ). The role of initial dual therapy of phenobarbital with an opioid has been described, but has not been compared in a large number of patients. The value of this approach has not been established, but anecdotally may provide benefit in infants with polysubstance use.
Opioids
Cochrane reviews, the American Academy of Pediatrics, and expert reviews identify opioid replacement as the ideal treatment for the withdrawal symptoms associated with in utero exposure to opiates. Opioid replacement as a first-line agent (1) improves weight gain but lengthens hospitalizations compared with supportive care, (2) reduces seizure rates and possibly duration of therapy compared with phenobarbital, and (3) reduces treatment failure compared with diazepam. Although many of the studies cited in the Cochrane review had methodological flaws, the standard of care practice that has developed since the 1970s has generally supported this approach. There is limited evidence from which definitive recommendations can be made regarding differential safety and efficacy of specific opioids. Moreover, high-quality data on the optimal dose regimens or comparative effectiveness on use of adjunctive agents are lacking. Current practice patterns for these have been developed empirically, and remain an area that would benefit from investigations of higher quality.
Morphine
Morphine is the most commonly used opioid for replacement therapy. Paregoric, a previously commonly used morphine source, was never subject to any formal evaluation by the Food and Drug Administration, and is no longer available. Diluted, deodorized tincture of opium (DTO) has a morphine concentration of 0.4 mg/mL and an ethanol concentration of 0.19%. This agent has been largely replaced by an ethanol-free morphine solution of 0.4 mg/mL concentration. Preservative-free morphine hydrochloride solution for neonatal administration is stable at 4°C for at least 6 months. Because of the relatively short half-life of morphine, best outcomes have been demonstrated when morphine doses are given no longer than 4 hours apart. Accordingly, infants who are sleeping at the nominal dose time should be awakened for drug administration.
Morphine in humans is metabolized primarily to morphine-3-glucuronide (M3G) and morphine-6-glucuronide (M6G) via uridine 5′-diphosphate glucuronosyl transferase (UGT)-2B7. The ontogeny of UGT development is dynamic in the immediate postpartum period, demonstrated by reduced M6G/morphine ratio in neonates younger than 7 days as well as an associated reduced postsurgical morphine requirement in comparison with older neonates. Nonlinear mixed-effects models have been used to estimate active metabolite formation as well as elimination. Clearance generally correlates with glomerular filtration, with minimal fecal elimination or metabolism to normorphine. The large interpatient and intrapatient variability of intravenous morphine pharmacokinetics and pharmacodynamics in neonates is due in part to a dynamic acquisition of metabolic enzymes, renal function, and changes in fat and extracellular fluid balance. Of note, the pharmacokinetics profile of orally administered morphine in neonates is currently unknown. An area of therapeutic need would be the characterization of the concentration-response relationship. Such a relationship, created with modeling and simulation, would be of use in designing an optimized dose regimen.
The initial dose of morphine was 0.12 to 0.6 mg/kg/d in a survey of 17 pediatric units in the United Kingdom. These investigators had the opinion that a higher initial dose may be associated with better control of symptoms, but acknowledged that evidence to support this intuition was lacking. A dose of 0.24 mg/kg/d was recommended by the 1998 report of the American Academy of Pediatrics (AAP), although this protocol outlined drop-unit doses that would make fine titration difficult. Neither the 2012 AAP Committee on Drugs NAS report nor the Cochrane review of the topic identified a favored specific dose. There is no generally accepted maximum dose of morphine used for NAS. A survey of neonatal units in the United Kingdom revealed that typical maximum doses were up to 1.3 mg/kg/d, and that one-third determined dose according to symptom control rather than a maximum predefined level. Specific protocols for dose titration are based either on a weight-based increase in dose based on scores above a specific NAS score, or a weight-independent dose based on a graded severity of NAS score. Table 1 outlines these 2 commonly used approaches.
| Weight Based | Symptom Based | |
|---|---|---|
| Initial Dose: For first elevated score >8, rescore in 1 h to verify. If still elevated | |
| Single NAS score | Dose every 4 h (mg) | |
| 9–12 | 0.04 | |
| 13–16 | 0.08 | |
| 17–20 | 0.12 | |
| 21–24 | 0.16 | |
| ≥25 | 0.20 | |
| Doses are fixed and not based on infant weight Dose Increase: | ||
| Single NAS score | Increase Dose (mg) | |
| 0–9 | None | |
| 9–12 | 0.02 | |
| 13–16 | 0.04 | |
| 17–20 | 0.06 | |
| ||
| Two NAS scores | Increase Dose (mg) | |
| 9–12 | 0.01 | |
| 13–16 | 0.02 | |
| 17–20 | 0.04 | |
| ||
a Phenobarbital loading dose of 20 mg/kg followed by 5 mg/kg/d, or clonidine.
Methadone
Methadone is a long-acting opioid commonly used for abstinence treatment. The longer half-life of methadone provides less of a flux between peak and trough levels, while also providing ease of administration at less frequent intervals. Oral bioavailability in adults is high, but variable. The pharmacokinetics of methadone in the pediatric and neonatal populations has been simulated using physiologic-based pharmacokinetic modeling that suggests significant interpatient and developmental variability, but decreased systemic exposure with age. This model has not been validated by rich patient-level data. There is scant published clinical trial evidence to guide its use in the neonatal population. In a single small study, outcomes with methadone were similar to those for phenobarbital or diazepam. Comparisons with oral morphine are limited to a single retrospective review of 46 patients, in which there was no significant difference in length of stay between treatments. A standard dose has not been established, but the protocol used by Lainwala and colleagues is provided in Box 1 . Methadone use remains relatively uncommon, ranging from fewer than 2% of units in the United Kingdom to as many as 20% in the United States. The long dosing interval has led some sites to use methadone as extended outpatient dosing. Compared with full inpatient treatment, infants discharged home on methadone have shorter hospitalizations but longer duration of therapy, though at least in one study having a similar total of methadone administered. Because of the likely variability of pharmacokinetics, frequent outpatient follow-up is required to allow careful monitoring and dose titration based on symptoms.
